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
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1573700410.1371/journal.pmed.0020049Correspondence and Other CommunicationsOtherGeneral MedicineClinical TrialsMedical JournalsEditorial Policies (Including Conflicts of Interest)Open Access to Trials Register CorrespondenceMcCabe Susanne E-mail: [email protected]
Competing Interests: The author declares that she has no competing interest but that she does have a longstanding interest in research and publication ethics.
2 2005 22 2 2005 2 2 e49Copyright: © 2005 Susanne McCabe.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.
From Registration to Publication
==== Body
I find the arguments raised by the PLoS Medicine editors very useful [1] as I had not considered that a scientific community would tolerate barring access to registers of trials. It leaves huge gaps for exploitation by privileged groups.
It is not only colleagues in research and allied professions who need access but the global community, including members of the public wherever they live, those who participate in trials and those who will be on the receiving end of their outcomes.
The annual reports of research ethics committees (RECs) are supposedly in the public domain after approval by Strategic Health Authorities in the UK. But very few members of the public know of their existence or how to access them. Approaches to individual committees even now can meet with varied reactions, from suspicious, defensive, or hostile—reluctantly sending one report, quizzing as to which organisation the enquirer belongs to and why they should want one—to extremely welcoming of interest and discussion.
The annual reports should be easily accessible online by now, surely, but they are not. The activities of RECs and information on what research is being carried out in the name of society as a whole largely remain hidden from public view.
There is no information about public access on COREC (Central Office for Research Ethics Committees; www.corec.org.uk) or OREC (Office for Research Ethics Committees; www.orecni.org.uk). COREC has not been open about dealing with issues of concern raised with them in the past. They do state that public interest is welcome now, so it would show a real commitment to making research activity more open if they would show support for totally open access to a register and to promote that through their Web site.
Citation: McCabe S (2005) Open access to trials register. PLoS Med 2(2): e49.
==== Refs
References
From registration to publication PLoS Med 2004 1 e46 [No authors listed] 15578113
| 15737004 | PMC549596 | CC BY | 2021-01-05 11:13:38 | no | PLoS Med. 2005 Feb 22; 2(2):e49 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020049 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1573700510.1371/journal.pmed.0020050Research ArticleEpidemiology/Public HealthHealth PolicyHIV/AIDSMedical EthicsHealth PolicyResource allocation and rationingHIV Infection/AIDSMedicine in Developing CountriesEthicsDesigning Equitable Antiretroviral Allocation Strategies in Resource-Constrained Countries Equitable Treatment Allocation StrategyWilson David P
1
Blower Sally M
1
*1Department of Biomathematics and UCLA AIDS Institute, School of MedicineUniversity of California, Los Angeles, CaliforniaUnited States of AmericaCarr Andrew Academic EditorSt. Vincent's HospitalAustralia
Competing Interests: The authors have declared that no competing interests exist.
Author Contributions: DPW and SMB designed the study, analyzed the data, and contributed to writing the paper.
SMB is a member of the editorial board of PLOS Medicine.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 22 2 2005 2 2 e5016 9 2004 21 12 2004 Copyright: © 2005 Wilson and Blower.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.
Equitable Allocation of Antiretrovirals in Resource-Constrained Countries
Designing an Equitable Strategy for Allocating Antiretroviral Treatments
Background
Recently, a global commitment has been made to expand access to antiretrovirals (ARVs) in the developing world. However, in many resource-constrained countries the number of individuals infected with HIV in need of treatment will far exceed the supply of ARVs, and only a limited number of health-care facilities (HCFs) will be available for ARV distribution. Deciding how to allocate the limited supply of ARVs among HCFs will be extremely difficult. Resource allocation decisions can be made on the basis of many epidemiological, ethical, or preferential treatment priority criteria.
Methods and Findings
Here we use operations research techniques, and we show how to determine the optimal strategy for allocating ARVs among HCFs in order to satisfy the equitable criterion that each individual infected with HIV has an equal chance of receiving ARVs. We present a novel spatial mathematical model that includes heterogeneity in treatment accessibility. We show how to use our theoretical framework, in conjunction with an equity objective function, to determine an optimal equitable allocation strategy (OEAS) for ARVs in resource-constrained regions. Our equity objective function enables us to apply the egalitarian principle of equity with respect to access to health care. We use data from the detailed ARV rollout plan designed by the government of South Africa to determine an OEAS for the province of KwaZulu–Natal. We determine the OEAS for KwaZulu–Natal, and we then compare this OEAS with two other ARV allocation strategies: (i) allocating ARVs only to Durban (the largest urban city in KwaZulu–Natal province) and (ii) allocating ARVs equally to all available HCFs. In addition, we compare the OEAS to the current allocation plan of the South African government (which is based upon allocating ARVs to 17 HCFs). We show that our OEAS significantly improves equity in treatment accessibility in comparison with these three ARV allocation strategies. We also quantify how the size of the catchment region surrounding each HCF, and the number of HCFs utilized for ARV distribution, alters the OEAS and the probability of achieving equity in treatment accessibility. We calculate that in order to achieve the greatest degree of treatment equity for individuals with HIV in KwaZulu–Natal, the ARVs should be allocated to 54 HCFs and each HCF should serve a catchment region of 40 to 60 km.
Conclusion
Our OEAS would substantially improve equality in treatment accessibility in comparison with other allocation strategies. Furthermore, our OEAS is extremely different from the currently planned strategy. We suggest that our novel methodology be used to design optimal ARV allocation strategies for resource-constrained countries.
A mathematical model that allows to determine a strategy for distribution of limited medical resources (in this case antiretrovirals) such that each individual in need will have an equal chance of receiving treatment
==== Body
Introduction
The HIV/AIDS epidemic is having a devastating impact in sub-Saharan Africa and other resource-constrained regions. Recently, the World Health Organization and other organizations have committed to expand access to antiretrovirals (ARVs) in the developing world, the United States government has pledged to provide $15 billion for AIDS in Africa and the Carribean, and drug prices have fallen [1]. However, even if these resources are provided for the global treatment of HIV, the number of individuals in need of treatment will far exceed the supply of ARVs [1]. Thus, difficult decisions will have to be made as to how to design HIV treatment strategies with these scarce resources. Resource allocation decisions can be made on the basis of many different epidemiological, ethical, or preferential treatment priority criteria. Many diverse groups have been suggested for treatment priority in resource-limited regions, including the following: only men, pregnant women, children, the sickest, the most economically productive, individuals in the military, or even individuals of the dominant ethnic group [2]. It has also been proposed that a lottery would be the only fair approach to allocating ARVs [3]. Only a limited number of ARVs will be available, and only a fixed number of health-care facilities (HCFs) can be used for ARV distribution. Thus, the resource allocation decisions that need to be made are extremely complex.
Here, we use operations research to address this important resource allocation problem and to design ARV allocation strategies that are rational and equitable. The allocation decisions that we make here are based on ethical criteria, and not on epidemiological or preferential treatment priority criteria. Specifically, we determine the optimal allocation strategy that would ensure that each individual with HIV has an equal chance of receiving ARVs. We present a novel spatial mathematical model of treatment accessibility that we use in conjunction with an equity objective function to determine an optimal equitable allocation strategy (OEAS) for ARVs in a resource-constrained region. We quantify how changing the size of the catchment region surrounding each HCF, and the number of HCFs utilized for ARV distribution, alters the OEAS. Specifically, we use data from the detailed ARV rollout plan designed by the government of South Africa to determine an OEAS (based upon a variety of assumptions) for the province of KwaZulu–Natal. We also discuss how our proposed ARV allocation strategy differs from the currently proposed plan.
Our current analysis is applied to the South African province of KwaZulu–Natal, although our methodology could be applied to any resource-constrained setting. KwaZulu–Natal is the largest province in South Africa with a population of approximately 9.4 million and has more people infected with HIV than any other province (approximately 21% of all cases in South Africa [4]). We use data from 51 communities (cities, towns, and villages) in the province of KwaZulu–Natal; we exclude communities with a population of less than 500 people. Data are not available on the number of individuals with HIV in each specific community, and thus we use the estimated HIV prevalence in the region (approximately 13% in urban areas and 9% in rural areas [4]) to estimate the number of infected people in each community. See Figure 1 and Table 1 for the population sizes and spatial locations of each of the 51 communities used in our analysis. For our analysis the quantity of ARVs available for distribution to the HCFs is sufficient to treat 10% of the total number of infected people, which is a realistic level during the incremental scale-up of ARV therapy over the next few years. The government of South Africa has selected 17 HCFs to participate in the ARV rollout that began in April 2004. These 17 HCFs are distributed throughout the province (see Figure 1 and Table 2). Some communities are close to HCFs, whilst others are a great distance from any HCF, with a range of 0–90 km (Figure 2A). Hence, this spatial distribution of HCFs produces large heterogeneity in accessibility to treatment. Inequality in access to health care is a common characteristic of resource-constrained regions [5,6,7,8,9,10]. We explicitly consider heterogeneity in treatment accessibility in our analysis of ARV allocation strategies.
Figure 1 Map of South Africa Indicating the Location of the KwaZulu–Natal Province and Map of KwaZulu–Natal
Black crosses indicate the location of the 17 HCFs that have been designated for ARV rollout by the South African government, and the spatial distribution of communities distinguished by the number of individuals infected with HIV (by both size and color). Durban (represented by the large red diamond) is the capital city of the province and has more individuals with HIV than any other community. Pietermaritzburg and Newcastle (represented by orange diamonds) have the next greatest numbers of individuals with HIV.
Figure 2 Accessibility of Communities to HCFs
(A) A histogram indicating heterogeneity in the distance from communities in KwaZulu–Natal to the closest HCF. The treatment accessibility function used in our model is a Gaussian distribution, exp(−kd
2), indicating that accessibility is strongly related to distance (d), and k is a dispersal length scale parameter.
(B) The catchment region is shown with an effective radius of 20 km for coverage from each HCF (k = 0.0151).
(C) The catchment region is shown with an effective radius of 40 km for coverage from each HCF (k
= 0.003786).
(D) The catchment region is shown with an effective radius of 60 km for coverage from each HCF (k = 0.00168).
In each case, the red dots indicate the location of the HCF, the green circles represent the locations where treatment accessibility has been reduced to 50% relative to someone located at the HCF, and the blue circles represent the locations where treatment accessibility has been reduced to 1% relative to someone located at the HCF. The locations of communities are presented as black diamonds. The large black diamonds denote large communities (with population greater than 10,000 people), and the small black diamonds denote small communities (with population less than 10,000 people). Substantially more area of the province is covered if HCFs have catchment regions of 60-km radius, relative to catchment regions of 40-km radius, and substantially less area of the province is covered if HCFs have a catchment region of only 20-km radius. However, the proportion of people with access does not differ greatly between the different catchment sizes because of the great spatial heterogeneity in the prevalence of people with HIV.
Table 1 Communities in KwaZulu–Natal with Populations of at Least 500 People
The location and population of each community (city, town, or village) is indicated. Sources: [33] and the World Gazetteer (http://www.world-gazetteer.com)
Table 2 HCFs Proposed for Use in the Rollout of ARVs in KwaZulu–Natal
Source: KwaZulu–Natal Department of Health (http://www.kznhealth.gov.za/default.htm)
We have developed a novel spatial mathematical model of treatment accessibility that we use to determine an OEAS for ARVs in a resource-constrained region. To the best of our knowledge, this is the first analysis to address how to deal with the extremely difficult problem of allocating a scarce supply of ARVs in order to design a rational and equitable allocation strategy. We model the “spatial diffusion of treatment” to the locations of disease, rather than modeling the “spatial diffusion of disease,” which is the conventional approach [11,12,13,14,15,16]. Our spatial model includes HCFs and the HIV-infected communities surrounding these HCFs; we refer to the region around each HCF as the catchment region. Thus, the radius of the catchment region specifies the approximate maximum distance that we assume infected people would be willing (or able) to travel for treatment. Each HCF can serve many communities, and some communities can access multiple HCFs; our model sums the number of people with HIV in each HCF's catchment region who could potentially travel to the HCF to receive ARVs (we define this number as the “effective demand” on that specific HCF). Thus, the “effective demand” on each HCF is a direct function of the number of individuals with HIV in the catchment region, weighted by their distance from the HCF. By including a weighting function we explicitly model heterogeneity in accessibility to treatment based on distance from the HCF. Here, the distance from a HCF becomes the main determinant influencing whether or not an individual with HIV has access to treatment.
We developed an equity objective function to assess how the limited supply of ARVs should be allocated to each HCF to ensure that an equal proportion of infected people in each community receive treatment. To apply our theoretical framework to KwaZulu–Natal we model the specific location of the 17 HCFs and the 51 communities of 500 or more individuals (see Figure 1); for these conditions we determine an OEAS. We compared our OEAS with two other allocation strategies: (i) allocating ARVs only to Durban, the major urban area (i.e., concentrating ARVs where there is the best health-care infrastructure) and (ii) allocating ARVs equally to all 17 HCFs. We conduct our analysis assuming three different radii of catchment regions: 20 km, 40 km, and 60 km. We then extend this analysis and recalculate the OEAS assuming that more than 17 HCFs are available to distribute ARVs. This analysis is useful because there is a second potential pool of 27 ARV-implementation HCFs in the South African operational plan for ARV rollout [17]. We analyze this case, in which 27 HCFs are utilized in the ARV rollout, and we also analyze how optimal ARV allocation would change if all 54 hospitals in KwaZulu–Natal were operational for the rollout of ARVs.
Methods
Calculating Demand and Treatment Access
We assume that the number of people with HIV who will travel to a specific HCF is directly proportional to the number of individuals with HIV in that particular community, but that the probability of an individual traveling to receive ARVs (i.e., the treatment accessibility) decreases with distance from the HCF. We define di,j as the distance from community i to HCF j, f(di,j) as a weighting function that determines the treatment accessibility to a HCF based upon distance di,j, and Ii as the number of people with HIV in community i. The distance, dij, between community i and HCF j is based on the longitude (lon) and latitude (lat) of each location and is determined by
where R is the radius of the earth, taken to be 6,371 km, and the angles are in radian measure. We calculate the “effective demand” of community i on HCF j to be the number of people with HIV in community i that will travel to HCF j for ARV regimes, namely, f(di,j)Ii. Thus, demand on HCFs for ARVs is reduced by the treatment accessibility function. Our model is conceptually similar to the “gravity” models that have been used to predict retail travel [18], plan land use [19], and determine accessibility of primary care [20]. However, this is to our knowledge the first time this approach has been used to calculate ARV allocations. We use a Gaussian to model treatment accessibility, f(d) = exp(−kd
2), where k is a dispersal length scale parameter determining the radius of the catchment region. The size of the actual catchment regions is unknown, but based upon distances from communities to HCFs in KwaZulu–Natal (see Figure 2A) we assume that individuals are likely to travel a maximum distance of approximately 40 km to a HCF (k = 0.003786). We vary the catchment region by considering a 20-km radius (k = 0.0151) and a 60-km radius (k = 0.00168). The different catchment regions that we simulate (with radii of 20 km, 40 km, and 60 km) for each HCF are illustrated in Figure 2B–2D. The number of people with HIV throughout the province that have access to HCFs is approximately 86% of the total number of people with HIV for the case of a 20-km catchment region, 89% for a 40-km catchment region, and 93% for a 60-km catchment region.
Modeling the Distribution of Treatment
To determine how many ARVs should be allocated to each HCF, we first calculate how a given supply of ARVs will be distributed from each HCF to the surrounding communities in the catchment region. We calculate the “effective demand” on HCF j, Dj, to be
which sums the “effective demand” of all communities on HCF j (where there are m communities). Then, we model the distribution of ARVs from a HCF to each community within the catchment region as the proportion of the “effective demand” on HCF j that is contributed by the respective community. Accordingly, ARVs will be distributed from HCF, j, to each community as the ratio
Therefore, the number of people treated in community i by the drug supply allocated to HCF j is
where Sj is the number of regimes allocated to HCF j. Hence, the total number of people with HIV treated in community i,Ti, summing over all n HCFs is
The Equity Objective Function
We establish an equity objective function to determine the optimal equitable allocation of ARVs to each HCF so that all individuals with HIV have an equal chance of receiving treatment. To obtain the same fraction of treated individuals in each community, given that there are A ARV regimes for a total of
individuals with HIV, the resulting objective function to minimize (based on least squares) becomes
Our goal is to minimize E, by solving for the number of ARVs to be allocated to each HCF (S
1, S
2,…, S
n), whilst enforcing the following three constraints: (i) ensure that the total number of ARVs available is equal to the sum of the supply allocated to all HCFs,
(ii) ensure that only a positive number of ARVs are allocated to each HCF (Sj ≥ 0, j = 1…n); and (iii) ensure that the number of people treated in each community is not greater than the number of people with HIV in the community (Ti ≤ Ii, i = 1…m). We note that if a different objective is required, then all of our preceding analysis still holds and only the functional form of the objective function needs to be altered. To solve the problem, and determine the OEAS, we used successive linear programming operations research techniques [21].
Results
The OEAS of ARVs in KwaZulu–Natal that we determined is complex (see Figure 3A and 3B). According to our OEAS, the majority of ARVs should be allocated to HCFs in Durban, and the remaining ARVs should be allocated to the other HCFs throughout the province (with two non-Durban HCFs receiving 5%–15% of the total ARVs and the remaining non-Durban HCFs each receiving less than 5% of the total ARVs available). We note that our OEAS does not produce perfect equality; however, our optimal strategy significantly improves equality in obtaining treatment over the two other allocation strategies that we analyzed for comparison: (i) ARVs allocated only to one HCF (in the largest city, Durban) (see Figure 3D and 3E), and (ii) equal quantities of ARVs allocated to each HCF throughout the province (see Figure 3G and 3H). For comparison of allocation strategies (in Figure 3) we used an effective catchment radius of 40 km (k = 0.003786). The proportion of infected individuals that are treated at each location is displayed graphically in Figure 3 for our OEAS (Figure 3C) and the two comparison allocation strategies (Figure 3F and 3I). The best achievable outcome, given the limited treatment resources available, is that 10% of people with HIV are treated in each community throughout the province, yielding the map shown in Figure 3C, 3F, and 3I, but with dark blue/magenta over the entire province. Whilst our OEAS does not fully achieve this, it is considerably better than both of the comparison ARV allocation strategies. Furthermore, the equity objective function evaluates to E = 0.27 for our OEAS, compared with (i) E = 0.50 and (ii) E = 133.88 for the comparison allocation strategies. There is large diversity in the fraction of individuals with HIV treated per community when equal quantities of ARVs are given to each HCF, evidenced by an inter-quartile range of 0.025%–41.746% compared with inter-quartile ranges of 0%–0% and 0.011%–9.982% for the first comparison strategy and our OEAS, respectively. Therefore, equal access is not obtained if equal quantities of ARVs are allocated to each HCF. Obviously, allocating to only one HCF (the first comparison strategy) could also be considered unequal because although the inter-quartile range is minimal, effectively only one community (Durban) receives ARVs. Our OEAS, while not perfect, achieves the best equality possible given the accessibility constraints and limited ARV supply.
Figure 3 Pie Charts of the Three Strategies for Allocating ARVs to HCFs
The three strategies considered are as follows: allocation of ARVs according to the results of minimizing our objective function (first row) allocation of ARVs only to one HCF in Durban (second row), allocation of ARVs equally to each of the 17 HCFs (third row). The proportion of ARVs allocated by these strategies to the 17 different HCFs is indicated in (A), (D), and (G), with each HCF represented by a different color. The spatial allocation of ARVs is shown in (B), (E), and (H), respectively. The respective percentage of infected people that are treated throughout the KwaZulu–Natal province is simulated in (C), (F), and (I). Here, the x–y plane represents spatial location, and the shaded color at a location refers to the proportion of individuals with HIV that are treated at the specified location. The plots were obtained by generating an interpolating surface where the z-ordinate, colored by magnitude, represents the proportion of treated individuals, and then orientating the view of the surface normal to the x–y plane. We performed surface data interpolation using the method of translates [32].
The catchment region for HCFs is a factor of large uncertainty. We considered three catchment region sizes: radii of 20 km, 40 km, and 60 km. We also simulated two additional cases with increased numbers and locations of HCFs (27 HCFs as suggested in South Africa's official ARV rollout operational plan [17]; and all 54 hospitals in KwaZulu–Natal). In Figure 4 we present box plots of the percentage of infected people that obtain treatment per community for the three sets of HCFs and the three catchment region sizes we simulate. For each specified condition we calculate the OEAS. It is apparent that equality in access to ARVs is improved substantially if the radius of each catchment region is increased and/or the number of HCFs is increased (Figure 4). Our results show that the number of HCFs utilized is of greater importance than the size of the catchment region. If 54 HCFs are used, then even a (small) catchment radius of 20 km results in the ideal median proportion of 10% of people with HIV in each community receiving ARVs. In the case of 27 HCFs, 88% of all people with HIV have access to HCFs for a 20-km catchment region, 91% for a 40-km catchment region, and 96% for a 60-km catchment region. In the case of 54 HCFs, 90% of all people with HIV in the province have access to HCFs for a 20-km catchment region, 94% for a 40-km catchment region, and 99% for a 60-km catchment region. Therefore, increasing the number of HCFs available for an ARV rollout is effective in significantly increasing equality in treatment accessibility as shown in Figure 4. Furthermore, if catchment regions actually have a radius of 60 km, or can be increased to this size through improvements in transportation, this would enable access to HCFs for almost all people in the province, as shown in Figure 4. The actual HCF allocations determined by our model and optimization for the cases of 17, 27, and 54 HCFs (and for all catchment sizes we consider) are presented as pie charts in Figure 5. It is clear from our analysis that the equality criterion, such that each individual with HIV in KwaZulu–Natal has an equal chance of receiving ARVs, can best be satisfied by utilizing all 54 HCFs for ARV distribution and ensuring that each HCF serves a catchment region of 40 to 60 km.
Figure 4 Percentage of People with HIV That Obtain Treatment per Community for Various Approaches
Box plots of the percentage of infected people that obtain treatment per community for the three different sets of HCFs simulated in our analysis for ARV rollout, namely, using the 17 HCFs likely to be used, the 27 HCFs suggested by the South African government as potential implementation points, and all of the 54 hospitals in the KwaZulu–Natal province. These cases are represented for each of the three catchment region sizes we considered (with radii of 20 km, 40 km, or 60 km) and referenced against the ideal fraction treated (dotted blue line) under perfect conditions of egalitarian distribution, given the limited ARV supply. The red crosses indicate the median percentage of people with HIV that obtain treatment per community.
Figure 5 Actual Allocation of ARVs to HCFs
These pie charts show ARV allocation to HCFs according to our model and optimization for the cases of 17 , 27 , and 54. The allocation is shown for each of the catchment region sizes considered: 20-km radius, 40-km radius, and 60-km radius.
Discussion
We have established an elegant and simple theoretical framework for determining an equitable and rational allocation of ARVs to HCFs in resource-constrained countries. To the best of our knowledge, this is the first analysis to address this very difficult problem. We determined that increasing the size of the catchment region of each HCF can improve access to HCFs considerably for rural populations. We suggest that studies be performed to collect data on the distance that individuals with HIV are willing and able to travel for treatment. This will facilitate discussions of this important issue, which must be considered in the making of policy decisions. A database consisting of such information has been proposed for South Africa [22]. In an effort to provide equal access to communities with relatively little access to ARV therapy, the concept of a mobile clinic that would travel between communities to take health-care workers and resources to the location of the demand is a new initiative in Nigeria (S. Agwale, personal communication) that could also be considered in other regions.
We calculated the optimal allocation of ARVs to available HCFs so that all infected individuals will have as close as possible to an equal chance of obtaining treatment. We have shown that increasing the number of HCFs involved in ARV distribution can improve equality of access to ARVs substantially. The current plan in KwaZulu–Natal is to use only 17 HCFs. However, our results clearly show that in order to achieve an optimal equitable allocation strategy, all existing infrastructure (i.e., all 54 HCFs) should be used. The strategy that we are advising may be fairly easy to accomplish at the policy level because the health-care infrastructure (specifically these HCFs) already exists, although consideration must be made for issues such as the training and transportation that is necessary, which may be costly. In contrast, increasing the size of catchment regions may be very difficult. Obviously, increasing both the number of HCFs and the size of the catchment region each services would substantially increase equality of access to health care in KwaZulu–Natal.
Future modeling studies could extend our work by not making the simplifying assumption that all patients have similar ease of travel over the same distance and by including weighting functions on distance impedance for different communities (based on the quality of the road infrastructure, for example, and the availability of transportation) (D. P. Wilson, J. O. Kahn, S. M. Blower, unpublished data). Here, we have shown how to calculate optimal ARV allocation strategies based upon the principle of equity. Future research is necessary to compare ARV allocation strategies based upon the principle of efficiency (i.e., allocating ARVs to maximize epidemic reduction) in order to determine whether utilizing different principles for optimization would result in similar (or different) allocation strategies.
The World Health Organization and the Joint United Nations Programme on HIV/AIDS have identified three core principles that should underlie the effort to fairly distribute ARVs, namely: urgency, equity, and sustainability [23]. They state that policy decisions for the fair distribution of ARVs should be based upon the following ethical principles: (i) the principle that like cases should be treated alike, (ii) the utilitarian principles of maximizing overall societal benefits, (iii) the egalitarian principles of equity (distributing resources, such as health care, equally among different groups), and (iv) the Maximin principle (which prioritizes individuals that are the least advantaged) [24]. Here, we investigated the level of decision-making associated with allocating ARVs to HCFs, and we have applied the egalitarian principle of equity with respect to access to health care. We suggest that allocating ARVs to HCFs to achieve equality in accessibility could be carried out, and then individual-level ethical considerations could be thought out at the next level of deliberation. Future research is necessary to identify alternative (and more detailed) ethical ARV allocation strategies.
Although we have focused on one equitable strategy, there are many other ARV allocation strategies that are ethical. Uneven access to HIV treatment has the very real potential to fracture social and political structures and could lead to intrastate and/or interstate conflict [2]. Government decisions on ARV allocation have potentially socially destabilizing ramifications because essentially the decisions determine who lives and who dies. Resource allocation decisions will have to be made at a number of levels: it must be decided what proportion of the available ARVs should be allocated to each province; then it must be decided how many ARVs should be allocated to each HCF within each region; and finally, particular groups of individuals may be chosen to have treatment priority.
Treatment priority decisions for individuals could be based on many different criteria, including disease progression (CD4 cell counts and viral load), socioeconomic status, ethnicity, and who is thought to have the greatest risk of transmitting infections (for example, pregnant women with HIV or female sex workers). Although it could be argued that behavioral core groups should be targeted to receive ARVs because this may have the greatest epidemiological impact, such an allocation strategy would be neither feasible nor practical to implement. For example, sex workers are an obvious behavioral core group, but many women would likely claim to be sex workers if they knew that ARVs were only available to sex workers. Additionally, the ethics of targeting such groups in favor of other societal groups must be questioned. It could also be argued that, to maximize the preventative effect of ARV therapy, ARVs should be concentrated in virological core groups (i.e., people with the highest viral load) [25,26]; this novel approach of targeting the virological core group has recently been proposed for controlling HSV-2 epidemics [27]. Identifying individuals in the virological core group would be far easier than identifying individuals in the behavioral core group. These individuals are likely to be the sickest and those with evidence of disease-related symptoms. Treatment allocation strategies could also be designed based on reducing the future epidemic impact and disregarding treatment equality amongst currently infected people. Such strategies place different social value on currently infected people in comparison with future infected people; such strategies therefore may not be ethical even though they may be epidemiologically sound (also, it is important to note that any epidemic predictions have large uncertainty ranges [28,29]).
Our model has been applied to the South African province of KwaZulu–Natal, but it can be applied by government health officials in any resource-constrained country. In many of the countries worst affected by the HIV pandemic, scarcity of resources will mean that not everyone that could potentially benefit from ARVs will be able to access them. Many of the decisions that must be made to develop an effective response to the HIV/AIDS epidemic are inevitably underpinned by ethical considerations. Leadership in most resource-constrained regions cannot avoid these decisions. Whilst there has been considerable attention given to South Africa, many other countries worldwide either have plans in place (e.g., Brazil, Thailand, and Botswana) or are in the process of developing national programs for ARV distribution through the public health system (e.g., Mozambique, Malawi, and Kenya) [1]. Legitimate authorities in each nation must come to their own consensus on the priorities and objectives of an ARV rollout, which is not a trivial matter [1,30]. Our objective function and model can be used to calculate allocation strategies that provide equity in access (compensating for geographical isolation), but if authorities in a given nation prioritize a different goal for ARV rollout, then an objective function to optimize can be formulated to reflect the specific national policy goal. Our model can be used by policy makers to determine an optimal scientifically based allocation strategy, based upon the specific objective function. As the ARV rollout commences in KwaZulu–Natal, difficult decisions will have to be made as to how to allocate scarce resources. We have shown that it is possible to obtain a mathematical solution to an equity problem. We suggest that our novel approach could be used to determine optimal equitable allocation strategies for many other resource-constrained countries that are just beginning to receive ARVs [31].
Patient Summary
Background
Antiretroviral drugs can change the lives of patients with HIV/AIDS. Their high price, however, means that many poor countries do not have enough of these drugs to treat all the people who need them. The decision of who will get treatment is very difficult, and different ways to come up with ethical solutions to the problem have been proposed.
Why Was This Study Done?
One of the approaches is to try to make sure that every infected person has the same chance to get antiretroviral drugs. David Wilson and Sally Blower, the authors of this study, wanted to find a scientific strategy to achieve this goal of equal access.
What Did the Researchers Do?
They used mathematical models to calculate how to distribute available drugs among hospitals and doctor's offices so that each patient in a particular area had an equal chance to get treated.
What Did They Find?
When they used their approach on a real example, the South African province of KwaZulu–Natal, they found that making some changes to the current plans for drug distribution would lead to more equal access among all of the individuals with HIV in the province. Instead of only 17 out of the 54 health care facilities in KwaZulu–Natal distributing the drugs (which is the current plan of the South African government), Wilson and Blower calculate that it would be fairer if all 54 facilities distributed the medicines.
What Does This Mean?
Mathematical models like the one used here are always based on assumptions and simplifications. As a consequence, they are never perfect matches for a real-life situation, but they can help to guide complicated decisions. This article suggests that the approach Wilson and Blower developed could help to determine strategies for equitable allocation of limited HIV treatment resources.
What Next?
The authors hope that the tools they developed will be used by policy makers in resource-poor countries to guide their strategies. They are keen to work with these policy makers to adapt and optimize the method to local settings and priorities.
More Information Online
Report by the World Health Organization and the Joint United Nations Programme on HIV/AIDS on ethics and equitable access to HIV/AIDS treatment: http://www.who.int/hiv/pub/advocacy/en/ethicsmeetingreport_e.pdf
Ruth Macklin's report on ethics and equity in access to HIV treatment: http://www.who.int/ethics/en/background-macklin.pdf
The Pro-Poor Health Policy Team's report on priority in HIV/AIDS treatment: http://www.who.int/ethics/en/background-pro-poor3.pdf
News article from the World Health Organization Bulletin on the South African HIV/AIDS treatment program: http://www.who.int/bulletin/volumes/82/1/en/news.pdf
Acknowledgments
The authors acknowledge the financial support of the National Institutes of Health National Institute of Allergy and Infectious Diseases (RO1 AI041935). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Citation: Wilson DP, Blower SM (2005) Designing equitable antiretroviral allocation strategies in resource-constrained countries. PLoS Med 2(2): e50.
Abbreviations
ARVantiretroviral
HCFhealth-care facility
OEASoptimal equitable allocation strategy
==== Refs
References
Institute of Medicine Committee on Examining the Probable Consequences of Alternative Patterns of Widespread Antiretroviral Drug Use in Resource-Constrained Settings Scaling up treatment for the global AIDS pandemic: Challenges and opportunities 2004 Washington (D.C.) National Academies Press 325
Cheek R Playing god with HIV: Rationing HIV treatment in South Africa. African Security Rev 10. Available: http://www.iss.org.za/PUBS/ASR/10No4/Cheek.html
2001 Accessed 10 January 2005
Ingham R Health-AIDS-Africa-drugs: Fair access to HIV drugs in Africa is “political timebomb.” Agence France-Presse. Available: http://www.aegis.com/news/afp/2003/AF0309D3.html
2003 September 5 Accessed 10 January 2005
Shisana O Simbayi L Full report: Nelson Mandela/HSRC Study of HIV/AIDS 2002 Cape Town Human Sciences Research Council 121
Long K The concept of health. Rural perspectives Nurs Clin North Am 1993 28 123 130 8451203
Benatar S Bioethics: Power and injustice: IAB presidential address Bioethics 2003 17 387 398 14870761
Stierle F Kaddar M Tchicaya A Schmidt-Ehry B Indigence and access to health care in sub-Saharan Africa Int J Health Plann Manage 1999 14 81 105 10538937
Castro A Farmer P Infectious disease in Haiti: HIV/AIDS, tuberculosis and social inequalities EMBO Reports 2003 4 S20 S23 12789400
Hjortsberg C Why do the sick not utilise health care? The case of Zambia Health Econ 2003 12 755 770 12950094
Noor A Zurovac D Hay S Ochola S Snow R Defining equity in physical access to clinical services using geographical information systems as part of malaria planning and monitoring in Kenya Trop Med Int Health 2003 8 917 926 14516303
Murray J Stanley E Brown D On the spatial spread of rabies among foxes Proc R Soc Lond B Biol Sci 1986 229 111 150 2880348
Torres-Sorando L Rodriguez D Models of spatio-temporal dynamics in malaria Ecol Modell 1997 104 231 240
Grenfell B Bolker B Cities and villages: Infection hierarchies in a measles metapopulation Ecol Lett 1998 1 63 70
Grenfell B Bjørnstad O Kappey J Travelling waves and spatial hierarchies in measles epidemics Nature 2001 414 716 723 11742391
Filipe J Gibson G Comparing approximations to spatio-temporal models for epidemics with local spread Bull Math Biol 2001 63 603 624 11497160
Xia Y Bjørnstad O Grenfell B Measles metapopulation dynamics: A gravity model for epidemiological coupling and dynamics Am Nat 2004 164 267 281 15278849
South Africa Department of Health Operational plan for comprehensive HIV and AIDS care, management, and treatment for South Africa. South African Government Information. Available: http://www.info.gov.za/otherdocs/2003/aidsplan.pdf
2003 Accessed 10 January 2005
Reilly WJ The law of retail gravitation 1931 New York Knickerbocker Press 183
Hansen W How accessibility shapes land use J Am Inst Plann 1959 25 73 76
Guagliardo M Spatial accessibility of primary care: Concepts, methods and challenges Int J Health Geogr 2004 3 3 14987337
Sarker, R.A. & Gunn, E.A. A simple SLP algorithm for solving a class of nonlinear programs European Journal of Operational Research 1997 101 140 154
Busgeeth K Rivett U The use of a spatial information system in the management of HIV/AIDS in South Africa Int J Health Geogr 2004 3 13 15239839
World Health Organization, Joint United Nations Programme on HIV/AIDS Treating 3 million by 2005: Making it happen. Geneva: World Health Organization. Available: http://www.who.int/3by5/publications/documents/en/3by5StrategyMakingItHappen.pdf
2003 Accessed 4 January 2005
Macklin R Ethics and equity in access to HIV treatment: 3 by 5 initiative. Geneva: World Health Organization. Available: http://www.who.int/ethics/en/background-macklin.pdf
2004 Accessed 4 January 2005
Hyman JM Li J Stanley EA The differential infectivity and staged progression models for the transmission of HIV Math Biosci 1999 155 77 109 10067074
Hyman JM Li J Stanley EA Modeling the impact of random screening and contact tracing in reducing the spread of HIV Math Biosci 2003 181 17 54 12421551
Blower S Wald A Gershengorn H Wang F Corey L Targeting virological core groups: A new paradigm for controlling HSV-2 epidemics J Infect Dis 2004 190 1610 1617 15478066
Blower S Aschenbach A Gershengorn H Kahn J Predicting the unpredictable: Transmission of drug-resistant HIV Nat Med 2001 7 1016 1020 11533704
Blower S Gershengorn H Grant R A tale of two futures: HIV and antiretroviral therapy in San Francisco Science 2000 287 650 654 10649998
Singler J Farmer P Treating HIV in resource-poor settings JAMA 2002 288 1652 1653 12350202
Blower S Bodine E Kahn J McFarland W The antiretroviral rollout and drug-resistant HIV in Africa: Insights from empirical data and theoretical models AIDS 2005 19 1 14 15627028
Light W Vail M Extension theorems for spaces arising from approximation by translates of a basic function J Approx Theory 2002 114 164 200
Falling Rain Genomics Directory of cities and towns in province of KwaZulu–Natal, South Africa. Available: http://www.fallingrain.com/world/SF/2/
2004 Accessed 28 April 2004
| 15737005 | PMC549597 | CC BY | 2021-01-05 10:39:33 | no | PLoS Med. 2005 Feb 22; 2(2):e50 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020050 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020052SynopsisGenetics/Genomics/Gene TherapyInfectious DiseasesVirologyEpidemiology/Public HealthInfectious DiseasesEpidemiologyGeneticsMass Spectometry–Based SARS Genotyping Synopsis2 2005 22 2 2005 2 2 e52Copyright: © 2005 Public Library of Science.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.
SARS Transmission Pattern in Singapore Reassessed by Viral Sequence Variation Analysis
==== Body
To quickly control infectious disease outbreaks, extensive information is required to identify the source and transmission routes, and to evaluate the effect of containment policies. Traditionally, scientists have used travel- and contact-tracing methods, but the recent SARS epidemic showed that sequence-based techniques for pathogen detection can also be important tools to help understand outbreaks. Jianjun Liu and colleagues adapted mass spectrometry (MS)–based genotyping, already used as a high-throughput way of detecting single nucleotide polymorphisms in human DNA, to the analysis of the SARS virus from clinical samples.
The major breakthroughs against SARS were the discovery of the SARS coronavirus (SARS-CoV) as the etiological agent and the sequencing of the SARS genome. Liu's colleagues at the Genome Institute of Singapore had previously shown that common genetic variants in the SARS-CoV genome could be used as molecular fingerprints to help trace the route of infection. However, as “sequence analysis of large numbers of clinical samples is challenging, cumbersome, and expensive,” they felt that “what is needed is a rapid, sensitive, high throughput, and cost-effective screening method.” Towards this goal, Liu and colleagues now demonstrate that an MS-based technique can quickly yield accurate information on clinical isolates (in this case from the 2003 SARS outbreak in Singapore).
Two transmission routes for the Singapore SARS outbreak
The scientists demonstrate the sensitivity of the assay in detecting SARS-CoV variations and test it further in cultured viral isolates and uncultured lung tissue samples of SARS-CoV. They analyzed isolates taken from 13 patients with SARS at different stages of the Singapore outbreak, identified nine sequence variations, and discovered a new primary route of introduction of the virus into the Singapore population. They also found a Singaporean origin for a German case of SARS, a result that could not be derived from standard sequencing methods. The analysis of the uncultured lung tissue also found different sequences in a single patient, which suggested the presence of multiple viral sequence variants in one host.
The study suggests that MS-based genotyping can be used for large-scale genetic characterization of viral DNA from clinical samples. The researchers found that the method was accurate and sensitive, with a 95% success rat e for detecting sequence variations at low virus concentrations. The MS-based assay allows high-throughput analysis and complements the “gold standard” direct sequence analysis method, which is used to identify new sequence variations. As such, it is particularly useful for investigating agents for which extensive sequence information exists.
Liu and colleagues propose that the most efficient method for a large-scale population investigation would be initial characterization of a genome sequence by direct sequence analysis in a subset of samples, followed by MS-based analysis of informative genetic variations. Altogether, their results suggest that MS-based genetic analysis can help real-time investigations in disease outbreaks.
| 0 | PMC549598 | CC BY | 2021-01-05 10:39:33 | no | PLoS Med. 2005 Feb 22; 2(2):e52 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020052 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020053SynopsisImmunologyInfectious DiseasesEpidemiology/Public HealthPediatricsInfectious DiseasesImmunology and allergyPublic HealthPediatricsTowards Better Evaluation of Pneumococcal Vaccines Synopsis2 2005 22 2 2005 2 2 e53Copyright: © 2005 Public Library of Science.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.
Use of Procalcitonin and C-Reactive Protein to Evaluate Vaccine Efficacy against Pneumonia
==== Body
Pneumonia remains the leading cause of death worldwide in children. Several vaccines against pneumococcal pneumonia are at various stages of development, but the testing of their efficacy is hampered by the lack of noninvasive tests that are sensitive and specific for the disease. Diagnosis is usually based on chest radiographs, which are not very specific for pneumococcal disease.
In their quest for a more specific diagnostic test, Shabir Madhi and colleagues—who are conducting clinical trials on pneumococcal vaccines in children—examined whether serum concentrations of procalcitonin and C-reactive protein could improve the specificity of chest radiographs to diagnose pneumococcal pneumonia and thus be useful in the future evaluation of pneumococcal vaccines. Elevated levels of both proteins are associated with bacterial disease. They might therefore help to differentiate bacterial from nonbacterial causes of pneumonia, and thus allow to “enrich” the analyzed disease cases for those of pneumococcal origin, against which the vaccine is potentially active.
This study represents a first step, in which the researchers tested whether adding information about procalcitonin and C-reactive protein levels to data from a completed vaccine trial would affect the outcome regarding vaccine efficacy. When reanalyzing previous trial data under these conditions, the vaccine appeared more efficacious compared with placebo when either elevated procalcitonin or elevated C-reactive protein levels were taken into account. The efficacy estimate was greatest when cases of pneumonia that had elevated levels of both procalcitonin and C-reactive protein were compared against placebo.
These data suggest that elevated levels of C-reactive protein and procalcitonin, in conjunction with chest radiography, could improve the specificity of a diagnosis of pneumococcal pneumonia over that of chest radiography alone. This combined diagnostic test could be useful for further evaluation of pneumococcal vaccines. The hope is that among patients identified as having pneumonia by the combined test, a higher proportion would have pneumonia of pneumococcal origin. As a consequence, there would be less “background noise” caused by other forms of pneumonia, and this should make it easier to assess the efficacy of vaccine candidates. However, as the researchers point out, this analysis was not a primary objective of the present trial. This analysis can therefore serve only as a hypothesis-generating study, and as such the hypothesis must be tested in other study settings.
The study was sponsored by Wyeth, manufacturers of the pneumococcal vaccine used.
| 0 | PMC549599 | CC BY | 2021-01-05 10:39:36 | no | PLoS Med. 2005 Feb 22; 2(2):e53 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020053 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020054SynopsisHIV/AIDSHIV Infection/AIDSMedicine in Developing CountriesInjections and HIV in Rural Zimbabwe Synopsis2 2005 22 2 2005 2 2 e54Copyright: © 2005 Public Library of Science.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.
Individual Level Injection History: A Lack of Association with HIV Incidence in Rural Zimbabwe
==== Body
Of the 40 million people worldwide with HIV, 30 million live in the developing world. By far the worst hit region is sub-Saharan Africa, where nearly four million children have lost one or both parents to HIV/AIDS since 2000. Is heterosexual transmission the driving force behind the HIV epidemic in sub-Saharan Africa? In a controversial debate, some researchers have suggested that other factors such as unsafe medical injection practices may also be to blame, and that by overlooking, and even suppressing, analysis of this possible route of transmission, the current focus on preventing sexual transmission may be misguided. In this month's PLoS Medicine, Ben Lopman and colleagues argue that although it is right to criticize the lack of evidence on unsafe medical injection, field data are hard to collect. They note that in the only published study addressing this issue, Kiwanuka and colleagues found no link between unsafe injections and HIV spread in rural Uganda. In an effort to “inform the debate” further, Lopman and colleagues looked at the association between HIV and unsafe injection practices in rural Zimbabwe.
Are medical injections an important cause of HIV in rural Africa?
The team analyzed data from adults in Manicaland, a rural part of Zimbabwe, who were taking part in the Manicaland HIV/STD Prevention Study. In 1999 and 2000, eligible patients were tested for HIV and surveyed (86.7% were HIV negative at the start of the study), and were followed up three years later. The team collected survey data on injections in the patients, who were male and female adults aged 15 to 54 years old, and tested for an association between injection exposure and HIV infection. In 2002 and 2003, 505 of the men and 1,342 of the women, representing a 69.7% follow-up, were again interviewed and tested for HIV infection. Of these, 40% reported having had an injection or needle prick during the study period. A total of 67 patients developed HIV during the study; of these 13 (19%) said they had not had sex during the study period and 40 (60%) said they had not had an injection. The statistical analysis found no significant association between injections and HIV infection in men or women.
Patients who had HIV when the study began did not have higher rates of injections. Instead, injections were highly associated with childbirth and pregnancy. But since HIV–positive women have reduced fertility, a reduction in the use of maternal services may partially explain why injections were not more common in these HIV-positive patients. In this study, the strongest predictor of HIV infection was symptoms of sexually transmitted disease.
Despite problems of recall bias and under-reporting of sexual activity—a particularly difficult problem in studies in Africa—sexual behavior is consistently linked with HIV incidence. Where does this leave the debate over injections in Africa? Certainly, for this community, they do not seem to be a major source of HIV infection, and local policy-makers would therefore do best to concentrate on the prevention of sexually transmitted infections.
| 0 | PMC549600 | CC BY | 2021-01-05 10:39:38 | no | PLoS Med. 2005 Feb 22; 2(2):e54 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020054 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1573701010.1371/journal.pmed.0020055EditorialPathologyStatisticsPathologyStatisticsResearch DesignWhy Bigger Is Not Yet Better: The Problems with Huge Datasets EditorialThe PLoS Medicine Editors 2 2005 22 2 2005 2 2 e55Copyright: © 2005 Public Library of Science.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.Traditional two dimensional journal publishing may no longer be enough, and may be hindering, the interpretation of the complex datasets and analyses generated by microarray experiments in medicine
==== Body
“Publishing results in traditional paper based way in a journal hides too much information.” This is the verdict of Markus Ruschhaupt and colleagues who, in a paper in Statistical Applications in Genetics and Molecular Biology (3: article 37), discuss a paradigm for the presentation of complex data—in this case, from microarray analyses. The title of the article, “A Compendium to Ensure Reproducibility in High-Dimensional Classification Tasks,” may not lend itself easily to a clinical audience, but the underlying message to clinicians could not be more important: that, currently, studies involving large datasets, especially ones that have a clinical outcome, are so poorly reported (or possibly so poorly done) that many are not reproducible. This problem was also the topic of a recent meeting in Heidelberg, “Best Practice in Microarray Studies” (http://www.biometrie.uni-heidelberg.de/workshops/bestpractice/index.htm).
As microarrays have become mainstream research tools in biology and medicine, the large datasets and complex analyses from these studies have presented challenges: for authors in analyzing the data, for reviewers and editors in deciding on the suitability of papers for publication, for journals in determining how much data needs to be presented within the paper itself, for other researchers in reproducing the data, and, finally, for readers in deciding how to assess the data presented. The results from several high-profile papers have already proved difficult to reproduce, even by those with sufficient time and computing expertise.
Where do such analyses leave the new science of molecular pathology? Ruschhaupt and colleagues comment that “the literature on the induction of prognostic profiles from microarray studies is a methodological wasteland.” Much the same could be said of other applications of molecular biology to clinical samples. A systematic review of molecular and biological tumor markers in neuroblastoma (Clin Cancer Res 10: 4–12) found that its conclusions were limited by “small sample sizes, poor statistical reporting, large heterogeneity across studies…and publication bias.” John Ioannidis and colleagues (Lancet 362: 1439–1444) did a similar analysis of 30 microarray studies with major clinical outcomes in cancer. They showed that the studies were small—median sample size was 25 patients, and validation was incomplete in most studies. They recommended that molecular prognostic studies be classified as phase 1 (early exploratory probing associations), phase 2 (exploratory with extensive analyses), or phase 3 (large confirmatory studies with pre-stated hypotheses and precise quantification of the magnitude of the effect), and that only studies that had undergone phase 3 testing should be considered robust enough for use in clinical practice. Most current studies should be considered as phase 1 or, at best, phase 2.
So, despite considerable hype, the published studies are far from the level of evidence that would be accepted for virtually any other medical test. In a review in 2003 (Hematology [Am Soc Hematol Educ Program] 2003: 279–293), Rita Braziel and colleagues concluded, “rapid identification and neutralization of spurious results is essential to prevent them from becoming accepted facts.”
But these problems are not new in medical research. In 1994 (BMJ 308: 283–284), Doug Altman, who was instrumental in developing the CONSORT guidelines for reporting of clinical trials, said that “huge sums of money are spent annually on research that is seriously flawed through the use of inappropriate designs, unrepresentative samples, small samples, incorrect methods of analysis, and faulty interpretation,” and “that quality control needs to be built in from the start rather than the failures being discarded.”
So how can we ensure that the wealth of data pouring out of microarray and other molecular diagnostic studies is turned into meaningful knowledge? The Microarray Gene Expression Data Society has proposed a set of guidelines (MIAME) for the reporting of microarray data, and that all such data should be deposited in public databases. But as Ruschhaupt and others have shown, disclosure of results and data is not enough, since there is little consensus on the appropriate statistical analyses and many are developed on a case by case basis, which may not be reproducible, even by the authors. Some researchers advocate the use of standard statistical packages, which allows the reader to repeat an entire analysis quickly and, hence, assess the robustness of the results. Some authors have produced a transcript of their statistical analyses as a supplement to their articles (e.g., Nucleic Acids Res 32: e50). At the very least authors should have a protocol with a prespecified plan for patient selection and statistical analysis—accepted practice for clinical trials, but not yet for other medical research. An ultimate aim for reporting would be the type of compendium discussed by Ruschhaupt and colleagues—“an interactive document that bundles primary data, statistical processing methods, figures, and derived data together with the textual documentation and conclusions.” One such compendium is illustrated in a paper by Robert Gentleman (Stat Appl Genet Mol Biol 4: article 2). PLoS Medicine is keen to work with authors towards making such reporting possible. But although the time might have gone when the two-dimensional journal article could suffice for complex papers, clinicians should nonetheless apply the same critical assessment that they would for any other clinical tool. If a result is too good to be true, it probably is.
Virginia Barbour is on the advisory board of the Microarray Gene Expression Data Society.
| 15737010 | PMC549601 | CC BY | 2021-01-05 11:13:38 | no | PLoS Med. 2005 Feb 22; 2(2):e55 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020055 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020056SynopsisDiabetes/Endocrinology/MetabolismDiabetesDiabetic Renal DiseaseWhy Blood Glucose Control Matters for the Kidney Synopsis2 2005 22 2 2005 2 2 e56Copyright: © 2005 Public Library of Science.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.
Multiple Metabolic Hits Converge on CD36 as Novel Mediator of Tubular Epithelial Apoptosis in Diabetic Nephropathy
==== Body
One of the most common and most serious complications of both type 1 and type 2 diabetes is diabetic nephropathy. It occurs in around 30% of patients with type 1 diabetes and 10% to 40% of patients with type 2 diabetes. Diabetic nephropathy is the leading cause of renal failure in the developed world. The main effect of diabetic nephropathy is proteinuria, initially in very small amounts but which increases, leading to nephrotic syndrome and end-stage renal disease in most cases.
Apoptotic renal tubular cells in diabetic nephropathy
Various risk factors in individuals with diabetes are known to increase the chance of developing diabetic nephropathy, including South Asian or African background, male sex, long history of diabetes, poor blood sugar control, high blood pressure, and smoking. One early change associated with diabetic nephropathy is degeneration of the renal tubular epithelium, but the exact cause of this at the cellular level is unclear. Erwin Böttinger and colleagues have dissected out one key point in the progression to diabetic nephropathy. They looked at cell lines of renal tubular cells from humans and mice and kidney biopsies from patients with diabetic nephropathy, patients with non-diabetic renal disease, and mice with genetic and induced diabetes. In the human cell lines they showed that glucose induced the expression of CD36, a receptor known to have a role in adhesion and signal transduction (in addition to being the receptor for malaria-infected erythrocytes). They then went on to show that apoptosis of these cells occurred in the presence of glycated (glucose-modified) albumins or free fatty acids, which are present in increased amounts in patients with diabetes, and that CD36 was essential for the apoptosis to occur. They then examined how CD36 triggered apoptosis and found that it involved src kinase, p38 MAP kinase, and caspase 3.
Comparing mice and humans, the researchers found that the two species are not alike: diabetic mice did not show an increase in tubular expression of CD36—even though the gene is present in mice—and had normal tubular epithelium and no tubular apoptosis. They confirmed this difference between humans and mice by showing that normal mouse epithelial cell lines were resistant to apoptosis caused by the glycated albumins; however, artificially expressing CD36 in these lines made them susceptible to apoptosis by these modified albumins.
These results provide insight into one of the crucial steps in diabetic nephropathy and, in humans at least, might help to explain why high blood glucose is so damaging to the kidney, hence providing a good reason—if another is needed—for encouraging patients to control blood glucose as tightly as possible.
| 0 | PMC549602 | CC BY | 2021-01-05 11:13:38 | no | PLoS Med. 2005 Feb 22; 2(2):e56 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020056 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020057SynopsisInfectious DiseasesEpidemiology/Public HealthHealth PolicyHIV/AIDSMedical EthicsPublic HealthHIV Infection/AIDSMedicine in Developing CountriesHealth PolicyEquitable Allocation of Antiretrovirals in Resource-Constrained Countries Synopsis2 2005 22 2 2005 2 2 e57Copyright: © 2005 Public Library of Science.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.
Designing Equitable Antiretroviral Allocation Strategies in Resource- Constrained Countries
Designing an Equitable Strategy for Allocating Antiretroviral Treatments
==== Body
Antiretroviral drugs change the lives of patients with HIV/AIDS—if they have access to them. Most patients in resource-poor countries cannot afford the drugs. Major initiatives are under way to expand access to antiretrovirals in developing countries, but the number of individuals in need of the drugs currently vastly exceeds the supply, and will continue to do so for the foreseeable future. These circumstances make for difficult decisions about treatment allocation. David Wilson and Sally Blower have shown how it is possible to design an equitable antiretroviral allocation strategy, that is, to come up with a plan that would give each individual with HIV an equal chance of receiving antiretrovirals. Their novel spatial model enables them to model the “spatial diffusion” of antiretrovirals in a resource-constrained country.
Modeling allocation of antiretrovirals in KwaZulu–Natal
Based on the premise that only a limited number of drugs will be available and only a limited number of health-care facilities can be used for drug distribution (each of them serving the population in a specific area), they determine an optimal equitable allocation strategy. They then apply this approach to a practical example—the equitable allocation of antiretrovirals to patients with HIV/AIDS in the South African province of KwaZulu–Natal. Using data from a detailed rollout plan for antiretrovirals designed by the South African government, they come up with an allocation strategy that differs substantially from the current governmental plan for the province.
KwaZulu–Natal has a total of 54 health-care facilities, of which 17 are assigned to allocate antiretrovirals under the current plan. It is the largest province in South Africa, with a population of about 9.4 million, and it has more people with HIV than any other province (about 21% of all cases in South Africa). Wilson and Blower assume that the available amount of antiretrovirals can treat 10% of the individuals with HIV in KwaZulu–Natal. Modeling the 17 health-care facilities and the 51 communities of individuals with HIV, they determine the amount of drugs to allocate to each facility to achieve equitable access by patients throughout the province. They then extend the analysis assuming that additional health-care facilities could be made available to distribute drugs. They conclude that in order to achieve the greatest degree of treatment equality, all 54 health-care facilities should be used, and they should, on average, each serve the population within a radius of 50 km.
Wilson and Blower discuss how their model can be adjusted and therefore used by policy makers in resource-constrained countries to determine a scientifically based allocation strategy for limited resources based on a number of specific objectives. They also recognize that there are other considerations that influence ethical treatment allocation besides equity, for example, the desire to maximize epidemic reduction, or the imperative to give priority to the least advantaged individuals They believe that their model can be adjusted and therefore “used by policy makers to determine an optimal scientifically based allocation strategy” for a number of specific objectives. Another possibility would be to apply the equity strategy to allocate drugs to particular health-care facilities (thereby achieving equality in accessibility), and then take additional ethical considerations into account at the community level.
| 0 | PMC549603 | CC BY | 2021-01-05 11:13:38 | no | PLoS Med. 2005 Feb 22; 2(2):e57 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020057 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1573701310.1371/journal.pmed.0020069PerspectivesEpidemiology/Public HealthHealth EconomicsHealth PolicyHIV/AIDSMedical EthicsResource allocation and rationingEthicsInternational healthHIV Infection/AIDSMedicine in Developing CountriesDesigning an Equitable Strategy for Allocating Antiretroviral Treatments PerspectivesCapron Alexander M Reis Andreas *Alexander M. Capron is Director of the Department of Ethics, Trade, Human Rights, and Health Law of the World Health Organization, Geneva, Switzerland. Andreas Reis is Associate Professional Officer in the Department of Ethics, Trade, Human Rights, and Health Law of the World Health Organization, Geneva, Switzerland. The paper reflects the authors' own views, which may not necessarily be those of the World Health Organization.
Competing Interests: The authors declare that they have no competing interests.
*To whom correspondence should be addressed: E-mail: [email protected] 2005 22 2 2005 2 3 e69Copyright: © 2005 Capron and Reis.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.
Designing Equitable Antiretroviral Allocation Strategies in Resource- Constrained Countries
Equitable Allocation of Antiretrovirals in Resource-Constrained Countries
A study by Wilson and Blower in the February issue of PLoS Medicine addressed the issue of ensuring equity in distributing AIDS medications. Reis and Capron discuss the study's implications
==== Body
Background
Of the roughly 40 million people living with HIV [1], an estimated 6 million in developing countries urgently need life-saving antiretroviral therapy (ART) [2]. Yet when the 3 by 5 Initiative was launched by the World Health Organization (WHO) and the Joint United Nations Programme on HIV/AIDS (UNAIDS) in December 2003—an initiative that aims to treat 3 million people with HIV in developing countries with ART by the end of 2005—most people with HIV in these countries did not even know their HIV status, and less than 8% were receiving ART. Moreover, even the success of the initiative would still mean that fewer than half of the people who could benefit from such treatment would be receiving it.
Given this gap between what can be done and what needs to be done, the people who set policies and administer programs to provide ART in high-burden countries are faced with difficult questions of distributive justice. Decisions regarding the pricing of ARTs and other care for patients with HIV/AIDS, the distribution of treatment centers, and potential measures to overcome barriers for vulnerable populations will determine who will get access to treatment and who will die. In order to deal with these crucial issues, decision-makers need guidance on how to design policies on equitable access to ART that respect human rights norms and ethical standards.
Calculating Equitable Access: A New Study
A new study by Wilson and Blower, published in the February issue of PLoS Medicine [3], addresses an important dimension of equity in AIDS treatment, namely, the accessibility of health facilities to persons in need. Most published measures of spatial accessibility to health care can be classified into four categories based on the measure of accessibility they use: provider-to-population ratios, distance to nearest provider, average distance to a set of providers, and gravitational (which shows the potential interaction between any population point and all service points within a reasonable distance) [4]. Wilson and Blower developed a mathematical model of the last type that could inform policy-makers' decisions regarding the optimal distribution of treatment sites to ensure equal access by all individuals infected with HIV. Applying this tool to the South-African province of KwaZulu–Natal, Wilson and Blower were able to confirm mathematically the intuitive assumption that using a maximum number of centers, at the least possible distance from most affected populations, would lead to the greatest fairness in the geographical distribution of ART.
Strengths and Weaknesses of the Study
While the authors suggest that their method could be adapted to take other objectives into account, here they have taken an exclusively egalitarian approach to equity. Although this notion of equity is broadly accepted, other important approaches could have been taken into consideration. Simple equality in access can actually produce inequities (because a fair approach would differentiate among groups in the population according to their different needs); further, under some theories, those who are least advantaged generally should receive a disproportionate share of newly distributed benefits (the maximin principle) [5]. In geographic terms, this goal could be reached by setting up treatment sites preferentially in neglected rural areas or urban slums. Conversely, utilitarian ethics would favor locating treatment sites so as to maximize overall benefits to the population, such as by concentrating treatment in already existing sites that could scale up treatment volume at the lowest cost per patient.
In determining equitable spatial accessibility for the application of their model to KwaZulu–Natal, the authors used a rather rough estimation of HIV prevalence (13% in urban areas and 9% in rural areas). As prevalence greatly varies between specific communities, future studies would certainly benefit from using more disaggregated data where available (see, for example, [6]). Similarly, as the authors recognize, the geographic accessibility of treatment not only is a function of distance, but may be strongly influenced by other factors, such as available transportation options. The concept of “catchment regions” is a valuable one, though still a factor of great uncertainty. Further research is needed to examine the ability and willingness of patients with HIV to travel, taking into account factors such as disease stage, travel times and transportation prices, and socioeconomic factors.
Place matters, but spatial accessibility is only one factor to be overcome in ensuring equitable access to health services. Studies show that even when services are available at a near distance, factors such as temporal accessibility, disease perception, stigmatization, and outright discrimination heavily influence “effective demand” [7]. Moreover, several studies have shown that the price of ARTs may be one of the greatest barriers to access and adherence [8], as even small fees at point of service can prove prohibitive for many people.
The Future
Wilson and Blower have developed a mathematical model to determine the fair geographical distribution of ART treatment sites and have applied it to the specific setting of KwaZulu–Natal. Despite some methodological and data limitations, such studies can inform policy-makers' decisions regarding the location of HIV services. Since distance to a treatment center is strongly determinant of patients' ability to access care, WHO is developing a service availability mapping tool to monitor relative equity between districts and identify major gaps in service availability, for example, availability of ART and prevention of mother-to-child transmission programs.
Not only is further research needed to refine the spatial accessibility model presented by the authors but careful attention must be paid to other factors that affect access to HIV services and to the underlying assumptions as to what would constitute fair distribution. In a recent guidance document, WHO and UNAIDS recommended that ART programs include special measures to ensure access of vulnerable and marginalized populations and women to ART [9]. The decision-making processes regarding who will get treatment and who won't must be closely monitored for transparency and inclusiveness. Evaluators should also be able to determine the extent to which the scaling-up of HIV/AIDS programs are reaching the target populations and producing equitable results (see Figure 1). To ensure that this process is robust and evenhanded, the guidance document recommends that national AIDS commissions and programs appoint ethics advisory bodies. These ethics committees are to make sure that issues of equity receive attention alongside technical considerations, such as the manner in which ART programs are integrated into the general health system and the identification and training of health personnel, whose absence is often the greatest barrier to adequate HIV care [10].
Figure 1 Steps to Equitable Access—The Policy Development Cycle at a Glance
IDU, intravenous drug user; NGO, non-governmental organization; PLWHA, people living with HIV/AIDS. (Source: [9])
Citation: Capron AM, Reis A (2005) Designing an equitable strategy for allocating antiretroviral treatments. PLoS Med 2(3): e69.
Abbreviations
ARTantiretroviral therapy
UNAIDSJoint United Nations Programme on HIV/AIDS
WHOWorld Health Organization
==== Refs
References
UNAIDS, WHO AIDS epidemic update: 2004 2004 Available: http://www.unaids.org/wad2004/report.html . Accessed 27 January 2005
WHO, UNAIDS Treating 3 million by 2005: Making it happen: The WHO strategy 2003 Geneva WHO Available: http://www.who.int/3by5/publications/documents/en/3by5StrategyMakingItHappen.pdf . Accessed 27 January 2005
Wilson DP Blower SM Designing equitable antiretroviral allocation strategies in resource-constrained countries PLoS Med 2005 2 e50 15737005
Guagliardo MF Spatial accessibility of primary care: Concepts, methods and challenges Int J Health Geogr 2004 3 3
Rawls J A theory of justice, revised ed 1999 Oxford Oxford University Press 560
UNAIDS, WHO Epidemiological fact sheets on HIV/AIDS and sexually transmitted infections, 2004 update: South Africa 2004 Available: http://globalatlas.who.int/GlobalAtlas/PDFFactory/HIV/EFS_PDFs/EFS2004_ZA.pdf . Accessed 28 January 2005
Mashamba A Robson E Youth reproductive health services in Bulawayao, Zimbabwe Health Place 2002 8 273 283 12399216
Desclaux A Lanièce I Ndoye I Taverne B Calandra S Furness A Karafin A The Senegalese antiretroviral drug access initiative: An economic, social, behavioral, and biomedical analysis Agence Nationale de Recherches sur le Sida 2004 Paris Available: http://www.ird.sn/activites/sida/Thesenegalese.pdf . Accessed 27 January 2005
WHO, UNAIDS Guidance on ethics and equitable access to HIV treatment and care 2004 Geneva WHO Available: http://www.who.int/ethics/en/ethics_equity_HIV_e.pdf . Accessed 27 January 2005
Chen L Evans T Anand S Boufford JI Brown H Human resources for health: Overcoming the crisis Lancet 2004 364 1984 1990 15567015
| 15737013 | PMC549605 | CC BY | 2021-01-05 10:40:26 | no | PLoS Med. 2005 Mar 22; 2(3):e69 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020069 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1573701410.1371/journal.pmed.0020073Research ArticleCancer BiologyPharmacology/Drug DiscoveryOncologyPathologySurgeryCancer: LungChemotherapyDrugs and Adverse Drug ReactionsOncologyPathologyAcquired Resistance of Lung Adenocarcinomas to Gefitinib or Erlotinib Is Associated with a Second Mutation in the EGFR Kinase Domain Resistance to EGFR Kinase InhibitorsPao William
1
2
*Miller Vincent A
2
Politi Katerina A
1
Riely Gregory J
2
Somwar Romel
1
Zakowski Maureen F
3
Kris Mark G
2
Varmus Harold
1
1Program in Cancer Biology and Genetics, Memorial Sloan-Kettering Cancer CenterNew York, New YorkUnited States of America2Thoracic Oncology Service, Department of MedicineMemorial Sloan-Kettering Cancer Center, New York, New YorkUnited States of America3Department of Pathology, Memorial Sloan-Kettering Cancer CenterNew York, New YorkUnited States of AmericaLiu Ed T. Academic EditorGenome Institute of SingaporeSingapore
Competing Interests: See Acknowledgments.
Author Contributions: See Acknowledgments.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 22 2 2005 2 3 e7320 1 2005 31 1 2005 Copyright: © 2005 Pao 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.
How Tumor Cells Acquire Resistance to Kinase Inhibitors
EGFR Inhibition in Non-Small Cell Lung Cancer: Resistance, Once Again, Rears Its Ugly Head
Background
Lung adenocarcinomas from patients who respond to the tyrosine kinase inhibitors gefitinib (Iressa) or erlotinib (Tarceva) usually harbor somatic gain-of-function mutations in exons encoding the kinase domain of the epidermal growth factor receptor (EGFR). Despite initial responses, patients eventually progress by unknown mechanisms of “acquired” resistance.
Methods and Findings
We show that in two of five patients with acquired resistance to gefitinib or erlotinib, progressing tumors contain, in addition to a primary drug-sensitive mutation in EGFR, a secondary mutation in exon 20, which leads to substitution of methionine for threonine at position 790 (T790M) in the kinase domain. Tumor cells from a sixth patient with a drug-sensitive EGFR mutation whose tumor progressed on adjuvant gefitinib after complete resection also contained the T790M mutation. This mutation was not detected in untreated tumor samples. Moreover, no tumors with acquired resistance had KRAS mutations, which have been associated with primary resistance to these drugs. Biochemical analyses of transfected cells and growth inhibition studies with lung cancer cell lines demonstrate that the T790M mutation confers resistance to EGFR mutants usually sensitive to either gefitinib or erlotinib. Interestingly, a mutation analogous to T790M has been observed in other kinases with acquired resistance to another kinase inhibitor, imatinib (Gleevec).
Conclusion
In patients with tumors bearing gefitinib- or erlotinib-sensitive EGFR mutations, resistant subclones containing an additional EGFR mutation emerge in the presence of drug. This observation should help guide the search for more effective therapy against a specific subset of lung cancers.
A specific secondary mutation in the kinase domain of the epidermal growth factor receptor can render cells insensitive to the two kinase inhibitors. This mutation was found in resistant tumors from three of six patients studied
==== Body
Introduction
Somatic gain-of-function mutations in exons encoding the epidermal growth factor receptor (EGFR) tyrosine kinase domain are found in about 10% of non-small cell lung cancers (NSCLCs) from the United States [1,2,3], with higher percentages observed in east Asia [2,4,5,6]. Some 90% of NSCLC-associated mutations occur as either multi-nucleotide in-frame deletions in exon 19, involving elimination of four amino acids, Leu-Arg-Glu-Ala, or as a single nucleotide substitution at nucleotide 2573 (T→G) in exon 21, resulting in substitution of arginine for leucine at position 858 (L858R). Both of these mutations are associated with sensitivity to the small-molecule kinase inhibitors gefitinib or erlotinib [1,2,3]. Unfortunately, nearly all patients who experience marked improvement on these drugs eventually develop progression of disease. While KRAS mutations have been associated with some cases of primary resistance to gefitinib or erlotinib [7], mechanisms underlying “acquired” or “secondary” resistance are unknown.
Acquired resistance to kinase-targeted anticancer therapy has been most extensively studied with imatinib, an inhibitor of the aberrant BCR-ABL kinase, in chronic myelogenous leukemia (CML). Mutations in the ABL kinase domain are found in 50%–90% of patients with secondary resistance to the drug (reviewed in [8]). Such mutations, which cluster in four distinct regions of the ABL kinase domain (the ATP binding loop, T315, M351, and the activation loop), interfere with binding of imatinib to ABL [9,10,11]. Crystallographic studies of various ABL mutants predict that most should remain sensitive to inhibitors that bind ABL with less stringent structural requirements. Using this insight, new small-molecule inhibitors have been identified that retain activity against the majority of imatinib-resistant BCR-ABL mutants [12,13].
Although imatinib inhibits different kinases in various diseases (BCR-ABL in CML, KIT or PDGFR-alpha in gastrointestinal stromal tumors [GISTs], and PDGFR-alpha in hypereosinophilic syndrome [HES]) (reviewed in [14]), some tumors that become refractory to treatment with imatinib appear to have analogous secondary mutations in the kinase-coding domain of the genes encoding these three enzymes. For example, in CML, a commonly found mutation is a C→T single nucleotide change that replaces threonine with isoleucine at position 315 (T315I) in the ABL kinase domain [9,10,11]. In GIST and HES, respectively, the analogous T670I mutation in KIT and T674I mutation in PDGFR-alpha have been associated with acquired resistance to this drug [15,16].
To determine whether lung cancers that acquire clinical resistance to either gefitinib or erlotinib display additional mutations in the EGFR kinase domain, we have examined the status of EGFR exons 18 to 24 in tumors from five patients who initially responded but subsequently progressed while on these drugs. These exons were also assessed in tumor cells from a sixth patient whose disease rapidly recurred while on gefitinib therapy after complete gross tumor resection. Because of the association of KRAS mutations with primary resistance to gefitinib and erlotinib [7], we also examined the status of KRAS in tumor cells from these six patients. In an effort to explain the selective advantage of cells with a newly identified “resistance” mutation in EGFR—a T790M amino acid substitution—we further characterized the drug sensitivity of putatively resistant EGFR mutants versus wild-type or drug-sensitive EGFR mutants, using both a NSCLC cell line fortuitously found to contain the T790M mutation and lysates from cells transiently transfected with wild-type and mutant EGFR cDNAs.
Methods
Tissue Procurement
Tumor specimens, including paraffin blocks, fine needle biopsies, and pleural effusions, were obtained through protocols approved by the Institutional Review Board of Memorial Sloan-Kettering Cancer Center (protocol 92–055 [7] and protocol 04–103 [Protocol S1]). All patients provided informed consent.
Mutational Analyses of EGFR and KRAS in Lung Tumors
Genomic DNA was extracted from tumor specimens, and primers for EGFR (exons 18–24) and KRAS2 (exon 2) analyses were as published [3,7]. All sequencing reactions were performed in both forward and reverse directions, and all mutations were confirmed at least twice from independent PCR isolates.
A specific exon 20 mutation (T790M) was also detected by length analysis of fluorescently labeled (FAM) PCR products on a capillary electrophoresis device (ABI 3100 Avant, Applied Biosystems, Foster City, California, United States), based on a new NlaIII restriction site created by the T790M mutation (2369 C→T), using the following primers: EGFR Ex20F, 5′-FAM-
CTCCCTCCAGGAAGCCTACGTGAT-3′ and EGFR Ex20R 5′-
TTTGCGATCTGCACACACCA-3′. Using serially mixed dilutions of DNA from NSCLC cell lines (H1975, L858R- and T790M-positive; H-2030, EGFR wild-type) for calibration, this assay detects the presence of the T790M mutation when H1975 DNA comprises 3% or more of the total DNA tested, compared to a sensitivity of 6% for direct sequencing (data not shown).
RT-PCR
The following primers were used to generate EGFR cDNA fragments spanning exon 20: EGFR 2095F 5′-
CCCAACCAAGCTCTCTTGAG-3′ and EGFR 2943R 5′-
ATGACAAGGTAGCGCTGGGGG-3′. PCR products were ligated into plasmids using the TOPO TA-cloning kit (Invitrogen, Carlsbad, California, United States), as per manufacturer's instructions. Minipreps of DNA from individual clones were sequenced using the T7 priming site of the cloning vector.
Functional Analyses of Mutant EGFRs
Two numbering systems are used for EGFR. The first denotes the initiating methionine in the signal sequence as amino acid −24. The second, used here, denotes the methionine as amino acid +1. Commercial suppliers of antibodies, such as the Y1068-specific anti-phospho-EGFR, use the first nomenclature. To be consistent, we consider Y1068 as Y1092. Likewise, the T790M mutation reported here has also been called T766M. Mutations were introduced into full-length wild-type and mutant EGFR cDNAs using a QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, California, United States) and cloned into expression vectors as described [3]. The following primers were used to generate the deletion (del) L747–E749;A750P mutant: forward 5′-
TAAAATTCCCGTCGCTATCAAGGAGCCAACATCTCCGAAAGCCAACAAGG-3′ and reverse 5′-
CCTTGTTGGCTTTCGGAGATGTTGGCTCCTTGATAGCGACGGGAATTTTA-3′. The following primers were used to introduce the T790M mutation: forward 5′-
AGCTCATCATGCAGCTCAT-3′ and reverse 5′-
ATGAGCTGCATGATGAGCT-3′. The L858R mutant cDNA was generated previously [3]. All mutant clones were fully re-sequenced bidirectionally to ensure that no additional mutations were introduced. Various EGFRs were transiently expressed in 293T human embryonic kidney cells as published [3]. Cells were treated with different concentrations of gefitinib or erlotinib.
Immunoblotting
See Methods and supplementary methods in [3] for details on cell lysis, immunoblotting, and antibody reagents. At least three independent experiments were performed for all analyses.
Cell Culture
The NSCLC cell lines H1650, H1975, H2030, H2347, H2444, H358, and H1734 were purchased from American Type Culture Collection (Manassas, Virginia, United States). H3255 was a gift of B. Johnson and P. Janne. Cells were grown in complete growth medium (RPMI-1640; American Type Culture Collection catalog no. 30–2001) supplemented with 10% fetal calf serum, 10 units/ml penicillin, and 10 μg/ml streptomycin) at 37 °C and 5% CO2. For viability studies, cells were seeded in complete growth medium in black 96-well clear bottom ViewPlates (PerkinElmer, Wellesley, Massachusetts, United States) at a density of 5,000 (H1975 and H2030) or 7,500 cells per well (H3255). Following overnight incubation, cells were grown for 24 h in the supplemented RPMI-1640 medium with 0.1% serum. Cells (in supplemented RPMI-1640 medium containing 0.1% serum) were then incubated for 48 h in the continued presence of gefitinib or erlotinib.
Viability Assay
Cell viability was assayed using Calcein AM (acetoxymethyl ester of Calcein, Molecular Probes, Eugene, Oregon, United States). Following incubation with gefitinib or erlotinib, monolayers were washed twice with PBS (containing calcium and magnesium) and incubated with 7.5 μmol Calcein AM in supplemented RPMI-1640 (no serum) for 30 min. Labeling medium was removed, and cells were washed three times with PBS. Calcein fluorescence (Ex, 485 nm; Em, 535 nM) was detected immediately using a Victor V multi-label plate reader (PerkinElmer). Three independent experiments were performed for each cell line; each experiment included four to eight replicates per condition.
Results
Case Reports
We identified secondary EGFR mutations in three of six individuals whose disease progressed on either gefitinib or erlotinib (Table 1). Brief case histories of these three patients are presented below.
Table 1 Specimens Analyzed in This Study for Mutations in the EGFR Tyrosine Kinase Domain (Exons 18 to 24) and KRAS (Exon 2)
The transbronchial biopsy in patient 1 had scant tumor cells; sequencing analysis revealed only wild-type sequence (see text)
aPercent tumor cells is defined by assessment of corresponding histopathological slides
n/a, not applicable
Patient 1
This 63-y-old female “never smoker” (smoked less than 100 cigarettes in her lifetime) initially presented with bilateral diffuse chest opacities and a right-sided pleural effusion. Transbronchial biopsy revealed adenocarcinoma. Disease progressed on two cycles of systemic chemotherapy, after which gefitinib, 250 mg daily, was started. Comparison of chest radiographs obtained prior to starting gefitinib (Figure S1A, left panel) and 2 wk later (Figure S1A, middle panel) showed dramatic improvement. Nine months later, a chest radiograph revealed progression of disease (Figure S1A, right panel). Subsequently, the patient underwent a computed tomography (CT)–guided biopsy of an area in the right lung base (Figure 1A, left panel). Despite continued treatment with gefitinib, either with chemotherapy or at 500 mg daily, the pleural effusion recurred, 12 mo after initiating gefitinib (Figure 1A, right panel). Pleural fluid was obtained for molecular studies. In total, this patient had three tumor specimens available for analysis: the original lung tumor biopsy, a biopsy of the progressing lung lesion, and pleural fluid. However, re-review of the original transbronchial biopsy showed that it had scant tumor cells (Table 1).
Figure 1 Re-Biopsy Studies
(A.) Patient 1. CT-guided biopsy of progressing lung lesions after 10 mo on gefitinib (left panel). Two months later, fluid from a right-sided pleural effusion (right panel) was collected for molecular analysis.
(B) Patient 2. CT-guided biopsy of a progressing thoracic spine lesion (left panel) and fluoroscopic-guided biopsy of a progressing lung lesion (right panel). The biopsy needles are indicated by white arrows.
Patient 2.
This 55-y-old woman with a nine pack-year history of smoking underwent two surgical resections within 2 y (right lower and left upper lobectomies) for bronchioloalveolar carcinoma with focal invasion. Two years later, her disease recurred with bilateral pulmonary nodules and further progressed on systemic chemotherapy. Thereafter, the patient began erlotinib, 150 mg daily. A baseline CT scan of the chest demonstrated innumerable bilateral nodules (Figure S1B, left panel), which were markedly reduced in number and size 4 mo after treatment (Figure S1B, middle panel). After 14 mo of therapy, the patient's dose of erlotinib was decreased to 100 mg daily owing to fatigue. At 23 mo of treatment with erlotinib, a CT scan demonstrated an enlarging sclerotic lesion in the thoracic spine. The patient underwent CT-guided biopsy of this lesion (Figure 1B, left panel), and the erlotinib dose was increased to 150 mg daily. After 25 mo of treatment, she progressed within the lung (Figure S1B, right panel). Erlotinib was discontinued, and a fluoroscopically guided core needle biopsy was performed at a site of progressive disease in the lung (Figure 1B, right panel). In total, this patient had three tumor specimens available for analysis: the original resected lung tumor, the biopsy of the enlarging spinal lesion, and the biopsy of the progressing lung lesion (Table 1).
Patient 3
This 55-y-old female “never smoker” was treated for nearly 4.5 y with weekly paclitaxel and trastuzumab [17] for adenocarcinoma with bronchioloalveolar carcinoma features involving her left lower lobe, pleura, and mediastinal lymph nodes. Treatment was discontinued owing to fatigue. Subsequently, the patient underwent surgical resection. Because of metastatic involvement of multiple mediastinal lymph nodes and clinical features known at that time to be predictive of response to gefitinib (female, never smoker, bronchioloalveolar variant histology), she was placed on “adjuvant” gefitinib 1 mo later (Figure S1C, left panel). This drug was discontinued after 3 mo when she developed a new left-sided malignant pleural effusion (Figure S1C, middle panel). Despite drainage and systemic chemotherapy, the pleural effusion recurred 4 mo later (Figure S1C, right panel), at which time pleural fluid was collected for analysis. In total, this patient had two clinical specimens available for analysis: tumor from the surgical resection and pleural fluid (Table 1).
Patients' Tumors Contain EGFR Tyrosine Kinase Domain Mutations Associated with Sensitivity to EGFR Tyrosine Kinase Inhibitors
We screened all available tumor samples from these three patients for previously described drug-sensitive EGFR mutations, by direct DNA sequencing of exons 19 and 21 [3]. Tumor samples from patient 1 showed a T→G change at nucleotide 2573, resulting in the exon 21 L858R amino acid substitution commonly observed in drug-responsive tumors. This mutation was present in the biopsy material from the progressing lung lesion (Figure S2A, upper panels) and from cells from the pleural effusion (Figure S2A, lower panels), both of which on cytopathologic examination consisted of a majority of tumor cells (Table 1). Interestingly, comparisons of the tracings suggest that an increase in copy number of the mutant allele may have occurred. Specifically, while the ratio of wild-type (nucleotide T) to mutant (nucleotide G) peaks at position 2573 was approximately 1:1 or 1:2 in the lung biopsy specimen (Figure S2A, upper panels), sequencing of DNA from the pleural fluid cells demonstrated a dominant mutant G peak (Figure S2A, lower panels). Consistent with this, a single nucleotide polymorphism (SNP) noted at nucleotide 2361 (A or G) demonstrated a corresponding change in the ratios of A:G, with a 1:1 ratio in the transbronchial biopsy, and a nearly 5:1 ratio in the pleural fluid (Figure 2A). Notably, we did not detect the 2573 T→G mutation in the original transbronchial biopsy specimen (Table 1; data not shown). As stated above, this latter specimen contained scant tumor cells, most likely fewer than needed for detection of an EGFR mutation by direct sequencing (see [7]).
Figure 2 Sequencing Chromatograms with the T790M EGFR Exon 20 Mutation in Various Clinical Specimens and the NSCLC Cell Line H1975
(A–C) In all three patients—patient 1 (A), patient 2 (B), and patient 3 (C)—the secondary T790M mutation was observed only in lesions obtained after progression on either gefitinib or erlotinib.
(D) Cell line H1975 contains both an exon 21 L858R mutation (upper panel) and the exon 20 T790M mutation (lower panel). The asterisks indicate a common SNP (A or G) at nucleotide 2361; the arrows indicate the mutation at nucleotide 2369 (C→T), which leads to substitution of methonine (ATG) for threonine (ACG) at position 790. In the forward direction, the mutant T peak is blue. In the reverse direction, the mutant peak is green, while the underlying blue peak represents an “echo” from the adjacent nucleotide.
All three specimens from patient 2, including the original lung tumor and the two metastatic samples from bone and lung, showed an exon 19 deletion involving elimination of 11 nucleotides (2238–2248) and insertion of two nucleotides, G and C (Figure S2B, all panels; Table 1). These nucleotide changes delete amino acids L747–E749 and change amino acid 750 from alanine to proline (A750P). A del L747–E749;A750P mutation was previously reported with different nucleotide changes [2]. In all samples from patient 2, the wild-type sequence predominated at a ratio of about 3:1 over the mutant sequence.
Both of the available tumor samples from patient 3 contained a deletion of 15 nucleotides (2236–2250) in exon 19 (Table 1; data not shown), resulting in elimination of five amino acids (del E746–A750). This specific deletion has been previously reported [3]. The ratio of mutant to wild-type peaks was approximately 1:1 in both specimens (data not shown).
Collectively, these results demonstrate that tumors from all three patients contain EGFR mutations associated with sensitivity to the tyrosine kinase inhibitors gefitinib and erlotinib. In addition, these data show that within individual patients, metastatic or recurrent lesions to the spine, lung, and pleural fluid contain the same mutations. These latter observations support the idea that relapsing and metastatic tumor cells within individuals are derived from original progenitor clones.
A Secondary Missense Mutation in the EGFR Kinase Domain Detected in Lesions That Progressed while on Treatment with Either Gefitinib or Erlotinib
To determine whether additional mutations in the EGFR kinase domain were associated with progression of disease in these patients, we performed direct sequencing of all of the exons (18 through 24) encoding the EGFR catalytic region in the available tumor specimens.
Analysis of patient 1's pre-gefitinib specimen, which contained scant tumor cells (Table 1; see above), not surprisingly showed only wild-type EGFR sequence (Table 1; data not shown). However, careful analysis of the exon 20 sequence chromatograms in both forward and reverse directions from this patient's lung biopsy specimen obtained after disease progression on gefitinib demonstrated an additional small peak at nucleotide 2369, suggesting a C→T mutation (Figure 2A, upper panels; Table 1). This nucleotide change leads to substitution of methionine for threonine at position 790 (T790M). The 2369 C→T mutant peak was even more prominent in cells from the patient's pleural fluid, which were obtained after further disease progression on gefitinib (Figure 2A, lower panels; Table 1). The increase in the ratio of mutant to wild-type peaks obtained from analyses of the lung specimen and pleural fluid paralleled the increase in the ratio of the mutant G peak (leading to the L858R mutation) to the wild-type T peak at nucleotide 2573 (see above; Figure S2A), as well as the increase in the ratio of the A:G SNP at position 2361 (Figure 2A). Collectively, these findings imply that the exon 20 T790M mutation was present on the same allele as the exon 21 L858R mutation, and that a subclone of cells harboring these mutations emerged during drug treatment.
In patient 2, the tumor-rich sample obtained prior to treatment with erlotinib did not contain any additional mutations in the exons encoding the EGFR tyrosine kinase domain (Figure 2B, upper panels; Table 1). By contrast, her progressing bone and lung lesions contained an additional small peak at nucleotide 2369, suggesting the existence of a subclone of tumor cells with the same C→T mutation observed in patient 1 (Figure 2B, middle and lower panels; Table 1). The relative sizes of the 2369 T mutant peaks seen in these latter two samples appeared to correlate with the relative size of the corresponding peaks of the exon 19 deletion (Figure S2B). Interestingly, the SNP at nucleotide 2361 (A or G) was detected in specimens from patient 2 before but not after treatment with erlotinib, suggesting that one EGFR allele underwent amplification or deletion during the course of treatment (Figure S2B).
Patient 3 showed results analogous to those of patient 2. A tumor-rich pre-treatment specimen did not demonstrate EGFR mutations other than the del E746–A750 exon 19 deletion; specifically, in exon 20, no secondary changes were detected (Figure 2C, upper panels; Table 1). However, analysis of DNA from cells in the pleural effusion that developed after treatment with gefitinib showed the C→T mutation at nucleotide 2369 in exon 20 (Figure 2C, lower panels; Table 1), corresponding to the T790M mutation described above. There was no dramatic change between the two samples in the ratio of the A:G SNP at position 2361. The mutant 2369 T peak was small, possibly because gefitinib had been discontinued in this patient for 4 mo at the time pleural fluid tumor cells were collected; thus, there was no selective advantage conferred upon cells bearing the T790M mutation.
To determine whether the 2369 C→T mutation was a previously overlooked EGFR mutation found in NSCLCs, we re-reviewed exon 20 sequence tracings derived from analysis of 96 fresh-frozen resected tumors [3] and 59 paraffin-embedded tumors [7], all of which were removed from patients prior to treatment with an EGFR tyrosine kinase inhibitor. We did not detect any evidence of the T790M mutation in these 155 tumors (data not shown; see Discussion). Collectively, our results suggest that the T790M mutation is associated with lesions that progress while on gefitinib or erlotinib. Moreover, at least in patients 1 and 2, the subclones of tumor cells bearing this mutation probably emerged between the time of initial treatment with a tyrosine kinase inhibitor and the appearance of drug resistance.
In three additional patients (case histories not described here) with lung adenocarcinomas who improved but subsequently progressed on therapy with either gefitinib or erlotinib, we examined DNA from tumor specimens obtained during disease progression. In all three patients, we found EGFR mutations associated with drug sensitivity (all exon 19 deletions). However, we did not find any additional mutations in exons 18 to 24 of EGFR, including the C→T change at position 2369 (data not shown). These results imply that alternative mechanisms of acquired drug resistance exist.
Patients' Progressive Tumors Lack KRAS Mutations
Mutations in exon 2 of KRAS2 occur in about one-fourth of NSCLCs. Such mutations rarely, if ever, accompany EGFR mutations and are associated with primary resistance to gefitinib or erlotinib [7]. To evaluate the possibility that secondary KRAS mutations confer acquired resistance to these drugs, we performed mutational profiling of KRAS2 exon 2 from tumor specimens from patients 1 to 3, as well as the three additional patients lacking evidence of the T790M mutation. None of the specimens contained any changes in KRAS (Table 1; data not shown), indicating that KRAS mutations were not responsible for drug resistance and tumor progression in these six patients.
An Established NSCLC Cell Line Also Contains Both T790M and L858R Mutations
We profiled the EGFR tyrosine kinase domain (exons 18 to 24) and KRAS exon 2 in eight established NSCLC lines (Table 2). Surprisingly, one cell line—H1975—contained the same C→T mutation at position 2369 (T790M) as described above (Figure 2D, lower panel). This cell line had previously been shown by others to contain a 2573 T→G mutation in exon 21 (L858R) [18], which we confirmed (Figure 2D, upper panel); in addition, H1975 was reported to be more sensitive to gefitinib inhibition than other lung cancer cell lines bearing wild-type EGFR [18]. Only exons 19 and 21 were apparently examined in this published study.
Table 2 Status of NSCLC Cell Lines Analyzed for EGFR Tyrosine Kinase Domain (Exons 18 to 24) and KRAS (Exon 2) Mutations
See Methods for further details
In our own analysis of H1975 (exons 18 to 24), the mutant 2369 T peak resulting in the T790M amino acid substitution was dominant, suggesting an increase in copy number of the mutant allele in comparison to the wild-type allele. The ratio of mutant to wild-type peaks was similar to that of the mutant 2573 G (corresponding to the L858R amino acid substitution) to wild-type T peaks (Figure 2D, all panels), implying that the T790M and L858R mutations were in the same amplified allele. To further investigate this possibility, we performed RT-PCR to generate cDNAs that spanned exon 20 of EGFR and included sequences from exon 19 and 21. PCR products were then cloned, and individual colonies were analyzed for EGFR mutations. Sequencing chromatograms of DNA from four of four clones showed both the 2369 C→T and 2573 T→G mutations, confirming that both mutations were in the same allele (data not shown).
Other NSCLC cell lines carried either EGFR or KRAS mutations, but none had both (Table 2). As reported, H3255 contained an L858R mutation [19] and H1650 contained an exon 19 deletion [18]. No other cell lines analyzed contained additional mutations in the exons encoding the EGFR tyrosine kinase domain.
A Novel PCR Restriction Fragment Length Polymorphism Assay Independently Confirms the Absence or Presence of the T790M Mutation
As stated above, the mutant peaks suggestive of a T790M mutation in exon 20 were small in some sequence chromatograms. To eliminate the possibility that these peaks were due to background “noise,” we sought to confirm the presence of the 2369 C→T mutation in specific samples, by developing an independent test, based on a fluorescence detection assay that takes advantage of a PCR restriction fragment length polymorphism (PCR-RFLP) generated by the specific missense mutation. After PCR amplification with exon-20-specific primers spanning nucleotide 2369, wild-type sequence contains specific NlaIII sites, which upon digestion yield a 106-bp product (see Methods;
Figure 3A). Presence of the mutant 2369 T nucleotide creates a new NlaIII restriction digest site, yielding a slightly shorter product (97 bp), readily detected by fluorescent capillary electrophoresis. This test is about 2 -fold more sensitive than direct sequencing (see Methods; data not shown).
Figure 3 A Novel PCR-RFLP Assay Independently Confirms Presence of the T790M Mutation in Exon 20 of the EGFR Kinase Domain
(A) Design of the assay (see text for details). “F” designates the fluorescent label, FAM. At the bottom of this panel, the assay demonstrates with the 97-bp NlaIII cleavage product the presence of the T790M mutation in the H1975 cell line; this product is absent in H2030 DNA. The 106-bp NlaIII cleavage product is generated by digestion of wild-type EGFR.
(B) The PCR-RFLP assay demonstrates that pre-drug tumor samples from the three patients lack detectable levels of the mutant 97-bp product, while specimens obtained after disease progression contain the T790M mutation. Pt, patient.
We first used DNA from the H1975 cell line (which contains both T790M and L858R mutations) to confirm the specificity of the PCR-RFLP assay. As expected, analysis of these cells produced both the 97- and 106-bp fragments. By contrast, analysis of DNA from H2030 (which contains wild-type EGFR;
Table 2) showed only the 106-bp fragment (Figure 3A). These data show that this test can readily indicate the absence or presence of the mutant allele in DNA samples. However, this test was only semi-quantitative, as the ratio of the mutant 97-bp product versus the wild-type 106-bp product varied in independent experiments from approximately 1:1 to 2:1.
We next used this PCR-RFLP assay to assess various patient samples for the presence of the specific 2369 C→T mutation corresponding to the T790M amino acid substitution. DNA from the progressing bone and lung lesions in patient 1 produced both the 97- and 106-bp fragments, but DNA from the original lung tumor did not (Figure 3B). The ratio of mutant to wild-type products was higher in the cells from the pleural fluid, consistent with the higher peaks seen on the chromatograms from direct sequencing of exon 20 (see Figure 2A). Likewise, DNA from progressive lesions from patients 2 and 3 yielded both 97- and 106-bp fragments in the PCR-RFLP assay (Figure 3B), whereas the pre-treatment specimens did not produce the 97-bp product. Collectively, these data from an independent assay confirm that the T790M mutation was present in progressing lesions from all three patients. We were also unable to detect the T790M mutation in any specimens from the three additional patients with acquired resistance that failed to demonstrate secondary mutations in EGFR exons 18 to 24 by direct sequencing (data not shown).
Biochemical Properties of EGFR Mutants
To determine how the T790M mutation would affect EGFR proteins already containing mutations associated with sensitivity to EGFR tyrosine kinase inhibitors, we introduced the specific mutation into EGFR cDNAs that encoded the exon 21 and 19 mutations found in patients 1 and 2, respectively. Corresponding proteins ([i] L858R and L858R plus T790M, [ii] del L747–E749;A750P and del L747–E749;A750P plus T790M, and [iii] wild-type EGFR and wild-type EGFR plus T790M) were then produced by transient transfection with expression vectors in 293T cells, which have very low levels of endogenous EGFR [3]. Various lysates from cells that were serum-starved and pre-treated with gefitinib or erlotinib were analyzed by immunoblotting. Amounts of total EGFR (t-EGFR) were determined using an anti-EGFR monoclonal antibody, and actin served as an indicator of relative levels of protein per sample. To assess the drug sensitivity of the various EGFR kinases in surrogate assays, we used a Y1092-phosphate-specific antibody (i.e., phospho-EGFR [p-EGFR]) to measure the levels of “autophosphorylated” Tyr-1092 on EGFR in relation to levels of t-EGFR protein. We also assessed the global pattern and levels of induced tyrosine phosphorylation of cell proteins by using a generalized anti-phosphotyrosine reagent (RC-20).
Gefitinib inhibited the activity of wild-type and L858R EGFRs progressively with increasing concentrations of drug, as demonstrated by a reduction of tyrosine-phosphorylated proteins (Figure 4A) and a decrease in p-EGFR:t-EGFR ratios (Figure 4B). By contrast, wild-type and mutant EGFRs containing the T790M mutation did not display a significant change in either phosphotyrosine induction or p-EGFR:t-EGFR ratios (Figure 4A and 4B). Similar results were obtained using erlotinib against wild-type and del E747–L747;A750P EGFRs in comparison to the corresponding mutants containing the T790M mutation (Figure 4C). These results suggest that the T790M mutation may impair the ability of gefitinib or erlotinib to inhibit EGFR tyrosine kinase activity, even in EGFR mutants (i.e., L858R or an exon 19 deletion) that are clinically associated with drug sensitivity.
Figure 4 EGFR Mutants Containing the T790M Mutation Are Resistant to Inhibition by Gefitinib or Erlotinib
293T cells were transiently transfected with plasmids encoding wild-type (WT) EGFR or EGFR mutants with the following changes: T790M, L858R, L858R + T790M, del L747–E749;A750P, or del L747–E749;A750P + T790M. After 36 h, cells were serum-starved for 24 h, treated with gefitinib or erlotinib for 1 h, and then harvested for immunoblot analysis using anti-p-EGFR (Y1092), anti-t-EGFR, anti-phosphotyrosine (p-Tyr), and anti-actin antibodies as described in Methods. The EGFR T790M mutation, in conjunction with either wild-type EGFR or the drug-sensitive L858R EGFR mutant, prevents inhibition of tyrosine phosphorylation (A) or p-EGFR (B) by gefitinib. Analogously, the T790M mutation, in conjunction with the drug-responsive del L747–E749;A750P EGFR mutant, prevents inhibition of p-EGFR by erlotinib (C).
Resistance of a NSCLC Cell Line Harboring Both T790M and L858R Mutations to Gefitinib or Erlotinib
To further explore the functional consequences of the T790M mutation, we determined the sensitivity of various NSCLC cells lines grown in the presence of either gefitinib or erlotinib, using an assay based upon Calcein AM. Uptake and retention of this fluorogenic esterase substrate by vehicle- versus drug-treated live cells allows for a comparison of relative cell viability among cell lines [20]. The H3255 cell line, which harbors the L858R mutation and no other EGFR TK domain mutations (Table 2), was sensitive to treatment with gefitinib, with an IC50 of about 0.01 μmol (Figure 5). By contrast, the H1975 cell line, which contains both L858R and T790M mutations (Table 2), was approximately 100-fold less sensitive to drug, with an IC50 of about 1 μmol (Figure 5). In fact, the sensitivity of H1975 cells was more similar to that of H2030, which contains wild-type EGFR (exons 18 to 24) and mutant KRAS (Figure 5). Very similar results were obtained with erlotinib (Figure S3).
Figure 5 Sensitivity to Gefitinib Differs Among NSCLC Cell Lines Containing Various Mutations in EGFR or KRAS
The three indicated NSCLC cell lines, H3255 (L858R mutation), H1975 (both T790M and L858R mutations), and H2030 (wild-type EGFR, mutant KRAS) (see Table 2), were grown in increasing concentrations of gefitinib, and the density of live cells after 48 h of treatment was measured using a Calcein AM fluorescence assay. Fluorescence in vehicle-treated cells is expressed as 100%. Results are the mean ± standard error of three independent experiments in which there were four to eight replicates of each condition. Similar results were obtained with erlotinib (see Figure S3).
Discussion
Specific mutations in the tyrosine kinase domain of EGFR are associated with sensitivity to either gefitinib or erlotinib, but mechanisms of acquired resistance have not yet been reported. Based upon analogous studies in other diseases with another kinase inhibitor, imatinib, a single amino acid substitution from threonine to methionine at position 790 in the wild-type EGFR kinase domain was predicted to lead to drug resistance, even before the association of exon 19 and 21 mutations of EGFR with drug responsiveness in NSCLC was reported. The T790M mutation was shown in vitro in the context of wild-type EGFR to confer resistance to gefitinib [21] and a related quinazoline inhibitor, PD153035 [22].
We show here, through molecular analysis of tumor material from three patients and one NSCLC cell line, as well as additional biochemical studies, that acquired clinical drug resistance to gefitinib or erlotinib is indeed associated with the T790M mutation. Importantly, we find that the T790M mutation confers drug resistance not just to wild-type EGFR but also to mutant EGFRs associated with clinical responsiveness to EGFR tyrosine kinase inhibitors [1,2,3]. Our results further demonstrate that an analogous mechanism of acquired resistance exists for imatinib and EGFR tyrosine kinase inhibitors (Table 3), despite the fact that the various agents target different kinases in distinct diseases.
Table 3 Analogous Mutations in Four Kinases Associated with Resistance to Kinase Inhibitors
In tumors from patients not treated with either gefitinib or erlotinib, the 2369 C→T mutation (T790M) appears to be extremely rare. We have not identified this mutation in 155 tumors (see above), and among nearly 1,300 lung cancers in which analysis of EGFR exons 18 to 21 has been performed [1,2,3,4,5,6], only one tumor (which also harbored an L858R mutation) was reported to contain the T790M mutation. Whether the patient from which this tumor was resected had received gefitinib or erlotinib is unclear, and the report did not note an association with acquired resistance to either drug [5].
How tumor cells bearing the T790M mutation emerge within gefitinib- or erlotinib-treated patients is a matter of investigation. Subclones bearing this mutation could arise de novo during treatment. However, based upon analogous studies in CML, it is also possible that NSCLC subclones bearing this secondary mutation pre-exist within the primary tumor clone in individual patients, albeit at low frequency [23]. In either scenario, treatment with gefitinib or erlotinib subsequently allows these resistant subclones to become apparent, because most cells bearing sensitivity-conferring mutations die, while cells with the T790M mutation persist.
From analysis of the crystal structure of the EGFR kinase domain bound to erlotinib, it is has been shown that the wild-type threonine residue at position 790 is located in the hydrophobic ATP-binding pocket of the catalytic region, where it forms a critical hydrogen bond with the drug [24]. The related compound, gefitinib, is predicted to interact with this threonine residue as well. Substitution of the threonine at position 790 by a larger residue like methionine would probably result in steric clash with the aromatic moieties on these two drugs [25]. By contrast, ATP would likely not depend on the accessibility of the same hydrophobic cavity and is therefore probably not affected by the incorporation of a bulky methionine side chain [25]. Consistent with this, the T790M mutation has been shown not to abrogate the catalytic activity of wild-type EGFR [22].
The T790M mutation could also affect the kinase activity or alter the substrate specificity of mutant EGFRs, such that a proliferative advantage would be conferred upon cells bearing the mutation. Consistent with this, the H1975 NSCLC cell line reported here to contain both T790M and L858R did not to our knowledge undergo any prior treatment with gefitinib or erlotinib; the doubly mutated cells must have become dominant over time through multiple passages in vitro. This scenario could explain the seemingly contradictory report by others who found the H1975 cell line to be highly sensitive to gefitinib [18]; our H1975 cells could represent a subclone that emerged over time. Analysis of earlier passages of H1975 cells for the T790M mutation would be informative in this regard.
Recently, new small-molecule inhibitors have been identified that retain activity against the majority of imatinib-resistant BCR-ABL mutants. The new drugs bind to ABL in an “open” conformation, as opposed to imatinib, which binds ABL in a “closed” conformation [12,13]. Analogously, it may be possible to find EGFR tyrosine kinase inhibitors that bind to the EGFR kinase domain in different ways than gefitinib and erlotinib. For example, the crystal structure of another EGFR inhibitor, lapatinib (GW572016), was recently solved bound to EGFR [26]. This study revealed that the quinazoline rings of erlotinib and lapatinib interact differently with the EGFR kinase domain, suggesting that while the T790M mutation may affect inhibition by erlotinib and gefitinib, it may not affect inhibition of EGFR by compounds similar to lapatinib. To our knowledge, no NSCLC patient who initially responded to but then progressed on either gefitinib or erlotinib has yet been treated with lapatinib.
In some of the patient specimens analyzed, the actual sequencing peaks demonstrating the T790M mutation were smaller than originally anticipated. These results differ from those of acquired resistance mutation in CML [10], GIST [15,27], and HES [16]. However, in contrast to all of these diseases, in which tumor cells are readily accessible, lung-cancer-related tumors are more difficult to access, as illustrated by the limited manner in which we were able to obtain tumor cells from various sites of disease (see Figure 1). Moreover, re-biopsy of patients with lung cancer is not routinely performed. The use of position emission tomography scans to identify the most metabolically active lesions for biopsy could possibly circumvent this factor in the future, as long as such lesions are resectable. Additionally, as more molecularly tailored treatment options become available for lung cancer, re-biopsy of progressive sites of disease should become a standard procedure, especially for patients on clinical trials of targeted agents.
Since tumor specimens from three additional patients with acquired resistance to EGFR tyrosine kinase inhibitors did not demonstrate the T790M mutation, this specific lesion does not account for all mechanisms of acquired resistance to gefitinib or erlotinib. Given the paradigm established with imatinib, other drug-resistance mutations in EGFR, either within or outside the tyrosine kinase domain, are likely to exist. It is also possible that EGFR amplification itself plays a role in acquired resistance, since imatinib-resistant clones have been shown to lack resistance mutations but contain amplified copies of BCR-ABL [11,28]. Nonetheless, studies presented here provide a basis for the rational development of “second generation” kinase inhibitors for use in NSCLC.
Supporting Information
Figure S1 Imaging Studies from Patients 1, 2, and 3
(A) Patient 1. Serial chest radiographs from before (day 0) and during gefitinib treatment (14 d and 9 mo), demonstrating initial response and subsequent progression.
(B) Patient 2. Serial CT studies of the chest before (day 0) and during erlotinib treatment (4 mo and 25 mo), demonstrating initial response and subsequent progression.
(C) Patient 3. Serial chest radiographs before (day 0) and during adjuvant gefitinib treatment (3 mo), following complete resection of grossly visible disease. The left-sided pleural effusion seen at 3 mo recurred 4 mo later, at which time fluid was collected for molecular analysis.
(951 KB PPT).
Click here for additional data file.
Figure S2 Sequencing Chromatograms with the EGFR Exon 19 and 21 Mutations Identified in Patients 1 and 2
(A) Status of EGFR exon 21 in tumor specimens from patient 1. DNA from the growing lung lesion and the pleural effusion demonstrated a heterozygous T→G mutation at position 2573, leading to the common L858R amino acid substitution.
(B) All three specimens from patient 2 showed the same heterozygous exon 19 deletion, removing residues 747–749 and changing the alanine at position 750 to proline.
(104 KB PPT).
Click here for additional data file.
Figure S3 Sensitivity to Erlotinib Differs among NSCLC Cell Lines Containing Various Mutations in EGFR or KRAS
See legend for Figure 5.
(153 KB PPT).
Click here for additional data file.
Protocol S1 Memorial Sloan-Kettering Cancer Center IRB Protocol 04–103
(566 KB PDF).
Click here for additional data file.
Accession Numbers
The LocusLink (http://www.ncbi.nlm.nih.gov/LocusLink/) accession number for the KRAS2 sequence discussed in this paper is 3845; the GenBank (http://www.ncbi.nlm.nih.gov/Genbank/) accession number for the KRAS2 sequence discussed in this paper is NT_009714.16. Reference EGFR sequence was obtained from LocusLink accession number 1956 and GenBank accession number NT_033968.
Patient Summary
Background
Normal cells in our body have safety mechanisms that keep them from growing out of control. Tumor cells have somehow found ways around these safety mechanisms, in some cases through activating particular growth-promoting genes. One of these, the EGFR gene, is often activated in lung cancer. Two drugs, gefitinib (also known as Iressa) and erlotinib (also called Tarceva), have been developed to inhibit activated EGFR, and studies have shown that they can shrink tumors in some patients. Most patients who respond to these drugs have tumors that carry an alteration (or mutation) in the EGFR gene, which somehow makes their tumors responsive to the drugs.
Why Was This Study Done?
In those patients in whom the drugs work, the tumors shrink initially, but after a while they stop responding and the cancer comes back. The cancer has, as researchers describe it, become resistant to the drugs. Understanding how tumors become resistant is important to develop new and better drugs.
What Did the Researchers Do?
They asked patients who initially responded to erlotinib or gefitinib but then became resistant to consent to studies allowing further analysis of tumor tissue during and after drug treatment. They then re-examined the EGFR gene in these tumor samples.
What Did They Find?
They found that tumors from all patients carried mutations in the EGFR gene that are known to make them responsive to the drugs. In addition, three of the post-treatment tumors had an identical second mutation in their EGFR gene. Biochemical studies showed that these secondary alterations made the original drug-sensitive EGFR less sensitive to drug treatment. The numbers are small but suggest that this secondary resistance mutation could be quite common. Tumor cells from the three other patients didn't have this mutation, which suggests that there are other ways for lung cancers to become resistant to gefitinib and erlotinib.
What Next?
Larger studies are needed to confirm that this particular mutation is a major cause of resistance against the two drugs. It is also important to find out what causes resistance in the other cases. And knowing about this resistance mutation will help researchers to develop drugs that will work even against tumors with the mutation.
More Information Online
The following pages contain some information on the EGFR kinase inhibitors.
U. S. Food and Drug Administration information page on Iressa (gefitinib): http://www.fda.gov/cder/drug/infopage/iressa/iressaQ&A.htm
Cancer Research UK information page about erlotinib (Tarceva): http://www.cancerhelp.org.uk/help/default.asp?page=10296
The work was performed in the laboratory of HV. We thank all the patients who participated in this study; J. Doherty for PCR and sequencing expertise; T. Wang for help with sequencing; J. Somar for PCR-RFLP analyses; M. Ladanyi for advice and critical reading of the manuscript; C. Azzoli, A. Chiang, L. Tyson, members of the interventional radiology service, and multiple others for assistance in obtaining patient samples; B. Johnson and P. Janne for providing H3255 cells; R. Heelan for radiologic evaluation; P. Yurttas and N. Pavletich for helpful discussions about EGFR crystal structures; M. McClellan and R. Wilson from the Genome Sequencing Center at Washington University in St. Louis for re-reviewing exon 20 sequence chromatograms from 96 fresh-frozen lung tumor specimens; D. Wong, M. Blackman, and D. Tabarini from the Memorial Sloan-Kettering Cancer Center (MSKCC) DNA Sequencing Core Facility; A. Ciro and H. Djaballah from the MSKCC High-Throughput Screening Core Facility for assistance with cell viability assays; and AstraZeneca and Genentech for providing gefitinib and erlotinib, respectively. This work was supported by an anonymous donor. KAP received support from the Labrecque Foundation and the American Cancer Society (PF-05–078-01-MGO); GJR from the National Institutes of Health (T32 CA 09207), RS from the Canadian Institutes of Health Research, and WP from the CHEST Foundation of the American College of Chest Physicians and the LUNGevity Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author contributions. WP conceived and designed the study; acquired, analyzed, and interpreted the data; and drafted the article and revised the manuscript. VAM conceived and designed the clinical aspects of the study, analyzed and interpreted the data, and helped draft the article. KAP helped design aspects of the study; acquired, analyzed, and interpreted the data; and helped draft the article. GJR helped acquire the clinical data and specimens and critically read the manuscript. RS designed and performed the cell viability studies and critically read the manuscript. MFZ acquired the pathologic data and critically read the manuscript. MGK helped acquire the clinical data and specimens and critically read the manuscript. HV contributed to the conception and design of the study and to the interpretation of the data, and edited the article and the revised manuscript.
Competing Interests. VAM has received research funding from Genentech (co-developer of erlotinib) and has received honoraria from AstraZeneca (maker of gefitinib) for consultancy. MGK has received research funding from AstraZeneca and research funding and consulting fees from Genentech and has represented AstraZeneca before the United States Food and Drug Administration. WP, VAM, MFZ, and HV, represented by the Sloan-Kettering Institute for Cancer Research, filed on June 1, 2004, a provisional patent application entitled “Use of mutations in EGFR kinase as an indicator of therapeutic efficacy of erlotinib in the treatment of NSCLC,” patent 60/576,275. HV is Co-founder and Chair of the Board of Directors of the Public Library of Science.
Citation: Pao W, Miller VA, Politi KA, Riely GJ, Somwar R, et al. (2005) Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain. PLoS Med 2(3): e73.
Abbreviations
CMLchronic myelogenous leukemia
CTcomputed tomography
deldeletion
EGFRepidermal growth factor receptor
GISTgastrointestinal stromal tumor
HEShypereosinophilic syndrome
NSCLCnon-small cell lung cancer
p-EGFRphospho-EGFR
PCR-RFLPPCR restriction fragment length polymorphism
SNPsingle nucleotide polymorphism
t-EGFRtotal EGFR
==== Refs
References
Lynch TJ Bell DW Sordella R Gurubhagavatula S Okimoto RA Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib N Engl J Med 2004 350 2129 2139 15118073
Paez JG Janne PA Lee JC Tracy S Greulich H
EGFR mutations in lung cancer: Correlation with clinical response to gefitinib therapy Science 2004 304 1497 1500 15118125
Pao W Miller V Zakowski M Doherty J Politi K EGF receptor gene mutations are common in lung cancers from “never smokers” and are associated with sensitivity of tumors to gefitinib and erlotinib Proc Natl Acad Sci U S A 2004 101 13306 13311 15329413
Huang SF Liu HP Li LH Ku YC Fu YN High frequency of epidermal growth factor receptor mutations with complex patterns in non-small cell lung cancers related to gefitinib responsiveness in Taiwan Clin Cancer Res 2004 10 8195 8203 15623594
Kosaka T Yatabe Y Endoh H Kuwano H Takahashi T Mutations of the epidermal growth factor receptor gene in lung cancer: Biological and clinical implications Cancer Res 2004 64 8919 8923 15604253
Shigematsu H Lin L Takahashi T Nomura M Suzuki M Clinical and biological features of epidermal growth factor receptor mutations in lung cancers J Natl Cancer Inst 2004 In press
Pao W Wang TY Riely GJ Miller VA Pan Q
KRAS mutations and primary resistance of lung adenocarcinomas to gefitinib or erlotinib PLoS Medicine 2005 2 e17 15696205
Deininger M Buchdunger E Druker BJ The development of imatinib as a therapeutic agent for chronic myeloid leukemia 2004 Blood: Epub ahead of print
Al-Ali HK Heinrich MC Lange T Krahl R Mueller M High incidence of BCR-ABL kinase domain mutations and absence of mutations of the PDGFR and KIT activation loops in CML patients with secondary resistance to imatinib Hematol J 2004 5 55 60 14745431
Gorre ME Mohammed M Ellwood K Hsu N Paquette R Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification Science 2001 293 876 880 11423618
Shah NP Nicoll JM Nagar B Gorre ME Paquette RL Multiple BCR-ABL kinase domain mutations confer polyclonal resistance to the tyrosine kinase inhibitor imatinib (STI571) in chronic phase and blast crisis chronic myeloid leukemia Cancer Cell 2002 2 117 125 12204532
O'Hare T Pollock R Stoffregen EP Keats JA Abdullah OM Inhibition of wild-type and mutant Bcr-Abl by AP23464, a potent ATP-based oncogenic protein kinase inhibitor: Implications for CML Blood 2004 104 2532 2539 15256422
Shah NP Tran C Lee FY Chen P Norris D Overriding imatinib resistance with a novel ABL kinase inhibitor Science 2004 305 399 401 15256671
Sawyers C Targeted cancer therapy Nature 2004 432 294 297 15549090
Tamborini E Bonadiman L Greco A Albertini V Negri T A new mutation in the KIT ATP pocket causes acquired resistance to imatinib in a gastrointestinal stromal tumor patient Gastroenterology 2004 127 294 299 15236194
Cools J DeAngelo DJ Gotlib J Stover EH Legare RD A tyrosine kinase created by fusion of the PDGFRA and FIP1L1 genes as a therapeutic target of imatinib in idiopathic hypereosinophilic syndrome N Engl J Med 2003 348 1201 1214 12660384
Krug LM Miller VA Crapanzano J Ng KK Pizzo B Randomized phase II trial of trastuzumab (tras) plus either weekly docetaxel (doc) or paclitaxel (pac) in previously untreated advanced non-small cell lung cancer (NSCLC) Proc Am Soc Clin Oncol 2001 20 1328
Sordella R Bell DW Haber DA Settleman J Gefitinib-sensitizing EGFR mutations in lung cancer activate anti-apoptotic pathways Science 2004 305 1163 1167 15284455
Tracy S Mukohara T Hansen M Meyerson M Johnson BE Gefitinib induces apoptosis in the EGFRL858R non-small cell lung cancer cell line H3255 Cancer Res 2004 64 7241 7244 15492241
Bozyczko-Coyne D McKenna BW Connors TJ Neff NT A rapid fluorometric assay to measure neuronal survival in vitro J Neuroscience Meth 1993 50 205 216
Blencke S Zech B Engkvist O Greff Z Orfi L Characterization of a conserved structural determinant controlling protein kinase sensitivity to selective inhibitors Chem Biol 2004 11 691 701 15157880
Blencke S Ullrich A Daub H Mutation of threonine 766 in the epidermal growth factor receptor reveals a hotspot for resistance formation against selective tyrosine kinase inhibitors J Biol Chem 2003 278 15435 15440 12594213
Kreuzer KA Le Coutre P Landt O Na IK Schwarz M Preexistence and evolution of imatinib mesylate-resistant clones in chronic myelogenous leukemia detected by a PNA-based PCR clamping technique Ann Hematol 2003 82 284 289 12692682
Stamos J Sliwkowski MX Eigenbrot C Structure of the epidermal growth factor receptor kinase domain alone and in complex with a 4-anilinoquinazoline inhibitor J Biol Chem 2002 277 46265 46272 12196540
Daub H Specht K Ullrich A Strategies to overcome resistance to targeted protein kinase inhibitors Nat Rev Cancer 2004 3 1001 1010
Wood ER Truesdale AT McDonald OB Yuan D Hassell A A unique structure for epidermal growth factor receptor bound to GW572016 (Lapatinib): Relationships among protein conformation, inhibitor off-rate, and receptor activity in tumor cells Cancer Res 2004 64 6652 6659 15374980
Chen LL Trent JC Wu EF Fuller GN Ramdas L A missense mutation in KIT kinase domain 1 correlates with imatinib resistance in gastrointestinal stromal tumors Cancer Res 2004 64 5913 5919 15342366
Gorre ME Sawyers CL Molecular mechanisms of resistance to STI571 in chronic myeloid leukemia Curr Opin Hematol 2002 9 303 307 12042704
| 15737014 | PMC549606 | CC BY | 2021-01-05 10:39:38 | no | PLoS Med. 2005 Mar 22; 2(3):e73 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020073 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020074SynopsisCancer BiologyPharmacology/Drug DiscoveryOncologyPathologySurgeryOncologyDrugs and Adverse Drug ReactionsCancer: LungChemotherapyPathologyHow Tumor Cells Acquire Resistance to Kinase Inhibitors Synopsis3 2005 22 2 2005 2 3 e74This is an open-access article distributed under the terms of the Creative Commons Public Domain Declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose2005This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.
Acquired Resistance of Lung Adenocarcinomas to Gefitinib or Erlotinib Is Associated with a Second Mutation in the EGFR Kinase Domain
EGFR Inhibition in Non-Small Cell Lung Cancer: Resistance, Once Again, Rears Its Ugly Head
==== Body
Acquired resistance to chemotherapy is a major obstacle to successful cancer treatment. Understanding the mechanisms by which tumors become resistant to a particular agent is key to identifying new drugs or combination regimens.
Kinases are signaling molecules that control many aspects of cell behavior, including cell proliferation, i.e., whether and how fast cells divide. Abnormally active kinases promoting tumor growth are found in many cancers and are a focus of rational cancer drug design. One target for kinase inhibitors is the epidermal growth factor receptor (EGFR). Two EGFR inhibitors, gefitinib and erlotinib, showed therapeutic benefits in a subset of patients with non-small cell lung cancer. Recent work has helped us understand why some patients respond and some don't: responsive tumors usually harbor activating mutations in the EGFR gene, which somehow make the tumors sensitive to treatment. Nearly all patients whose tumors initially respond to EGFR inhibitors, however, eventually become resistant to the drugs and progress despite continued therapy.
William Pao and colleagues examined tumors from six patients with non-small cell lung cancer who initially responded to gefitinib or erlotinib but subsequently relapsed. Tumors from all six patients carried activating mutations in the EGFR gene. In addition, in three out of the six cases, the resistant tumor cells carried an identical second mutation in the EGFR gene. Whereas the activating mutation was present in tumor cells before treatment with erlotinib or gefitinib, the second mutation was not found in pre-treatment biopsies from these patients, nor in over 150 lung cancer samples from patients who had not been treated with either drug. Additional cell culture studies supported the notion that the secondary mutation causes resistance to gefitinib or erlotinib. It is clear, though, that this is only one mechanism of resistance, because in the three other cases resistance occurred in the absence of the second mutation. What caused the resistance in those tumors is not known.
All kinases share some common features, and a resistance mutation very similar to the one identified here has also been found in other kinase genes from tumors with acquired resistance to imatinib, another kinase inhibitor. As Gary Gilliland and colleagues point out in an accompanying Perspective (DOI: 10.1371/journal.pmed.0020075), the initial identification three years ago of resistance mutations against imatinib led to the rapid development of alternative kinase inhibitors that work even against tumors with the resistance mutation. Similarly, the results by Pao and colleagues should help researchers develop second generation drugs for lung cancer.
Re-biopsy of progressing lung lesion
| 0 | PMC549607 | CC0 | 2021-01-05 11:13:38 | no | PLoS Med. 2005 Mar 22; 2(3):e74 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020074 | oa_comm |
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1573698910.1371/journal.pmed.0020075PerspectivesCancer BiologyMolecular Biology/Structural BiologyPharmacology/Drug DiscoveryOncologyPathologyRespiratory MedicineSurgeryCancer: LungOncologyChemotherapyDrugs and Adverse Drug ReactionsPathologyEGFR Inhibition in Non-Small Cell Lung Cancer: Resistance, Once Again, Rears Its Ugly Head PerspectivesClark Jennifer Cools Jan Gilliland D. Gary *Jennifer Clark and D. Gary Gilliland are Instructor and Professor, respectively, at Harvard Medical School, Brigham and Women's Hospital, and the Dana-Farber Cancer Institute, Boston, Massachusetts, United States of America. Jan Cools is Research Fellow at the University of Leuven, Belgium (VIB).
Competing Interests: The authors declare that they have no competing interests.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 22 2 2005 2 3 e75Copyright: © 2005 Clark 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.
Acquired Resistance of Lung Adenocarcinomas to Gefitinib or Erlotinib Is Associated with a Second Mutation in the EGFR Kinase Domain
How Tumor Cells Acquire Resistance to Kinase Inhibitors
Most patients with non-small cell lung cancer who initially respond to gefitinib or erlotinib (tyrosine kinase inhibitors) ultimately develop resistance and disease relapse. What is the mechanism for this resistance?
==== Body
Kinase Inhibition for Treatment of Cancer
Uncontrolled proliferation of tumor cells is a hallmark of cancer. In many types of cancer, mutations in genes that activate cellular signal transduction pathways contribute to enhanced proliferation and survival of cancer cells. One well-characterized example is mutation in tyrosine kinases, enzymes that regulate the growth and survival of cells. Tyrosine kinase activity is tightly regulated in normal cells, but is dysregulated due to mutation in some cancers, including lung cancer, resulting in enhanced proliferation and survival of cancer cells. The tyrosine kinases are attractive candidates for molecularly targeted therapy in cancer, because cancers become dependent on growth signals from the mutant tyrosine kinases. Tyrosine kinases require ATP for their enzymic activity, and thus small molecules that mimic ATP can bind to mutant kinases and inactivate them.
The paradigm for tyrosine kinase inhibition as treatment for cancer using small-molecule inhibitors was first established in the context of chronic myelogenous leukemia (CML) associated with the BCR-ABL gene rearrangement [1]. Imatinib (Gleevec), a 2-phenylaminopyrimidine, is a competitive inhibitor of ATP binding to the ABL kinase, thereby inhibiting the constitutively activated BCR-ABL tyrosine kinase. Imatinib induces complete remission in most patients with CML in stable phase [1], and also has activity in CML that has progressed to blast crisis [2].
Imatinib is also a potent inhibitor of the ARG, KIT, PDGFRA, and PDGFRB tyrosine kinases. As a consequence, there have been additional dividends from the United States Federal Drug Administration approval of imatinib for treatment of BCR-ABL-positive CML. For example, imatinib is effective in treatment of chronic myelomonocytic leukemia with gene rearrangements that constitutively activate PDGFRB [3], of hypereosinophilic syndrome with activating mutations in PDGFRA [4], and of gastrointestinal stromal cell tumors associated with activating mutations in KIT [5] (all reviewed in [6]).
More recently, this paradigm has been extended to treatment of non-small cell lung cancer (NSCLC). Several mutations have been identified in the context of epidermal growth factor receptor (EGFR) in patients with NSCLC that are associated with clinical response to the small-molecule EGFR inhibitors gefitinib (Iressa) or erlotinib (Tarceva) [7,8,9], including in-frame deletions such as del L747–E749;A750P in exon 19, or L858R in exon 21. Although responses are often dramatic, most responding patients ultimately develop clinical resistance and relapse of disease [7,8,9]. The basis for resistance had not been known, in part owing to the difficulty in obtaining tissue from re-biopsy at time of relapse.
Resistance to Small-Molecule Tyrosine Kinase Inhibitors
As might have been anticipated in treatment of cancer with any single agent, resistance to small-molecule tyrosine kinase inhibitors has emerged as a significant clinical problem. This was first appreciated in patients with CML treated with imatinib whose tumors developed resistance, and has been most extensively studied in that context. Although there are many potential mechanisms for development of clinical resistance, most cases of imatinib-resistant CML are due to point mutations in the BCR-ABL kinase domain itself, including T315I [10,11]. Similar mutations in the homologous residues of the kinase domains of PDGFRA (T674I) and KIT (T670I) account for imatinib resistance in some patients with hypereosinophilic syndrome and gastrointestinal stromal cell tumors, respectively [4,12]. These findings suggest strategies to overcome resistance that include the use of alternative small-molecule inhibitors. Indeed, about three years after the recognition of imatinib resistance mutations in BCR-ABL-positive CML, new drugs are now in clinical trials that are potent inhibitors of imatinib-resistant BCR-ABL mutants [13,14].
A Basis for Resistance to Small-Molecule EGFR Inhibitors in NSCLC
In an elegant new study in PLoS Medicine, Pao and colleagues have identified acquired mutations in patients with NSCLC that appear to explain clinical resistance to gefitinib or erlotinib [15]. The mechanism of resistance in three patients was acquisition of a T790M substitution in EGFR that was not present at time of diagnosis, but was detected with progression of disease after initial response to gefitinib or erlotinib. T790M in the context of either transiently expressed wild-type EGFR or the mutant alleles del L474–E749;A750P or L858R impairs inhibition by gefitinib or erlotinib as assessed by autophosphorylation. Furthermore, the NSCLC cell line H1975 harbors both the L858R and T790M mutations, and is resistant to inhibition by gefitinib or erlotinib, unlike cell lines that express the L858R allele alone. In the H1975 cell line, it was possible to obtain adequate quantities of RNA to confirm that the L858R and T790M mutations are present on the same allele, as would be predicted if T790M confers resistance to inhibition of the L858R allele.
Structural models of EGFR provide structural insights into these biological data. A ribbon structure of erlotinib bound to the EGFR kinase domain (Figure 1) shows the threonine residue at position 790 in green and the positions of the exon 19 and L858R gain-of-function mutations. Substitution of methionine for threonine at position 790 would be predicted to result in steric hindrance of erlotinib binding to EGFR (Figure 2).
Figure 1 Erlotinib Bound to the EGFR Kinase Domain
Schematic representation of the wild-type EGFR tyrosine kinase domain (cyan) bound to erlotinib (orange) from the Protein Data Bank (http://www.rcsb.org/pdb/) entry 1M17. The threonine 790 side chain is shown in green. The positions of the phosphate-binding loop (P-loop), the αC-helix, and the activation loop (conserved structural features in kinase domains) are shown for reference. Sites of common lung-cancer-associated drug-sensitive mutations (exon 19 deletion [del] and L858R) are also depicted.
(Figure: Nikola Pavletich, Structural Biology Program, Memorial Sloan-Kettering Cancer Center)
Figure 2 Structural Models of EGFR Showing the T790M Resistance Mutation
(A) Space-filling representation of the wild-type kinase active site (cyan) with the viewer looking down the vertical axis. The structure above the plane of the figure is omitted for clarity. The threonine 790 side chain is green, and erlotinib's molecular surface is shown as a yellow net.
(B) The threonine 790 side chain is replaced by the corresponding methionine side chain from the structure of the insulin receptor tyrosine kinase (Protein Data Bank entry 1IRK). The EGFR and insulin receptor have a similar structure in this region of the active site. The methionine side chain would sterically clash with erlotinib, as shown, as well as with the related kinase inhibitor gefitinib (not shown).
(Figure: Nikola Pavletich, Structural Biology Program, Memorial Sloan-Kettering Cancer Center)
These observations provide convincing evidence that, at least in some patients with NSCLC, resistance to gefitinib or erlotinib can be attributed to acquisition of a T790M mutation in the context of EGFR. However, three additional patients with clinical resistance to gefitinib or erlotinib did not have the T790M mutation, nor did they have mutant KRAS alleles that have previously been shown by these same authors to confer resistance to these inhibitors [9]. Thus, mechanisms of resistance are heterogeneous.
Next Steps, and Lessons Learned
It will be important to identify alternative small-molecule inhibitors for the T790M resistance mutation. Structural data suggest that one compound, lapatinib, may subserve this purpose [16], but it has not been tested for biological activity in this context. New chemical screens and/or rational drug design to identify alternative inhibitors is warranted. In addition, only half of this small cohort of patients with NSCLC with clinical resistance to gefitinib or erlotinib had the T790M substitution. Efforts to identify alternative mechanisms for resistance may be guided by experience with imatinib resistance in the context of BCR-ABL, and should include full-length sequencing of EGFR to identify other resistance mutations, and analysis for evidence of gene amplification, as well as investigation of other well-characterized mechanisms of drug resistance such as drug efflux or increased drug metabolism.
Pao and colleagues' superb study also highlights several important points that may guide development of kinase-targeted therapies in the future. It is clear that, to the extent that small-molecule kinase inhibitors are effective as single agents in treatment of cancer, resistance will develop. Furthermore, based on previous experience, some of these patients are likely to harbor acquired point mutations in the target kinase that confer resistance. Resistance mutations identified via in vitro screens have shown a high degree of correlation with those that develop in vivo, as shown in screens for imatinib-resistant BCR-ABL mutants [11] and PKC412-resistant FLT3 mutants [17], as well as the T790M resistance mutation to gefitinib in the context of EGFR [18]. Thus, in vitro screens for mutations that confer resistance to kinase inhibitors are warranted, followed by efforts to identify drugs that overcome resistance. This proactive approach should shorten the time frame for new drug development.
These findings also emphasize the critical need for re-biopsy of patients with cancer treated with molecularly targeted therapies at time of relapse. Tissue acquisition is more challenging in solid tumors than for hematopoietic malignancies, and may entail risk. Nonetheless, it is clear that data derived from such analyses will be essential to inform approaches to improving therapy for NSCLC and other solid tumors.
Citation: Clark J, Cools, J, Gilliland DG (2005) EFGR inhibition in non-small cell lung cancer: Resistance, once again, rears its ugly head. PLoS Med 2(3): e75.
Abbreviations
CMLchronic myelogenous leukemia
EGFR
epidermal growth factor receptor
NSCLCnon-small cell lung cancer
==== Refs
References
Druker BJ Sawyers CL Kantarjian H Resta DJ Reese SF Activity of a specific inhibitor of the BCR-ABL tyrosine kinase in the blast crisis of chronic myeloid leukemia and acute lymphoblastic leukemia with the Philadelphia chromosome N Engl J Med 2001 344 1038 1042 11287973
Sawyers CL Hochhaus A Feldman E Goldman JM Miller CB Imatinib induces hematologic and cytogenetic responses in patients with chronic myelogenous leukemia in myeloid blast crisis: Results of a phase II study Blood 2002 99 3530 3539 11986204
Apperley JF Gardembas M Melo JV Russell-Jones R Bain BJ Response to imatinib mesylate in patients with chronic myeloproliferative diseases with rearrangements of the platelet-derived growth factor receptor beta N Engl J Med 2002 347 481 487 12181402
Cools J DeAngelo DJ Gotlib J Stover EH Legare RD A tyrosine kinase created by fusion of the PDGFRA and FIP1L1 genes as a therapeutic target of imatinib in idiopathic hypereosinophilic syndrome N Engl J Med 2003 348 1201 1214 12660384
Joensuu H Roberts PJ Sarlomo-Rikala M Andersson LC Tervahartiala P Effect of the tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor N Engl J Med 2001 344 1052 1056 11287975
Sawyers C Targeted cancer therapy Nature 2004 432 294 297 15549090
Lynch TJ Bell DW Sordella R Gurubhagavatula S Okimoto RA Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib N Engl J Med 2004 350 2129 2139 15118073
Paez JG Janne PA Lee JC Tracy S Greulich H EGFR mutations in lung cancer: Correlation with clinical response to gefitinib therapy Science 2004 304 1497 1500 15118125
Pao W Miller V Zakowski M Doherty J Politi K EGF receptor gene mutations are common in lung cancers from “never smokers” and are associated with sensitivity of tumors to gefitinib and erlotinib Proc Natl Acad Sci U S A 2004 101 13306 13311 15329413
Gorre ME Mohammed M Ellwood K Hsu N Paquette R Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification Science 2001 293 876 880 11423618
Azam M Latek RR Daley GQ Mechanisms of autoinhibition and STI-571/imatinib resistance revealed by mutagenesis of BCR-ABL Cell 2003 112 831 843 12654249
Tamborini E Bonadiman L Greco A Albertini V Negri T A new mutation in the KIT ATP pocket causes acquired resistance to imatinib in a gastrointestinal stromal tumor patient Gastroenterology 2004 127 294 299 15236194
Shah NP Tran C Lee FY Chen P Norris D Overriding imatinib resistance with a novel ABL kinase inhibitor Science 2004 305 399 401 15256671
Weisberg E Manley PW Breitenstein W Bruggen J Cowan-Jacob SW Characterization of AMN107, a selective inhibitor of wildtype and mutant BCR-ABL Cancer Cell 2005 In press
Pao W Miller VA Politi KA Riely GJ Somwar R Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain PLoS Med 2005 2 e73 15737014
Wood ER Truesdale AT McDonald OB Yuan D Hassell A A unique structure for epidermal growth factor receptor bound to GW572016 (Lapatinib): Relationships among protein conformation, inhibitor off-rate, and receptor activity in tumor cells Cancer Res 2004 64 6652 6659 15374980
Cools J Mentens N Furet P Fabbro D Clark JJ Prediction of resistance to small molecule FLT3 inhibitors: Implications for molecularly targeted therapy of acute leukemia Cancer Res 2004 64 6385 6389 15374944
Blencke S Zech B Engkvist O Greff Z Orfi L Characterization of a conserved structural determinant controlling protein kinase sensitivity to selective inhibitors Chem Biol 2004 11 691 701 15157880
| 15736989 | PMC549608 | CC BY | 2021-01-05 10:39:33 | no | PLoS Med. 2005 Mar 22; 2(3):e75 | utf-8 | PLoS Med | 2,005 | 10.1371/journal.pmed.0020075 | oa_comm |
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-61570117910.1186/1471-2156-6-6Research ArticleGenetic variation analysis of the Bali street dog using microsatellites Irion Dawn N [email protected] Alison L [email protected] Sherry [email protected] Alan N [email protected] Niels C [email protected] Veterinary Genetics Laboratory, Center for Veterinary Genetics, School of Veterinary Medicine, University of California, Davis, California 95616, USA2 Yayasan Yudisthira, Sanur, Bali 80228, Indonesia3 School of Biotechnology and Biomolecular Sciences, University of New South Wales, Sydney NSW 2052, Australia2005 8 2 2005 6 6 6 10 8 2004 8 2 2005 Copyright © 2005 Irion et al; licensee BioMed Central Ltd.2005Irion 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
Approximately 800,000 primarily feral dogs live on the small island of Bali. To analyze the genetic diversity in this population, forty samples were collected at random from dogs in the Denpasar, Bali region and tested using 31 polymorphic microsatellites. Australian dingoes and 28 American Kennel Club breeds were compared to the Bali Street Dog (BSD) for allelic diversity, heterozygosities, F-statistics, GST estimates, Nei's DA distance and phylogenetic relationships.
Results
The BSD proved to be the most heterogeneous, exhibiting 239 of the 366 total alleles observed across all groups and breeds and had an observed heterozygosity of 0.692. Thirteen private alleles were observed in the BSD with an additional three alleles observed only in the BSD and the Australian dingo. The BSD was related most closely to the Chow Chow with a FST of 0.088 and also with high bootstrap support to the Australian dingo and Akita in the phylogenetic analysis.
Conclusions
This preliminary study into the diversity and relationship of the BSD to other domestic and feral dog populations shows the BSD to be highly heterogeneous and related to populations of East Asian origin. These results indicate that a viable and diverse population of dogs existed on the island of Bali prior to its geographic isolation approximately 12,000 years ago and has been little influenced by domesticated European dogs since that time.
==== Body
Background
Bali, a province of the Republic of Indonesia, is an island just 87 km from north to south and 142 km from east to west and home to more than 2.9 million people [1]. Approximately 800,000 stray dogs (Fig. 1) also live on the island based on a survey conducted by the Bali Street Dog Foundation (personal communication). Only a small percentage of these dogs live in homes or are provided routine veterinary care [2].
Figure 1 Typical Balinese street dogs. Their phenotypic appearance is similar to that described for randomly breeding feral dog subpopulations in other parts of the world.
More than 90% of the residents of Bali are Hindu [3] with myth and ritual playing a vital part of daily life [1]. The dog is also an important part of Balinese life and mythology. A popular tale from the Mahabharata [4] describes King Yudisthira's journey to Heaven's Gate, and his love for a dog that befriended him on his arduous and tragic journey (Fig. 2). As a direct result of such mythology, BSDs are treated with a degree of reverence and are often provided ceremonial food offerings [2]. The deliberate killing of street dogs is not typically practiced, because Balinese people believe that all things should be allowed to die naturally [2]. These cultural mores have contributed to the current overpopulation of dogs on the island.
Figure 2 The Story of Yudisthira [4].
As a result of overpopulation, many BSDs suffer from chronic skin diseases, internal parasites, parvo- and distemper-virus infections, and malnutrition. In an effort to reduce the dog population and to care for their medical needs, the Bali Street Dog Foundation (Yayasan Yudisthira Swarga) was founded in 1998 [2]. They provide emergency care, treatment for skin disease and parasites, sterilization, public education on the plight of feral dogs, and improved veterinarian training. Twenty to 30 dogs are sterilized each day, with more than 9,000 dogs sterilized to date.
The BSD population is of interest for both its genetic diversity and historical relationships. It is also a population that has bred more or less randomly for thousands of years with limited genetic influx, due mainly to geographic barriers and a strict rabies control program in effect since 1926. The present study is concerned with the genetic diversity of this unique canine population and its relationship to other canine subpopulations in Asia and throughout the world. Data presented herein was derived from the DNA testing of 40 BSD samples from the Denpasar city region of Bali with 31 polymorphic microsatellite loci. The genetic diversity of the BSD was compared to that of the Australian dingo and 28 American Kennel Club (AKC) breeds.
Results
Locus diversity
Analysis of locus diversity across all 30 subpopulations revealed that the number of observed alleles ranged from six to 20 with a total of 366 for all loci (Table 1). Overall heterozygosity of the loci was high, with an average of 0.779, and all but four loci having HT values greater than 0.700. Average HS was 0.577 for the 30 subpopulations, with all but three loci having HS values greater than 0.500. The HS and HT values were closest for C23.123 and farthest for C22.279 and C10.404. HWE analysis revealed that all but one locus had at least one population out of equilibrium for the 30 populations sampled. C01.424, C31.646 and CPH16 had 7 populations out of HWE and AHT130 did not have any populations with p values below 0.05. The level of locus diversity attributable to subpopulation structure was evaluated with two statistics – RST and FST. Both statistics gave similar average values at 0.230 and 0.236 respectively. However, RST ranged from 0.098 to 0.486 while FST ranged from 0.179 to 0.328.
Table 1 Observed number of alleles, average total heterozygosity (HT), average subpopulation heterozygosity (HS), number of populations out of HWE, average p values, RST, FST, RST/FSTratio, GSTand pairwise FSTvalues for 31 loci.
BSD Pairwise FST by Locus
Chr. Num. Observed Alleles HT HS Num. Loci with p value <0.05 Average p value RST FST RST/FST × Dingo × Chow
CPH16 CFA20 11 0.829 0.610 7 0.407 0.098 0.233 0.420 0.005 0.132
C08.618 CFA08 9 0.744 0.553 4 0.481 0.148 0.229 0.646 0.071 0.185
FH2001 CFA23 13 0.791 0.593 4 0.433 0.167 0.225 0.741 0.083 0.100
C20.446 CFA20 10 0.729 0.553 3 0.541 0.173 0.215 0.804 0.123 0.105
C01.424 CFA01 9 0.716 0.476 7 0.475 0.258 0.320 0.807 0.125 0.110
CPH02 CFA32 9 0.693 0.520 4 0.499 0.194 0.223 0.871 0.148 0.066
FH2004 CFA11 18 0.809 0.611 3 0.522 0.187 0.214 0.873 0.057 0.060
AHT137 CFA11 14 0.861 0.672 2 0.427 0.176 0.197 0.893 0.064 0.188
C03.877 CFA03 12 0.730 0.509 1 0.524 0.268 0.275 0.977 0.244 0.114
C06.636 CFA06 12 0.652 0.483 6 0.435 0.233 0.237 0.982 0.069 0.024
AHT121 CFA13 18 0.865 0.632 1 0.511 0.255 0.251 1.016 0.098 0.063
VIASD10 CFA07 9 0.759 0.555 2 0.525 0.249 0.233 1.066 0.268 0.030
C31.646 CFA31 14 0.814 0.566 7 0.410 0.301 0.281 1.075 0.088 0.036
RVC1 CFA15 9 0.774 0.569 4 0.458 0.270 0.242 1.119 0.355 0.216
LEI002 CFA27 11 0.737 0.551 4 0.534 0.259 0.229 1.127 0.107 0.149
LEI004 CFA37 13 0.667 0.509 4 0.510 0.246 0.219 1.128 0.133 0.051
C28.176 CFA28 10 0.735 0.546 6 0.400 0.270 0.234 1.152 0.249 0.013
C22.279 CFA22 11 0.836 0.529 2 0.441 0.262 0.214 1.226 0.201 0.109
PEZ02 Unlinked 12 0.762 0.600 1 0.567 0.232 0.187 1.242 0.134 0.027
FH2054 CFA12 10 0.848 0.654 4 0.524 0.251 0.199 1.261 0.055 0.069
C23.123 CFA23 8 0.766 0.636 4 0.402 0.351 0.278 1.263 0.110 0.002
CPH08 CFA19 11 0.765 0.582 4 0.465 0.284 0.222 1.276 0.082 0.056
C14.866 CFA14 10 0.840 0.604 3 0.473 0.330 0.255 1.293 0.180 0.112
PEZ08 CFA17 17 0.859 0.684 5 0.465 0.235 0.179 1.312 0.121 0.079
AHT130 CFA18 11 0.829 0.614 0 0.539 0.313 0.235 1.331 0.116 0.105
AHT111 CFA02 11 0.785 0.582 4 0.371 0.348 0.246 1.414 0.214 0.006
C10.404 CFA10 13 0.865 0.558 3 0.526 0.486 0.328 1.479 0.177 0.160
C09.250 CFA09 10 0.830 0.582 1 0.491 0.412 0.272 1.513 0.016 0.042
FH2140 CFA05 20 0.795 0.621 2 0.560 0.297 0.192 1.546 0.119 0.072
AHT139 CFA15 6 0.664 0.508 4 0.452 0.330 0.206 1.604 0.085 0.078
CPH03 CFA06 15 0.815 0.612 2 0.526 0.394 0.229 1.724 0.017 0.180
All 366 0.779 0.577 108 0.481 0.230 0.236 1.135 0.126 0.088
Bali street dog diversity
Overall, the BSD was the most genetically diverse population surveyed here, displaying 239 total alleles out of the 366 seen in all 30 subpopulations or 65.3% of the total observed alleles (Table 2). The Australian dingo displayed 144 alleles and the AKC breeds displayed 138.8 on average. Analysis of expected (HE) and observed (HO) heterozygosities (Table 2) revealed that the BSD had a 44.0% higher HE than the Australian dingo (0.736 vs. 0.511) and a 28.4% higher HE than the average AKC breed (0.573). HO was also highest in the BSD at 0.692, versus 0.426 in the Australian dingo and 0.563 in the average AKC breed.
Table 2 Total number of alleles (NA) observed, range of the lowest and highest number of observed alleles per locus, expected heterozygosity (HE), observed heterozygosity (HO), FIS, number of loci out of HWE and average p values for all 31 loci for the BSD, a bootstrapped sampling of the BSD, the Australian dingo, the American Kennel Club breeds and for all subpopulations.
NA Observed NA Range HE HO FIS Num. Loci with p value <0.05 Average p value
Bali Street Dog 239 3 – 14 0.736 0.692 0.097 4 0.357
Bali Street Dog20 214.6 --- 0.727 0.692 --- --- ---
Australian dingo 144 2 – 9 0.511 0.426 0.194 12 0.284
AKC Breeds 138.8 2.2 – 7.8 0.573 0.563 0.137 3.29 0.492
All Populations 366 6 – 20 0.577 0.562 0.136 3.60 0.492
In order to evaluate the bias of sampling twice the number of BSDs, twenty samples were taken at random from the total pool of 40 for bootstrap determinations, and the number of observed alleles, HE, and HO were calculated. This process was repeated for 10,000 iterations and the average value for each measurement was determined. The average bootstrap value for the number of observed alleles for 20 BSDs was 214.6. The average bootstrapped HE and HO values for the BSD were 0.727 and 0.692 respectively.
To understand the loss of approximately 24 observed alleles after bootstrapping, the allele frequencies for the BSD at each locus was examined. While the BSD had the highest number of alleles, they also had the highest number of alleles with a frequency below 5% (67 out of 239, data not shown).
FIS estimates were calculated to assess the level of inbreeding for each subpopulation (Table 2). The BSD had the lowest value at 0.097 and the Australian dingo had the highest at 0.194 with the average of AKC breeds at 0.137.
Private alleles
Allele frequency analysis also revealed that 10 private alleles were observed in the BSD as well as three alleles shared only with the Australian dingo (Table 3). The majority of private alleles in the BSD were below 5% in frequency with the exception of AHT121 where the private alleles had a combined frequency of 18.75%. The BSD and the Australian dingo also shared three alleles not seen in any AKC breed at locus C10.404 with a combined frequency of 16.25% in the BSD and 75% in the Australian dingo.
Table 3 Private alleles for the BSD and Australian dingo subpopulations relative to the 28 comparison AKC breeds.
Locus Allele Pop Freq. Pop Freq.
AHT111 92 BSD 0.013
AHT121 82 BSD 0.075
AHT121 90 BSD 0.113
C06.636 158 BSD 0.013
C10.404 168 BSD 0.038 Dingo 0.075
C10.404 170 BSD 0.063 Dingo 0.450
C10.404 172 BSD 0.063 Dingo 0.225
C10.404 174 BSD 0.038
C23.123 154 BSD 0.013
FH2140 160 BSD 0.013
FH2140 171 BSD 0.025
PEZ02 144 BSD 0.013
VIASD10 94 BSD 0.013
Asian alleles
Several additional unique alleles were found only in the BSD, Australian dingo, Chow Chow and Akita; demonstrating a closer relationship of the BSD to Asian versus non-Asian dogs (data not shown). Appearing at the highest frequency was allele 201 of locus CPH08 in the BSD, Australian dingo and Chow Chow at frequencies of 21.3%, 10% and 20%, respectively. Further, the BSD, Australian dingo, Chow Chow and Akita share allele 113 of locus C22.279 at frequencies of 23.8%, 67.5%, 15% and 5%, respectively. Results for locus PEZ08 demonstrate a lack of influence of European alleles where a high frequency of deviations from n+4 alleles were observed in the AKC breeds sampled yet no deviation from n+4 alleles was observed in the BSD or the Australian dingo.
A pairwise FST analysis was performed for each locus between the BSD and the two closest subpopulations: the Australian dingo and Chow Chow (Table 1). The BSD was most similar to the Australian dingo at locus CPH16 with a FST of 0.005 and to the Chow Chow at locus C23.123 with a FST of 0.002. The BSD was most dissimilar to the Australian dingo and Chow Chow at locus RVC1 with a FST of 0.355 and 0.216 respectively. Several loci had similar distances for both population pairs, such as locus C01.424 or locus C20.446 and may indicate areas of the genome that are neutral to either environmental or human selection.
Genetic distance relationships
Further distance analysis was performed for all 31 loci between all 30 subpopulations using both Nei's DA distance and pairwise FST estimates (Table 4). Across all loci, the BSD shared allele frequencies most closely with the Chow Chow (DA = 0.242, FST = 0.088) and the Australian dingo (DA = 0.242, FST = 0.126), and least closely with the Airedale Terrier (DA = 0.454, FST = 0.258).
Table 4 Nei's DA distance (lower triangle) and mean FSTestimates (upper triangle) between each pair of 9 dog subpopulations represented graphically in Figure 3.
BSD Dingo Chow Akita AES AS AT BCO BLT MBT BMD BS BT BU BX BZ DP GH GR JRT KE LR NE PG PM PPN PWC RR WM YT
BSD 0.13 0.09 0.13 0.14 0.14 0.26 0.19 0.26 0.25 0.21 0.15 0.17 0.18 0.33 0.18 0.26 0.17 0.14 0.11 0.17 0.17 0.18 0.23 0.13 0.11 0.15 0.15 0.16 0.13
Dingo 0.24 0.23 0.26 0.27 0.27 0.41 0.32 0.42 0.40 0.37 0.30 0.29 0.29 0.50 0.31 0.41 0.30 0.26 0.25 0.31 0.30 0.31 0.39 0.27 0.25 0.29 0.29 0.29 0.25
Chow 0.24 0.40 0.19 0.22 0.21 0.34 0.26 0.33 0.32 0.29 0.24 0.23 0.24 0.41 0.26 0.34 0.26 0.21 0.17 0.23 0.25 0.25 0.31 0.20 0.17 0.22 0.22 0.23 0.17
Akita 0.29 0.41 0.36 0.19 0.19 0.33 0.26 0.35 0.32 0.31 0.20 0.22 0.25 0.40 0.23 0.30 0.24 0.18 0.17 0.23 0.22 0.26 0.30 0.20 0.17 0.20 0.20 0.22 0.19
AES 0.28 0.43 0.43 0.37 0.10 0.29 0.16 0.33 0.30 0.23 0.14 0.16 0.19 0.32 0.24 0.24 0.13 0.12 0.10 0.14 0.16 0.18 0.25 0.12 0.10 0.17 0.13 0.16 0.13
AS 0.27 0.42 0.38 0.36 0.20 0.27 0.15 0.32 0.29 0.24 0.16 0.17 0.18 0.35 0.21 0.23 0.16 0.13 0.10 0.17 0.16 0.19 0.23 0.13 0.10 0.16 0.13 0.17 0.12
AT 0.45 0.55 0.50 0.45 0.40 0.36 0.35 0.46 0.44 0.37 0.34 0.32 0.34 0.50 0.31 0.46 0.32 0.29 0.26 0.34 0.32 0.32 0.42 0.27 0.28 0.34 0.32 0.31 0.28
BCO 0.35 0.49 0.48 0.44 0.26 0.25 0.45 0.39 0.37 0.24 0.21 0.21 0.25 0.41 0.29 0.30 0.22 0.21 0.15 0.24 0.23 0.25 0.30 0.17 0.16 0.20 0.22 0.21 0.16
BLT 0.43 0.55 0.51 0.52 0.46 0.42 0.51 0.54 0.10 0.43 0.35 0.40 0.35 0.49 0.40 0.44 0.36 0.30 0.25 0.34 0.32 0.35 0.46 0.33 0.27 0.36 0.36 0.36 0.30
MBT 0.41 0.52 0.48 0.48 0.41 0.38 0.48 0.50 0.09 0.54 0.52 0.54 0.38 0.46 0.50 0.52 0.48 0.43 0.36 0.42 0.41 0.34 0.44 0.30 0.26 0.35 0.34 0.33 0.29
BMD 0.39 0.51 0.47 0.49 0.36 0.35 0.45 0.36 0.54 0.42 0.23 0.28 0.30 0.42 0.35 0.36 0.28 0.23 0.18 0.29 0.25 0.28 0.37 0.21 0.20 0.26 0.24 0.26 0.21
BS 0.32 0.46 0.45 0.38 0.28 0.28 0.48 0.35 0.52 0.34 0.36 0.20 0.24 0.35 0.23 0.28 0.17 0.15 0.13 0.19 0.15 0.22 0.30 0.15 0.11 0.18 0.15 0.19 0.17
BT 0.36 0.48 0.44 0.41 0.29 0.30 0.43 0.35 0.59 0.37 0.42 0.35 0.23 0.40 0.24 0.28 0.19 0.20 0.16 0.22 0.24 0.23 0.28 0.17 0.18 0.22 0.18 0.24 0.17
BU 0.31 0.41 0.42 0.42 0.29 0.27 0.41 0.39 0.42 0.31 0.41 0.40 0.36 0.36 0.27 0.30 0.22 0.22 0.17 0.26 0.24 0.28 0.32 0.22 0.16 0.24 0.19 0.26 0.19
BX 0.47 0.58 0.55 0.54 0.37 0.40 0.51 0.48 0.49 0.46 0.46 0.45 0.50 0.36 0.44 0.44 0.40 0.33 0.31 0.39 0.39 0.40 0.46 0.35 0.31 0.40 0.36 0.40 0.31
BZ 0.35 0.46 0.45 0.41 0.40 0.33 0.39 0.44 0.52 0.37 0.47 0.36 0.39 0.36 0.50 0.32 0.21 0.23 0.17 0.27 0.23 0.28 0.34 0.20 0.19 0.24 0.25 0.27 0.20
DP 0.45 0.57 0.55 0.48 0.39 0.37 0.60 0.44 0.57 0.41 0.47 0.44 0.43 0.41 0.48 0.46 0.29 0.28 0.23 0.29 0.30 0.34 0.37 0.24 0.23 0.29 0.26 0.32 0.26
GH 0.35 0.46 0.46 0.44 0.27 0.27 0.41 0.34 0.50 0.35 0.40 0.30 0.32 0.35 0.49 0.33 0.43 0.19 0.16 0.21 0.21 0.23 0.28 0.17 0.15 0.22 0.15 0.23 0.16
GR 0.33 0.43 0.42 0.39 0.25 0.26 0.37 0.33 0.43 0.29 0.37 0.31 0.35 0.35 0.40 0.37 0.44 0.34 0.11 0.18 0.13 0.18 0.31 0.15 0.10 0.17 0.16 0.16 0.14
JRT 0.25 0.43 0.32 0.34 0.21 0.20 0.37 0.27 0.38 0.24 0.30 0.28 0.30 0.27 0.39 0.31 0.38 0.29 0.25 0.15 0.11 0.15 0.26 0.11 0.08 0.14 0.13 0.16 0.07
KE 0.33 0.50 0.42 0.41 0.27 0.28 0.44 0.35 0.47 0.32 0.42 0.36 0.36 0.38 0.46 0.42 0.42 0.36 0.34 0.28 0.22 0.25 0.32 0.15 0.15 0.22 0.17 0.23 0.16
LR 0.33 0.45 0.46 0.39 0.29 0.28 0.42 0.36 0.42 0.31 0.36 0.26 0.38 0.37 0.46 0.35 0.44 0.32 0.26 0.24 0.37 0.20 0.33 0.15 0.12 0.23 0.19 0.17 0.16
NE 0.35 0.45 0.44 0.45 0.28 0.30 0.42 0.38 0.45 0.45 0.37 0.37 0.38 0.39 0.46 0.43 0.46 0.34 0.33 0.28 0.40 0.34 0.36 0.19 0.17 0.21 0.21 0.22 0.20
PG 0.39 0.54 0.48 0.45 0.38 0.35 0.51 0.44 0.60 0.56 0.50 0.45 0.43 0.44 0.52 0.51 0.52 0.43 0.49 0.41 0.48 0.46 0.51 0.27 0.25 0.30 0.28 0.30 0.27
PM 0.26 0.46 0.39 0.38 0.23 0.25 0.41 0.29 0.49 0.45 0.35 0.29 0.29 0.34 0.45 0.34 0.41 0.30 0.30 0.22 0.27 0.28 0.34 0.43 0.11 0.19 0.16 0.14 0.13
PPN 0.26 0.42 0.37 0.35 0.22 0.21 0.39 0.28 0.42 0.40 0.32 0.26 0.34 0.28 0.38 0.32 0.37 0.30 0.24 0.20 0.27 0.24 0.31 0.39 0.24 0.15 0.13 0.13 0.11
PWC 0.31 0.45 0.41 0.39 0.30 0.27 0.47 0.32 0.50 0.48 0.41 0.33 0.37 0.39 0.49 0.37 0.48 0.37 0.31 0.27 0.35 0.34 0.35 0.45 0.32 0.28 0.17 0.21 0.17
RR 0.28 0.43 0.40 0.37 0.25 0.22 0.39 0.33 0.49 0.45 0.36 0.29 0.27 0.29 0.45 0.38 0.41 0.27 0.29 0.26 0.27 0.30 0.32 0.38 0.28 0.24 0.31 0.21 0.16
WM 0.32 0.45 0.41 0.41 0.29 0.28 0.39 0.34 0.47 0.44 0.37 0.31 0.36 0.40 0.47 0.40 0.45 0.37 0.29 0.29 0.36 0.26 0.34 0.41 0.28 0.27 0.35 0.28 0.19
YT 0.30 0.44 0.37 0.41 0.29 0.22 0.40 0.31 0.44 0.42 0.35 0.31 0.32 0.31 0.40 0.32 0.40 0.29 0.30 0.18 0.30 0.29 0.34 0.43 0.26 0.25 0.31 0.29 0.32
Genetic distance relationships amongst the five Asian subpopulations were further explored using neighbor-joining dendograms with four non-Asian subpopulations for comparison (Fig. 3). The BSD, Chow Chow, Australian dingo and Akita clustered together in 90% of the trees. The BSD, Chow Chow and Australian dingo further clustered in 87% of the trees. The BSD and Australian dingo maintained their relationship within the larger cluster in 84% of the trees. In the remainder of the tree, the Rhodesian Ridgeback, Greyhound, Airedale Terrier and Borzoi maintained a relationship in 51% of the trees and the Airedale Terrier/Borzoi cluster was seen in 63% of the trees. The Pug did not maintain a relationship with any other breed in this analysis, but was intermediate to the Asian and non-Asian subpopulations.
Figure 3 a. Unrooted neighbor-joining dendogram showing the genetic relationships among 9 dog subpopulations based on DA genetic distance. b. Rooted neighbor-joining dendogram showing the genetic relationships among 9 dog subpopulations based on DA genetic distance. In both versions of the dendogram the Pug did not cluster with any population but is placed intermediate between the Asian and non-Asian subpopulations.
Discussion
Population diversity
Microsatellites have been previously used to assess genetic diversity and relationships in feral dog subpopulations [6,7]. Kim et al. [6] found that HO was high in three feral dog subpopulations of Korea, Sakhalin and Taiwan, ranging from 0.539 in the Taiwanese to 0.717 in the Korean dogs. Given that the loci used in that study had an average allele number of 7.75, these values are similar to the HO of 0.692 observed in the BSD. Wilton et al. [7] surveyed a population of Australian dingoes and found an average HO of 0.387 using microsatellites with an average allele number of 6.93, similar to the HO of 0.426 for the Australian dingoes reported herein with an average allele number of 11.8.
Given the size of the island of Bali, it is extraordinary that 800,000 feral dogs can thrive and maintain such high levels of genetic diversity. Of all the subpopulations surveyed here, the BSD has the highest number of observed alleles, the highest heterozygosity, the fewest number of loci out of Hardy-Weinberg equilibrium and the lowest FIS. Even after adjusting for sample size, the BSD maintains their status as the most heterogeneous population in the study. Unlike the Australian dingo which exhibits a much lower level of diversity, the BSD findings suggest either a large founding population on Bali and/or a consistent genetic influx since the geographic isolation of ~12,000 years ago. This data also supports that the BSD appears to approximate a randomly breeding population with little selection pressure.
When comparing the heterogeneity of the BSD to that observed within the AKC breeds some caveats should be addressed. One may initially expect long established, well-defined dog breeds to be much less heterogeneous than reported here. While some breeds do have a low HE, such as the Boxer with a HE of 0.320, breeds like the Jack Russell Terrier have a high HE of 0.713 and overall their HE is higher than that of the dingo. Of first note, the selection of the dogs that contribute to a breed composition mostly occurs prior to official breed recognition primarily by genetic drift due to geographic isolation and selection for particular working or physical characteristics. After official breed recognition future breeding choices are based primarily on the availability of sires and dams that approximate the breed standard. As a result, there is a founding population that proceeds to breed mostly by convenience. Also, many breed standards have changed considerably over the years resulting in retention of a certain level of diversity within each breed, some breeds retaining much more than others. Finally, dogs comprising the comparison AKC breeds were sampled from across the United States, removing any geographical bias of the genotypes observed and slightly elevating the heterozygosities.
Locus diversity
The average allelic diversity of the loci used in the present study was 11.8 alleles per locus, versus 7.75 in the Kim et al work [6]. However, the average number of alleles observed is 4.6 among the subpopulations in the present study and the average HT is 0.577. The average values for the 11 subpopulations surveyed in the Kim et al [6] work were 4.34 and 0.547, respectively. The higher total allelic diversity in the present study is likely due to the fact that nearly three times more subpopulations were studied.
RST and FST values were nearly identical across all subpopulations and all loci, indicating that approximately 23% of the differences observed in allele frequencies can be attributed to differences between subpopulations. FST provides an unbiased estimate of genetic drift between subpopulations by comparing alleles identical by state. RST takes advantage of the stepwise mutation model, which assumes that mutations most often occur as whole repeat unit losses or gains from the original allele size. As a result, the number of mutations provides an estimate of time from divergence. It is interesting, therefore, to compare RST and FST values by locus. Eighteen of the 31 loci studied have an RST to FST ratio greater than 1.1 (Table 1) indicating that the populations have been separated for a sufficient amount of time for mutations to impact genetic structure. An interesting exception is observed at CPH16 where the ratio is 0.420. CPH16 may have a mutation pattern where both stepwise additions and subtractions occur at equal and high frequency. Of note, the average pairwise RST value between the BSD and each of the 29 comparison subpopulations is 0.056 at locus CPH16. The highest RST to FST ratio occurs at locus CPH03 with a value of 1.724. Interestingly, the BSD and the Australian dingo have a pairwise RST value of 0.017 at CPH03, whereas the average value of the BSD compared to the other 28 subpopulations has a value of 0.254. The distance between the BSD and the Australian dingo at CPH03 may support that those two populations were isolated most recently from each other relative to the other 28 subpopulations.
Bali street dog origin
The origin of the people of Bali is clouded by myth and a scarcity of archeological findings. Therefore, the origin of the dog on Bali is also speculative. Nonetheless, a hypothesis can be formed based on known human and dog histories. Current evidence points to an early migration of humans from Africa through Indonesia and into Australia approximately 60,000 to 70,000 years ago [8,9]. Recent excavations have also revealed that there was a great expansion into Indonesia from China between 4,000 and 5,000 years ago that could have contributed to a population pre-existing on Bali [1]. Supportive evidence that Indonesia was populated prior to 5,000 years ago is a higher degree of heterogeneity in the Indonesian population than seen in the North Asian population, suggesting that the Indonesia was populated earlier than regions to the North [10]. The "Slow Boat Model" for the peopling of Polynesia also suggests a prolonged mixing of Southeast Asians with Indonesians, which predated migration to the East [11]. In short, Indonesia appears to be a human genetic melting pot with genetic influences over tens of thousands of years.
The dog on the island of Bali may also be a parallel "canine genetic melting pot." While the domestication date of the dog is in much dispute [12], approximately 14,000 years ago is accepted as a late date. During the earliest human migrations through Indonesia however, it is highly possible that wolf packs or feral dogs traveled the same routes, establishing a feral population on Bali in the process. Even if humans were not capable of taming the dog at that time, dogs could still have benefited from close proximity to humans. Figure 4 shows a superimposition of the proposed geographic origin for five Asian and four non-Asian dog subpopulations presented herein and the major theorized human migration routes. It is noteworthy that the BSD, Chow Chow and Australian dingo, related breeds by genetic analysis, all share one proposed human migration route.
Figure 4 Human migration patterns proposed in "Tracing the road down under" [8], a summary of the Modern human origins: Australian perspectives conference at the University of New South Wales, September 2003 with locations of origin for 5 Asian and 4 non-Asian dog subpopulations.
If a feral dog population was established on the island of Bali more than 14,000 years ago, then that population became isolated approximately 10,000 years ago when the sea levels drastically rose, submerging the land bridges of the Indonesia archipelago [13]. Geographic isolation was unlikely to have been absolute; genetic diversity of the BSD was invariably enhanced at various times by the influx of new dogs. At the time humans migrated to Indonesia from China, dogs were known to be domesticated and undoubtedly accompanied people as companions [17]. Mitochondrial DNA sequencing evidence suggests that the dingo was introduced into Australia about that time from the Indonesian archipelago [15,8,9]. Bali's documented history of repeated war and trade spanning the last 2,000 years [1,16,17] represents actions that are often associated with the introduction of new animals. Indeed, a somewhat free movement of dogs probably occurred up to 1926, when the import of dogs to Bali was greatly curtailed as a means to prevent the introduction of rabies [5]. This policy greatly reduced, though not eliminated, new outside introductions of new dogs to the island. In contrast to the Australian dingo population, which appears to have undergone a severe population bottleneck or founder effect based on microsatellite alleles and mtDNA [18], the BSD population maintains a high level of genetic variation. There is no evidence for a genetic bottleneck or small founding population for the BSD.
The relatedness of the BSD to the Australian dingo and the Chow Chow is evidenced by common unique alleles and allele frequencies despite the very different levels of genetic diversity between the subpopulations. According to the hypothesis presented herein, one could imagine that feral dog subpopulations were established throughout Indonesia with much mixing until ~12,000 years ago. At that time, each population became closed with little influx of new genetic material until humans migrated south from Asia between 4,000 and 5,000 years ago. The degree of influx since that period would have been influenced by the frequency of trade and conflict, factors determined by accessibility, available natural resources, and political structure. The island of Bali is historically a less visited island than it's neighbor Java and therefore the indigenous dog population would have been subjected to less influence.
Conclusions
This study into the diversity and relationship of the BSD to other domestic and feral dog populations shows the BSD to be highly diverse and related to populations of East Asian origin. These results indicate that a viable and diverse population of dogs existed on the island of Bali prior to its geographic isolation approximately 12,000 years ago and has been little influenced by domesticated European dogs since that time. It would be of interest to study feral subpopulations on other islands in the archipelago to determine if the same level of diversity is observed elsewhere, or if the situation on Bali is truly unique. Y-chromosome, mitochondrial and MHC marker typing on the BSD, as well as feral dogs from other regions, would help to determine if indeed dogs followed the same migration routes as their likely human companions.
Methods
Animal selection
BSDs were randomly captured and taken to a BSD Foundation field clinic for treatment or sterilization and simultaneously sampled for DNA collection with buccal swabs. Familial relationships of the BSDs sampled could not be easily determined; therefore the sample population was doubled (40 vs. 19–20 samples) over that of other study groups. Blood samples from the Australian dingo were taken from captive animals in Australia. Australian dingoes were known to be unrelated by at least one generation.
Dogs from 28 American Kennel Club (AKC) breeds, equally representing the AKC group designations, were sampled with buccal swabs for a previous study [19]. Twenty dogs were tested for each breed, with the exception of two breeds (Doberman Pinscher and the Border Collie) that comprised 19 individuals. The 28 breeds included were: Airedale Terrier, Akita, American Eskimo, Australian Shepherd, Belgian Tervuren, Bernese Mountain Dog, Border Collie, Borzoi, Boxer, Brittany, Bull Terrier, Bulldog, Chow Chow, Doberman Pinscher, Golden Retriever, Greyhound, Jack Russell Terrier, Keeshond, Labrador Retriever, Miniature Bull Terrier, Norwegian Elkhound, Papillon, Pembroke Welsh Corgi, Pomeranian, Pug, Rhodesian Ridgeback, Weimaraner, and Yorkshire Terrier. Dogs within each breed were unrelated by at least one generation.
Marker selection
Thirty-one of the 100 microsatellites multiplexed into 12 PCRs by the Veterinary Genetics Laboratory [20] had been previously used to evaluate the Australian dingo samples (unpublished data). For comparison purposes, those same 31 microsatellites were selected for use in the present study. All markers but one (PEZ02) were mapped on either the 1999 canine genetic linkage map [21] or the Radiation hybrid map [22]. Loci selected for study represented 25 of the 38 autosomes of the dog, with five autosomes represented by two loci. The average distance for the markers on chromosomes CFA06, CFA11, CFA20 and CFA23 is 23.5 cM and 23.4 Mb between AHT139 and RVC1 on CFA15. As a result, only 25 loci are known to be unlinked. PEZ02 has not been mapped and may be linked to a marker in the study.
Forward primers were synthesized and dye labeled with either Fam, Hex or Vic, or Tamra or Ned (Applied Biosystems, Inc. (ABI), Foster City, CA). Reverse primers were synthesized by Operon (Alameda, CA). Primer sequences and concentrations for all markers are available as Additional file 1.
Sample preparation and PCR conditions
BSD and AKC breed DNA was derived from buccal cells harvested from the inside of the cheek with nylon bristle cytology brushes (Medical Packaging Corp., Camarillo, CA). Samples were collected by owners or field volunteers and submitted directly to the laboratory. DNA was extracted by heating a single swab for 10 min at 95°C in 400 μl 50 mM NaOH and then neutralized with 140 μl 1 M Tris-HCl, pH 8.0. Australian dingo DNA was extracted from blood using a standard sodium hydroxide digest.
A 2 μl aliquot of extract was used in each PCR which equates to approximately 50 ng DNA. All markers and DNAs were amplified with a PCR reagent mix of 1X PCR buffer (ABI), 4.17 mM MgCl2, 200 μM of each dNTP (Hoffmann-La Roche Inc, Nutley, NJ), 0.6 unit AmpliTaq (ABI), and 2% DMSO (Sigma) then covered with 15 ul Chill-out™ Liquid Wax (MJ Research, Inc., Waltham, MA) to prevent evaporation. One of five thermal cycler programs was used for each primer mix ranging from 56° to 64° degrees for the annealing temperature. All PCR work was done in polycarbonate 96-well v-bottom microtiter plates (USA Scientific, Ocala, FL) on MJ Research PTC-100 thermal cyclers (MJ Research, Inc., Waltham, MA). Protocols are also available in Additional 1.
Gel electrophoresis conditions and DNA fragment analysis
One μl aliquots of PCR product were mixed with 2 μl Fluorescent Ladder (CXR) 60–400 (Promega 400) or Internal Lane Standard 600 (Promega 600) (Promega, Madison, WI) fluorescent size standard, denatured on MJ Research PTC-100 thermal cyclers for three minutes at 95°C, then held at 5°C or placed on ice for at least one minute before gel loading. Two μl aliquots were then loaded onto a 6% denaturing polyacrylamide gel and run on an ABI 377 Automated Sequencer using ABI 10" × 7 1/8" short plates (12 cm). Gels were run at 1.10 kV (constant) voltage, 60.0 mA current, 200 W power, 51°C and 40.0 mW (constant) laser power for up to 2 hours when using Promega 400, and up to 3 hours using Promega 600. DNA fragment analysis was performed with in-house designed STRand software [23], which replaces ABI Genotyper and Genescan software. This data was then transferred to an in-house database compatible with the STRand software.
Statistical analysis
Allelic diversity and observed heterozygosities (HO) were determined by direct counting for each of the 30 subpopulations. Hardy-Weinberg equilibrium (HWE) tests were performed using Genepop version 3.4 [24]. Pairwise FST estimates and subpopulation expected heterozygosities (HE) for the 30 breeds or dog groups were performed using Genepop version 3.4 [24]. FIS estimates (inbreeding coefficient of each subpopulation) for each allele following Weir and Cockerham [25] were calculated using Genepop version 3.4 and are presented as averages across all loci.
Gene diversity or total population heterozygosity (HT) and its associated parameters, HS (average heterozygosity among subpopulations) and GST (coefficient of genetic differentiation), were calculated across all loci using the public domain software, DISPAN [26]. Two additional measures of variance, FST [25] and RST [27,28] were calculated using Genepop version 3.4. A pairwise genetic distance matrix using Nei's DA distance was also created using DISPAN with bootstrapping. Genotype data for all populations is available in Additional file 2.
Phylogenetic tree construction
Allele frequencies were used to compute a matrix of genetic distances [29], which were then used to construct a phylogenetic tree of relationships among 5 Asian and 4 non-Asian dog subpopulations. Takezaki's [30] POPTREE program was used to create a neighbor joining tree using DA distances with 1000 bootstrap replications. The output of POPTREE was then converted to the New Hampshire format for editing in the stand alone program TREEVIEW version 1.6.6 [31] and bootstrap values were added.
Abbreviations
BSD: Bali Street Dog
FIS, FST RST, GST: F-statistics indices
HS, HT, HE, HO: Heterozygosity indices
HWE: Hardy-Weinberg equilibrium
NA: Number of alleles
AES: American Eskimo Dog
AS: Australian Shepherd
AT: Airedale Terrier
BCO: Border Collie
BLT: Bull Terrier
BMD: Bernese Mountain Dog
BS: Brittany Spaniel
BT: Belgian Tervuren
BU: Bulldog
BX: Boxer
BZ: Borzoi
Chow: Chow Chow
Dingo: Australian dingo
DP: Doberman Pinscher
GH: Greyhound
MBT: Miniature Bull Terrier
PG: Pug
RR: Rhodesian Ridgeback
GR: Golden Retriever
JRT: Jack Russell Terrier
KE: Keeshond
LR: Labrador Retriever
NE: Norwegian Elkhound
PG: Pug
PM: Pomeranian
PPN: Papillon
PWC: Pembroke Welsh Corgi
RR: Rhodesian Ridgeback
WM: Weimaraner
YT: Yorkshire Terrier
Authors' contributions
DNI performed the majority of data acquisition and analysis, wrote first draft of the manuscript and prepared the final draft for submission. ALS performed the majority of sample processing, assisted in data acquisition and the writing of the subsequent drafts of the manuscript as well as final draft preparation. SG sampled the dogs tested, provided background for the manuscript and assisted in the final draft preparation. ANW provided the Australian dingo data for comparison and assisted in the subsequent drafts of the manuscript. NCP directed the research and assisted in the writing of the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
"MS 2784776144318916 Supplement1.xls" and contains the primer sequences, expected size range, primer concentration used and annealing temperatures used. In a separate sheet within the same file the protocols for each PCR reaction are listed.
Click here for file
Additional File 2
"MS 2784776144318916 Supplement2.xls" and contains the individual genotype data for each animal used to derive the statistics and phylogenetic results presented herein.
Click here for file
Acknowledgements
We thank the Yayasan Yudisthira Swarga (Bali Street Dog Foundation) and director Sherry Grant for bringing this interesting dog population to our attention and their effort in sampling feral dogs.
==== Refs
Hobart A Ramseyer U Leemann A The peoples of Bali 1996 Oxford: Blackwell Publishers
Yayasan Yudisthira Swarga (Bali Street Dog Foundation)
Vishva Hindu Parishad
Buck W Mahabhrata 1973 Berkeley: University of California Press
Department of Foreign Affairs of the Republic of Indonesia State Gazette 1912, No 452 and Ministry of Agriculture Decree No 892/Kpts/The Nation56/1997
Kim KS Tanabe Y Park CK Ha JH Genetic variability in East Asian dogs using microsatellite loci analysis Journal of Heredity 2001 92 398 403 11773246 10.1093/jhered/92.5.398
Wilton AN Steward DJ Zafiris K Microsatellite variation in the Australian dingo Journal of Heredity 1999 90 108 11 9987915 10.1093/jhered/90.1.108
Dayton L Modern human origins meeting: "Tracing the road down under" Science 2003 302 555 Oct 24
Dayton L Modern human origins meeting: "On the trail of the first Australian dingo" Science 2003 302 555 6 Oct 24
Faradz SM Pattiiha MZ Leigh DA Jenkins M Leggo J Buckley MF Holden JJ Genetic diversity at the FMR1 locus in the Indonesian population Annals of Human Genetics 2000 64 329 39 11415517 10.1046/j.1469-1809.2000.6440329.x
Gibbons A The peopling of the Pacific Science 2001 291 1735 7 Mar 2 11249818 10.1126/science.291.5509.1735
Pennisi E Canine evolution: A shaggy dog history Science 2002 298 1540 2 Nov 22 12446885 10.1126/science.298.5598.1540
Hanebuth T Stattegger K Grootes PM Rapid flooding of the sunda shelf: A late-glacial sea-level record Science 2000 288 1033 5 May 12 10807570 10.1126/science.288.5468.1033
Savolainen P Zhang YP Luo J Lundeberg J Leitner T Genetic evidence for an East Asian origin of domestic dogs Science 2002 298 1610 3 Nov 22 12446907 10.1126/science.1073906
Savolainen P Leitner T Wilton AN Matisoo-Smith E Lundeberg J A detailed picture of the Origin of the Australian dingo, obtained from the study of Mitochondrial DNA Proceedings of the National Academy of Sciences of the USA 2004 101 12387 12390 15299143 10.1073/pnas.0401814101
Ardika IW Archaeological research in northeastern Bali, Indonesia PhD thesis 1991 Australian National University
Hartaningsih N Dharma DMN Rudiyanto MD Anjing Bali: Pemuliabiakan dan pelestarian (Bali Street Dogs: breeding and conservation) 1999 Yogyakarta: Kanisius Press
Wilton AN Dickman CR, Lunney D DNA Methods of Assessing Australian dingo Purity A Symposium on the Australian dingo 2001 Mosman NSW: Royal Zoological Society of New South Wales 49 55
Irion DN Schaffer AL Famula TR Eggleston ML Hughes SS Pedersen NC Analysis of genetic variation in 28 dog breed subpopulations with 100 microsatellite markers Journal of Heredity 2003 94 81 7 12692167 10.1093/jhered/esg004
Eggleston ML Irion DN Schaffer AL Hughes SS Draper JE Robertson KR Millon LV Pedersen NC PCR multiplexed microsatellite panels to expedite canine genetic disease linkage analysis Animal Biotechnology 2002 13 223 235 12517076 10.1081/ABIO-120016191
Neff MW Broman KW Mellersh CS Ray K Acland GM Aguirre GD Ziegle JS Ostrander EA Rine J A second-generation genetic linkage map of the domestic dog, Canis familiaris Genetics 1999 151 803 20 9927471
Guyon R Lorentzen TD Hitte C Kim L Cadieu E Parker HG Quignon P Lowe JK Renier C Gelfenbeyn B Vignaux F DeFrance HB Gloux S Mahairas GG Andre C Galibert F Ostrander EA A 1-Mb Resolution Radiation Hybrid Map of the Canine Genome Proceedings of the National Academy of Sciences of the USA 2003 100 5286 5291 12672950 10.1073/pnas.0831002100
STRand Nucleic Acid Analysis Software
Raymond M Rousset F GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism Journal of Heredity 1995 86 248 249
Weir BS Cockerham CC Estimating F-statistics for the analysis of population structure Evolution 1984 38 1358 1370
Ota T Program DISPAN: Genetic distance and phylogenetic analysis 1993 Copyright of the Pennsylvania State University, University Park, PA
Rousset F Equilibrium values of measure of population subdivision for stepwise mutation processes Genetics 1996 142 1357 1362 8846911
Michalakis Y Excoffier L A generic estimation of population subdivision using distances between alleles with special interest to microsatellite loci Genetics 1996 142 1061 1064 8849912
Nei M Molecular evolutionary genetics 1987 New York: Columbia University Press
Takezaki N POPTREE version for DOS
Page RDM TreeView version 1.6.6 Distributed by the Division of Environmental and Evolutionary Biology, Institute of Biomedical and Life Sciences 2001 University of Glasgow, Glasgow G12 8QQ, Scotland, UK
| 15701179 | PMC549630 | CC BY | 2021-01-04 16:38:18 | no | BMC Genet. 2005 Feb 8; 6:6 | utf-8 | BMC Genet | 2,005 | 10.1186/1471-2156-6-6 | oa_comm |
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-61570306710.1186/1471-2474-6-6Research ArticleIsometric force production parameters during normal and experimental low back pain conditions Descarreaux Martin [email protected] Jean-Sébastien [email protected] Normand [email protected] Faculté de Médecine, Division de Kinésiologie, U. Laval, Canada2 Département de Chiropratique, Université du Québec à Trois-Rivières, Canada2005 9 2 2005 6 6 6 5 8 2004 9 2 2005 Copyright © 2005 Descarreaux et al; licensee BioMed Central Ltd.2005Descarreaux 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 control of force and its between-trial variability are often taken as critical determinants of motor performance. Subjects performed isometric trunk flexion and extension forces without and with experiment pain to examine if pain yields changes in the control of trunk forces. The objective of this study is to determine if experimental low back pain modifies trunk isometric force production.
Methods
Ten control subjects participated in this study. They were required to exert 50 and 75% of their isometric maximal trunk flexion and extension torque. In a learning phase preceding the non painful and painful trials, visual and verbal feedbacks were provided. Then, subjects were asked to perform 10 trials without any feedback. Time to peak torque, time to peak torque variability, peak torque variability as well as constant and absolute error in peak torque were calculated. Time to peak and peak dF/dt were computed to determine if the first peak of dF/dt could predict the peak torque achieved.
Results
Absolute and constant errors were higher in the presence of a painful electrical stimulation. Furthermore, peak torque variability for the higher level of force was increased with in the presence of experimental pain. The linear regressions between peak dF/dt, time to peak dF/dt and peak torque were similar for both conditions. Experimental low back pain yielded increased absolute and constant errors as well as a greater peak torque variability for the higher levels of force. The control strategy, however, remained the same between the non painful and painful condition. Cutaneous pain affects some isometric force production parameters but modifications of motor control strategies are not implemented spontaneously.
Conclusions
It is hypothesized that adaptation of motor strategies to low back pain is implemented gradually over time. This would enable LBP patients to perform their daily tasks with presumably less pain and more accuracy.
==== Body
Background
There are many possible explanations for the origin and consequences of acute and chronic low back pain but the transition from acute to chronic low back conditions needs to be clarified [1,2]. There is a number of recent evidences suggesting that chronic LBP patients exhibit deficits in proprioception and trunk motor control. For example, changes in postural control[3], delayed muscle responses to sudden trunk loading[4], increased trunk movement detection threshold[5] and increased repositioning errors in patients with LBP have all been reported[6,7].
For example, Oddsson et al.[8] used spectral parameters of the surface electromyographic (EMG) signal from lumbar back muscles assessed during a fatiguing isometric contraction to classify LBP and healthy subjects. They observed more activation imbalances in chronic LBP subjects and proposed that these changes would eventually become «normal» behavior for the chronic LBP individuals[8,9]. They suggested that changes in the trunk muscular activity could result from subtle postural adjustments that were developed during the acute phase to avoid pain. These observed changes imply a certain level of adaptation to the initial acute pain. Motor control in chronic subjects can be influenced by the presence of chronic pain but also by other phenomenon like type II fiber atrophy, degenerative changes and decreased trunk muscle force and endurance[10].
Clinical studies of chronic LBP patients involve heterogeneous populations and the effect of chronic pain cannot be differentiated from other degenerative and functional changes occurring in the lumbar spine. Experimentally induced LBP eliminates some of these uncertainties and could allow an examination of the effect of pain per se[11]. To examine the effect of experimental pain on the sensori-motor control of the lumbar spine, two different protocols have been used in the past: (1) lumbar cutaneous pain induced by electrical or mechanical stimulation[11] and (2) deep lumbar pain induced by saline injection of lumbar muscles[11,12]. Zedka et al.[11] noted increased stretch reflex responses in the presence of cutaneous electrical and mechanical stimulations. They also noted an increase in EMG amplitude (during extension) as well as several changes in the motor patterns (trunk velocity and range of motion) in presence of muscle pain induced by a saline injection. Hodges and is colleagues recently demonstrated that feedforward recruitment of trunk muscles is altered in presence of experimental and clinical LBP[13,14]. In a postural task where the subjects were asked to rapidly flex the upper limb, they noted delayed transversus abdominis muscles activation in the presence of experimental pain[14] or chronic low back pain[13]. They concluded that some changes in motor control that occur in LBP patients (experimental or chronic) may be caused by pain. These modifications do not resolve spontaneously with alleviation of symptoms since the effect was still observed after a 10-min delay.
In a previous study, we have observed that two different control strategies were used by chronic LBP subjects to produce accurate trunk isometric forces[15]. One subgroup of LBP subjects used an open-loop control strategy similar to that used by healthy control subjects whereas a second subgroup of subjects used a less open-loop control strategy characterized in part by a longer time to peak force. Both LBP subgroups, however, were able to produce isometric trunk forces as accurately as the healthy subjects.
The aim of the present study was to evaluate if a painful stimulation, induced by cutaneous electrical stimulation, would spontaneously yield a change in the control strategy or the variability of trunk isometric force production. To explore whether isometric trunk forces are similarly programmed with and without experimental pain, we used the model proposed by Gordon and Ghez[16]. If motor planning is affected by lumbar cutaneous pain, a more closed-loop mode of control characterized by an increased time-to-peak force and a lack of relationship between the peak of dF/dt and the peak force should be observed. On the other hand, the absence of such a change in the mode of control could yield a more variable force production resulting from sensory and motor effects of pain on the motor response.
Methods
Force production parameters were measured in 10 healthy subjects with no history of chronic or recurrent LBP (10 men, age: 25.9 years). Each subject gave their written inform consent and the study was approved by the local ethics committee. All subjects were university students. Force data (torque) were obtained from an isometric testing apparatus (Loredan Biomedical, West Sacramento, USA) and recorded at a sampling rate of 500 Hz. Torque data were digitally filtered with a seventh-order dual pass Butterworth filter (7 Hz low pass cut-off frequency). The first time derivative of torque was calculated using a finite difference algorithm (window 25 ms).
Superficial pain was elicited by electrical stimulation of the skin over the spinous process of L3 using bipolar surface Ag-AgCl electrodes (Beckman electrodes, 1 cm diameter). This site of stimulation was chosen to ensure that no direct muscular activation could result from the electrical stimulus. The range of voltage used during our experiment was 135–140 Volts. This stimulation created a focal cutaneous painful stimulus with very limited current spread. The technique for inducing the pain stimulation was inspired from the work of Arendt-Nielsen et al.[17]. Each stimulus consisted of a standard 3-second constant-voltage pulse train of 1-ms pulses delivered at 10 Hz (ISI = 100 ms; S88 Grass stimulator with SIU8T constant voltage isolation unit, USA). The amplitude used for the stimulation was determined when the subjects quoted the pain intensity of the stimulation between 7.5 and 8.5 on a 0–10 scale. The intensity of pain was monitored throughout the experiment and adjusted accordingly to ensure a constant pain level.
Testing was done in a neutral standing posture (no trunk flexion or extension). First, maximal isometric flexion and extension torques of trunk muscles were collected. The higher torque value obtained in three consecutive 4-second trials was used as the reference for maximal voluntary contraction. Then, four experimental conditions of trunk torque production were evaluated without and with experimental pain: 50 and 75% of the maximal isometric torque in both extension and flexion. Conditions were presented by block with the order of presentation being randomized across subjects. For each condition, trials without pain were presented first followed by the trials with pain. For each trial, subjects were instructed to produce a trunk isometric force as quickly as possible following an auditory signal. They were encouraged to produce a single impulse ("shoot and release") and to make no attempt at correcting the force once the contraction was initiated. For each condition, a learning phase was provided. During this phase, after each trial, subjects were given visual accuracy feedback through an oscilloscope located in front of them. Subjects were specifically asked to produce peak torques that were within 10% of the goal target. This learning phase stopped when five consecutive contractions were within the 10% margin. Following these learning trials, subjects performed 10 consecutive trials without any visual feedback. The pain condition followed. A second learning sequence with feedbacks and without pain was given to the subjects. This procedure was used to insure that no differences between the control and pain conditions would reflect a pain effect and not a loss of calibration after a block of trials without pain. Hence, if any differences between the control and pain conditions were found, these would reflect a pain effect and not a learning effect. For the pain trials, the stimulation was initiated 0.5 sec before an auditory tone indicating the subjects to initiate the contraction. All dependent variables were derived from the behavior observed for the 10 trials without feedback without and with the experimental pain.
For each experimental trial, the onset of torque and peak torque were determined. Using this information, time to peak torque, time to peak torque variability, peak torque variability as well as the constant and absolute error in peak torque were calculated for each condition. Constant error represents the positive or negative difference between the peak torque reached and the target torque. Absolute error in peak torque represents the positive difference between the reached peak torque and the goal peak torque whereas time to peak represents the period of time between the beginning of rising torque and the maximal torque obtained in the trial. Peak dF/dt was also computed to examine if the first peak of dF/dt could predict the peak torque achieved. Linear regressions were calculated for each subject and a high r2, indicating that the first peak of dF/dt could predict the peak torque, was taken as an indication of a preprogrammed or open-loop mode of control[16].
Results
On average, the maximal voluntary contraction in flexion and extension were 236.2 Nm and 346.5 Nm, respectively. Table 1 presents a summary of the statistical analyses for all dependant variables. The ANOVA for absolute errors yielded a main effect of Pain. Absolute errors were higher in the painful condition than in the normal condition (30.7 Nm vs 23.9 Nm respectively; F1-9 = 8.29, p = 0.018). The main effects of Direction, Force level and all interactions were not significant (ps > 0.05). Similar observations were made for the constant errors as the ANOVA yielded a main effect of Pain (F1-9 = 6.22, p = 0.035). The main effects of Direction, Force level and all interactions also were not significant (ps > 0.05). The painful stimulus yielded increased constant and absolute errors indicating that subjects, on average, overshot the target by 25.9 Nm (13.9 Nm for the normal condition).
Table 1 Statistical analyses for all dependant variables.
Pain (P) Direction (D) Force Level (F) P × D D × F P × F
Time to peak force F = 0.017
p = 0.901 F = 1.094
p = 0.323 F = 0.541
p = 0.481 F = 1.011
p = 0.341 F = 2.009
p = 0.190 F = 2.628
p = 0.139
Time to peak force variability F = 8.763
p = 0.016* F = 0.083
p = 0.779 F = 0.005
p = 0.943 F = 0.115
p = 0.742 F = 3.904
p = 0.080 F = 0.028
p = 0.870
Peak force variability F = 7.756
p = 0.021* F = 1.183
p = 0.209 F = 0.487
p = 0.503 F = 0.032
p = 0.861 F = 0.096
p = 0.764 F = 8.047
p = 0.020*
Constant error F = 6.222
p = 0.034* F = 0.530
p = 0.485 F = 5.02
p = 0.052 F = 5.018
p = 0.053 F = 0.273
p = 0.614 F = 0.027
p = 0.873
Absolute error F = 8.289
p = 0.018* F = 1.1
p = 0.482 F = 0.537
p = 0.482 F = 0.909
p = 0.365 F = 2.222
p = 0.170 F = 1.158
p = 0.310
*p < 0.05
Figure 1 illustrates, for one subject, the mean and variability of ten consecutive flexion trials (50% of maximal flexion torque) without and with pain. With pain, the torque-time curves exhibit greater variability around the peak.
Figure 1 Typical torque-time curves illustrating the mean (SD is represented by the dashed line) of ten consecutive flexion trials (without feedback) in the control condition. (b) Typical torque-time curves illustrating the mean (SD is represented by the dashed line) of ten consecutive flexion trials (without feedback) in the experimental pain condition.
Figure 2 illustrates peak torque variability with and without experimental pain for both levels of force. The ANOVA for peak torque variability showed a significant main effect of Pain (F 1,9 = 7.76, p = 0.021) and an interaction of Pain × Force (F 1,9 = 8.05, p = 0.020) but no main effect of Force (ps > 0.05). A decomposition of the interaction showed that peak torque variability increased with the painful stimulation only for the higher level of force. For the higher level of force, the variability was 12.6 Nm in the control condition and 18.6 Nm in the presence of electrical stimulation (Tukey: p = 0.027). For the lower level of force, peak torque variability was similar for both conditions (15.0 Nm; p > 0.05).
Figure 2 Mean (SD) peak torque variability with and without electrical stimulation for both levels of force.
The average time to peak torque was not affected by the lumbar electrical stimulation. On average, the time to peak torque was 240 ms. The main effects of Pain, Direction, Force level and all interactions were not significant (ps > 0.05). The time to peak torque variability, however, was increased in the painful condition than in the normal condition (81 ms vs 58 ms respectively; F1-9 = 8.76, p = 0.016). Again, the main effects of Direction, Force level and all interactions were not significant (ps > 0.05).
The dF/dt curves for both conditions are characterized by a single peak in the first phase of isometric force production. On average, peak dF/dt explained 73.5 and 74.3 percent of the variance observed in peak torque for the normal and the painful conditions, respectively (p > 0.05). This suggests a similar control strategy for both the normal and the painful conditions.
Discussion
The presence of an experimental cutaneous lumbar pain altered the production of isometric trunk forces in various ways. Specifically, when exposed to the painful cutaneous electrical stimulation, subjects showed greater absolute and constant errors in isometric trunk torque production. The effects of a lumbar cutaneous painful stimulation on the isometric trunk force production yielded an overestimation of the learned level of force to be performed in both flexion and extension. This observation argues against a specific modification of the trunk flexor or extensor motoneuronal pool following the painful stimulation. Rather it appears that, independently of the direction of the force production required, the pain stimulation yielded an increased excitability of the agonist motor pathways. Our results, however, cannot discriminate if these modifications occurred at the programming stage or at the execution stage (upper and lower motorneurons). All of these mechanisms have previously been suggested to explain the modifications induced by experimental pain[11,14,18]. Across the painful trials, subjects also showed greater torque variability for the higher level of forces. Time to peak torque variability was also greater with than without pain. Linear regressions between peak dF/dt and peak torque however, were similar in both conditions indicating that the subjects used a similar strategy force control strategy without and with pain. In a previous study, chronic low back pain patients demonstrated longer time to peak values suggesting a shift from an open loop strategy of control to a more close loop strategy of control[15]. Results of the present experiment suggest that, in the presence of experimental cutaneous pain, subjects maintained an open loop control strategy to perform the task. Such an absence of modification in the control strategy could be specifically related to the task used in the present experiment. Subjects were explicitly told to perform the isometric force production as quickly and accurately as possible.
Numerous authors have showed that chronic LBP patients exhibit deficits in proprioception and trunk motor control [3-7]. The link between persistent pain and subsequent adaptations to low back symptoms remains unclear although some hypotheses have been formulated. Lund et al. first proposed a pain adaptation model that could occur in the presence of persistent pain[19]. This adaptation is characterized by an increased motoneuron output when the muscle is acting as an antagonist and by a decreased motoneuron output when the muscle is acting as an agonist. Luoto et al., throughout a series of study, have shown that motor control deficits observed in chronic LBP subjects can be, at least in part, due to impairment in central processing. According to the authors, pain would consist of an irrelevant sensory input that cannot be ignored but that is hampering central processing[20,21]. Even though pain probably causes peripheral adaptations, central impairment must also be considered. Oddsson and his colleagues[9] suggested that chronic low back patients, for whom acute pain reactions are no longer present, could develop a new strategy (postural adjustments) to avoid the sensation of pain. Although experimentally induced pain is different from clinical pain, some authors reported motor changes similar to those observed in LBP patients when inducing experimental pain in control subjects[14]. Hodges et al. observed that experimental and recurrent low back pain induced similar delays in transversus abdominis activation[4,13,14]. Zedka et al[11] observed a decrease in velocity and range of trunk motion after saline injection of lumbar muscles similar to those observed in patients with low back pain. They also demonstrated an increased excitability in the long latency lumbar response after a painful cutaneous electrical stimulation. These changes were attributed to the interactions between nociceptives afferents and motor neuron pool excitability[17]. Overall, it appears that both clinical and experimental low back pain can influence trunk muscle activations suggesting that the sole presence of pain is detrimental to motor performance.
The painful stimulus used in the present experiment consisted of trains of 1 ms electrical pulses applied to the spinous process of the third lumbar vertebra. Although the experimenter could not observe nor palpate any muscular contractions and the subjects did not report any involuntary muscular contractions, the possibility exists that spreading of the current over the surrounding tissues activated the neuromuscular junction of the nearby lumbar paraspinal muscles. Consequently, imperceptible muscular contractions could have occurred, particularly at the high level of Voltage used to elicit the perception of pain in our subjects. Muscular activation of the paraspinal muscles due to the current spreading while performing an isometric force reproduction task could certainly lead to a deterioration of the subjects' performance (both the accuracy and variability) under the experimental pain condition. However, the overestimation of the learned level of force was observed in both flexion and extension – whereas the painful stimulation was always applied to the spinous process (presumably biasing only the paraspinal muscles) – argues against this factor playing a critical role in the findings reported in the present manuscript. Further work, however, is needed to quantify the performance of healthy subjects performing an isometric force reproduction task using a different experimental pain protocol (e.g. muscle saline injection).
In a previous study, we observed that some chronic LBP patients, compared to healthy control subjects, adopted a more close loop control strategy of trunk isometric force production to maintain a particular level of performance[15]. Both experiments used a similar protocol of isometric force production but the present results failed to reveal any changes in the control strategy adopted by the subjects. On the other hand, in the presence of experimental cutaneous pain, a less accurate isometric force production was observed. Therefore our results suggest that, even if some modifications occurred directly in the presence of pain[11,14], adaptation to low back pain and modifications of motor control strategies are not implemented spontaneously. It seems that the modification in control strategy observed for chronic LBP subjects could be an adaptation to limit the variability of force production. For control subjects, the "rise time regulation" strategy or variations thereof have been suggested to help in reducing response variability[16,22,23]. Also, it has been suggested that, in the presence of persistent experimental or chronic low back pain, subjects need to adapt their motor control strategies in order to limit exacerbation of pain symptoms[9,19]. Whether chronic LBP subjects adopt a new control strategy to limit their pain symptoms or to minimize their force production errors remains to be determined.
Conclusion
The present data indicate that trunk isometric force production can be affected by experimental cutaneous low back pain. While the motor control strategy remained the same between the non painful and painful condition, subjects showed less accuracy and more variability in the painful condition. Experimental cutaneous low back pain is different from deep tissue pain and the observed changes. This precludes any generalization to acute low back pain. It is hypothesized, however, that adaptation of motor strategies to low back pain is implemented gradually over time. This would enable LBP patients to perform their daily tasks with presumably less pain and more accuracy.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MD and JS performed all the testing and data analyses. NT acted as the thesis director of MD and JS and participated in the design and coordination of the study. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgment
MD was supported by FCQ and FCAR and JSB by CIHR-FCQ.
==== Refs
Dwyer AP Backache and its prevention Clin Orthop 1987 35 43 2957137
Fransen M Woodward M Norton R Coggan C Dawe M Sheridan N Risk factors associated with the transition from acute to chronic occupational back pain Spine 2002 27 92 98 11805644 10.1097/00007632-200201010-00022
Mientjes MI Frank JS Balance in chronic low back pain patients compared to healthy people under various conditions in upright standing Clin Biomech (Bristol, Avon) 1999 14 710 716 10545625 10.1016/S0268-0033(99)00025-X
Hodges P Richardson C Jull G Evaluation of the relationship between laboratory and clinical tests of transversus abdominis function Physiother Res Int 1996 1 30 40 9238721
Taimela S Kankaanpaa M Luoto S The effect of lumbar fatigue on the ability to sense a change in lumbar position. A controlled study Spine 1999 24 1322 1327 10404574 10.1097/00007632-199907010-00009
Newcomer KL Laskowski ER Yu B Johnson JC An KN Differences in repositioning error among patients with low back pain compared with control subjects Spine 2000 25 2488 2493 11013501 10.1097/00007632-200010010-00011
Newcomer K Laskowski ER Yu B Larson DR An KN Repositioning error in low back pain. Comparing trunk repositioning error in subjects with chronic low back pain and control subjects Spine 2000 25 245 250 10685490 10.1097/00007632-200001150-00017
Oddsson LI De Luca CJ Activation imbalances in lumbar spine muscles in the presence of chronic low back pain J Appl Physiol 2003 94 1410 1420 12482760
Oddsson LI Giphart JE Buijs RJ Roy SH Taylor HP De Luca CJ Development of new protocols and analysis procedures for the assessment of LBP by surface EMG techniques J Rehabil Res Dev 1997 34 415 426 9323645
Ebenbichler GR Oddsson LI Kollmitzer J Erim Z Sensory-motor control of the lower back: implications for rehabilitation Med Sci Sports Exerc 2001 33 1889 1898 11689740
Zedka M Prochazka A Knight B Gillard D Gauthier M Voluntary and reflex control of human back muscles during induced pain J Physiol 1999 520 Pt 2 591 604 10523425 10.1111/j.1469-7793.1999.00591.x
Moe-Nilssen R Ljunggren AE Torebjork E Dynamic adjustments of walking behavior dependent on noxious input in experimental low back pain Pain 1999 83 477 485 10568856 10.1016/S0304-3959(99)00153-0
Hodges PW Changes in motor planning of feedforward postural responses of the trunk muscles in low back pain Exp Brain Res 2001 141 261 266 11713638 10.1007/s002210100873
Hodges PW Moseley GL Gabrielsson A Gandevia SC Experimental muscle pain changes feedforward postural responses of the trunk muscles Exp Brain Res 2003 151 262 271 12783146 10.1007/s00221-003-1457-x
Descarreaux M Blouin JS Teasdale N Force production parameters in patients with low back pain and healthy control study participants Spine 2004 29 311 317 14752355 10.1097/01.BRS.0000105983.19980.A8
Gordon J Ghez C Trajectory control in targeted force impulses. II. Pulse height control Exp Brain Res 1987 67 241 252 3622687
Arendt-Nielsen L Graven-Nielsen T Svarrer H Svensson P The influence of low back pain on muscle activity and coordination during gait: a clinical and experimental study Pain 1996 64 231 240 8740599 10.1016/0304-3959(95)00115-8
Le Pera D Graven-Nielsen T Valeriani M Oliviero A Di Lazzaro V Tonali PA Arendt-Nielsen L Inhibition of motor system excitability at cortical and spinal level by tonic muscle pain Clin Neurophysiol 2001 112 1633 1641 11514246 10.1016/S1388-2457(01)00631-9
Lund JP Donga R Widmer CG Stohler CS The pain-adaptation model: a discussion of the relationship between chronic musculoskeletal pain and motor activity Can J Physiol Pharmacol 1991 69 683 694 1863921
Luoto S Aalto H Taimela S Hurri H Pyykko I Alaranta H One-footed and externally disturbed two-footed postural control in patients with chronic low back pain and healthy control subjects. A controlled study with follow-up Spine 1998 23 2081 9; discussion 2089-90 9794052 10.1097/00007632-199810010-00008
Luoto S Taimela S Hurri H Alaranta H Mechanisms explaining the association between low back trouble and deficits in information processing. A controlled study with follow-up Spine 1999 24 255 261 10025020 10.1097/00007632-199902010-00011
Gordon J Ghez C Trajectory control in targeted force impulses. III. Compensatory adjustments for initial errors Exp Brain Res 1987 67 253 269 3622688
Newell KM Carlton LG Hancock PA Kinetic analysis of response variability Psychol Bull 1984 96 133 151 10.1037//0033-2909.96.1.133
| 15703067 | PMC549631 | CC BY | 2021-01-04 16:32:05 | no | BMC Musculoskelet Disord. 2005 Feb 9; 6:6 | utf-8 | BMC Musculoskelet Disord | 2,005 | 10.1186/1471-2474-6-6 | oa_comm |
==== Front
J BiolJournal of Biology1478-58541475-4924BioMed Central London jbiol101534503610.1186/jbiol10Research ArticleThe phosphatidylserine receptor has essential functions during embryogenesis but not in apoptotic cell removal Böse Jens [email protected] Achim D [email protected] Laura [email protected] Stefanie [email protected] Ivonne [email protected] Martin [email protected] Marianne [email protected]öntgen Frank [email protected] Andreas [email protected] Junior Research Group Infection Genetics, German Research Center for Biotechnology (GBF), Mascheroder Weg 1, 38124 Braunschweig, Germany2 Department of Pathology, School of Veterinary Medicine Hannover, Bünteweg 17, 30559 Hannover, Germany3 Department of Experimental Immunology, German Research Center for Biotechnology (GBF), Mascheroder Weg 1, 38124 Braunschweig, Germany4 Ozgene Pty. Ltd., Canning Vale, WA 6970, Australia2004 23 8 2004 3 4 15 15 14 5 2004 16 7 2004 21 7 2004 Copyright © 2004 Böse 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
Phagocytosis of apoptotic cells is fundamental to animal development, immune function and cellular homeostasis. The phosphatidylserine receptor (Ptdsr) on phagocytes has been implicated in the recognition and engulfment of apoptotic cells and in anti-inflammatory signaling. To determine the biological function of the phosphatidylserine receptor in vivo, we inactivated the Ptdsr gene in the mouse.
Results
Ablation of Ptdsr function in mice causes perinatal lethality, growth retardation and a delay in terminal differentiation of the kidney, intestine, liver and lungs during embryogenesis. Moreover, eye development can be severely disturbed, ranging from defects in retinal differentiation to complete unilateral or bilateral absence of eyes. Ptdsr -/- mice with anophthalmia develop novel lesions, with induction of ectopic retinal-pigmented epithelium in nasal cavities. A comprehensive investigation of apoptotic cell clearance in vivo and in vitro demonstrated that engulfment of apoptotic cells was normal in Ptdsr knockout mice, but Ptdsr-deficient macrophages were impaired in pro- and anti-inflammatory cytokine signaling after stimulation with apoptotic cells or with lipopolysaccharide.
Conclusion
Ptdsr is essential for the development and differentiation of multiple organs during embryogenesis but not for apoptotic cell removal. Ptdsr may thus have a novel, unexpected developmental function as an important differentiation-promoting gene. Moreover, Ptdsr is not required for apoptotic cell clearance by macrophages but seems to be necessary for the regulation of macrophage cytokine responses. These results clearly contradict the current view that the phosphatidylserine receptor primarily functions in apoptotic cell clearance.
==== Body
Background
Programmed cell death, or apoptosis, is required for the normal development of almost all multicellular organisms and is a physiological mechanism for controlling cell number; as a result, structures that are no longer needed are deleted during development and abnormal cells are eliminated [1,2]. Most of the cells produced during mammalian embryonic development undergo physiological cell death before the end of the perinatal period [3]. Apoptotic cells are removed rapidly and efficiently as intact cells or apoptotic bodies by professional phagocytes or by neighboring cells. This highly regulated process prevents the release of potentially noxious or immunogenic intracellular materials and constitutes the fate of most dying cells throughout the lifespan of an organism [4,5]. Phagocytosis of apoptotic cells is very distinct from other engulfment processes that result, for example, in the clearance of microorganisms, because engulfment of apoptotic cells triggers the secretion of potent anti-inflammatory and immunosuppressive mediators, whereas pathogen recognition causes the release of pro-inflammatory signals [6].
Almost all cell types can recognize, respond to, and ingest apoptotic cells by using specific sets of phagocytic receptors that bind to specific ligands on apoptotic cells. Detailed genetic studies in Drosophila and Caenorhabditis elegans have recently yielded evidence that basic phagocytic mechanisms and pathways for the recognition and engulfment of apoptotic cells are highly conserved throughout phylogeny [7,8]. In vertebrates, a number of receptors have been identified that can mediate phagocytosis of apoptotic cells. These include, for example, scavenger receptors and pattern recognition receptors such as CD36, SR-A and CD14, integrins such as the vitronectin receptor αvβ3, and members of the collectin family and their receptors CD91 and calreticulin [9-13]. The individual roles of these molecules in binding, phagocytosis or transduction of anti-inflammatory signals upon apoptotic cell recognition have not been well defined, however [5,6,14]. The importance of efficient mechanisms for apoptotic cell clearance in vivo is supported by the observation that autoimmune responses can be provoked in mice when key molecules for apoptotic cell recognition and uptake are missing. This has been reported for knockout mice lacking the complement protein C1q [15], for mice with a mutation in the tyrosine kinase receptor gene Mer [16] and, more recently, in mice lacking transglutaminase 2 or milk fat globule epidermal growth factor 8 (MFG-E8) [17,18].
The exposure of the phospholipid phosphatidylserine (PS) in the outer leaflet of the plasma membrane of apoptotic cells has been described as one of the hallmarks of the induction of apoptosis and is considered to be one of the most important signals required for apoptotic cell recognition and removal [19]. A number of cell-surface and bridging molecules can interact with exposed PS on apoptotic cells. These include the serum proteins β2-glycoprotein 1 and protein S [20,21], the growth-arrest-specific gene product GAS-6 [22], complement activation products [23], the milk fat globule protein MFG-E8 [24], and annexin I [25]. In most cases the receptors on phagocytes that recognize these PS-bridging molecules have not been defined, but it has been reported that GAS-6 is a ligand for the tyrosine kinase receptor Mer and that MFG-E8 can bind to the vitronectin receptor αvβ3 [16,24]. Other molecules that bind PS with varying specificity are the lectin-like oxidized low-density lipoprotein receptor-1 (LOX-1) and the scavenger receptors CD36 and CD68 (for review see [5] and references therein).
The best-characterized molecule so far that binds PS in a stereo-specific manner is the phosphatidylserine receptor (Ptdsr) [26]. In vitro, it has been shown that the Ptdsr can mediate the uptake of apoptotic cells and that such Ptdsr-mediated phagocytosis can be inhibited through addition of PS liposomes, the PS-binding molecule annexin V or an anti-Ptdsr antibody [26]. Moreover, the binding of Ptdsr to PS on apoptotic cells has been reported to be important for the release of anti-inflammatory mediators, including transforming growth factor-β1 (TGF-β1), platelet-activating factor (PAF), and prostaglandin E2 [26,27]. These data supported the hypothesis that Ptdsr fulfils a role as a crucial signaling switch after the engagement of macrophages with apoptotic cells and is thereby fundamental for preventing local immune responses to apoptotic cells before their clearance [28].
Very recently, Ptdsr has been found in the cell nucleus. Its nuclear localization is mediated by five independent nuclear localization signals, each of which alone is capable of targeting Ptdsr to the cell nucleus [29]. Moreover, an additional study performed recently in Hydra showed an exclusively nuclear localization for the Ptdsr protein [30]. Most interestingly, the nuclear localization of Ptdsr in Hydra epithelial cells did not change upon phagocytosis of apoptotic cells. These reports challenge the original hypothesis, according to which Ptdsr is an exclusively transmembrane receptor for apoptotic cell recognition and anti-inflammatory signaling.
To examine further the role of Ptdsr in vivo, we performed gene-expression and gene-targeting studies in mice. A perinatally lethal phenotype was observed in Ptdsr-knockout mice, and Ptdsr-deficient embryos displayed multiple defects in tissue and organ differentiation. While this work was in progress, both Li et al. [31] and Kunisaki et al. [32] also reported the generation and phenotypic characterization of Ptdsr-knockout mice. Of note, although some of their results were confirmed in our study, we found a fundamentally different phenotype with regard to clearance of apoptotic cells. Moreover, our study revealed marked and unexpected findings in Ptdsr-deficient mice that are not related to apoptosis.
Results
Generation of Ptdsr-deficient mice
To investigate in vivo the functions of the phosphatidylserine receptor Ptdsr, we generated a null allele in the mouse by gene targeting (Figure 1a,1b,1c). In contrast to previously described Ptdsr-knockout mice [31,32], we used Bruce4 embryonic stem (ES) cells for gene targeting [33], thus generating a Ptdsr-null allele in a pure, isogenic C57BL/6J genetic background. The newly established knockout mouse line was named Ptdsrtm1Gbf (hereafter referred to as Ptdsr -/-). Heterozygous Ptdsr+/- mice were viable and fertile and showed no obvious abnormalities. Ptdsr +/- mice were intercrossed to generate homozygous Ptdsr-deficient mice. The absence of Ptdsr expression in Ptdsr -/- embryos was confirmed by RT-PCR (data not shown), and by northern and western blotting analyses (Figure 1d,e). Interbreeding of heterozygous mice showed that the mutation was lethal, since homozygous mutants were not detected in over 100 analyzed litters at weaning. To determine the stages of embryonic development affected by the Ptdsrtm1Gbf mutation, timed breedings were followed by PCR genotyping (Figure 1c) of embryos. We recovered fewer than the expected number of homozygous embryos from intercrosses of Ptdsr+/-mice. From a total of 1,031 embryos analyzed between gestational day (E) 9.5 and E18.5, 198 (19.2%) Ptdsr-deficient homozygous embryos were harvested, indicating that the introduced mutation is associated with a low rate of embryonic lethality in utero.
From E9.5 to E12.5, Ptdsr -/- embryos were viable and of normal size. At E13.5 and thereafter, however, most Ptdsr -/-embryos showed morphological abnormalities (Table 1). All homozygous embryos harvested were growth-retarded from E13.5 onwards, had a pale appearance, and displayed multiple developmental dysmorphologies. These included various head and craniofacial malformations, such as exencephaly, cleft palate and abnormal head shape (Figure 1f,g). Gross inspection revealed that eye development was severely affected in 14.1% of homozygous embryos. The affected animals displayed a complete unilateral or bilateral absence of the eyes (Table 1) that was never detected in Ptdsr +/+ or Ptdsr +/- littermates. Furthermore, homozygous embryos harvested between E12.5 and E15.5 had subcutaneous edema (Figure 1f,g). Because we were able to recover Ptdsr -/- embryos until E18.5, we investigated whether Ptdsr-knockout mice could be born alive. Careful observation of timed matings allowed us to recover Ptdsr -/- neonates, but homozygous pups died during delivery or within minutes after birth. Ptdsr-deficient neonates were also growth-retarded, had a pale appearance and displayed various malformations. These included cleft palate, abnormal head shape, absence of eyes and edematous skin (Figure 1h). Thus, deletion of the Ptdsr gene resulted in perinatal lethality with variable severity and penetrance of phenotypes.
Expression of Ptdsr during embryogenesis and in adult tissues
The observed perinatal lethality indicates that Ptdsr plays an important role during development. Analysis by RT-PCR (data not shown) showed that Ptdsr is expressed early in development, because we were able to detect Ptdsr transcripts in ES cells and embryos at all developmental stages. To analyze in more detail the temporal and spatial expression patterns of Ptdsr, and to correlate expression patterns with observed pathological malformations, we made use of a Ptdsr-β-geo gene-trap reporter mouse line generated from a Ptdsr gene-trap ES cell clone. This line has an insertion of β-galactosidase in the 3' region of the gene (Figure 2a).
We first examined Ptdsr expression by X-Gal staining in heterozygous embryos staged from E9.5 to E12.5. These developmental stages were chosen so as to investigate Ptdsr expression in affected organs prior to the onset of pathological malformations in Ptdsr -/- embryos. At E9.5 we found Ptdsr expression in the developing neural tube, somites, heart, gut and branchial arches (Figure 2b). At E10.5, Ptdsr expression remained high in the developing nervous system, with most intense staining in the forebrain, hindbrain and neural tube. At this stage of embryogenesis, high levels of Ptdsr expression could also be detected in the developing limb buds and eyes (Figure 2b). Ptdsr expression was altered at E12.5, with most intensive β-galactosidase staining in the eyes, developing condensations of the limb buds, neural tube and brain (Figure 2b). Transverse sections of X-Gal-stained embryos at E12.5 showed an asymmetric expression pattern in the neural tube with intense staining of the central mantle layer but no expression in the dorsal part of the neural tube (for example, the roof plate; Figure 2c). Expression in dorsal root ganglia lateral to the neural tube and in the somites was observed; Ptdsr was expressed throughout the somite structure (myotome, dermatome and sclerotome; Figure 2d). Expression boundaries between somites were evident, with no expression in the segmental interzones, which correspond to the prospective intervertebral discs (Figure 2d). Transverse sections of the developing eye at E12.5 revealed strong Ptdsr expression in the inner layer of the neural cup, which will later develop into the neural retina. Furthermore, Ptdsr expression was detected in the primary lens fiber cells of the developing lens (Figure 2e). We carefully investigated whether Ptdsr is expressed from E10.5 to E12.5 in the developing kidney and lungs, but no expression could be detected indicating that Ptdsr expression is required only at later stages in the development of these organs (see below).
Hybridization of a multiple-tissue northern blot revealed a single transcript of about 1.8 kb in almost every tissue analyzed in adult mice (Figure 2f). The most prominent expression was observed in testis, thymus, kidney, liver and skin, with moderate to low expression in lung, small intestine, spleen, stomach and skeletal muscle. Thus, Ptdsr is ubiquitously expressed throughout embryogenesis and in adult tissues, although at different levels.
Ptdsr is required for normal tissue and organ differentiation
We next examined the role of Ptdsr in organ development. Serial histological sections of Ptdsr -/- and control embryos were taken to perform a detailed morphological analysis of all organ systems during development. A significant delay in organ and tissue differentiation was observed at E16.5 in lungs, kidneys and intestine. Lungs of control littermates were properly developed with expanding alveoli (Figure 3a). Terminal bronchi and bronchioles were already well developed, and terminally differentiated epithelial cells with cilia on the luminal cell surface were present. In contrast, almost no alveoli or bronchioles were present in Ptdsr -/- lungs, indicating a delay or arrest in lung sacculation and expansion. Instead, we observed an abundance of mesenchyme that appeared highly immature (Figure 3g). A similar delay in tissue differentiation of Ptdsr -/- embryos was found in the kidneys (Figure 3h). Kidneys from Ptdsr+/+ embryos were well developed at E16.5, showing terminally differentiated glomeruli with Bowman's capsule and collecting tubules lined with cuboidal epithelial cells (Figure 3b). In contrast, Ptdsr-deficient kidneys had only primitive glomeruli at E16.5, and collecting tubules were less well-developed. Instead, a large amount of undifferentiated mesenchyme was present in Ptdsr -/- kidneys (Figure 3h). A delay in tissue differentiation was also found in the intestine at this stage of development. Ptdsr -/- embryos displayed improperly developed villi and an underdeveloped or absent submucosa (Figure 3i). In wild-type embryos (Figure 3c), intestinal cellular differentiation was already highly organized, with intramural ganglion cells between the external and internal muscular layers. Such neuronal cells were absent from the intestine of Ptdsr -/- embryos (Figure 3i), however.
Some Ptdsr -/- mice (4.5 %) also displayed extensive brain malformations that resulted in externally visible head abnormalities, with occasional ectopic tissue outside the skull or exencephaly (Figure 1f,h). Histological analysis revealed an extensive hyperplasia of brain tissue with herniation of brain tissue either through the skull-cap or through the ventral skull (Figure 3d,j). In the most severe cases, expansion of brain tissue in mutant mice resulted in further perturbations of cortical structures (Figure 3d,j). Of note, a similar brain phenotype was observed in the Ptdsr-deficient mouse line generated by Li and colleagues [31].
In contrast to the study of Li et al. [31], however, we found almost normally developed lungs at birth. Ptdsr -/-lungs showed, in comparison to wild-type, only a slight delay in maturation and were fully ventilated in neonates in most cases (Figure 3e,k). This demonstrates that Ptdsr-deficient mice can overcome the delay in embryonic lung differentiation and display normal lung morphology at birth. Thus, it would appear highly unlikely that Ptdsr -/- mice die from respiratory failure. Consistent with the observations of Kunisaki and colleagues [32], we found severely blocked erythropoietic differentiation at an early erythroblast stage in the liver (Figure 3f,3l), suggesting an explanation for the grossly anemic appearance that we observed in our Ptdsr -/- mice.
Loss of Ptdsr activity is associated with defects in ocular development and can lead to formation of ectopic eye structures
By gross morphology we could differentiate two classes of Ptdsr mutants: those that appeared normal with both eyes present (Figure 4) and those that were severely affected and displayed uni- or bilateral anophthalmia (Figure 5).
Analysis of normal or mildly affected embryos revealed no differences between mutant and wild-type embryos in the differentiation of the developing eye until E16.5. In both genotypes, inner and outer layers of the retina displayed a comparable differentiation status, as shown, for example, at E12.5 (Figure 4a,e). At day E16.5, however, retinal layers in Ptdsr -/- embryos were much thinner than in wild-type embryos, contained fewer cells and were greatly reduced in size (Figure 4b,f). Comparison of the retinal structures of Ptdsr +/+ and Ptdsr -/- embryos revealed that all four retinal layers were present in Ptdsr-knockout mice at E16.5 (Figure 4b,f). At E18.5 (Figure 4c,g) and in neonatal animals (postnatal day P0; Figure 4d,h), the differences in retinal differentiation between Ptdsr+/+ and Ptdsr -/- mice were still evident, but the size reduction of the retinal layers was less pronounced in the knockout mice. Ptdsr-deficient animals seem to have compensated for the marked delay in cellular differentiation and expansion of retinal layers. Close examination of retinal structures revealed that the inner granular layer was still less expanded in Ptdsr-deficient animals, however, and that it contained fewer cells and was still severely underdeveloped in comparison with the corresponding retinal layer in control animals (Figure 4c,4g and 4d,4h). Thus, even mildly affected Ptdsr -/- mutants had ocular malformations with defects in differentiation of retinal structures.
We next examined Ptdsr -/- embryos that displayed unilateral or bilateral absence of eyes (Figure 5a) by serial sectioning of whole embryos. These embryos showed complex malformations of the optical cup, including absence of the lens (Figure 5b). Most surprisingly, we found pigmented epithelial cells in the nasal cavity of all Ptdsr-knockout mice with anophthalmia that were analyzed histopathologically. We could identify black-colored pigmented cells embedded in the epithelium of the maxillary sinus that resembled presumptive retinal-pigmented epithelium (Figure 5b,c). Examination of consecutive serial sections revealed the formation of a primitive eye structure, with induction and subsequent proliferation of ectopic mesenchymal tissue immediately adjacent to the displaced pigmented epithelium (Figure 5d). This structure was clearly induced ectopically, and we failed to identify similar changes in any of the wild-type embryos. In summary, we observed a wide range of ocular malformations in Ptdsr-deficient mice that ranged from differentiation defects in retinal cell layers (for example, the inner granular layer) in mildly affected homozygotes to anophthalmia in severely affected Ptdsr -/- mice that was associated with induction of ectopic eye structures in nasal cavities.
Phagocytosis and clearance of apoptotic cells is normal in Ptdsr-deficient mice
We next tested whether Ptdsr is functionally required for the clearance of apoptotic cells. We started with an investigation of cell death in vivo in the interdigital areas of the developing limbs. Apoptosis of interdigital cells in the distal mesenchyme of limb buds occurs most prominently from developmental stages E12.0 to E13.5 and can be easily examined in situ by whole-mount terminal deoxynucleotide transferase-mediated UTP end-labeling (TUNEL). We compared the pattern of interdigital cell death in fore and hind limb buds from Ptdsr -/- (n = 3) and Ptdsr +/+ (n = 3) mice at E12.5 and E13.5. No differences in accumulation of TUNEL-positive cell corpses were observed between the two genotypes (Figure 6a). The kinetics of cell death occurrence and regression of the interdigital web was similar in wild-type and mutant littermates, providing no evidence that Ptdsr-deficiency is associated with impaired clearance of apoptotic interdigital cells during limb development.
To investigate further whether removal of apoptotic cells is impaired in Ptdsr -/- mice, we stained immunohistochemically for activated caspase 3 (aCasp3) and analyzed additional organs and tissues where apoptosis plays a crucial role in tissue remodeling during development. Starting at E12.5, we analyzed and compared the number and distribution of aCasp3-positive cells in over 140 serial sections of three wild-type and six Ptdsr -/- embryos in consecutive and corresponding sections. The sagittal sections were separated by 5 μm, allowing a detailed analysis of apoptosis in several organs and tissues. Tissue restructuring by programmed cell death occurred most notably within the ventral part of the neural tube (Figure 6b,f) and in the developing paravertebral ganglia (Figure 6d,h) with many apoptotic cells being present. In these tissues Ptdsr is highly expressed at E12.5 (Figure 2c) but we observed no difference in the number or distribution of apoptotic cells in Ptdsr+/+ and Ptdsr -/- embryos. The same was true for the developing kidney: apoptotic cells were present in Ptdsr+/+ and Ptdsr -/- embryos, in limited numbers, but we failed to detect any differences in the number of apoptotic cells between the genotypes (Figure 6c,6g). Furthermore, when we continued our analysis of apoptotic cell clearance in vivo at E16.5, E17.5 and E18.5 of embryonic development as well as in neonatal mice, the number and distribution of apoptotic cells was similar in both genotypes. As already observed at E12.5, analysis of aCasp3-stained sections of the developing thymus, heart, diaphragm, genital ridge, eyes and retina convincingly showed that there was no impairment in apoptotic cell removal in Ptdsr -/- mice. Moreover, because Li and colleagues [31] reported impaired clearance of dead cells during lung development in Ptdsr-deficient mice, we examined the rate of apoptosis induction and cell clearance in our Ptdsr-knockout mice in the lung. Analysis of aCasp3-stained lung tissue from Ptdsr+/+ and Ptdsr -/- mice at E17.5 and P0 demonstrated that apoptosis was an extremely rare event during lung morphogenesis at this stage. In addition, there were no differences in the number or distribution of apoptotic cells in Ptdsr -/- and Ptdsr +/+ mice. Furthermore, we were unable to detect any evidence of tissue necrosis in lungs from Ptdsr-deficient mice. In contrast to the report of Li et al. [31], we never observed recruitment of neutrophils or other signs of pulmonary inflammation at any stage of development in our Ptdsr-deficient mice.
To analyze whether macrophages are recruited into areas where apoptosis is prominent during embryogenesis, we stained consecutive serial sections either with the macrophage surface marker F4/80 or with aCasp3. Surprisingly, there was no co-localization of macrophages with apoptotic cells. In virtually all embryonic tissues, apoptotic cells and macrophages were localized in different compartments (Figure 6e,6i; and see also Additional data file 1, Figure S1). This suggests that at this stage of development it is mainly neighboring cells that are involved in removal of apoptotic cells, rather than professional macrophages. In summary, our analysis in vivo did not reveal any impairment in apoptotic cell clearance in Ptdsr-deficient embryos during development and further suggests that phagocytosis of apoptotic cells is mainly mediated by non-professional 'bystander' cells.
To determine whether macrophages from Ptdsr-knockout mice were impaired in the efficacy of apoptotic cell uptake in vitro, we performed phagocytosis assays with fetal-liver-derived macrophages (FLDMs) and quantified their phagocytosis rates. Phagocytosis of apoptotic thymocytes was investigated at 60, 90 and 120 minutes after addition of target cells in the absence of serum. Analysis of phagocytosis rates by flow cytometric analysis (FACS) revealed no differences in the efficacy of apoptotic cell uptake between Ptdsr -/- and Ptdsr+/+ macrophages and demonstrated no differences in apoptotic cell engulfment between selected time points (data not shown). To re-examine and further independently validate the result of normal apoptotic cell uptake by Ptdsr -/- macrophages, we performed phagocytosis assays for 60 min and determined the percentage of macrophages that had engulfed apoptotic cells, in a total of at least 300 macrophages counted by fluorescence microscopy. Phagocytosed, 5-carboxytetramethylrhodamine- (TAMRA-) labeled apoptotic cells were identified as being engulfed by inclusion in F4/80-labeled macrophages. Analysis was done independently by three investigators who were not aware of macrophage genotypes (Ptdsr -/- or Ptdsr+/+). Again, no differences were found in the percentage of macrophages that had engulfed apoptotic cells (Figure 7a,c,e) or in the relative number of phagocytosed apoptotic cells per macrophage (phagocytotic index; Figure 7f). Moreover, single Ptdsr -/-macrophages could be identified that had engulfed even more apoptotic target cells than had wild-type macrophages (Figure 7b,d). Thus, Ptdsr-deficient macrophages had a normal ability to ingest apoptotic cells and were not impaired in recognition or phagocytosis of cells that had undergone programmed cell death.
Ptdsr-deficiency results in reduced production of pro- and anti-inflammatory cytokines after macrophage stimulation
In addition to its suggested importance for phagocytosis of apoptotic cells, it has been proposed that Ptdsr fulfils a second crucial role in regulating and maintaining a non-inflammatory environment upon the recognition of apoptotic cells by macrophages [26]. We therefore tested whether Ptdsr -/- macrophages were able to release anti-inflammatory cytokines after ingestion of apoptotic cells. We examined levels of TGF-β1 and interleukin-10 (IL-10) after stimulation of FLDMs with lipopolysaccharide (LPS), with and without co-culture of apoptotic cells. Quantification of TGF-β1 and IL-10 levels after 22 hours of culture demonstrated that Ptdsr -/- macrophages were able to secrete these anti-inflammatory cytokines upon ingestion of apoptotic cells, although at a slightly lower level than wild-type (Figure 8a,b). This indicates that ablation of Ptdsr function does not compromise in general the ability of macrophages to release immune-suppressive cytokines after recognition and engulfment of apoptotic cells.
To analyze whether pro-inflammatory signaling is affected in Ptdsr -/- macrophages, we stimulated FLDMs from Ptdsr+/+ and Ptdsr -/- mice with LPS and measured levels of tumor necrosis factor-α (TNF-α) at different time points after stimulation (Figure 8c). Ptdsr -/- macrophages produced significantly less TNF-α than did wild-type macrophages. The difference in TNF-α secretion was first visible after 3 h of LPS stimulation and became more prominent during the course of the experiment (for example, after 9 h and 12 h of LPS stimulation; Figure 8c). To analyze whether TNF-α release by Ptdsr -/- macrophages can be affected by engulfment of apoptotic cells, we stimulated FLDMs with LPS, apoptotic cells or both. Quantification of TNF-α levels by ELISA after 22 h showed that Ptdsr-deficient macrophages release less TNF-α after stimulation with LPS alone, and also after double stimulation of macrophages with LPS and apoptotic cells (Figure 8d). Moreover, the double stimulation demonstrated that the LPS-induced TNF-α release by Ptdsr -/- macrophages could be inhibited by co-administration of apoptotic cells to an extent comparable to that seen in wild-type macrophages. Similar results were obtained when other pro-inflammatory cytokines, such as interleukin-6 and monocyte chemoattractant protein-1, were analyzed (data not shown). These results indicate that Ptdsr is not required in macrophages for the inhibition of pro-inflammatory signaling after recognition and engulfment of apoptotic cells. Ptdsr-deficiency does, however, affect the overall release of pro- and anti-inflammatory cytokines after stimulation with LPS and after double treatment with LPS and apoptotic cells, indicating that Ptdsr-deficient macrophages have a reduced capacity to produce or secrete pro- and anti-inflammatory cytokines.
Discussion
Ptdsr is required for the differentiation of multiple organ systems during development
In this study, we have generated a null mutation in the phosphatidylserine receptor (Ptdsr) gene in C57BL/6J mice. We show that ablation of Ptdsr results in profound differentiation defects in multiple organs and tissues during embryogenesis, although with variable penetrance. While this work was in progress, two other groups reported the generation of Ptdsr-deficient mice [31,32]. In all three knockout mouse lines, the first two exons ([31] and this study) or exons one to three [32] were deleted by replacement with a neomycin-selection cassette. The Ptdsr-knockout mouse lines differ in the genetic background in which the mutation was generated and maintained, however. In our case, the Ptdsr-null allele was generated in an isogenic C57BL/6J background, whereas Li et al. [31] and Kunisaki et al. [32] investigated the phenotype of their Ptdsr-knockout mice in a mixed 129 × C57BL/6 background. The ablation of Ptdsr function results in perinatal lethality in all cases, but there are interesting differences in severity or expressivities of phenotypes among the different Ptdsr-deficient mouse lines. This might be due either to differences in genetic background or because the phenotypes that have been investigated in this study have not been analyzed in such detail before.
In the Ptdsr-knockout mouse line reported here, growth retardation started from E12.5 onwards and was associated with delayed differentiation in several organs in which Ptdsr is expressed either during embryogenesis or later in adulthood. At E16.5 almost no branching morphogenesis of the lung epithelium was observed in Ptdsr -/- lungs. Similarly, epithelial structures were only partially developed in mutant kidneys, without terminal differentiation of Bowman's capsule and with a severe reduction in the number of differentiated collecting tubules. Likewise, the differentiation of the intestine was also severely delayed at this developmental stage. When compared with wild-type controls, intestinal tissues of Ptdsr knockout mice appeared unstructured, with an absence of enteric ganglia and of differentiated smooth muscle tissue. Interestingly, defects in kidney and intestine differentiation were not described in the Ptdsr-knockouts generated by Li et al. [31] and Kunisaki et al. [32]. Surprisingly, when we examined Ptdsr-/- embryos shortly before birth (E18.5) or neonatally, we found only mild differentiation delays in organs that appeared severely affected at mid-gestation. This 'recovery' was most visible in Ptdsr -/- lungs: at P0 we found expanded lungs in the knockout mice that showed normal branching patterns, with differentiated alveoli and bronchioles.
We investigated the occurrence of programmed cell death during lung development in wild-type and Ptdsr-knockout mice throughout embryogenesis (E16.5 to P0). Comparative immunohistochemistry for aCasp3 revealed that apoptosis is a rare event during lung morphogenesis. Furthermore, we failed to detect any differences in the number of apoptotic cells in Ptdsr-knockout and wild-type animals in the rare cases where we could detect apoptotic cells within lung tissues. These findings are contrary to the results reported by Li et al. [31], who suggested that impaired clearance of apoptotic mesenchymal and epithelial cells causes a failure in lung morphogenesis in Ptdsr-deficient mice. In contrast, our findings are in line with the current view on lung development during embryogenesis. Accordingly, formation of the epithelial lung via branching morphogenesis can be subdivided into a series of sequential steps that involve: first, formation of the organ anlage in the form of a placode; second, primary bud formation by placode invagination; third, branch initiation and branch outgrowth; fourth, further reiteration of the branching process; and fifth, terminal differentiation of organ-specific proximal and distal structures [34,35]. In contrast to other invagination processes during embryogenesis, such as mammary gland formation, the lumen of the lungs is expanded by successive branching events, branch outgrowth and elongation, rather than by apoptosis [34,36]. Finally, because the lungs of Ptdsr -/-neonates were almost fully expanded and appeared normal in structure in comparison to wild-type littermates, it is highly unlikely that Ptdsr mutants die of respiratory lung failure. In addition, Li and colleagues [31] demonstrated that surfactant expression is normal in Ptdsr-deficient animals, supporting the idea of normal maturation of surfactant-producing type II alveolar epithelial cells and lung function. Other defects must therefore be responsible for the death of Ptdsr-mutant mice. The frequently observed subcutaneous edema of various extents in Ptdsr-deficient homozygotes gave us a hint that Ptdsr-deficiency and lethality might be associated with cardiovascular problems. Indeed, very recently we have obtained strong evidence that Ptdsr-knockout mice die as a result of defects in heart development that are associated with specific cardiopulmonary malformations; (J.E. Schneider, J.B., S.D. Bamfort, A.D.G., C. Broadbent, K. Clarke, S. Neubauer, A.L. and S. Battacharya, unpublished observations).
In addition, we demonstrate that eye development requires a functional Ptdsr gene. Ptdsr-deficient embryos can be roughly divided into two categories. The first, severely affected group develops anophthalmia that correlates with formation of ectopic retinal-pigmented epithelium and induction of proliferation of underlying mesenchyme in the nasal cavity. This phenotype represents a completely novel lesion that to our knowledge has not been described before in any other mouse mutant. The second group shows normal external eye structures, although in this case retinal development is temporally delayed during mid-gestation, with persistent, abnormal morphogenesis of the inner granular retinal layer at later stages of embryogenesis. A possible explanation for these two phenotypes can be found in the expression pattern of the Ptdsr gene. Initially, Ptdsr is expressed throughout the whole developing nervous system, with exceptionally high levels in the anterior part of the forebrain. Later expression becomes more restricted to the developing retina and lens. Thus, Ptdsr might play an important role in early events of ocular morphogenesis, such as establishment and bisection of eye fields and formation of optic cups. These early eye-formation steps are closely interconnected with development of the forebrain [37,38] and the nose [39-41].
Interestingly, we occasionally observed serious malformations of forebrain and nasal structures in Ptdsr-knockout embryos that were associated with bilateral anophthalmia (see for example the mutant embryo in Figure 1g). This suggests that Ptdsr is involved in the regulation of differentiation processes within forebrain regions, and that ablation of Ptdsr function might secondarily affect early eye formation. Li et al. [31] found smaller lenses in Ptdsr-knockout mice and described the formation of retinal protrusions, although anophthalmia and specific differentiation defects of retinal cell layers were not reported in their study. Li et al. proposed [31] that the eye phenotype they observed could be explained by failed removal of apoptotic cells during eye development, but we think that the observed defects are unrelated to a failure of apoptotic cell clearance. A recent comprehensive kinetic analysis of apoptosis induction during mouse retinal development described four major peaks of apoptotic cell death [42]. This study demonstrated that there is an initial phase of cell death during the invagination of the optic cup (E10.5), followed by subsequent waves of apoptosis induction immediately before and after birth (E18.5 to postnatal day P2), and from postnatal days P9 to P10 and P14 to P16 [42]. Thus, besides the formation of the inner and outer layers of the optic cup in early eye development, other major phases of retinal cell apoptosis take place only postnatally and correspond to important periods in the establishment of neuronal connections. Furthermore, cell death during normal retinal development occurs in retinal layers distinct from the inner granular layer where we observed the most pronounced differentiation defects in the Ptdsr -/- mutants described here. Other studies that connect the postnatal elimination of apoptotic photoreceptor cells to Ptdsr-mediated macrophage engulfment [43] should be interpreted with extreme caution as these studies were based on the monoclonal anti-Ptdsr antibody mAb 217G8E9 [26,43] (see below).
Consistent with the results of Li et al. [31], we found particular brain malformations in our Ptdsr -/- mice. Exencephaly and hyperplastic brain phenotypes were observed at a low penetrance in Ptdsr-mutant mice (less then 4.5% of homozygotes), but these do not resemble to any extent the brain-overgrowth phenotypes of caspase- or Apaf1-knockout mice ([44], and references therein) in that we failed to identify any differences in the number or distribution of apoptotic cells or pyknotic cell clusters in the neuroepithelium of Ptdsr -/- and Ptdsr+/+ mice. Thus, reduced cell death or diminished clearance of apoptotic neural progenitor cells is unlikely to be the cause of the brain hyperplasia.
In summary, our studies demonstrate that Ptdsr is required for normal tissue differentiation, especially during the mid-gestation period when we observed the most severe differentiation delays in several organs of Ptdsr-knockout mice. The multiple defects in tissue differentiation cannot be explained by failure of apoptotic cell clearance, as this process is normal in our Ptdsr-knockout line. This result therefore indicates that Ptdsr has a novel, hitherto unexpected, role in promoting tissue maturation and terminal differentiation. Additional studies with conditionally targeted Ptdsr-deficient mice are required to investigate the role of spatial and temporal Ptdsr expression and function during tissue differentiation.
Ptdsr is not essential for the clearance of apoptotic cells
Our studies demonstrate that Ptdsr is not a primary receptor for the uptake of apoptotic cells. Investigation of apoptotic cell clearance in vivo in Ptdsr -/- embryos conclusively showed that removal of apoptotic cells is not compromised by ablation of Ptdsr function. Comparative analysis of ten different tissues and organs in Ptdsr+/+ and Ptdsr -/- animals at several stages of embryonic development and in neonates failed to identify impaired uptake of apoptotic cells at any time during development. Furthermore, phagocytosis assays in vitro demonstrated a completely normal uptake of apoptotic cells by Ptdsr -/- macrophages, with some knockout macrophages showing loads even higher than wild-type of engulfed dead cells. These results are contrary to the expected role of Ptdsr in apoptotic cell clearance and to the reported findings of Li et al. [31] and Kunisaki et al. [32], as well as to a study done with a phosphatidylserine receptor null allele in C. elegans [45]. In previous studies in the mouse, the distribution and amount of apoptotic cells in Ptdsr-knockout and control animals were investigated in only a few tissues and at one [31] or two [32] developmental stages. Li et al. [31] examined lung, midbrain and retina at day E17.5 of gestation and identified apoptotic cells by TUNEL staining. Their findings must be interpreted with caution because remodeling of cellular structures by apoptosis in specific retina layers is known to occur mainly postnatally [42], and apoptosis plays an important physiological role in the maintenance and homeostasis of lung epithelium after birth or in pathological conditions involving pulmonary inflammation and not during lung development [46]. This postnatal role for apoptosis is in accordance with our data, as we rarely observed apoptotic cells in retina or lung tissue throughout embryogenesis in Ptdsr+/+ and Ptdsr -/- mice. Kunisaki et al. [32] analyzed TUNEL-stained sections of liver and thymus at days E13.5 and E16.5 of development in Ptdsr+/- and Ptdsr -/- embryos and found reduced rather than increased numbers of TUNEL-positive cells in Ptdsr-deficient embryos. Using co-localization of TUNEL-positive cells with F4/80-positive macrophages they suggested that Ptdsr -/- embryos exhibited a three-fold increase in the frequency of unphagocytosed TUNEL-positive cells together with a severely reduced number of F4/80-positive cells. These results must be interpreted very carefully, however, as it is technically difficult to unambiguously identify engulfed target cells in individual macrophages in solid tissues by fluorescence microscopy.
In addition, our data suggest that during embryogenesis, macrophage-mediated clearance of apoptotic cells is not the only - or even the primary - mechanism for the removal of apoptotic cells. In many tissues where programmed cell death occurs as a prominent event during embryogenesis, such as remodeling of the genital ridge during gonad morphogenesis and differentiation of the neural tube, we found almost no co-localization of apoptotic cells and macrophages. This indicates that in these cases clearance of apoptotic cells is directly mediated by neighboring 'bystander' cells rather than by macrophages that have been recruited into areas where apoptosis occurs. Obviously these in vivo clearance mechanisms are not compromised by Ptdsr-deficiency in our knockout mutant. This finding is in line with studies in macrophageless Sfpi1-knockout embryos that are deficient for the hematopoietic-lineage-specific transcription factor PU.1. Here, the phagocytosis of apoptotic cells during embryogenesis is taken over by 'stand-in' mesenchymal neighbors [47]. As recognition of phosphatidylserine is thought to be a universal engulfment mechanism for all cells that are able to phagocytose apoptotic cells, it is very striking that apoptotic cell clearance mediated by non-professional bystander cells is also not compromised by Ptdsr-deficiency.
In contrast to Li et al. [31], we did not observe any impairment in the uptake of apoptotic cells by Ptdsr -/- macrophages in vitro. We performed phagocytosis assays in vitro with fetal-liver-derived macrophages, while in their assays, Li and colleagues used thioglycollate-elicited peritoneal macrophages after adoptive transfer of Ptdsr -/- hematopoietic stem cells. The different results obtained in the two studies are puzzling; they might be due to the use of different macrophage or cell populations. We and Kunisaki et al. [32] found that Ptdsr-deficiency is to some extent associated with defects in hematopoiesis. Thus, it seems possible that recruitment and activation/differentiation of macrophages after adoptive transfer and thioglycollate elicitation are affected by Ptdsr-deficiency. We do not think that the different results observed in Ptdsr-knockout mice in a mixed C57BL/6 × 129 background and in a pure C57BL/6J background can be attributed to genetic background effects: comparison of apoptotic cell engulfment efficacies of thioglycollate-elicited macrophages from 129P2/OlaHsd and C57BL/6J mice did not show any differences in apoptotic cell uptake (J.B. and A.L., unpublished observations). Moreover, in contrast to our studies, neither Li et al. [31] nor Kunisaki et al. [32] determined phagocytotic engulfment indexes for Ptdsr-deficient macrophages.
Interestingly, we observed differences between Ptdsr+/+ and Ptdsr -/- macrophages in the secretion of pro- and anti-inflammatory cytokines after stimulation with LPS and apoptotic cells. This provides evidence that cellular activation and effector mechanisms are impaired in Ptdsr-deleted macrophages. It remains to be determined which classical pathways of macrophage activation and function involve Ptdsr. This is especially important in light of recent findings that demonstrated nuclear localization of the Ptdsr protein [29].
Most strikingly, the recently published data regarding the genetic ablation or perturbation of phosphatidylserine receptor function in C. elegans are also contradictory. Wang et al. [45] reported that psr-1, the C. elegans homolog of Ptdsr, is important for cell-corpse engulfment, whereas psr-1 RNAi studies performed by Arur et al. [25] yielded, in this respect, no phenotype. Moreover, Wang and colleagues hypothesized on the basis of their data that psr-1 might act to transduce an engulfment signal upstream of Ced-2 (Crk II), Ced-5 (Dock 180), Ced-10 (Rac 1) and Ced-12 (Elmo) in one of the two cell-corpse engulfment pathways in the worm [45]. But the loss-of-function phenotype of psr-1 mutants and the complementation phenotypes in overexpressing transgenic worms shown by Wang et al. [45] are rather weak as compared to the classical C. elegans engulfment mutants [8].
Many previous functional studies that reported a requirement for Ptdsr for the phagocytosis of apoptotic cells used the monoclonal anti-Ptdsr antibody mAb 217G8E9 [26]. This antibody was used in Ptdsr binding and blocking experiments, as well as in subcellular localization studies, which led to the conclusion that Ptdsr is a transmembrane receptor critical for signal transduction at the engulfment interface. More recently it was used in binding assays to show that the human and worm Ptdsr molecules can recognize phosphatidylserine [45]. In the course of the study presented here, we stained immunohistochemically for Ptdsr with mAb 217G8E9 on wild-type and Ptdsr-deficient macrophages and fibroblasts (see Additional data file 1, Figure S2 and data not shown). To our surprise, we observed similar staining patterns with cells of both genotypes. Furthermore, using a Ptdsr-peptide array we found that mAb 217G8E9 can bind weakly to a Ptdsr peptide, explaining the original isolation of Ptdsr cDNA clones by phage display [26]; but the antibody mainly recognizes additional, as-yet unknown, membrane-associated protein(s) (see Additional data file 1, Figure S2). Experiments that have used this antibody should therefore be interpreted with great caution as they might come to be viewed in a different light.
Conclusion
Our results demonstrate that Ptdsr is essential for the differentiation and maturation of multiple tissues during embryogenesis. Ablation of Ptdsr function results in neonatal lethality and severe defects in the morphogenesis of several organs. The developmental malformations cannot be explained by impaired clearance of apoptotic cells, a process that proved to be normal in Ptdsr-deficient mice. This opens up the possibility either that there is an as-yet unknown Ptdsr receptor, which might act as a primary phosphatidylserine recognition receptor, or that recognition of phosphatidylserine and subsequent apoptotic cell engulfment and anti-inflammatory signaling are mainly mediated through phosphatidylserine bridging proteins and their cognate receptors. Although Ptdsr -/- macrophages were not impaired in their ability to phagocytose apoptotic cells, they showed reduced cytokine responses after stimulation. Further work will be required to determine the molecular mechanisms of these newly recognized Ptdsr functions during development.
Materials and methods
Construction of the targeting vector and generation of Ptdsr-knockout and gene-trap mice
Targeting vector
A Ptdsr-containing bacterial artificial chromosome (BAC) clone (GenBank accession number AC091694; RP-23-316F3) was isolated by sequence homology from a C57BL/6J genomic BAC library (RP-23; BACPAC Resources, Oakland, USA). A 14.5 kb KpnI/BamHI fragment containing the entire Ptdsr locus and 5' and 3' flanking regions was subcloned from this BAC clone and a 1.9 kb RsrII/AatII fragment containing exons I and II of the Ptdsr gene was replaced by a 1.2 kb loxP-flanked neomycin-resistance gene cassette (neo).
Homologous recombination in ES cells and generation of germ-line chimeras
Bruce4 ES cells were transfected with KpnI-linearized targeting vector and selected with G418. ES-cell clones resistant to G418 were isolated and analyzed by Southern blot analysis for homologous recombination events within the Ptdsr locus. Chimeric mice were produced by microinjection of two independent homologous recombinant (Ptdsr+/-) ES cells into BALB/c blastocysts and transfer to pseudopregnant foster mothers followed by development to term. Chimeric males were mated with C57BL/6J females. From the two selected ES-cell clones, one successfully contributed to the germ-line. Germ-line transmission of the mutant allele was verified by PCR and Southern blot of genomic DNA from black coat-color F1 offspring.
Ptdsr gene-trap and generation of germ-line chimeras
An ES-cell line carrying a β-geo gene-trap vector in the Ptdsr locus was identified by searching the BayGenomics database (BayGenomics, San Francisco, USA; [48]) with the full-length Ptdsr cDNA. A single ES-cell line was identified carrying the gene-trap in intron V, between exons V and VI of the Ptdsr gene. Chimeric mice were generated by microinjection into CB20 blastocysts and transfer to pseudopregnant foster mothers. Chimeric males were mated with 129P2/OlaHsd females. Germ-line transmission of the mutant gene-trap allele was verified by expression analysis using β-galactosidase staining and RT-PCR.
Genotype analysis
The genotypes of embryos or animals were determined by PCR analysis and confirmed by Southern blot. Genomic DNA for PCR was prepared from extraembryonic membranes or tail clips using a non-organic tail-DNA extraction protocol [49]. High molecular weight genomic DNA for Southern blotting was prepared according to standard protocols. For PCR analysis the wild-type Ptdsr allele was detected using forward primer 1 (5'-GACACTGTCCATGGCAAACAC-3') and reverse primer 2 (5'-TAAAGTCGCCTTCCAGAAGATT-3'). The primer 1 site is located 5' to the deletion and the primer 2 site within the deletion. This primer pair amplified a fragment of approximately 300 bp from wild-type and Ptdsr+/- mice but not from Ptdsr -/-mutants. To detect the mutant Ptdsr allele, genomic DNA was also amplified using primer 1 and reverse primer 3 (5'-CCACACGCGTCACCTTAATA-3'), which corresponds to a sequence in the neo cassette. In this case, a 500 bp fragment was detected in mice heterozygous or homozygous for the mutant allele, while no signal was detected in wild-type mice. For Southern blot analysis, genomic DNA (30 μg) was digested overnight with BamHI (30 U; Roche Diagnostics GmbH, Mannheim, Germany) and ScaI (30 U; Roche), fractionated on a 0.8 % agarose gel, transferred to a nylon membrane (Hybond N; Amersham Biosciences Europe GmbH, Freiburg, Germany) and hybridized with 5' and 3' flanking probes. The BamHI digest was hybridized with a Ptdsr-specific 5' flanking probe, and Southern blot gave a single 17.2 kb band for wild-type (+/+), an 11.6 kb band for homozygous (-/-) and both bands for heterozygous (+/-) mice. The ScaI digest was hybridized using a 3' flanking probe, and Southern blot gave a single 12.4 kb band for wild-type, a 17.2 kb band for homozygous and both bands for heterozygous mice.
Northern blot analysis
Total RNA was isolated from homogenized embryos using TRIZOL reagent (Invitrogen GmbH, Karlsruhe, Germany). For northern blots, either total RNA (30 μg) was extracted from embryos, electrophoresed and transferred to a nylon membrane (Hybond N; Amersham) or a polyA+ RNA northern blot (OriGene Technologies Inc., Rockville, USA) was hybridized using as the probe a Ptdsr fragment amplified from wild-type cDNA using forward primer 5'-GTTCCAGCTCGTCAGACTCG-3' and reverse primer 5'-TGCCCCTAAGACATGACCAC-3'. In all experiments the same membrane was re-hybridized with a β-actin probe (OriGene) to confirm that equivalent RNA levels were present in each lane. Northern blotting indicated that homozygous mutant embryos did not express Ptdsr mRNA and heterozygous mutant embryos expressed only reduced amounts of Ptdsr mRNA.
Western blot analysis
Embryos (E13.5) for protein isolation were homogenized in lysis buffer containing 1 × PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS and protease inhibitor cocktail (CompleteMini; Roche). Equal amounts (25 μg) of protein lysate were separated by SDS-polyacrylamide gel electrophoresis and transferred onto a PVDF membrane (Millipore, Billerica, USA) according to standard protocols. Western blots were done using a specific antibody to Ptdsr (PSR N-20, sc-11632; Santa Cruz Biotechnology Inc., Santa Cruz, USA) and β-actin (ab-6276; Abcam, Cambridge, UK) as described by the supplier. Secondary antibodies conjugated to horseradish peroxidase were from Santa Cruz and Abcam, used as described by the supplier, and detection was performed with an enhanced chemiluminescence system (ECLPlus; Amersham).
Animal experiments
Wild-type C57BL/6J and 129P2/OlaHsd mice were obtained from Jackson Laboratories (Bar Harbor, USA) and Harlan UK (Bicester, UK), respectively. All mice were housed in individually ventilated cages in a specific pathogen-free environment with a 12 h light-dark cycle and were fed a regular unrestricted diet. The GBF's routine surveillance program screened for selected pathogens. The Ptdsrtm1Gbf mutant was crossed to C57BL/6J mice to establish the co-isogenic C57BL/6J-Ptdsrtm1Gbf mouse line. All studies were approved by the appropriate authorities.
Isolation of embryos
Heterozygous male and female mice were intercrossed in order to obtain Ptdsr-deficient progeny. Females were daily monitored for vaginal plugs, and noon of the day of plug detection was defined as E0.5. Embryos at indicated time points were dissected in sterile PBS, washed in ice-cold PBS and transferred to cold fixative. Extra-embryonic membranes were kept and used for genotyping. Ptdsr -/- embryos and their wild-type littermates were used for experiments.
Histology, TUNEL staining and immunohistochemistry
Embryos for histology and immunohistochemistry were harvested and fixed in 10% neutral-buffered formalin, dehydrated through a graded series of alcohol, embedded in paraffin, sagittally sectioned at 5 μm intervals, and every fifth section was processed for hematoxylin and eosin (H&E) staining according to standard protocols. Remaining sections of wild-type and Ptdsr -/- specimens were used for immunohistochemistry. For detection of apoptotic cells and macrophages, anti-aCasp3 (an antibody specific for activated caspase 3; R&D Systems, Minneapolis, USA) and anti-F4/80 (Serotec GmBH, Düsseldorf, Germany; #MCA 1957) antibodies were used as described by the supplier. Detection was performed using indirect streptavidin with biotinylated secondary antibodies and cobalt-enhanced diaminobenzidine (brown) or fast-red (red) as chromogens. Sections were counterstained with hematoxylin. For whole-mount terminal deoxynucleotidyl transferase-mediated UTP end labeling (TUNEL), limb buds were dissected from E12.5 and E13.5 embryos, fixed in 4% paraformaldehyde and processed for analysis as previously described [50].
Preparation of fetal liver-derived macrophages (FLDMs)
Fetal livers were excised from embryos at E12.5 and E13.5, respectively, washed in PBS and dissociated enzymatically for 60 min at 37°C. The digestion buffer (150 μl per liver) comprised 0.6 U/ml dispase I (Roche), 0.1% collagenase D (Roche), 10 U DNase (Roche), and 20% FCS in PBS. X-Vivo 15 medium (Cambrex, East Rutherford, USA) was added to the resulting cell suspension, and after centrifugation (200 × g; 3 min) cells were resuspended in X-Vivo 15 medium supplemented with 50 ng/ml macrophage colony-stimulating factor (M-CSF; Sigma-Aldrich, St. Louis, USA) and cultured on non-treated tissue-culture dishes at 37°C with 5% CO2. Every second or third day the medium was changed by centrifugation. Following withdrawal of M-CSF on day 6 after excision, adherent cells were cultured for an additional 24-48 h in X-Vivo 15 medium.
Macrophage phagocytosis assays
For preparation of monolayer cultures of macrophages, FLDMs were plated on glass coverslips in 24 well plates (2 × 105 cells per well) in X-Vivo 15 medium. For preparation of apoptotic target cells, primary thymocytes were harvested from the thymus of 4- to 8-week-old C57BL/6J mice, stained with TAMRA for 15 min, and apoptosis was induced either by treating cells with 5 μM staurosporine in medium for 4 h at 37°C or by culturing cells in medium overnight. The efficacy of apoptosis induction was compared in thymic target cells and controls by FACS analysis. On average, 60% of the cells of the resulting population were apoptotic, with exposed PS on their surface, and less than 5% of the cells were necrotic, as confirmed by FITC-annexin V and propidium iodide staining. The apoptotic thymocytes obtained were washed with PBS and added to the prepared FLDM cultures (ratio 10:1). Phagocytosis was then allowed to proceed at 37°C and 5% CO2. After the indicated time periods, the uptake of apoptotic cells by FLDMs was stopped by intensive washing of co-cultures with cold PBS to remove unphagocytosed cells. To measure phagocytosis of apoptotic thymocytes, macrophages were further processed for immunofluorescence analysis. Cells were fixed in 4% paraformaldehyde, blocked in 0.5% BSA/PBS and stained with an anti-F4/80 antibody (Serotec) followed by a secondary antibody coupled to Alexa 488 (Molecular Probes Inc., Eugene, USA). Coverslips were mounted on slides and engulfed thymocytes were enumerated by fluorescence microscopy. The percentage of phagocytosis was calculated by counting at least 300 macrophages and determining the number of macrophages that had engulfed apoptotic thymocytes. The phagocytotic index was calculated according to the following formula: phagocytotic index = (total number of engulfed cells/total number of counted macrophages) × (number of macrophages containing engulfed cells/total number of counted macrophages) × 100. The experiments were performed at least three times, each time in triplicate, and the counting was done by three different investigators.
Measurement of macrophage cytokine production
Monolayer cultures of FLDMs and apoptotic thymocytes were prepared as described above. FLDMs were incubated with medium, LPS (10 ng/ml), apoptotic cells (ratio 1:10) or both for the determination of IL-10, TGF-β1 or TNF-α levels after co-culture for 22 h. For TNF-α quantification at various time points, FLDMs were cultured with a high concentration of LPS (100 ng/ml). Culture supernatants were harvested and TNF-α (Mouse TNF-α OptEIA set; BD Biosciences, Heidelberg, Germany) and TGF-β1 (Quantikine, TGF-β1 immunoassay; R&D Systems) were measured by ELISA as described by the supplier. IL-10 in culture supernatants was determined by a cytometric bead assay (Mouse inflammation CBA; BD Biosciences) as indicated in the manual. Data are presented as mean ± SEM from at least three independent experiments, each carried out in triplicate. Analysis of the results used the Wilcoxon-signed rank test; p values below 0.05 were considered significant.
Additional data files
Additional data file 1 contains: Figure S1 showing the localization of apoptotic cells and macrophages in the subcutis of developing embryos; and Figure S2 showing immunohistochemical staining of the Ptdsr protein in macrophages derived from wild-type and Ptdsr-knockout mice.
Supplementary Material
Additional data file 1
Figure S1 showing the localization of apoptotic cells and macrophages in the subcutis of developing embryos; and Figure S2 showing immunohistochemical staining of the Ptdsr protein in macrophages derived from wild-type and Ptdsr-knockout mice
Click here for additional data file
Acknowledgements
We thank Rudi Balling (GBF Research Center) and Shoumo Bhattacharya (University of Oxford) for many helpful and stimulating discussions. We thank Evi Wollscheid-Lengeling (GBF) for help with harvest of neonatal mice, Ronald Frank (GBF) for providing Ptdsr peptide arrays, Maria Ebel (GBF) for ES cell blastocyst injections, Manfred Rohde (GBF) for electron microscopy, Kurt Dittmar (GBF) for help with confocal microscopy and Bastian Pasche (GBF) for critical reading of the manuscript. We thank BayGenomics, a genomic consortium funded by the US National Heart, Lung, and Blood Institute, for providing the ES cell gene-trap line RRJ099. This work was supported in part by the EU project EUMORPHIA, "Understanding human molecular physiology and pathology through integrated functional genomics in the mouse model" (QLG2-CT-2002-00930).
Figures and Tables
Figure 1 Targeted inactivation of the phosphatidylserine receptor gene. (a) Ptdsr gene-targeting strategy. Homologous recombination in ES cells results in the deletion of exons I and II of the murine Ptdsr gene through replacement of a loxP-flanked neomycin phosphotransferase gene (neo), thereby ablating the reading frame of the encoded protein. Coding exons I-VI are shown as filled boxes, and deleted exons are colored green. Restriction sites are: A, AatII; B, BamHI; EI, EcoRI; EV, EcoRV; K, KpnI; R, RsrII; S, SacII; Sc, ScaI, X, XhoI. The probe sites are red boxes labeled: C, 5' outside probe; D, 3' outside probe. (b) Southern blot analysis of genomic DNA extracted from wild-type (+/+) and Ptdsr+/- (+/-) animals, digested with BamHI and hybridized with the 5' outside probe to confirm germ-line transmission of the mutant Ptdsr allele. 'Wild-type' indicates the BamHI fragment of 17.2 kb from the wild-type Ptdsr allele; 'mutant' indicates the BamHI fragment of 11.6 kb from the targeted Ptdsr allele. (c) PCR genotyping of embryos and animals from intercrosses of heterozygous Ptdsr+/- using a wild-type and a mutant allele-specific primer combination, respectively. (d) Northern blot analysis of total RNA isolated from E13.5 wild-type, Ptdsr+/- and Ptdsr -/- embryos. (e) Western blot analysis of protein from homogenates of E13.5 wild-type, Ptdsr+/- and Ptdsr -/- embryos using a Ptdsr-specific antibody. Developmental abnormalities at (f,g) E15.5 and (h) birth; in this and all subsequent figures wild-type littermates are located on the left and homozygous mutant mice on the right. The Ptdsr -/- embryos show exencephaly (f) or prosencephalic hernia in the forebrain region (arrowhead, neonate 2; h), uni-or bilateral absence of the eyes (f,g and neonate 2 in h, and arrow, neonate 3 in h), an abnormal head shape with proboscis (g), edema (arrowheads in f and g), and general anemia (asterisk, neonate 3 in h).
Figure 2 Expression analysis of Ptdsr during embryonic development. (a) Schematic representation of the construction of the Ptdsr gene-trap mouse line used for expression analysis at different embryonic stages. Gray and bright blue boxes represent regulatory elements of the gene-trap, and β-geo, the β-galactosidase/neomycin phosphotransferase fusion protein-expression cassette [48,51]. Restriction enzyme nomenclature is as in Figure 1 (b) Whole-mount β-galactosidase staining of heterozygous Ptdsr gene-trap embryos at mid-gestation. Expression of Ptdsr is highest in neural tissues and somites, in the branchial arches, the developing limbs, the heart, the primitive gut and the developing eye. (c-e) Sectioning of E12.5 β-galactosidase-stained embryos confirms expression of Ptdsr in (c) the neural tube; (inset in c) neural epithelium; (d) somites; and (e) eyes. Expression in the eye is restricted to developing neural retinal and lens cells. (f) Expression analysis of adult tissues by northern blot. Expression of Ptdsr in the muscle (asterisk) was detected only on long-term exposures of the filter (> 48 h). A β-actin hybridization was used to confirm equal loading of RNA samples. Scale bar, 100 μm.
Figure 3 Histological analysis of wild-type and Ptdsr -/-organs during embryogenesis. (a-f) Wild-type embryos and (g-l) Ptdsr -/- littermates were isolated at various embryonic stages, serially sectioned sagittally and analyzed for developmental abnormalities in detail after H&E staining. At E16.5, the lungs of (g) Ptdsr -/- embryos had sacculation just starting, and well-formed alveoli (asterisks) or epithelium-lined bronchioles (arrows) were scarce compared to (a) wild-type lungs. At E16.5, the glomeruli (arrows) in the kidney of (h) Ptdsr -/- embryos were underdeveloped compared to (b) wild-type, collecting tubules (arrowheads) were missing and undifferentiated blastemas (asterisks) were more abundant. The jejunum had no intramural ganglia in Ptdsr -/-embryos (i; and arrows in c); and a well-developed submucosa (asterisk in c) was missing. Brain sections at E18.5 show that (j) Ptdsr -/-embryos may have herniation (arrow) of the hypothalamus through the ventral skull (secondary palate), most likely through Rathke's pouch, and a severe malformation of the cortex (asterisks) compared to (d) wild-type embryos. At E18.5, (e) wild-type and (k) Ptdsr -/- lungs showed normal sacculation and formation of alveoli (asterisks) and bronchioles (arrow). (f) Wild-type neonatal liver had significant numbers of megakaryocytes (arrows), compared to (l) homozygous mutant littermates, and higher numbers of erythropoietic islands and of mature erythrocytes. Hepatocellular vacuoles are due to glycogen stores (asterisks) that were not metabolized in perinatally dying Ptdsr -/- animals, in contrast to wild-type newborns. Scale bar, 100 μm, except for (d) and (j), 1 mm.
Figure 4 Morphology of wild-type and Ptdsr -/- retinas. Serial sagittal sections of (a-d) wild-type and (e-h) Ptdsr -/- retina were analyzed for developmental abnormalities at (a,e) E12.5, (b,f) E16.5, (c,g) E18.5, and (d,h) P0. Normal patterning of the retina was observed in Ptdsr -/-embryos, with an outer granular layer (OGL), outer plexiform layer (OPL), inner granular layer (IGL) and inner plexiform layer (IPL). Note that the IGL in Ptdsr -/- retinas is less thick than that in wild-type littermates in comparing (c,g) and (d,h). Morphometric analysis (numbered lines) of wild-type and Ptdsr -/- retinas confirmed the initial finding of a thinner retina in Ptdsr -/- animals than in wild-type (all values in μm). Scale bar, 50 μm.
Figure 5 Histological analysis of eye development in severely affected eyeless Ptdsr -/- embryos. (a) In anophthalmic Ptdsr -/- embryos, unilateral or bilateral absence of the eyes could be detected. (b-d) Serial H&E-stained sagittal sections of homozygous mutant embryos at (b) E17.5 and (c,d) E18.5 show complex malformation of the optic cup and lack of any lens structure. Careful examination of adjacent sections (b-d) reveals an ectopic misplacement of retinal-pigmented epithelium in the maxillary sinus. Not only is the deposition of pigment clearly visible (higher magnification insets) but also the induction of proliferation of underlying tissues and the change in morphology of the maxillary sinus (d). Scale bar, 100 μm in (b-d).
Figure 6 Analysis of programmed cell death and involvement of macrophages in the removal of apoptotic cells in wild-type and Ptdsr -/-embryos. (a) Whole-mount TUNEL staining (blue) of limb buds from wild-type and Ptdsr -/- embryos at E13.5 show no differences in the amount or localization of apoptotic cells during the beginning regression of the interdigital web. Serial sagittal sections stained for activated caspase 3 (aCasp3; red) in (b-d) wild-type and (f-h) Ptdsr -/- embryos at E12.5 show apoptotic cells in the neural tube (b,f), the mesonephros (c,g) and the developing paravertebral ganglia (d,h). Tissue distribution and total number of apoptotic cells was indistinguishable between genotypes and was confirmed by the comparison of consecutive sections of wild-type and Ptdsr -/-embryos from different developmental stages. Analysis of macrophage numbers and location by F4/80 staining (brown) of consecutive sections in paravertebral ganglia of (e) wild-type and (i) homozygous mutant embryos revealed that macrophages (arrows) are not located close to apoptotic cells during embryonic development. (For comparison, see also Additional data file 1, Figure S1, with the online version of this article). Scale bar, 100 μm.
Figure 7 Phagocytosis of apoptotic cells by fetal liver-derived macrophages (FLDMs). FLDMs from (a,b) wild-type and (c,d) Ptdsr -/- embryos were cultured for 60 min with TAMRA-stained (red) apoptotic thymocytes (treated with staurosporine) from C57BL/6J mice and then stained with F4/80 (green). Macrophages of both genotypes have phagocytosed apoptotic cells (arrowheads). (e) Quantification of phagocytosis of apoptotic cells by wild-type or Ptdsr -/-macrophages revealed no differences in the percentage of macrophages that had engulfed apoptotic cells, whether or not apoptosis had been induced by staurosporine. Microscopic analysis (b,d) and quantification of the number of apoptotic cells phagocytosed by single macrophages and (f) calculation of the average number of cells phagocytosed per macrophage failed to reveal differences in the efficacy of removal of apoptotic cells between wild-type and Ptdsr -/- FLDMs.
Figure 8 Cytokine production by FLDMs upon stimulation with lipopolysaccharide (LPS) and apoptotic cells. FLDMs from wild-type and Ptdsr -/- embryos were incubated (a,b,d) with medium (0), LPS (10 ng/ml), apoptotic cells (ratio 1:10) or in combination with LPS and apoptotic cells or (c) with LPS (100 ng/ml) alone. Culture supernatants were harvested after 22 h (a,b,d) or at the indicated time points (c). TNF-α and TGF-β1 were quantified by ELISA and IL-10 by cytometric bead array (CBA) assay. Data are presented as mean ± SEM from at least three independent experiments, each carried out in triplicate. *, significant difference between genotypes, p < 0.05; **, significant difference between genotypes, p < 0.01; Wilcoxon-signed rank test.
Table 1 Penetrance of phenotypes in Ptdsr -/- mice from E9.5 to E18.5, as detected by gross morphology
Dysmorphic phenotypes Ratio in analyzed mice (affected/total) Penetrance (%)
Head malformations 9/198 4.5
cleft 4/198 2.0
others 5/198 2.5
Edema (E12.5-E15.5) 15/155 9.7
Pale appearance (= E14.5) 72/72 100
Ocular lesions 28/198 14.1
unilaterally absent eyes 21/198 10.6
right 16/198 8.1
left 5/198 2.5
bilaterally absent eyes 7/198 3.5
Subsets of the major categories of malformation are indicated by indentation.
==== Refs
Jacobson MD Weil M Raff MC Programmed cell death in animal development Cell 1997 88 347 354 9039261 10.1016/S0092-8674(00)81873-5
Baehrecke EH How death shapes life during development Nat Rev Mol Cell Biol 2002 3 779 787 12360194 10.1038/nrm931
Vaux DL Korsmeyer SJ Cell death in development Cell 1999 96 245 254 9988219 10.1016/S0092-8674(00)80564-4
Savill J Fadok V Corpse clearance defines the meaning of cell death Nature 2000 407 784 788 11048729 10.1038/35037722
Lauber K Blumenthal SG Waibel M Wesselborg S Clearance of apoptotic cells: getting rid of the corpses Mol Cell 2004 14 277 287 15125832 10.1016/S1097-2765(04)00237-0
Savill J Dransfield I Gregory C Haslett C A blast from the past: clearance of apoptotic cells regulates immune responses Nat Rev Immunol 2002 2 965 975 12461569 10.1038/nri957
Franc NC Heitzler P Ezekowitz RA White K Requirement for croquemort in phagocytosis of apoptotic cells in Drosophila Science 1999 284 1991 1994 10373118 10.1126/science.284.5422.1991
Gumienny TL Brugnera E Tosello-Trampont AC Kinchen JM Haney LB Nishiwaki K Walk SF Nemergut ME Macara IG Francis R CED-12/ELMO, a novel member of the CrkII/Dock180/Rac pathway, is required for phagocytosis and cell migration Cell 2001 107 27 41 11595183 10.1016/S0092-8674(01)00520-7
Fadok VA Warner ML Bratton DL Henson PM CD36 is required for phagocytosis of apoptotic cells by human macrophages that use either a phosphatidylserine receptor or the vitronectin receptor αvβ3 J Immunol 1998 161 6250 6257 9834113
Platt N Suzuki H Kurihara Y Kodama T Gordon S Role for the class A macrophage scavenger receptor in the phagocytosis of apoptotic thymocytes in vitro Proc Natl Acad Sci USA 1996 93 12456 12460 8901603 10.1073/pnas.93.22.12456
Devitt A Moffatt OD Raykundalia C Capra JD Simmons DL Gregory CD Human CD14 mediates recognition and phagocytosis of apoptotic cells Nature 1998 392 505 509 9548256 10.1038/33169
Savill J Hogg N Ren Y Haslett C Thrombospondin cooperates with CD36 and the vitronectin receptor in macrophage recognition of neutrophils undergoing apoptosis J Clin Invest 1992 90 1513 1522 1383273
Ogden CA deCathelineau A Hoffmann PR Bratton D Ghebrehiwet B Fadok VA Henson PM C1q and mannose binding lectin engagement of cell surface calreticulin and CD91 initiates macropinocytosis and uptake of apoptotic cells J Exp Med 2001 194 781 795 11560994 10.1084/jem.194.6.781
Grimsley C Ravichandran KS Cues for apoptotic cell engulfment: eat-me, don't-eat-me and come-get-me signals Trends Cell Biol 2003 13 648 656 14624843 10.1016/j.tcb.2003.10.004
Botto M Dell'Agnola C Bygrave AE Thompson EM Cook HT Petry F Loos M Pandolfi PP Walport MJ Homozygous C1q deficiency causes glomerulonephritis associated with multiple apoptotic bodies Nat Genet 1998 19 56 59 9590289
Scott RS McMahon EJ Pop SM Reap EA Caricchio R Cohen PL Earp HS Matsushima GK Phagocytosis and clearance of apoptotic cells is mediated by MER Nature 2001 411 207 211 11346799 10.1038/35075603
Szondy Z Sarang Z Molnar P Nemeth T Piacentini M Mastroberardino PG Falasca L Aeschlimann D Kovacs J Kiss I Transglutaminase 2-/- mice reveal a phagocytosis-associated crosstalk between macrophages and apoptotic cells Proc Natl Acad Sci USA 2003 100 7812 7817 12810961 10.1073/pnas.0832466100
Hanayama R Tanaka M Miyasaka K Aozasa K Koike M Uchiyama Y Nagata S Autoimmune disease and impaired uptake of apoptotic cells in MFG-E8-deficient mice Science 2004 304 1147 1150 15155946 10.1126/science.1094359
Fadok VA Voelker DR Campbell PA Cohen JJ Bratton DL Henson PM Exposure of phosphatidylserine on the surface of apoptotic lymphocytes triggers specific recognition and removal by macrophages J Immunol 1992 148 2207 2216 1545126
Balasubramanian K Schroit AJ Characterization of phosphatidylserine-dependent β2-glycoprotein I macrophage interactions. Implications for apoptotic cell clearance by phagocytes J Biol Chem 1998 273 29272 29277 9786940 10.1074/jbc.273.44.29272
Anderson HA Maylock CA Williams JA Paweletz CP Shu H Shacter E Serum-derived protein S binds to phosphatidylserine and stimulates the phagocytosis of apoptotic cells Nat Immunol 2003 4 87 91 12447359 10.1038/ni871
Nakano T Ishimoto Y Kishino J Umeda M Inoue K Nagata K Ohashi K Mizuno K Arita H Cell adhesion to phosphatidylserine mediated by a product of growth arrest-specific gene 6 J Biol Chem 1997 272 29411 29414 9367994 10.1074/jbc.272.47.29411
Mevorach D Mascarenhas JO Gershov D Elkon KB Complement-dependent clearance of apoptotic cells by human macrophages J Exp Med 1998 188 2313 2320 9858517 10.1084/jem.188.12.2313
Hanayama R Tanaka M Miwa K Shinohara A Iwamatsu A Nagata S Identification of a factor that links apoptotic cells to phagocytes Nature 2002 417 182 187 12000961 10.1038/417182a
Arur S Uche UE Rezaul K Fong M Scranton V Cowan AE Mohler W Han DK Annexin I is an endogenous ligand that mediates apoptotic cell engulfment Dev Cell 2003 4 587 598 12689596 10.1016/S1534-5807(03)00090-X
Fadok VA Bratton DL Rose DM Pearson A Ezekewitz RA Henson PM A receptor for phosphatidylserine-specific clearance of apoptotic cells Nature 2000 405 85 90 10811223 10.1038/35011084
Huynh ML Fadok VA Henson PM Phosphatidylserine-dependent ingestion of apoptotic cells promotes TGF-β1 secretion and the resolution of inflammation J Clin Invest 2002 109 41 50 11781349 10.1172/JCI200211638
Henson PM Bratton DL Fadok VA The phosphatidylserine receptor: a crucial molecular switch? Nat Rev Mol Cell Biol 2001 2 627 633 11483996 10.1038/35085094
Cui P Qin B Liu N Pan G Pei D Nuclear localization of the phosphatidylserine receptor protein via multiple nuclear localization signals Exp Cell Res 2004 293 154 163 14729065 10.1016/j.yexcr.2003.09.023
Cikala M Alexandrova O David CN Proschel M Stiening B Cramer P Bottger A The phosphatidylserine receptor from Hydra is a nuclear protein with potential Fe(II)-dependent oxygenase activity BMC Cell Biol 2004 5 26 15193161 10.1186/1471-2121-5-26
Li MO Sarkisian MR Mehal WZ Rakic P Flavell RA Phosphatidylserine receptor is required for clearance of apoptotic cells Science 2003 302 1560 1563 14645847 10.1126/science.1087621
Kunisaki Y Masuko S Noda M Inayoshi A Sanui T Harada M Sasazuki T Fukui Y Defective fetal liver erythropoiesis and T-lymphopoiesis in mice lacking the phosphatidylserine receptor Blood 2004 103 3362 3364 14715629 10.1182/blood-2003-09-3245
Köntgen F Suss G Stewart C Steinmetz M Bluethmann H Targeted disruption of the MHC class II Aa gene in C57BL/6 mice Int Immunol 1993 5 957 964 8398989
Affolter M Bellusci S Itoh N Shilo B Thiery JP Werb Z Tube or not tube: remodeling epithelial tissues by branching morphogenesis Dev Cell 2003 4 11 18 12530959 10.1016/S1534-5807(02)00410-0
Chuang PT McMahon AP Branching morphogenesis of the lung: new molecular insights into an old problem Trends Cell Biol 2003 13 86 91 12559759 10.1016/S0962-8924(02)00031-4
Debnath J Mills KR Collins NL Reginato MJ Muthuswamy SK Brugge JS The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini Cell 2002 111 29 40 12372298 10.1016/S0092-8674(02)01001-2
Chow RL Lang RA Early eye development in vertebrates Annu Rev Cell Dev Biol 2001 17 255 296 11687490 10.1146/annurev.cellbio.17.1.255
Graw J The genetic and molecular basis of congenital eye defects Nat Rev Genet 2003 4 876 888 14634635 10.1038/nrg1202
Lagutin OV Zhu CC Kobayashi D Topczewski J Shimamura K Puelles L Russell HR McKinnon PJ Solnica-Krezel L Oliver G Six3 repression of Wnt signaling in the anterior neuroectoderm is essential for vertebrate forebrain development Genes Dev 2003 17 368 379 12569128 10.1101/gad.1059403
Grindley JC Davidson DR Hill RE The role of Pax-6 in eye and nasal development Development 1995 121 1433 1442 7789273
Zhang L Mathers PH Jamrich M Function of Rx, but not Pax6, is essential for the formation of retinal progenitor cells in mice Genesis 2000 28 135 142 11105055 10.1002/1526-968X(200011/12)28:3/4<135::AID-GENE70>3.3.CO;2-G
Pequignot MO Provost AC Salle S Taupin P Sainton KM Marchant D Martinou JC Ameisen JC Jais JP Abitbo M Major role of BAX in apoptosis during retinal development and in establishment of a functional postnatal retina Dev Dyn 2003 228 231 238 14517994 10.1002/dvdy.10376
Hisatomi T Sakamoto T Sonoda KH Tsutsumi C Qiao H Enaida H Yamanaka I Kubota T Ishibashi T Kura S Clearance of apoptotic photoreceptors: elimination of apoptotic debris into the subretinal space and macrophage-mediated phagocytosis via phosphatidylserine receptor and integrin αvβ3 Am J Pathol 2003 162 1869 1879 12759244
Zheng TS Hunot S Kuida K Flavell RA Caspase knockouts: matters of life and death Cell Death Differ 1999 6 1043 1053 10578172 10.1038/sj.cdd.4400593
Wang X Wu YC Fadok VA Lee MC Gengyo-Ando K Cheng LC Ledwich D Hsu PK Chen JY Chou BK Cell corpse engulfment mediated by C. elegans phosphatidylserine receptor through CED-5 and CED-12 Science 2003 302 1563 1566 14645848 10.1126/science.1087641
Jyonouchi H Airway epithelium and apoptosis Apoptosis 1999 4 407 417 14634325 10.1023/A:1009607607603
Wood W Turmaine M Weber R Camp V Maki RA McKercher SR Martin P Mesenchymal cells engulf and clear apoptotic footplate cells in macrophageless PU.1 null mouse embryos Development 2000 127 5245 5252 11076747
Stryke D Kawamoto M Huang CC Johns SJ King LA Harper CA Meng EC Lee RE Yee A L'Italien L BayGenomics: a resource of insertional mutations in mouse embryonic stem cells Nucleic Acids Res 2003 31 278 281 12520002 10.1093/nar/gkg064
The Jackson Laboratory Induced Mutant Resource
Conlon RA Reaume AG Rossant J Notch1 is required for the coordinate segmentation of somites Development 1995 121 1533 1545 7789282
Mitchell KJ Pinson KI Kelly OG Brennan J Zupicich J Scherz P Leighton PA Goodrich LV Lu X Avery BJ Functional analysis of secreted and transmembrane proteins critical to mouse development Nat Genet 2001 28 241 249 11431694 10.1038/90074
Frank R Spot-synthesis: an easy technique for the positionally addressable, parallel chemical synthesis on a membrane support Tetrahedron 1992 48 9217 9232 10.1016/S0040-4020(01)85612-X
| 15345036 | PMC549712 | CC BY | 2021-01-04 16:05:44 | no | J Biol. 2004 Aug 23; 3(4):15 | utf-8 | J Biol | 2,004 | 10.1186/jbiol10 | oa_comm |
==== Front
J BiolJournal of Biology1478-58541475-4924BioMed Central London jbiol111534503510.1186/jbiol11Research ArticleAdaptive evolution of centromere proteins in plants and animals Talbert Paul B [email protected] Terri D 1Henikoff Steven [email protected] Howard Hughes Medical Institute, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N, Seattle, WA 98109-1024, USA2004 31 8 2004 3 4 18 18 25 5 2004 20 7 2004 22 7 2004 Copyright © 2004 Talbert 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
Centromeres represent the last frontiers of plant and animal genomics. Although they perform a conserved function in chromosome segregation, centromeres are typically composed of repetitive satellite sequences that are rapidly evolving. The nucleosomes of centromeres are characterized by a special H3-like histone (CenH3), which evolves rapidly and adaptively in Drosophila and Arabidopsis. Most plant, animal and fungal centromeres also bind a large protein, centromere protein C (CENP-C), that is characterized by a single 24 amino-acid motif (CENPC motif).
Results
Whereas we find no evidence that mammalian CenH3 (CENP-A) has been evolving adaptively, mammalian CENP-C proteins contain adaptively evolving regions that overlap with regions of DNA-binding activity. In plants we find that CENP-C proteins have complex duplicated regions, with conserved amino and carboxyl termini that are dissimilar in sequence to their counterparts in animals and fungi. Comparisons of Cenpc genes from Arabidopsis species and from grasses revealed multiple regions that are under positive selection, including duplicated exons in some grasses. In contrast to plants and animals, yeast CENP-C (Mif2p) is under negative selection.
Conclusions
CENP-Cs in all plant and animal lineages examined have regions that are rapidly and adaptively evolving. To explain these remarkable evolutionary features for a single-copy gene that is needed at every mitosis, we propose that CENP-Cs, like some CenH3s, suppress meiotic drive of centromeres during female meiosis. This process can account for the rapid evolution and the complexity of centromeric DNA in plants and animals as compared to fungi.
==== Body
Background
Centromeres are the chromosomal loci where kinetochores assemble to serve as attachment sites for the spindle microtubules that direct chromosome segregation during mitosis and meiosis. Despite this essential conserved function in all eukaryotes, centromere structure is highly variable, ranging from the simple short centromeres of budding yeast, which have a consensus sequence of approximately 125 base pairs (bp) on each chromosome, to holokinetic centromeres that span the entire length of a chromosome [1]. In plants and animals, centromeres are large and complex, typically comprising megabase-sized arrays of tandemly repeated satellite sequences that are rapidly evolving [2] and may differ significantly between closely related species [3-5]. The failure of conventional cloning and sequencing assembly tools to adequately characterize rapidly evolving satellite sequences at centromeres has made them the last regions of most eukaryotic genomes to be well understood [1].
Although there is no discernable conservation of centromeric DNA sequences in disparate eukaryotes, considerable progress has been made in identifying common proteins that form the kinetochore [6]. A universal protein component of centromeric chromatin found in all eukaryotes that have been examined is a centromere-specific variant of histone H3 (CenH3), which replaces canonical H3 in centromeric nucleosomes [7,8]. CenH3s are essential kinetochore components yet, like centromeric DNA, they are rapidly evolving [1]. In both Drosophila [9] and Arabidopsis [10], this rapid evolution of CenH3s is associated with positive selection (adaptive evolution), and involves regions of CenH3 that are predicted to contact the centromeric DNA [9,11,12].
The finding of positive selection in a protein that is required at every cell division is remarkable. Ancient proteins with conserved function are expected to be under negative selection because they typically have achieved an optimal sequence, so new mutations tend to produce deleterious variants that are quickly eliminated from populations. The canonical histones are extreme examples of this type of protein. In contrast, recurrent positive selection generally occurs as a consequence of genetic conflict, for example in the 'arms race' between pathogen surface antigens and the immune-cell proteins that recognize them. In this case, a mutation in a surface antigen that allows the pathogen to escape detection and proliferate will trigger selection for a new immune receptor to fight the mutated pathogen, which can then mutate again, and so on. The evidence for positive selection of CenH3 proteins specifically in the regions that contact DNA thus suggests a conflict between centromeric DNA and a histone component of the nucleosome that packages it. Is it commonplace for eukaryotes to have such a conflict at their centromeres? Is the conflict unique to centromere-specific histones, or are other proteins that bind centromeres also involved in this conflict? Is conflict responsible for centromere complexity? To answer these questions, we investigated the evolution of a second common DNA-binding kinetochore protein.
Of the handful of essential kinetochore proteins that are widely distributed among eukaryotes, only one class other than CenH3 has been shown to bind centromeric DNA: centromere protein C (CENP-C), a conserved component of the inner kinetochore in vertebrates [13-16]. Human CENP-C binds DNA non-specifically in vitro [17-19] and binds centromeric alpha satellite DNA in vivo [20,21]. Vertebrate CENP-C and the yeast centromere protein Mif2p [22,23] share a 24 amino-acid motif (CENPC motif) that has also been found in kinetochore proteins in nematodes [24] and plants [25]. As expected for kinetochore proteins, disruption or inactivation of genes encoding proteins containing a CENPC motif (CENP-Cs) results in the failure of proper chromosome segregation [16,23,24,26-28].
Other than the defining CENPC motif, these proteins are dissimilar in sequence across disparate phyla. Such a small stretch of sequence conservation, accounting for less than 5% of the length of these 549-943 amino-acid proteins, is unexpected considering that CENP-Cs are encoded by essential single-copy genes that are expected to be subject to strong negative selection. We therefore wondered whether the same evolutionary forces responsible for the rapid evolution of CenH3s cause divergence of CENP-Cs outside of the CENPC motif.
Here, we describe coding sequences from several unreported Cenpc genes and test whether Cenpc genes are in general, like CenH3 genes, subject to positive selection. We find evidence for adaptive evolution of CENP-C in plants and animals, but we find negative selection in yeasts. Our results provide support for a meiotic drive model of centromere evolution.
Results and discussion
CenH3s evolve under negative selection in some lineages
Previous work has shown that CenH3s are evolving adaptively in Drosophila and Arabidopsis [9,10], but their mode of evolution in mammals is not known. Selective forces acting on proteins can be measured by comparing the estimated rates of nonsynonymous nucleotide substitution (Ka) and synonymous substitution (Ks) between coding sequences from closely related species. These rates are expected to be equal if the coding sequences are evolving neutrally (Ka/Ks = 1). Negative selection is indicated by Ka/Ks < 1, and positive selection is indicated by Ka/Ks > 1.
To obtain a pair of closely related mammalian CenH3s, we used the sequence of the mouse (Mus musculus) CenH3, CENP-A [29], to query the High Throughput Genomic Sequences portion of the GenBank database [30] with a tblastn search, and identified a rat (Rattus norvegicus) genomic clone (AC110465) that contains the predicted rat CENP-A coding sequence. The predicted CENP-A protein is encoded in four exons and is 87% identical in amino-acid sequence to mouse CENP-A, excluding a 25 amino-acid insertion that appears to derive from a duplication of the amino terminus (Figure 1). This gene model is partially supported by an expressed sequence tag (EST; BF561223) that includes the first three exons, but which terminates in the predicted intron 3.
To determine whether Cenpa is evolving adaptively in rodents, we compared Ka and Ks between mouse and rat using K-estimator [31]. Positive selection in single-copy genes that are essential in every cell is expected to be localized and more difficult to detect than in nonessential genes or members of multigene families because of simultaneous negative selection to maintain their essential functions. In Drosophila and Arabidopsis, CenH3s are under positive selection in their tails, but also under negative selection in much of their histone-fold domains. We therefore used the sliding-window function of K-estimator to scan through the coding sequences using 99 bp windows every 33 bp in an effort to find regions of positive selection. This analysis detected statistically significant negative selection for all of the windows except one that failed to rule out neutrality, indicating that CENP-A is under negative selection (Ka = 0.11, Ks = 0.33; Ka < Ks with p < 0.001) in both the tail and the histone-fold domains. Similar results were obtained when comparing either sequence with the Cenpa gene from Chinese hamster (Cricetulus griseus) [32], although the greater divergence (Ks = 0.45 rat, 0.67 mouse) makes the statistical conclusion near the limit of reliability (Ks ≤ ~0.5) because of the increased likelihood of multiple substitutions. Thus, CENP-A appears to have been under negative selection throughout its length in multiple rodent lineages.
We also compared the human Cenpa gene [33] with the Cenpa gene from chimpanzee (Pan troglodytes). A blastn search of the Genome Sequencing Center's assembly of the chimpanzee genome [34] using human Cenpa identified the chimp Cenpa gene encoded in four exons in Contig 286.218. We searched the NCBI trace archives [35] to verify the sequence and the existence of appropriate putative intron splice sites. The predicted chimpanzee Cenpa gene differs from the human gene by six synonymous nucleotide substitutions and an indel (insertion or deletion) of two codons. This excess of synonymous substitutions indicates negative selection of CENP-A (p < 0.01). Overall negative selection of CENP-A appears also to extend to the bovine (CB455530) protein, given the relatively high degree of conservation seen for all regions, including the tail and Loop 1 regions that evolve adaptively in Drosophila (Figure 1a).
We also found overall negative selection in CenH3s of grasses. We used the CENH3 gene (AF519807) of maize (Zea mays) [36] to search ESTs [37] from sugarcane (Saccharum officinarum), and identified three that encode full-length CENH3 genes (CA119873, CA127217, and CA142604). The CenH3 proteins encoded by these ESTs differ from each other by 2-4 amino acids. Because sugarcane is thought to be octaploid, these variants may represent co-expressed homeologs. The coding regions of ESTs CA119873 and CA127217 differ by four synonymous and four nonsynonymous substitutions (Ks = 0.03, Ka = 0.01), suggesting negative selection. Comparison of either of these sequences with maize CENH3 by sliding-window analysis found that all windows had Ks > Ka, with overall negative selection (Ks = 0.24, Ka = 0.13; p < 0.01). Thus, in contrast to CenH3s in Arabidopsis and Drosophila, CenH3s of rodents, primates, and grasses appear not to be evolving adaptively.
The evident lack of positive selection on CenH3 in mammals and grasses raises the possibility that another kinetochore protein is evolving in conflict with centromeric DNA in these organisms, in which centromeric satellite sequences are known to be evolving rapidly [2,38]. We focused on CENP-C, which is found to co-localize with CenH3 to the inner kinetochore in humans [13] and maize [36].
Mammalian CENP-C is evolving adaptively
To address the possibility that CENP-C is adaptively evolving in mammals, we used the mouse sequence [14] as a query in a tblastn search to identify Cenpc ESTs from rat. From these ESTs (see Additional data file 1), we obtained and sequenced a full-length cDNA (see Additional data file 2), and compared its coding sequence with that of the mouse Cenpc gene (68% predicted amino-acid identity). We found positive selection over most of the amino-terminal two-thirds of the coding sequence, interrupted by one region of significant negative selection (mouse codons 208-273), one region of nearly significant negative selection (mouse 410-464), and three short regions without significant selection (Figure 2a; Table 1). Most of the carboxy-terminal one-third of the protein, including the CENPC motif and an additional region that is homologous to the budding yeast CENP-C protein Mif2p [22,23], has been under negative selection. We conclude that at least some regions of Cenpc genes are evolving adaptively in rodents.
To determine whether any of these regions is also under positive selection in primates, we identified the Cenpc gene of chimpanzee by using the human Cenpc coding sequence (GenBank accession number M95724) to search the assembled chimpanzee genome and the NCBI trace archives. We found that the chimpanzee genome contains a single copy of the Cenpc structural gene (contigs 375.88-375.100), as well as a processed Cenpc pseudogene (contigs 76.642-76.643), as has been found in humans [14,18,39]. The predicted chimpanzee Cenpc coding sequence differs by 17 nucleotide substitutions from the human cDNA sequence, with Ks = 0.0054 and Ka = 0.0063. The > 99% identity of the human and chimp coding sequences provides little opportunity to detect selection, but using sliding-window analysis we found a single region of significant positive selection (human codons 278-585) that overlaps the central regions of positive selection found in the more divergent rat-mouse comparison, indicating that the central portion of CENP-C is under positive selection in both rodents and primates.
To confirm these results, we applied the codeml program of PAML [40] to a multiple sequence alignment of mammalian CENP-Cs. PAML calculates the likelihood of models for neutral and adaptive evolution based on a tree and estimates Ka/Ks ratios. We compared the null model with two fixed site classes (Ka/Ks = 0 or 1) to a 'data-driven' model in which two classes of sites were estimated from the data. The data-driven model was found to be significantly more probable than the null model (χ2 = 8.7; p = 0.01) with Ka/Ks = 0.20 for 57% of the 685 sites in the multiple alignment and Ka/Ks = 1.64 for 43% of the sites (data not shown). Similar results were obtained using either a DNA- or a protein-based tree, or testing more complex models. When the same tests were applied to the core region of 11 aligned Brassicaceae (mustard family) CenH3s, only 17% of residues were estimated to be in the positive selection class (Ka/Ks = 2.54) ([11] and data not shown), which indicates that positive selection on mammalian CENP-C has occurred more extensively than on CenH3s.
Amino-acid sites of positive selection in mammalian CENP-Cs were identified as those with significant posterior probabilities. These were found to be scattered throughout the multiply aligned region with 5 of the 18 highly significant sites prominently clustered within 25 residues (human codons 424-448) in a region of positive selection identified by K-estimator analysis. Therefore, pairwise K-estimator and multiple PAML analyses yield similar results and reveal that large regions of mammalian CENP-Cs have been adaptively evolving.
Adaptively evolving regions overlap DNA-binding and centromere-targeting regions
The regions of positive selection in rodent and primate CENP-Cs overlap some protein landmarks identified in functional analyses of human CENP-C. The binding activity of human CENP-C to DNA in vitro has been mapped by two groups of investigators. Sugimoto and colleagues [17,18] found that the region including amino acids 396-498 bound DNA and was stabilized by including flanking amino acids on one or both sides (330-498 or 396-581; Figure 3a), suggesting that at least two regions in the central portion of the protein contribute to DNA binding. Yang and colleagues [19] identified two non-overlapping DNA-binding regions: amino acids 23-440 and 459-943. They found a weak DNA-binding activity at the carboxyl terminus in region 638-943, which includes the CENPC motif (737-759) and the conserved Mif2p-homologous region (890-941). This suggests that region 459-943 itself contains at least two DNA-binding regions, a weak one at region 638-943, and a stronger one that may correspond to region 396-581 described by Sugimoto and colleagues. Both the central region and the carboxyl terminus have been shown to bind DNA in vivo [21]. Comparison of the regions of positive selection found in rodents and primates with these DNA-binding regions reveals extensive overlap with the central DNA-binding regions (Figure 3a), including the cluster of highly significant sites between codons 424 and 448 identified by PAML analysis. This is consistent with previous evidence that adaptive evolution of CenH3s occurs in regions that have been implicated in DNA binding [9,11]. No positive selection was observed for the poorly mapped carboxy-terminal DNA-binding domain in our sliding-window analysis, suggesting either that this DNA-binding domain is not evolving adaptively or that strong negative selection on the CENPC motif can obscure detection by our sliding-window analysis of positive selection on nearby amino acids that contact centromeric DNA. In the DNA-binding Loop 1 region of Arabidopsis CenH3, adaptively evolving codons are found in close proximity to codons under strong negative selection [11].
In human CENP-C, three regions have been reported to confer centromere targeting. One targeting signal was recently reported in region 283-429 [41]. A second targeting region was mapped by mutation to region 522-534, with arginine 522 crucial for localization [42]. Targeting by the conserved carboxyl terminus (728-943) occurs for species as distant as Xenopus [21,41-43]. A segment that includes both the first and second targeting regions (1-584) failed to confer targeteting to centromeres in hamster BHK cells, however [43]. We find that these two targeting regions are within the region of positive selection in primates and overlap with three of the regions of positive selection in rodents. A correspondence between centromere targeting and adaptive evolution has been noted for Drosophila CenH3, where the adaptively evolving Loop 1 region has been shown to be necessary and sufficient for targeting when swapped between native and heterologous orthologs [44]. Therefore, the lack of centromeric targeting of a human CENP-C fragment containing the first and second targeting regions in the heterologous hamster system might be attributed to adaptive evolution of DNA-binding specificity in these regions.
Targeting of native CENP-C proteins depends on other centromere proteins that vary according to species [45], but the dependence of CENP-Cs on CenH3s for targeting appears to be universal [24,46-49]. This dependence suggests that CENP-C proteins contain a conserved CenH3-interacting region, for which the CENPC motif is the only obvious candidate. The first half of the CENPC motif is rich in arginines, whereas the second half has mixed chemical properties including three aromatic residues (Figure 3c). In the non-specific binding of nucleosome cores to DNA, 14 DNA contacts are made by arginines binding to the minor groove [50]. This suggests that the weak DNA binding of the carboxyl terminus of CENP-C may be mediated by the arginines of the CENPC motif, with the remainder of the motif contacting a conserved structural feature of centromeric nucleosomes.
Not all regions of CENP-C that display positive selection correspond to regions that bind DNA in vitro or that are sufficient for targeting centromeres. For example, the region comprising the most amino-terminal 200 or so amino acids of rodent CENP-C has been evolving adaptively, but the orthologous region in human CENP-C fails to bind DNA in a southwestern assay [17,19] or to localize to centromeres of human embryonic kidney cells [21]. This suggests that the amino-terminal region of CENP-C plays a supporting role in packaging centromeric chromatin. A parallel situation appears to hold for the adaptively evolving amino-terminal tail of Drosophila CenH3, which was found to be neither necessary nor sufficient for targeting in vivo to homologous centromeres. In this case, Loop 1 was identified as the targeting domain, and the amino-terminal tail was hypothesized to help stabilize higher-order chromatin structure by binding to linker DNA, similar to the known binding activity of canonical histone tails [44]. If CENP-C in mammals is subject to the same evolutionary forces that shape the adaptive evolution of the CenH3 tail in Drosophila, then CENP-C might be playing a comparable role in the stabilization of higher-order centromeric chromatin.
Positive selection in the central DNA-binding and centromere-targeting region of CENP-C offers an explanation for the lack of conservation of this region between chicken and mammals [51]: as positive selection acts on the amino acids that contact rapidly evolving centromeric satellites and that serve to target the protein to a specific but ever-changing substrate, it may eventually erase all recognizable homology in these protein regions.
Cenpc gene structure and conservation in plants
Our finding that adaptive evolution is occurring in animal CENP-Cs encouraged a similar survey of plant CENP-Cs, because centromeres from both animals and seed plants comprise rapidly evolving satellite sequences. At the time we began this study, Cenpc genes in plants had been characterized only in maize (Z. mays), so we needed first to identify Cenpc homologs from other plants to ascertain whether or not the gene is evolving adaptively.
Three Cenpc homologs have been described in maize: CenpcA, CenpcB, and CenpcC [25]. Immunological localization of CENP-CA to maize centromeres indicates that it is probably functional, so plant relatives of maize CENP-CA should also represent CENP-Cs. We used the CENP-CA protein sequence (AAD39434) as a query in a tblastn search of GenBank, and identified a single Cenpc homolog (AC013453, At1g15660) in the genome of Arabidopsis thaliana by sequence similarity at both protein termini (Figure 4). Isolation and sequencing of a full-length Cenpc cDNA (Additional data file 2) revealed that the 705 amino-acid CENP-C protein of Arabidopsis is encoded in 11 exons, with the CENPC motif encoded in exon 10 (Figure 5). Recently, Arabidopsis CENP-C has been found to localize to Arabidopsis centromeres [52].
We searched the GenBank EST database, querying with the predicted protein sequences of maize CENP-CA and Arabidopsis CENP-C. We identified ESTs from putative plant Cenpc genes in 20 angiosperm species representing eight families and in the moss Physcomitrella patens (see Additional data file 1). We obtained the cDNA clones corresponding to 16 of these ESTs and sequenced them completely (see Additional data file 2). An alignment of the carboxyl termini encoded by cDNAs representing six angiosperm families revealed that the final 80 or so amino acids of CENP-C, including the CENPC motif, are highly conserved in plants (Figure 4b). For comparison, the carboxyl termini of vertebrate CENP-C proteins have approximately 180 amino acids following the CENPC motif (Figure 3a), including a block of 52 amino acids that is conserved in yeast Mif2p [22,23], but not in nematodes [24]. The carboxyl termini of plant CENP-Cs do not show significant similarity to animal and fungal CENP-Cs except for the CENPC motif.
As an aid in identifying other conserved regions of angiosperm CENP-Cs, we developed gene models for full-length Cenpc cDNAs by aligning them with available genomic sequences (Additional data file 1). A full-length cDNA from barrel medic (Medicago truncatula) encodes a protein of 697 amino acids, which corresponds to a gene model of eleven exons when aligned to a genomic pseudogene (Figure 5). We also predicted gene models for Cenpc genes in the grasses using cDNAs and genomic sequences from rice (Oryza sativa), maize, and sorghum (Sorghum bicolor) (Figure 5). The maize gene model of 14 exons suggests an explanation for the anomalous maize cDNA 'CenpcC' (AF129859) [25], which differs from all other plant Cenpcs in encoding an unrelated carboxyl terminus. CenpcC is 99.9% identical to maize CenpcA until it diverges downstream of the CENPC motif at the point corresponding to the end of exon 13 in our gene model. On the basis of an overlap with maize and Sorghum genomic sequence that spans the intron between exons 13 and 14, we conclude that the divergent 3' end of CenpcC derives from the unspliced intron 13 of CenpcA, and that all angiosperm CENP-Cs share a highly conserved carboxyl terminus.
Comparing the gene models of Arabidopsis, barrel medic, maize, Sorghum, and rice, the limited conservation of the encoded amino-acid sequences and approximate correspondence of exon sizes suggest that the exons in the amino-terminal half and the final two exons of plant CENP-C are conserved (Figures 3,5). The middle region does not show conservation of intron position or encoded peptide sequence, indicating rapid evolution within angiosperms. We assumed conservation of the first five intron positions in the 5' half of the coding sequence to generate an amino-terminal alignment that represents five families, including the protein encoded by a beet (Beta vulgaris) cDNA that appears to contain an unspliced intron. Our alignment reveals short regions of conservation throughout the amino terminus, as well as a high relative incidence of the dipeptide SQ in the poorly conserved exon 5 (Figure 4).
Despite these short regions of conservation within angiosperms, no sequence similarity between plant and animal CENP-Cs could be detected outside of the CENPC motif. Nevertheless, plant and animal CENP-Cs appear to share an overall architecture (Figure 3). Both angiosperm and vertebrate CENP-Cs [16] have regions of conservation at the amino and carboxyl termini, with little or no conservation in the middle region of the protein. Remarkably, plant and animal CENP-Cs also share the same modular exon organization for the CENPC motif, which lies within a 105-108 bp exon (encoding 35-36 amino acids) that is spliced in the same frame in both plants and animals (see Additional data file 3). Considering the similar overall lengths of plant and animal CENP-Cs, the arrangement of conserved regions, and the common location of the CENPC module, it appears that corresponding regions of the protein are evolving similarly and may serve similar functions.
Recurrent exon duplications in the grasses
Multiple alignment of plant Cenpcs revealed that one region of the gene is subject to duplication, but only in grasses. One part of the poorly conserved middle region of the gene has been repeatedly duplicated and deleted, thus encoding proteins of different sizes. In rice, an ancestral pair of exons, corresponding to exons 9 and 10 in maize CenpcA, has been triplicated in tandem (Figure 5). To facilitate comparison with maize and other grasses, we designated the rice exons as 9a-10a, 9b-10b, and 9c-10c. Exon 9c has an additional internal tandem duplication of its first 14 codons. Consensus sequences derived from overlapping truncated ESTs (Additional data file 1) and cDNAs (Additional data file 2) from the closely related species wheat (Triticum aestivum) and barley (Hordeum vulgare) indicate that there are two tandem copies of exons 9 and 10 in these species (designated 9p-10p and 9q-10q in Figure 5). We confirmed the sequence of these exons by designing primers and amplifying the corresponding regions from wheat and barley genomic DNAs. Single copies of exons 9 and 10 were found in full-length cDNAs from sugarcane, Sorghum bicolor and Sorghum propinquum (Table 2; Figure 5).
Exon duplications were also found for Sorghum species but, surprisingly, these involved a different pair of exons, 11 and 12. One full-length cDNA from S. bicolor has only a single copy of exons 11 and 12, whereas a truncated pseudogene from S. bicolor and a full-length cDNA from S. propinquum are duplicated for exons 11 and 12 (designated 11a-12a and 11b-12b). The S. bicolor pseudogene has a deletion that joins sequences just upstream of the initiation codon in exon 1 to sequences upstream of exon 2. Despite the presence of tandemly duplicated exons, the S. bicolor truncated pseudogene is more closely related to the full-length S. bicolor gene than it is to the S. propinquum gene. Exons 11 and 12 in the S. bicolor full-length gene are identical to 11b-12b in the pseudogene, but have 7 differences from 11a-12a. This suggests that the duplication of exons 11 and 12 preceded the divergence of S. propinquum and S. bicolor, and that the full-length S. bicolor gene may have been derived by loss of exons 11a-12a from a full-length ancestral gene similar to the truncated pseudogene.
We wondered why two different pairs of exons, 9-10 and 11-12, were each independently subject to duplication in the grasses. When we examined multiple alignments of the peptide sequences encoded by both exon pairs in Logos format, it became apparent that they resembled each other in length and composition (Figure 6a). Exons 9 and 11 both encode peptides of 25-28 residues that are rich in acidic amino acids, whereas exons 10 and 12 encode peptides of 30-38 residues that are rich in basic amino acids. We compared alignments of exons 9 and 11 and alignments of exons 10 and 12 using the Local Alignment of Multiple Alignments (LAMA) program, and found that these exon pairs appear to be homologous (E < 0.0001 for both comparisons). We conclude that exon pairs 9-10 and 11-12 derive from a more ancient duplication event.
To trace the likely ancestry of these duplication events, we used an alignment of the exons from multiple species to construct phylogenetic trees of duplicates of exons 9-10 and 11-12 (Figure 6b). This phylogeny suggests that there have been numerous duplication events in the history of the grasses (Figure 6c and data not shown): first, a duplication generating exons 9-10 and 11-12 in an ancestor of the grasses; second, a duplication generating exons 9p-10p and 9q-10q; third, a duplication generating exons 11a-12a and 11b-12b in the Sorghum lineage; fourth, two duplications generating rice exons 9a-10a, 9b-10b, and 9c-10c all within the rice 9q-10q lineage; and fifth, a partial duplication in rice exon 9c.
There also appear to have been at least three losses of duplications: one of exons 11a-12a in the lineage leading to the full-length S. bicolor gene, one of exons 11b-12b in the sugarcane genes, and one of the hypothetical rice 9p-10p. Alternatively, it is possible that the latter loss and one of the rice-specific duplications resulted from gene conversion of rice 9p-10p by a derivative of rice 9q-10q. Regardless of the exact number of duplication and deletion events, it is clear that the exon pair ancestral to grass exons 9-10 and 11-12 has been subjected to repeated episodes of duplication and deletion.
Plant CENP-Cs are adaptively evolving
The delineation of gene models for plant Cenpcs allowed us to analyze them for evidence of adaptive evolution. First, we compared Cenpcs from Arabidopsis species in which we had previously found adaptively evolving CenH3s. Using the A. thaliana genomic sequence to design primers, we amplified, cloned, and sequenced a Cenpc cDNA from A. arenosa (Additional data file 2). Comparing this sequence with that of A. thaliana, the predicted proteins differ by 87 amino-acid subtitutions out of 703 alignable residues, plus five indels of 1-3 amino acids.
We applied the sliding window option of K-estimator to the aligned coding sequences of A. thaliana and A. arenosa Cenpc. At three regions, Ka exceeded its 99% confidence interval for the null hypothesis, indicating that these regions are under positive selection (Figures 2b,3). These regions correspond approximately to exon 5 (codons 178-221 in the A. thaliana sequence), the 3' half of exon 6 (codons 376-441), and exons 8 and 9 (codons 486-618). In addition, a region encompassing most of exons 1 and 2 (codons 24-89) was found to be under positive selection with p < 0.03. We also determined that the 5' half of exon 6 (codons 255-386) and the conserved exons 10 and 11 (codons 595-703) are under negative selection with p < 0.01.
Curiously, an indel at the beginning of exon 9, where the A. arenosa cDNA has a CAG (glutamine) codon that is absent in the A. thaliana cDNA, appears to be caused by the species-specific use of alternative acceptor splice sites, because the genomic sequence (data not shown) at this intron-exon boundary is identical in both species (...cag cag ^GAG GGT... or ...cag ^CAG GAG GGT...). The presence of species-specific alternative splicing of the same codon in an adaptively evolving region suggests that splicing variation can contribute to adaptive variability.
To examine whether positive selection in Cenpc is unique to Arabidopsis or occurs more generally in plants, we compared Cenpc genes from the two Sorghum species. We removed the duplicate exons 11a and 12a from the S. propinquum coding sequence in order to compare the sequence with the full-length gene from S. bicolor . For this comparison, Ka = 0.014 and Ks = 0.003. Ka exceeds the 99% confidence interval of the null hypothesis, and neutral evolution can be rejected in favor of positive selection. The limited divergence between these two sequences did not allow statistically significant conclusions to be drawn about positive selection in particular regions of the gene.
To address which regions are under positive selection, we compared the S. bicolor sequence with the maize CenpcA coding sequence (75% amino-acid identity). Between maize CenpcA and S. bicolor, Ka = 0.12, Ks = 0.14, and there are seven indels of 1-11 codons. We identified positive selection for a single window in exon 1, for a region including all of exon 5, for a region in the second half of exon 6, and for a region from the end of exon 12 through most of exon 14 (Table 2 and Figure 2c). Negative selection was found for a region from exons 1-4, a region in the middle one-third of exon 6, and a region from the end of exon 6 through exon 12 (Table 2 and Figure 2c). The regions of positive selection seen in exons 1 and 5 clearly overlap the corresponding regions in Arabidopsis (Figure 3b). Although the region of positive selection seen in exon 6 of the grasses cannot be aligned with that in exon 6 of Arabidopsis because of sequence divergence, they occur in the same general area of the protein.
The region of positive selection in exons 12-14 was somewhat surprising given the strong conservation around the CENPC motif, and we wondered if this selection was specific to the maize or Sorghum lineage. To test this possibility, we compared maize CenpcA and S. bicolor Cenpc with a Cenpc gene from sugarcane. Of three sugarcane cDNAs that we obtained (Additional data file 2), two had identical coding sequences (Cenpc1), and the third (Cenpc2) differed by 13 nucleotide substitutions, suggesting that Cenpc1 and Cenpc2 may be homeologous genes in the polyploid sugarcane genome. We compared Cenpc1 to maize CenpcA and S. bicolor Cenpc. Regions of positive or negative selection identified in the maize/Sorghum comparison were generally found to coincide with regions under selection in the corresponding direction in maize/sugarcane and Sorghum/sugarcane comparisons, although in a few cases the selection was not found at the p < 0.05 level of significance in all comparisons (Table 2). We conclude that these regions are subject to recurrent adaptive evolution.
In two cases, a region was found to be under significant selection in opposite directions in different comparisons. First, a region in exon 6 (maize codons 232-286) that was not under significant selection in the maize/S. bicolor comparison was under negative selection in the maize/sugarcane comparison, but under positive selection in the S. bicolor/sugarcane comparison; this suggests that positive selection in S. bicolor and negative selection in maize combined to give a non-significant result in the maize/S. bicolor comparison. Second, the region of positive selection in exons 12-14 identified from the maize/Sorghum comparison was under positive selection in the maize/sugarcane comparison, but under negative selection in the Sorghum/sugarcane comparison, indicating that the positive selection in this region is unique to the maize CenpcA lineage (Table 2, Figure 3b). Therefore, in some regions of CENP-C adaptive evolution appears to be episodic, as has been seen previously for primate lysozymes [53].
Phylogeny-based PAML analysis confirms that plant CENP-Cs are adaptively evolving and in an episodic fashion, consistent with inferences based on pairwise K-estimator analysis. As was found for mammalian CENP-C, the data-driven model for grass CENP-C was found to be significantly more probable than the null model (χ2 = 12.0; p = 0.003) with Ka/Ks = 0.00 for 51% of the 686 sites in the multiple alignment and Ka/Ks = 2.00 for 49% of the sites (data not shown). Using the PAML 'free-ratio' option that measures Ka/Ks differences between branches in a tree [54], we found that CENP-C is adaptively evolving (χ2= 10.0; p = 0.007 for the data-driven over the null model) along both Sorghum lineages (Ka/Ks = 2.6) and along the sugarcane Cenpc2 lineage (Ka/Ks = 1.3) but not detectably along the sugarcane Cenpc1 lineage (Ka/Ks = 0.23). PAML analysis also confirmed that the carboxy-terminal region of maize CenpcA is adaptively evolving (χ2 = 7.8; p = 0.02 for the data-driven model over the null model) with Ka/Ks = 1.4. Thus, different methods of analysis demonstrate that CENP-Cs are adaptively evolving in an episodic fashion in grasses that have multiple CENP-C copies.
Maize CenpcA is co-expressed with another Cenpc gene, CenpcB [25], for which incomplete sequence information is available (AF129858, AW062057, and AY109432). The available CenpcB sequence begins in exon 5 and continues through exon 14, and has seven in-frame indels relative to CenpcA. We found negative selection in a region from the end of exon 5 through the first half of exon 6 (CenpcA codons 205-358, p < 0.01), and positive selection from the end of exon 6 through exon 7 (codons 403-492, p < 0.01). Elsewhere the null hypothesis could not be rejected. Comparing CenpcB with S. bicolor, Ka = 0.13 and Ks = 0.11, and there are six in-frame indels between the sequences. We found positive selection in the first half of exon 6 (Sorghum codons 273-327, p < 0.03) and negative selection from exon 10 through the first few codons of exon 13 (codons 537-619, p < 0.02). Elsewhere the null hypothesis could not be rejected. In summary, regions under negative selection in other grass Cenpcs can be under positive selection in CenpcB, and regions under positive selection in other grass Cenpcs are under negative selection or evolving neutrally in CenpcB (Figure 3b), suggesting that CENP-CB has been subjected to different selective forces since its divergence from CENP-CA.
Just as gene duplication can result in different selective pressures on the two genes, duplications within a gene can lead to specialization and thus can change selective pressures on the region. Such specialization appears to have occurred between the anciently duplicated region encoded by exons 9-10 and 11-12 (Figure 6a). In maize, sugarcane, and Sorghum we detected negative selection for exons 9-12, but in the more recent duplication of exons 9 and 10 in wheat and barley we detected positive selection in a region from the last codon of the first copy of exon 10 to the first four codons of exon 12 (p < 0.01). Additional windows in exons 9p-10p and 12 had Ka >Ks (p > 0.05), suggesting that most of the duplicated region has been evolving adaptively (Figure 2d). In contrast, in the adjacent conserved carboxy-terminal region (corresponding to CenpcA codons 625-690), we detected only negative selection (p = 0.01), as though exon duplication allowed for adaptation.
We find an approximate correspondence between adaptively evolving regions in angiosperms and those in mammals that overlap DNA-binding and centromere-targeting regions. Although no DNA-binding regions have been experimentally determined for plant CENP-Cs, the correspondence with animal CENP-Cs suggests that the different regions play comparable roles. One of the corresponding adaptive regions is repeatedly duplicated in grasses, and the distribution of basic residues in exons 10 and 12 suggests that the repeat unit binds DNA. A parallel situation again appears to be found for the amino-terminal tails of some Drosophila CenH3s, which contain repeats of a minor groove-binding motif that are thought to provide DNA compositional preference [12]. Thus, both plant and animal CENP-Cs show adaptively evolving features that parallel those found in CenH3s.
Yeast MIF2 is under negative selection
If positive selection for CENP-Cs in plants and animals is related to centromere complexity, then we would expect conventional negative selection to operate in organisms such as budding yeast, which have simple centromeres. The MIF2 genes of Saccharomyces cerevisiae [26] and S. paradoxus [55] are 93% identical in amino-acid sequence, with Ka = 0.036 and Ks = 0.38. In sharp contrast to all pairwise comparisons of plant and animal Cenpc genes, Ka was much less than Ks for all of the 99 bp windows of yeast MIF2, indicating that it is under negative selection throughout its length (p < 0.001). In all pairwise comparisons among these two species and the additional species S. mikatae and S. bayanus [55], we consistently found evidence of negative selection with Ka << Ks (range of Ka, 0.036-0.093; range of Ks, 0.38-0.82). We also found strong negative selection for all 99 bp windows in pairwise comparisons of yeast CenH3 (Cse4p; data not shown). Thus, adaptive evolution of both CenH3s and CENP-Cs appears to be limited to organisms with complex centromeres.
Meiotic drive model of centromere evolution
We have demonstrated that CENP-C has been adaptively evolving in multiple lineages of both plants and animals, a feature that had been previously shown for some CenH3s. Thus, the occurrence of adaptive evolution appears to be a general feature of proteins that bind to complex centromeres. Recurrent adaptive evolution implies an arms race, and an arms race involving centromeric DNA-binding proteins is remarkable given that centromeres have a conserved function. But centromeric DNA is rapidly evolving in plants and animals, so adaptation of the major centromere DNA-binding proteins would maintain an interface with the conserved kinetochore machinery. Indeed, regions of CENP-C that show evidence of positive selection include DNA-binding and specificity regions, in parallel with previous findings for Drosophila and Arabidopsis CenH3 [9,11,44].
A 'meiotic drive' model has been proposed to explain the rapid evolution of centromeric DNAs and CenH3s [1]. According to this model, centromeres compete during female meiosis for inclusion in the single meiotic product that becomes the egg nucleus and so gets transmitted to the next generation. In both animals and seed plants, which of the four meiotic products becomes the egg nucleus is determined by its position in the female tetrad. A centromere variant will increase in the population if it achieves an orientation in female meiosis resulting in its inclusion in the egg nucleus more frequently than its competitors. For example, an expansion of a satellite array may lead to a 'stronger' centromere variant with an expanded kinetochore that attracts more microtubules and results in a slightly greater probability of a favorable orientation in female meiosis. The mechanism of such orientation is unknown, but in some insects and plants the female meiotic spindle has an asymmetric distribution of microtubules or is monopolar [56], so a stronger centromere variant might better capture the favored pole. The new variant will therefore increase in the population and eventually become fixed. This meiotic drive process ('centromere drive') can account for the rapid evolution and complex structure of centromeric DNA. As a rare new variant spreads in the population, however, disparities in centromere strength may interfere with fertility in males, where the four meiotic products contribute equally to the next generation. Mutations in CenH3 that restore centromere parity in meiosis will therefore be selected in males, resulting in the adaptive evolution of CenH3 and suppression of the meiotic drive of centromeric DNA. Recurrent cycles of meiotic drive by centromere variants, or centromere drive, and suppression by CenH3 mutations would result in the observed rapid evolution of both centromeres and CenH3s.
The lack of evidence for adaptive evolution in CenH3s from mammals and grasses does not seem to fit this scenario. But the extensive positive selection on the corresponding CENP-Cs provides a ready explanation for the absence of an adaptive signal for CenH3. The meiotic drive model predicts that over evolutionary time any mutation that restores centromere parity will be selected, suggesting that proteins besides CenH3 - and in particular other kinetochore proteins that contact centromeric DNA - may be positively selected to suppress centromere drive. Our demonstration of the adaptive evolution of CENP-C, especially in DNA-binding regions, fulfills this prediction of the centromere drive model. Apparently, in mammals and grasses CENP-C performs the function of a suppressor of meiotic drive.
The large size and lack of sequence conservation of CENP-Cs make them much larger mutational targets for suppression than CenH3s. Moreover, PAML analysis suggests that a larger proportion of CENP-C than CenH3 residues are evolving adaptively. Mammalian CENP-A consists of a well-conserved histone-fold domain with only a short unconstrained tail region. Conversely, Drosophila species have the longest CenH3 tails known [12] but lack any identifiable CENP-C homologs. It is tempting to speculate that the interaction of the long CenH3 tail of Drosophila with centromeric satellites compensates for the absence of CENP-C and permitted its loss. This might explain why Drosophila CenH3s localize in a species-specific manner [44], whereas human CENP-A can be functionally replaced by its budding yeast CenH3 counterpart [57].
Centromere drive may have important consequences for karyotypic evolution. Centromeres of two acrocentric chromosomes frequently fuse (Robertsonian translocations), and metacentrics often misdivide to yield two acrocentrics. In humans, there is a bias in favor of Robertsonian translocations over their homologous acrocentric pair when transmitted by females, and male carriers have reduced fertility [58]. This general sterility of Robertsonian males is consistent with centromere drive underlying post-zygotic reproductive isolation in emerging species [1]. Centromere drive provides a mechanism for the tendency of karyotypes to be either mostly metacentric or mostly acrocentric [59] and for the karyotype-specific accumulation of selfish B chromosomes in mammals [60]. Our finding that CENP-Cs, like CenH3s, evolve adaptively addresses a perceived shortcoming of the centromere drive model for post-zygotic reproductive isolation: mutations that rescued hybrid sterility did not map to the Drosophila CenH3 gene [61,62]. The fact that CenH3 is not the only adaptively evolving centromere protein indicates that there are multiple candidate drive suppressors that might rescue hybrid sterility when in a mutant form.
In contrast to CENP-Cs of plants and animals, yeast Mif2p appears to have evolved entirely under negative selection. This is consistent with Mif2p interacting with a stable centromere, rather than one that is rapidly evolving. In accordance with this observation, budding yeast centromeres are determined by the presence of a consensus DNA sequence that includes binding sites for the Cbf1 and CBF3 proteins [49]. The consensus DNA sequences and their binding proteins are recognizably similar in yeasts as distantly related as Candida glabrata and Kluyveromyces lactis, which have greater average divergence from budding yeast in protein sequences than mammals have from fish [63]. We attribute this extreme conservation of centromere sequence to optimization of the DNA-protein interactions at the centromere. Such optimization would be inevitable in fungi that produce equivalent gametes in a tetrad. No such optimization would occur when centromeres compete at female meiosis I for a favored orientation. Seed plants and animals evolved female meiosis independently, so the parallels that we see for evolution of CenH3 and CENP-C would reflect parallel evolutionary forces in these two ancient lineages.
Materials and methods
DNA clones and sequencing
Genomic DNAs and cDNAs were obtained from several sources (Additional data file 2). The A. thaliana cDNA was amplified from a cDNA pool from whole plants of the ecotype Columbia, and the A. arenosa cDNA was amplified from a cDNA pool from leaves of Care-1 [64]. Both of these cDNAs were amplified using the same primers: 5'-GGAATTTTCCGGTGATTTAGATG-3', which terminates in the initiation codon, and 5'-TGATCACAAGAGGATGGTTGA-3', from the 3' untranslated region of the A. thaliana genomic Cenpc sequence. Genomic DNAs from wheat and barley were generously provided by Andreas Houben. Exons 9p-10p and the intervening intron 9p were amplified from both wheat and barley using the primers 5'-AGATGAACCAATCCATCCAC-3' and 5'-AAATTCGTTTTCCTCTCTTTGCT-3'. Likewise, 9q-10q and intron 9q were amplified with the primers 5'-AGATAAGCCAATCCATACATCA-3' and 5'-CCCCTCTTTTCATTCTCTTCAA-3'. The first and last of these four primers were also used to amplify both exon pairs as a unit to confirm their contiguity and to determine the genomic sequence around the junctions of intron 10p with exons 10p and 9q. Amplifications used High Fidelity Platinum Taq polymerase (Invitrogen, Carlsbad, USA). The amplified fragments were cloned using the pCR2.1-TOPO-TA cloning kit (Invitrogen) according to the manufacturer's instructions. Sequencing was carried out using ABI Big Dye sequencing on both strands of all reported sequences. Sequencing primers were standard vector primers or were designed using Primer 3 [65]. Sequences were assembled using Sequencher 4.1.2 software [66]. Accession numbers of sequences are given in Additional data file 2.
Sequence analyses
Sequence similarities of genes and their encoded proteins were identified using the NCBI BLAST server [35,67], as well as by use of Gramene [68] and the TIGR Gene Indices [69]. Translations and sequence manipulations utilized the Sequence Manipulation Suite [70,71]. Alignments of coding and amino-acid sequences were performed using the European Bioinformatics Institute Clustal W Server [72,73], with adjustments by hand to take account of splice-site alignment. Conservation in alignments was displayed using MacBoxShade 2.1 (MD Baron, Institute for Animal Health, Surrey, UK). Protein blocks were made, displayed, and compared using the Multiple Alignment Processor, sequence Logos, and LAMA [74] programs on the Blocks WWW Server [75]. To make blocks from grass exons 9-12, gaps in ClustalW alignments were first filled with Xs, which do not appear in subsequent sequence Logos representations. Gene models of exon-intron boundaries were made by alignment of cDNAs with identical or homologous genomic sequences, as well as by splice-site prediction using the NetGene2 server [76,77].
K-estimator [31] was used to estimate Ka and Ks in comparisons of Cenpa/CENH3 or Cenpc genes from pairs of closely related species. Prior to analysis, gaps were removed from the coding sequences as indicated by the amino-acid alignments.
We estimated Ka and Ks for windows of 99 nucleotides, positioned every 33 nucleotides. For candidate regions of positive selection, we determined the confidence intervals of Ka and Ks under the null hypothesis that Ka is equal to Ks using the default parameters (1,000 replicates). For individual windows of 99 nucleotides, or for regions defined by contiguous groups of overlapping windows, limited trial and error suggested that statistically significant positive selection was not supported if Ka/Ks < 1.5. Therefore, to find evidence of positive selection, we determined the confidence intervals for regions defined by sets of overlapping or immediately adjacent 99 nucleotide windows with Ka/Ks ≥ 1.5. For regions with Ks = 0, one or more flanking windows with Ks > 0 were included in the region analyzed, regardless of the value of Ka, so that a value for Ka/Ks could be defined. Similarly, we looked for statistically significant negative selection for regions defined by overlapping or adjacent 99 nucleotide windows with Ka/Ks ≤ 0.67. The codeml program of PAML version 3.13d [40] was also used to test for positive selection and to estimate Ka/Ks ratios as previously described [11].
Additional data files
The following are provided as additional data files. Additional data file 1, containing Table S1, reports accession numbers for selected Cenpc ESTs and genomic sequences from GenBank. Additional data file 2, containing Table S2, reports accession numbers for Cenpc cDNAs and amplified genomic sequences. Additional data file 3, containing Figure S1, displays the conservation of the exon containing the CENPC motif.
Supplementary Material
Additional data file 1
Table S1 reports accession numbers for selected Cenpc ESTs and genomic sequences from GenBank
Click here for additional data file
Additional data file 2
Table S2 reports accession numbers for Cenpc cDNAs and amplified genomic sequences
Click here for additional data file
Additional data file 3
Figure S1 displays the conservation of the exon containing the CENPC motif
Click here for additional data file
Acknowledgements
We thank Harmit Malik, Jennifer Cooper, and Caro-Beth Stewart for helpful discussions and Jorja Henikoff for help with LAMA. We also thank the American Type Culture Collection, the Arizona Genomics Institute, Clemson University Genomics Institute, Marie-Michele Cordonnier-Pratt and Lee Pratt, Ze-Guang Han, the Institute of Chemistry of the University of São Paolo, the Japanese Rice Program of the National Institute of Agrobiological Sciences, the Samuel Roberts Noble Foundation, Benildo G. de los Reyes, Bento Soares, the United States Department of Agriculture, Bernd Weisshaar, and Ian Wilson for supplying cDNAs, and Andreas Houben for supplying wheat and barley genomic DNAs.
Figures and Tables
Figure 1 The rat CENP-A protein. (a) Alignment of predicted CENP-A proteins of mammals. Relative to other mammalian CENP-As, rat CENP-A has a 25 amino-acid insertion that arises from a duplication of the amino terminus, shown as over-lined regions. The boundary between the tail and the histone-fold domains (HFD) is indicated below the alignment, along with the position of Loop 1. (b) Alignment of duplicated regions of the rat Cenpa gene (rat1 and rat2) with Cenpa genes of mouse and Chinese hamster. The region that became duplicated in rat extends from upstream of the start codon to codon 22 in mouse and hamster, and is bounded by a conserved dodecamer repeat. The encoded amino acids are shown above (rat1) or below (rat2) the duplicated sequence.
Figure 2 Sliding-window analysis of Ka/Ks for selected pairs of Cenpc genes. Each point represents the value of Ks, Ka, or Ka/Ks for a 99 nucleotide (33 codon) window plotted against the codon position of the midpoint of the window. Ka/Ks is not defined where Ks = 0. The aligned coding sequence is represented at the top of each graph, with the CENPC motif represented by a filled rectangle; exons are also indicated for the plant sequences. Regions of statistically significant positive selection (black bars) and negative selection (gray bars) are marked. (a) Rat and mouse. The interrupted gray bar indicates that p = 0.06 for this region. (b) Arabidopsis thaliana and Arabidopsis arenosa. (c) Maize (CenpcA) and Sorghum bicolor. (d) Wheat and barley, exons 9p-14.
Figure 3 Comparisons of CENP-C proteins in animals, yeast and plants. The CENPC motif and conserved regions found at the termini of CENP-C proteins are indicated. For pairwise comparisons of protein-coding sequences, regions of positive and negative selection between the species compared are shown. (a) Alignment of animal and fungal CENP-Cs. Mammalian CENP-Cs align throughout their lengths, as do the two Saccharomyces Mif2p proteins, but others align only at conserved regions. Portions of the human CENP-C protein implicated in centromere-targeting (purple bars) and DNA-binding (black bars) are shown at the top. The scale bar at the top marks the length of human CENP-C in amino acids. (b) Alignment of plant CENP-Cs. Within angiosperm families, proteins align throughout their lengths. Between families, weak conservation is found at the amino terminus and strong conservation at the carboxyl terminus. (c) Logos representation of an alignment of the CENPC motif from human; mouse; cow; chicken; Caenorhabditis elegans; budding yeast; Schizosaccharomyces pombe; Physcomitrella patens; maize CenpcA; rice; A. thaliana; black cottonwood, soybean, and tomato.
Figure 4 Alignment of conserved regions of angiosperm CENP-C predicted proteins. (a) Short regions of conservation are encoded in the first six exons of Cenpc genes from five families. The dipeptide SQ (underlined) is relatively frequent in exon 5. (b) Multiple alignment reveals strong conservation in the carboxyl termini of encoded proteins from six families. The CENPC motif is indicated. At, A. thaliana; Mt, barrel medic; Os, rice; Zm, maize CENP-CA; St, potato; SLe, tomato; Bv, beet; Pbt, black cottonwood.
Figure 5 Gene models of selected plant Cenpc genes. Exon/intron structure is conserved across families from exon 1 through the beginning of exon 6, and for the final two exons and introns. Exon sizes are given to the nearest codon where genomic sequence is available to confirm predicted exons. Duplicated exons are indicated by gray shading.
Figure 6 CENP-C exon repeats in the grasses. (a) Alignments of copies of the duplicated exons 9, 10, 11, and 12 from the grass species in this study, excluding pseudogenes, are shown in Logos format. (b) A neighbor-joining phylogram (with gaps excluded) of the exon pairs 9-10 and 11-12 in grass species. A parsimony tree gave essentially the same topology. Dots indicate the locations of inferred duplication events in the tree. Presumed pseudogenes are marked with ψ. (c) Schematic representation of exon duplication events leading to various Cenpc gene structures, and examples of grass species with these structures. Pairs of arrows indicate duplication events; lines terminating in a filled circle indicate loss of an exon pair in derivatives.
Table 1 Pairwise comparison of mouse and rat Cenpc genes
Human Mouse Rat Selection
1-86 1-84 1-77 +*
109-248 107-218 100-236 +**
239-304 208-273 226-291 -**
294-353 263-321 281-335 +*
411-455 377-420 391-434 +**
445-497 410-464 424-478 - (0.06)
487-552 454-519 468-533 +**
565-670 531-633 545-643 +**
671-790 634-754 644-764 -**
858-934 821-897 831-907 -*
+ denotes Ka > Ks; -, Ka < Ks; * p < 0.05; ** p < 0.01. Number ranges represent codon positions based on the complete coding sequences prior to removal of indels for alignment. Human codon positions are given for comparison with previous functional studies. Number in parentheses is a p value greater than 0.05.
Table 2 Regions of selection in pairwise comparisons of maize CenpcA, Sorghum bicolor Cenpc, and sugarcane Cenpc1
Exons Direction of selection Maize vs. Sorghum Maize vs. sugarcane Sorghum vs. sugarcane
1 + 12-44 12-44 1-42
+ + (0.17) + (0.04)
1-5 - 34-165 23-176 87-163
- - -
4-6 + 155-253 166-253 153-317
+ + +
6 * 232-286
-
6 - 298-363 298-363 298-352
- - - (0.13)
6 + 353-409 342-407 397-431
+ + + (0.06)
6-12 - 410-621 397-630 432-579
- - -
12-14 * 611-687 609-685 591-700
+ + -
+, Ka > Ks; -, Ka < Ks; p ≤ 0.01 except where given in parentheses. * Direction of selection varies with lineage. Regions of selection are identified by codon positions based on the sequence of maize CenpcA.
==== Refs
Henikoff S Ahmad K Malik HS The centromere paradox: stable inheritance with rapidly evolving DNA Science 2001 293 1098 1102 11498581 10.1126/science.1062939
Schueler MG Higgins AW Rudd MK Gustashaw K Willard HF Genomic and genetic definition of a functional human centromere Science 2001 294 109 115 11588252 10.1126/science.1065042
Lohe A Roberts P Verma RS Evolution of satellite DNA sequences in Drosophila In Heterochromatin, Molecular and Structural Aspects 1988 Cambridge: Cambridge University Press 148 186
Haaf T Willard HF Chromosome-specific alpha-satellite DNA from the centromere of chimpanzee chromosome 4 Chromosoma 1997 106 226 232 9254724 10.1007/s004120050243
Heslop-Harrison JS Brandes A Schwarzacher T Tandemly repeated DNA sequences and centromeric chromosomal regions of Arabidopsis species Chromosome Res 2003 11 241 253 12769291 10.1023/A:1022998709969
Choo KH Domain organization at the centromere and neocentromere Dev Cell 2001 1 165 177 11702777 10.1016/S1534-5807(01)00028-4
Palmer DK O'Day K Wener MH Andrews BS Margolis RL A 17-kD centromere protein (CENP-A) copurifies with nucleosome core particles and with histones J Cell Biol 1987 104 805 815 3558482 10.1083/jcb.104.4.805
Yoda K Ando S Morishita S Houmura K Hashimoto K Takeyasu K Okazaki T Human centromere protein A (CENP-A) can replace histone 3 in nucleosome reconstitution in vitro Proc Natl Acad Sci USA 2000 97 7266 7271 10840064 10.1073/pnas.130189697
Malik HS Henikoff S Adaptive evolution of Cid, a centromere-specific histone in Drosophila Genetics 2001 157 1293 1298 11238413
Talbert PB Masuelli R Tyagi AP Comai L Henikoff S Centromeric localization and adaptive evolution of an Arabidopsis histone H3 variant Plant Cell 2002 14 1053 1066 12034896 10.1105/tpc.010425
Cooper JL Henikoff S Adaptive evolution of the histone fold domain in centromeric histones Mol Biol Evol 2004 21 1712 1718 15175412 10.1093/molbev/msh179
Malik HS Vermaak D Henikoff S Recurrent evolution of DNA-binding motifs in the Drosophila centromeric histone Proc Natl Acad Sci USA 2002 99 1449 1454 11805302 10.1073/pnas.032664299
Saitoh H Tomkiel J Cooke CA Ratrie H Maurer M Rothfield NF Earnshaw WC CENP-C, an autoantigen in scleroderma, is a component of the human inner kinetochore plate Cell 1992 70 115 125 1339310 10.1016/0092-8674(92)90538-N
McKay S Thomson E Cooke H Sequence homologies and linkage group conservation of the human and mouse Cenpc genes Genomics 1994 22 36 40 7959789 10.1006/geno.1994.1342
Burkin DJ Jones C Burkin HR McGrew JA Broad TE Sheep CENPB and CENPC genes show a high level of sequence similarity and conserved synteny with their human homologs Cytogenet Cell Genet 1996 74 86 89 8893808
Fukagawa T Brown WR Efficient conditional mutation of the vertebrate CENP-C gene Hum Mol Genet 1997 6 2301 2308 9361037 10.1093/hmg/6.13.2301
Sugimoto K Kuriyama K Shibata A Himeno M Characterization of internal DNA-binding and C-terminal dimerization domains of human centromere/kinetochore autoantigen CENP-C in vitro : role of DNA-binding and self-associating activities in kinetochore organization Chromosome Res 1997 5 132 141 9146917 10.1023/A:1018422325569
Sugimoto K Yata H Muro Y Himeno M Human centromere protein C (CENP-C) is a DNA-binding protein which possesses a novel DNA-binding motif J Biochem 1994 116 877 881 7883764
Yang CH Tomkiel J Saitoh H Johnson DH Earnshaw WC Identification of overlapping DNA-binding and centromere-targeting domains in the human kinetochore protein CENP-C Mol Cell Biol 1996 16 3576 3586 8668174
Politi V Perini G Trazzi S Pliss A Raska I Earnshaw WC Della Valle G CENP-C binds the alpha-satellite DNA in vivo at specific centromere domains J Cell Sci 2002 115 2317 2327 12006616
Trazzi S Bernardoni R Diolaiti D Politi V Earnshaw WC Perini G Della Valle G In vivo functional dissection of human inner kinetochore protein CENP-C J Struct Biol 2002 140 39 48 12490152 10.1016/S1047-8477(02)00506-3
Brown MT Sequence similarities between the yeast chromosome segregation protein Mif2 and the mammalian centromere protein CENP-C Gene 1995 160 111 116 7628703 10.1016/0378-1119(95)00163-Z
Meluh PB Koshland D Evidence that the MIF2 gene of Saccharomyces cerevisiae encodes a centromere protein with homology to the mammalian centromere protein CENP-C Mol Biol Cell 1995 6 793 807 7579695
Moore LL Roth MB HCP-4, a CENP-C-like protein in Caenorhabditis elegans, is required for resolution of sister centromeres J Cell Biol 2001 153 1199 1208 11402064 10.1083/jcb.153.6.1199
Dawe RK Reed LM Yu HG Muszynski MG Hiatt EN A maize homolog of mammalian CENPC is a constitutive component of the inner kinetochore Plant Cell 1999 11 1227 1238 10402425 10.1105/tpc.11.7.1227
Brown MT Goetsch L Hartwell LH MIF2 is required for mitotic spindle integrity during anaphase spindle elongation in Saccharomyces cerevisiae J Cell Biol 1993 123 387 403 8408221 10.1083/jcb.123.2.387
Tomkiel J Cooke CA Saitoh H Bernat RL Earnshaw WC CENP-C is required for maintaining proper kinetochore size and for a timely transition to anaphase J Cell Biol 1994 125 531 545 8175879 10.1083/jcb.125.3.531
Kalitsis P Fowler KJ Earle E Hill J Choo KH Targeted disruption of mouse centromere protein C gene leads to mitotic disarray and early embryo death Proc Natl Acad Sci USA 1998 95 1136 1141 9448298 10.1073/pnas.95.3.1136
Kalitsis P MacDonald AC Newson AJ Hudson DF Choo KH Gene structure and sequence analysis of mouse centromere proteins A and C Genomics 1998 47 108 114 9465302 10.1006/geno.1997.5109
NCBI High-throughput genomic sequences
Comeron JM K-Estimator: calculation of the number of nucleotide substitutions per site and the confidence intervals Bioinformatics 1999 15 763 764 10498777 10.1093/bioinformatics/15.9.763
Figueroa J Pendon C Valdivia MM Molecular cloning and sequence analysis of hamster CENP-A cDNA BMC Genomics 2002 3 11 12019018 10.1186/1471-2164-3-11
Sullivan KF Hechenberger M Masri K Human CENP-A contains a histone H3 related histone fold that is required for targeting to the centromere J Cell Biol 1994 127 581 592 7962047 10.1083/jcb.127.3.581
Chimpanzee Genome
NCBI BLAST
Zhong CX Marshall JB Topp C Mroczek R Kato A Nagaki K Birchler JA Jiang J Dawe RK Centromeric retroelements and satellites interact with maize kinetochore protein CENH3 Plant Cell 2002 14 2825 2836 12417704 10.1105/tpc.006106
Vettore AL da Silva FR Kemper EL Souza GM da Silva AM Ferro MI Henrique-Silva F Giglioti EA Lemos MV Coutinho LL Analysis and functional annotation of an expressed sequence tag collection for tropical crop sugarcane Genome Res 2003 13 2725 2735 14613979 10.1101/gr.1532103
Miller JT Jackson SA Nasuda S Gill BS Wing RA Jiang J Cloning and characterization of centromere specific DNA element from Sorghum bicolor Theor Appl Genet 1998 96 832 839 10.1007/s001220050809
Xie Y Heng HH FISH mapping of centromere protein C (CENPC) on human chromosome 4q31-q21 Cytogenet Cell Genet 1996 74 192 193 8941372
Yang Z PAML: a program package for phylogenetic analysis by maximum likelihood Comput Appl Biosci 1997 13 555 556 9367129
Suzuki N Nakano M Nozaki N Egashira S Okazaki T Masumoto H CENP-B interacts with CENP-C domains containing Mif2 regions responsible for centromere localization J Biol Chem 2004 279 5934 5946 14612452 10.1074/jbc.M306477200
Song K Gronemeyer B Lu W Eugster E Tomkiel JE Mutational analysis of the central centromere targeting domain of human centromere protein C, (CENP-C) Exp Cell Res 2002 275 81 91 11925107 10.1006/excr.2002.5495
Lanini L McKeon F Domains required for CENP-C assembly at the kinetochore Mol Biol Cell 1995 6 1049 1059 7579707
Vermaak D Hayden HS Henikoff S Centromere targeting element within the histone fold domain of Cid Mol Cell Biol 2002 22 7553 7561 12370302 10.1128/MCB.22.21.7553-7561.2002
Goshima G Kiyomitsu T Yoda K Yanagida M Human centromere chromatin protein hMis12, essential for equal segregation, is independent of CENP-A loading pathway J Cell Biol 2003 160 25 39 12515822 10.1083/jcb.200210005
Howman EV Fowler KJ Newson AJ Redward S MacDonald AC Kalitsis P Choo KH Early disruption of centromeric chromatin organization in centromere protein A (CenpA) null mice Proc Natl Acad Sci USA 2000 97 1148 1153 10655499 10.1073/pnas.97.3.1148
Oegema K Desai A Rybina S Kirkham M Hyman AA Functional analysis of kinetochore assembly in Caenorhabditis elegans J Cell Biol 2001 153 1209 1226 11402065 10.1083/jcb.153.6.1209
Van Hooser AA Ouspenski II Gregson HC Starr DA Yen TJ Goldberg ML Yokomori K Earnshaw WC Sullivan KF Brinkley BR Specification of kinetochore-forming chromatin by the histone H3 variant CENP-A J Cell Sci 2001 114 3529 3542 11682612
Westermann S Cheeseman IM Anderson S Yates JR 3rdDrubin DG Barnes G Architecture of the budding yeast kinetochore reveals a conserved molecular core J Cell Biol 2003 163 215 222 14581449 10.1083/jcb.200305100
Luger K Mader AW Richmond RK Sargent DF Richmond TJ Crystal structure of the nucleosome core particle at 2.8 Å resolution Nature 1997 389 251 260 9305837 10.1038/38444
Okamura A Pendon C Valdivia MM Ikemura T Fukagawa T Gene structure, chromosomal localization and immunolocalization of chicken centromere proteins CENP-C and ZW10 Gene 2001 262 283 290 11179694 10.1016/S0378-1119(00)00517-5
Shibata F Murata M Differential localization of the centromere-specific proteins in the major centromeric satellite of Arabidopsis thaliana J Cell Sci 2004 117 2963 2970 15161939 10.1242/jcs.01144
Messier W Stewart CB Episodic adaptive evolution of primate lysozymes Nature 1997 385 151 154 8990116 10.1038/385151a0
Yang Z Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution Mol Biol Evol 1998 15 568 573 9580986
Kellis M Patterson N Endrizzi M Birren B Lander ES Sequencing and comparison of yeast species to identify genes and regulatory elements Nature 2003 423 241 254 12748633 10.1038/nature01644
Pardo-Manuel de Villena F Sapienza C Nonrandom segregation during meiosis: the unfairness of females Mamm Genome 2001 12 331 339 11331939 10.1007/s003350040003
Wieland G Orthaus S Ohndorf S Diekmann S Hemmerich P Functional complementation of human centromere protein A (CENP-A) by Cse4p from Saccharomyces cerevisiae Mol Cell Biol 2004 24 6620 6630 15254229 10.1128/MCB.24.15.6620-6630.2004
Daniel A Distortion of female meiotic segregation and reduced male fertility in human Robertsonian translocations: consistent with the centromere model of co-evolving centromere DNA/centromeric histone (CENP-A) Am J Med Genet 2002 111 450 452 12210311 10.1002/ajmg.10618
Pardo-Manuel de Villena F Sapienza C Female meiosis drives karyotypic evolution in mammals Genetics 2001 159 1179 1189 11729161
Palestis BG Burt A Jones RN Trivers R B chromosomes are more frequent in mammals with acrocentric karyotypes: support for the theory of centromeric drive Proc R Soc Lond B Biol Sci 2004 271 S22 S24 15101408 10.1098/rsbl.2003.0084
Sainz A Wilder JA Wolf M Hollocher H Drosophila melanogaster and D. simulans rescue strains produce fit offspring, despite divergent centromere-specific histone alleles Heredity 2003 91 28 35 12815450 10.1038/sj.hdy.6800275
Coyne JA Orr HA Speciation 2004 Sunderland: Sinauer
Dujon B Sherman D Fischer G Durrens P Casaregola S Lafontaine I De Montigny J Marck C Neuveglise C Talla E Genome evolution in yeasts Nature 2004 430 35 44 15229592 10.1038/nature02579
Henikoff S Comai L A DNA methyltransferase homolog with a chromodomain exists in multiple forms in Arabidopsis Genetics 1998 149 307 318 9584105
Rozen S Skaletsky H Primer3 on the WWW for general users and for biologist programmers Methods Mol Biol 2000 132 365 386 10547847
Gene Codes Corporation
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
Gramene
TIGR Gene Indices
Stothard P The sequence manipulation suite: JavaScript programs for analyzing and formatting protein and DNA sequences Biotechniques 2000 28 1102 1104 10868275
Translate
Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 1994 22 4673 4680 7984417
EMBL/EBI ClustalW
Pietrokovski S Searching databases of conserved sequence regions by aligning protein multiple-alignments Nucleic Acids Res 1996 24 3836 3845 8871566 10.1093/nar/24.19.3836
Blocks WWW Server
Hebsgaard SM Korning PG Tolstrup N Engelbrecht J Rouze P Brunak S Splice site prediction in Arabidopsis thaliana pre-mRNA by combining local and global sequence information Nucleic Acids Res 1996 24 3439 3452 8811101 10.1093/nar/24.17.3439
NetGene2 Server
Bonaldo MF Lennon G Soares MB Normalization and subtraction: two approaches to facilitate gene discovery Genome Res 1996 6 791 806 8889548
Xiao HS Huang QH Zhang FX Bao L Lu YJ Guo C Yang L Huang WJ Fu G Xu SH Identification of gene expression profile of dorsal root ganglion in the rat peripheral axotomy model of neuropathic pain Proc Natl Acad Sci USA 2002 99 8360 8366 12060780 10.1073/pnas.122231899
Meat Animal Research Center
Samuel Roberts Noble Foundation
ATCC:The global bioresource center
Arizona Genomics Institute
CUGI: Clemson University Genomics Institute
Vettore AL da Silva FR Kemper EL Souza GM da Silva AM Ferro MI Henrique-Silva F Giglioti EA Lemos MV Coutinho LL Analysis and functional annotation of an expressed sequence tag collection for tropical crop sugarcane Genome Res 2003 13 2725 2735 14613979 10.1101/gr.1532103
Laboratory for Genomics and Bioinformatics
Rice Genome Research Program
Agriculture Research Service
cerealsDB.uk.net
| 15345035 | PMC549713 | CC BY | 2021-01-04 16:05:44 | no | J Biol. 2004 Aug 31; 3(4):18 | utf-8 | J Biol | 2,004 | 10.1186/jbiol11 | oa_comm |
==== Front
J BiolJournal of Biology1478-58541475-4924BioMed Central London jbiol161558831210.1186/jbiol16Research ArticleThe functional landscape of mouse gene expression Zhang Wen [email protected] Quaid D [email protected] Richard 1Shai Ofer 3Bakowski Malina A 1Mitsakakis Nicholas 1Mohammad Naveed 1Robinson Mark D 1Zirngibl Ralph 2Somogyi Eszter 2Laurin Nancy 2Eftekharpour Eftekhar 4Sat Eric 5Grigull Jörg 1Pan Qun 1Peng Wen-Tao 1Krogan Nevan 12Greenblatt Jack 12Fehlings Michael 46van der Kooy Derek 2Aubin Jane 2Bruneau Benoit G 27Rossant Janet 25Blencowe Benjamin J 12Frey Brendan J 3Hughes Timothy R [email protected] Banting and Best Department of Medical Research, University of Toronto, 1 King's College Circle, Toronto, ON M5S 1A8, Canada2 Department of Medical Genetics and Microbiology, University of Toronto, 1 King's College Circle, Toronto, ON M5S 1A8, Canada3 Department of Electrical and Computer Engineering, University of Toronto, 1 King's College Circle, Toronto, ON M5S 1A8, Canada4 Department of Surgery, University of Toronto, 1 King's College Circle, Toronto, ON M5S 1A8, Canada5 Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, ON M5G 1X5, Canada6 Division of Cell and Molecular Biology, Toronto Western Research Institute and Krembil Neuroscience Center, 399 Bathurst St., Toronto, ON M5T 2S8, Canada7 The Hospital for Sick Children, 555 University Ave., Toronto, ON M5G 1X8, Canada2004 6 12 2004 3 5 21 21 1 9 2004 13 10 2004 18 10 2004 Copyright © 2004 Zhang 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
Large-scale quantitative analysis of transcriptional co-expression has been used to dissect regulatory networks and to predict the functions of new genes discovered by genome sequencing in model organisms such as yeast. Although the idea that tissue-specific expression is indicative of gene function in mammals is widely accepted, it has not been objectively tested nor compared with the related but distinct strategy of correlating gene co-expression as a means to predict gene function.
Results
We generated microarray expression data for nearly 40,000 known and predicted mRNAs in 55 mouse tissues, using custom-built oligonucleotide arrays. We show that quantitative transcriptional co-expression is a powerful predictor of gene function. Hundreds of functional categories, as defined by Gene Ontology 'Biological Processes', are associated with characteristic expression patterns across all tissues, including categories that bear no overt relationship to the tissue of origin. In contrast, simple tissue-specific restriction of expression is a poor predictor of which genes are in which functional categories. As an example, the highly conserved mouse gene PWP1 is widely expressed across different tissues but is co-expressed with many RNA-processing genes; we show that the uncharacterized yeast homolog of PWP1 is required for rRNA biogenesis.
Conclusions
We conclude that 'functional genomics' strategies based on quantitative transcriptional co-expression will be as fruitful in mammals as they have been in simpler organisms, and that transcriptional control of mammalian physiology is more modular than is generally appreciated. Our data and analyses provide a public resource for mammalian functional genomics.
==== Body
Background
Tissue-specific gene expression has traditionally been viewed as a predictor of tissue-specific function: for example, genes specifically expressed in the eye are likely to be involved in vision. But microarray analysis in model organisms such as yeast and Caenorhabditis elegans has established that coordinate transcriptional regulation of functionally related genes occurs on a broader scale than was previously recognized, encompassing at least half of all cellular processes in yeast [1-5]. Consequently, gene expression patterns can be used to predict gene functions, thereby providing a starting point for the directed and systematic experimental characterization of novel genes [1-10]. As an example, it was observed in yeast that a group of more than 200 genes involved primarily in RNA processing and ribosome biogenesis is transcriptionally co-regulated, in addition to being constitutively expressed at some level [11]. Application of statistical inference methods led to the prediction that the uncharacterized genes in this co-regulated group were likely to be involved in RNA processing and/or ribosome biogenesis [5,9]. Subsequent experimental analysis using yeast mutants validated that many of these predictions were in fact accurate [9].
To date, this approach has only been extensively applied to relatively simple model organisms such as yeast and C. elegans. Its general utility in mammals has not yet been established with respect to the proportion of either genes or functional categories to which it can be effectively applied. Nor has it been formally examined how the use of quantitative transcriptional co-expression for inference of gene function compares to the more traditional approach of inferring functions on the basis of tissue-specific transcription. The extent and precision of hypotheses regarding gene functions that can be drawn from expression analysis in mammals is an important and timely question, given the current absence of knowledge of the physiological functions of at least half of all mammalian genes. Given that distinct and coordinate expression of a group of functionally related genes implies an underlying pathway-specific transcriptional regulatory mechanism, identification of such instances would also represent a step towards delineating mammalian transcriptional networks.
Here, in order to demarcate the general utility of using gene expression patterns to infer mammalian gene functions, and to use this information to begin characterizing genes discovered by sequencing the mouse genome [12], we used custom-built DNA oligonucleotide microarrays to generate an expression data set for nearly 40,000 known and predicted mouse mRNAs across 55 diverse tissues. Several criteria show that these data are reliable and consistent with other information about gene expression and tissue function. Cross-validation results from machine-learning algorithms show that patterns of gene co-expression within many functional categories are 'learnable' and distinct from patterns of other categories, thus proving that many functional categories are transcriptionally co-expressed and likely to be co-regulated. In contrast, tissue-specificity alone is a comparatively poor predictor of gene function, illustrating the importance of quantitative gene expression measurements. To exemplify this, we functionally characterized the highly conserved gene PWP1, which is widely expressed. PWP1 is co-expressed with many RNA-processing genes in mouse, and we show that its yeast homolog is required for rRNA biogenesis. The data and the associated analyses in this paper will be invaluable for directing experimental characterization of gene functions in mammals, as well as for dissecting the mammalian transcriptional regulatory hierarchy.
Results
Expression analysis of mouse XM gene sequences
In order to generate an extensive survey of mammalian gene expression, we analyzed mRNA abundance in 55 mouse tissues using custom-designed microarrays of 60-mer oligonucleotides [13] corresponding to 41,699 known and predicted mRNAs identified in the draft mouse genome sequence using gene-finding programs [12,14] (NCBI 'XM' sequences; approximately 39,309 are unique; for further details, see the Materials and methods section). Tissue collection was a collaborative effort among several labs in the Toronto area, each with expertise in distinct areas of physiology; consequently, the mouse tissues we analyzed were obtained from several different strains of mice which are typically used to study specific organs and cell types of interest (additional cell lines and fractionated cells from animals were also analyzed, but the results are not included here because the data appear to bear little relationship to the tissue of origin of the cells examined). Since it has previously been established that there is a high correlation in expression of orthologous genes between mice and humans [15], large variations in tissue-specific expression should not occur between individuals within the same species, although we cannot rule out subtle strain-specific differences. To maximize the fidelity of measurements, unamplified cDNA from at least 1 μg of polyA-purified mRNA was hybridized to each array, with fluor-reversed duplicates performed in each case. For most organs this required pooling RNA from multiple animals; for example, more than 50 mice were required to obtain sufficient prostate mRNA. Consequently, potential variations due to parameters such as circadian rhythms or individual dissections should have been minimized by averaging over multiple animals.
All hybridizations were performed in duplicate. Data processing and normalization are described in detail in the Materials and methods. The data were processed so that each measurement reflects the abundance of each transcript in each tissue relative to the median expression across all 55 tissues; although the microarray spot intensities were used to determine which genes were detected as expressed (see below), the figures herein show the normalized, arcsinh-transformed and median-subtracted data, which for convenience we refer to as ratios. All of the data, together with tables detailing correspondence to genes in other cDNA and EST databases, annotations and other features of the encoded proteins, probe sequences, and other files used in our analyses below, are available as Additional data files and without restriction on our website [16].
Validation of expression data
Four lines of evidence support the quality of our data and its consistency with existing knowledge of mammalian physiology and gene expression. First, we detected the expected patterns of expression for genes previously shown to be expressed specifically in each of the 55 tissues surveyed (Figure 1). This validates the accuracy of our dissections, and indicates that there was little cross-contamination between tissue samples.
Second, there is a clear correspondence, albeit not absolute, between our data and two other mouse microarray data sets [15,17], which surveyed a subset of the genes and tissues that we have examined. Thirteen tissues and 1,109 genes were unambiguously shared among the three studies (Figure 2a). Our data are more highly correlated with those of Su et al. [15], who also employed oligonucleotide array technology, whereas Bono et al. [17] used spotted cDNAs (Figure 2a). Furthermore, our data are more highly correlated with either of the two other studies than the two other studies are to one another. (It should be noted that these previous studies did not examine the use of transcriptional co-expression to predict gene function, which is the focus of the present study.)
Third, our array data are consistent with RT-PCR analysis. We tested for expected tissue-specific expression of 107 genes (a mixture of characterized and uncharacterized) in 18 selected tissues. In this analysis a single primer pair was tested for each gene. (It is possible that the predicted exon structures for many of the poorly characterized XM genes are incorrect: there was a clear correspondence between whether a product was obtained and whether there was an EST or cDNA in the public databases, which would indicate correct gene structures – see Materials and methods.) Among the 55 primer pairs that could result in amplification, 53 (96%) gave a correct-product size in the tissue(s) expected on the basis of our array data, and 47 (85%) produced amplification most strongly or exclusively in the expected tissue(s) (Figure 2b and data not shown). Although RT-PCR is semi-quantitative, there is an obvious correspondence between the left and right panels in Figure 2b, confirming that our microarray measurements are largely consistent with a more conventional expression analysis method.
Fourth, in the analyses detailed in the following sections, we show that the annotations of genes expressed preferentially in each tissue correspond in many cases to known physiological functions of the tissue, further confirming the accuracy of the dissections and the microarray measurements. Moreover, sets of functionally related genes were often observed to display uniform expression profiles, a result that is highly unlikely to occur by chance.
Definition of 21,622 confidently detected transcripts
In order to establish rigorously which genes are expressed in each tissue sample, we used the 66 negative-control spots on our arrays (corresponding to 30 randomly generated sequences, 31 mouse intergenic or intronic regions, and five yeast genes). We considered the XM genes to be 'expressed' only if their intensity exceeded the 99th percentile (that is, all but 1%) of intensities from the negative controls (Figure 3a). 21,622 transcripts satisfied this criterion in at least one sample. There were 1,790 transcripts that were detected in every sample, and manual inspection verified that many of these have traditional 'housekeeping' functions (for example, ribosomal proteins, actin and tubulin). There were 4,475 transcripts detected in only one of the 55 samples (Figure 3b). Most of the 21,622 genes, however, were expressed in multiple tissues (Figure 3b). Each of the tissues expressed fewer than half of the 21,622 genes (Figure 3c). The number of genes detected in each sample was slightly lower than the conventional estimate of 10,000 genes expressed per cell (for example, we detected 6,094 different transcripts in embryonic stem (ES) cells, the only pure cell population examined, whereas a recent study using sequence tags indicated approximately 8,400 different transcripts in human ES cells [18]). This level of detection is not unexpected, for several reasons. First, tissues are mixtures of cell types, such that low-abundance, cell-type-specific transcripts may be diluted below the array detection limits of 1 in 1,000,000 [13]; second, the arrays did not include every single mouse gene; and third, our threshold for expression was conservative. The full 21,622 × 55 data matrix is found in the Additional data files. Figure 4a shows a clustering analysis of the 21,622 expressed genes in the 55 surveyed tissues, which illustrates that distinct tissues with related physiological roles also tend to have similar overall gene expression profiles. For example, all components of the nervous system featured higher expression of a common subset of transcripts, as did all components of the lower digestive tract.
Correspondence between gene and tissue function
To examine the relationships among tissues and gene functions, we asked whether genes carrying specific Gene Ontology 'Biological Process' (GO-BP) categories, which reflect the physiological function of a gene, were preferentially expressed in each of the tissue samples, using a statistical test (Wilcoxon-Mann-Whitney; WMW). A selection of the WMW scores are shown in Figure 4b, and expression patterns of all genes in all GO-BP categories can be seen in the Additional data files and at the Toronto gene expressions website [19]. This analysis revealed that the preferentially expressed GO-BP categories typically reflected known functions of the tissue, sometimes with surprising resolution. For example, while the category 'synaptic transmission' scored highly in all neuronal tissues, 'learning and memory' was highest in cortex and striatum; 'locomotor behavior' was highest in cortex, midbrain, and spinal cord; 'response to temperature', in the trigeminal nucleus of the brainstem; and 'neurogenesis', in both adult central nervous system and embryonic heads (Figure 4d). While the WMW test may not have captured all of the categories relevant to each brain tissue, this finding does illustrate that our data contain differential expression of genes involved in distinct high-level neural functions. Further investigation of several tissue-associated GO-BP categories that were initially unanticipated revealed that they are easily rationalized; for instance, lung, bladder, skin, and intestines all express immune-related categories, presumably because they are exposed to the environment and infiltrated by immune cells (see for example, [20]).
Correspondence between gene function and transcriptional co-expression
An alternative way to ask whether gene regulation corresponds to gene function is to examine the correlations among the transcript levels of genes, independent of the tissue-source information. An initial confirmation that patterns of transcript abundance correspond to gene functions comes from simply examining the behavior of all genes within distinct functional categories. For example, Figure 5 shows the expression of individual genes in 17 categories that exemplify ways in which gene expression relates to gene function (similar diagrams for all GO-BP categories can be seen in the Additional data files and at the Toronto gene expressions website [19]). There are prominent patterns that are distinctive of a subset of genes in each category. The fact that not all of the genes within each annotation category conformed to a single pattern could result from imperfections in the annotations or the measurements, or could be due to the correspondence between gene function and gene expression being less than absolute. While highly tissue-specific expression of genes in a category was observed in some cases (such as 'pregnancy' genes in placenta or 'fertilization' genes in testis), it was much more common that genes within a category were expressed across multiple functionally related tissues (for example, 'bone remodeling' in all bone tissues), consistent with the results shown in Figure 4b. In other instances, genes within a single annotation category were subdivided into multiple expression patterns: for example, 'cell-cell adhesion' contains three distinct groups of genes with elevated expression in skin-containing samples, neural tissues, and digestive tract, respectively. Consistent with a previous study [21], we observed coordinate regulation of genes within distinct biochemical pathways; Figure 5 includes the examples 'polyamine biosynthesis' and 'serine biosynthesis'. Moreover, a number of functional categories corresponding to basic cellular or biochemical functions which are traditionally thought of as 'housekeeping' (since they are required for cell viability) were in fact coordinately regulated across tissues: Figure 5 shows genes in the category 'RNA splicing', which are expressed most highly in neural and embryonic tissues, perhaps reflecting the higher levels of gene expression and alternative mRNA splicing known to occur in these tissues. Interestingly, subsets of genes in the categories 'cytokinesis', 'microtubule-based movement', 'oxidative phosphorylation', and 'M phase', all of which might be considered as central to cellular physiology, were also expressed in distinctive patterns among mouse tissues.
We also asked more generally whether groups of co-expressed transcripts were associated with specific GO-BP categories. Figure 4c shows that this is indeed the case: any given 'cluster' of genes with correlated expression levels is more likely than not to be associated with a local enrichment of one or a few annotation categories, and manual analysis suggests that tissue-specific expression often reflects the known physiological role(s) of the tissues in which the genes are expressed (examples are shown in Figure 4d). False-discovery rate analysis (see the Materials and methods section) confirmed that over 58% of the 21,622 genes were co-regulated with a set of genes significantly enriched for at least one GO-BP category. For the 7,387 GO-BP annotated genes, over 66% were co-expressed with a set of genes significantly enriched for at least one GO-BP category; in over 25% of these instances, the most significant category was one of its existing annotations. Random permutation analysis (that is, repeating the analysis with randomized gene identities) established a false discovery rate [22] of less than 1% for these analyses (see Materials and methods for details). Hence, quantitative co-expression of functionally related genes appears to be a general phenomenon in mammals.
Using transcriptional co-expression to predict mouse gene functions
It stands to reason that a gene expressed in a specific tissue is likely to be functioning in that tissue. Therefore, we next asked how accurately mammalian gene functions can be predicted on the basis of gene expression profiles. There are many anecdotal examples in which the tissue-specific or cell-type-specific expression of a gene has been used to aid in discovering its function, and this approach has been advocated in previous analyses of mouse tissue expression data (see for example, [15]). Our data indicate that the expression of most mouse genes shows some degree of tissue restriction, but most of the genes are not expressed in a highly tissue-specific manner (Figure 3b). Furthermore, most tissues express genes from multiple functional categories (Figure 4b), and genes from many functional categories are expressed across many tissues (Figure 5), which could make it difficult to distinguish genes in these categories on the basis of expression in one or a few tissues. In addition, defining tissue specificity involves drawing thresholds to form lists, rather than using the quantitative expression information directly to draw functional inferences.
An alternative strategy is to generate functional predictions on the basis of transcriptional co-expression [23,24], which we show (above) often reflects gene function (Figure 5). This approach utilizes quantitative measurements and places no restriction on tissue-specificity, allowing all expressed genes to be treated equally in the analysis. Furthermore, the use of quantitative co-expression allows the application of sophisticated computational tools that have been optimized for the general problem of classification on the basis of features within a data matrix [25]. We examined the extent to which this approach is effective for our data, and we show (below) that it yields almost universally superior predictions of gene function in comparison to using information regarding simple tissue specificity or tissue restriction.
In this analysis, we used support vector machines (SVMs) [26]. An SVM is a machine-learning algorithm (a computer program) that has previously been shown to work well for the prediction of gene functions in yeast on the basis of microarray expression data [25] but which has not, to our knowledge, been used extensively to predict gene functions from mammalian expression-profiling data. The theory and implementation of SVMs have been described elsewhere in detail [25,26]. Briefly, an SVM outputs a 'discriminant value' for each gene in each category, and this value reflects relative confidence that the gene is in the category in question. The SVM considers each functional category separately, and the discriminant value is assigned on the basis of where the gene lies relative to other genes within the 'gene expression space' (for example, analysis of 55 samples results in 55 different coordinates). If the gene lies in a region where there is a high proportion of genes that are known to be in the category in question, this will lead to a high discriminant value. SVMs are conceptually related to clustering analysis in the sense that the discriminant values are derived from similarity among expression profiles. But in clustering analysis, genes are grouped solely on the basis of their expression levels; in contrast, SVMs use the known classifications (that is, knowledge regarding which genes are in the category and which are not) in order to map the initial gene expression space into a one-dimensional space (the discriminant values) in which the two classes are optimally distinguished.
Importantly, the discriminant values output by an SVM can be processed to obtain an estimate of the probability that the prediction for each gene in each category is correct (that is, an estimate of precision), on the basis of how well previously annotated genes in the given category can be distinguished from previously annotated genes that are not in the category. This is accomplished by a three-fold cross-validation strategy, in which the analysis is run three times, each time with a different one-third of the annotations masked so that the SVM algorithm does not know whether or not they are in the category when it is assigning discriminant values. Any given discriminant value is then converted to a precision value by simply asking what proportion of the masked genes with discriminant values above the given discriminant value really are in the category in question. The proportion of known genes in the category that are identified by the SVM as being in the category is also obtained at each discriminant value, and is referred to as recall. For all subsequent analyses we used precision and recall as our primary measures of success.
We trained separate SVMs for each of the 992 GO-BP categories. This revealed that genes in hundreds of categories could be recognized with precision greater than 50% (Figure 6a). Typically, not all of the genes in a category could be recognized (the curves in Figure 6a correspond to recall of 10% through 40%); this is due to the fact that not all genes within any given category display the characteristic expression pattern (Figure 5). As a control, when the gene labels were randomized, only zero to fifteen categories (depending on the randomization run) achieved 10% precision and 10% recall simultaneously (black dotted line at the bottom of Figure 6a). Therefore, this analysis demonstrates that, in a blind test, the known genes in many functional categories can be distinguished on the basis of the expression profiles of other genes that are members of the same functional category. This implies that there are distinct regulatory mechanisms that control these pathways, and indicates that correlation-based methods can be used to predict the functions of uncharacterized genes in mammals.
Predicted functions for unannotated genes are supported by sequence features
We next used these trained SVMs (Figure 6a) to predict functions for the 12,123 unannotated genes for which we detected expression in our data. The number of genes with at least one predicted function (that is, one GO-BP category) is shown in Figure 6b at varying precision thresholds (blue line). All of the predictions with precision above 15% are listed in the Additional data files. To make the outputs easier to peruse manually, we grouped 587 GO categories into 231 'superGO' categories, by combining categories that resulted in the same set of predicted genes and that were manually verified to be physiologically related. Figure 6b (red line) confirms that the number of unannotated genes that are predicted to have some function by an SVM with 'superGO' categories are similar to those with the original GO categories, although the number of categories has been compressed.
In order to provide a set of 'highest priority' predictions, we singled out those with the highest estimated precision. Among the unannotated genes (that is, those carrying no annotation in GO-BP), 1,092 (representing 117 superGO categories) were associated with precision values of 50% or greater; thus, on the basis of the analysis above, each of these genes is more than 50% likely to be involved in the given biological process. Figure 7 shows the original microarray data for these 1,092 genes, sorted by the predicted categories. Predictions were made for genes expressed in all of the tissues analyzed, and represent a wide spectrum of biological processes.
While some predictions correspond to expression in a single tissue (for example, the 56 genes predicted in 'vision' were predominantly expressed in the eye), such cases were unusual. Rather, most of the predictions were based on expression in multiple functionally related tissues (for example, the five genes predicted in 'regulation of cell migration' were characterized primarily by high expression in colon, large intestine, and small intestine) or more complex patterns (for example, genes predicted in 'CNS/brain development' were preferentially expressed in all adult neural tissues as well as in embryonic heads). Many predictions were found to be in categories related to the cell cycle and RNA processing. These genes tended to be expressed constitutively, but were most highly expressed in embryonic tissues, presumably because of rapid cell growth during development. However, many other predictions relate to neural functions, the immune response, muscle contraction, small-molecule metabolism, and other aspects of adult physiology. All of the individual predictions are provided in a table in the Additional data files, together with the expected precision and other information regarding the gene and the encoded protein, and these can be sorted by gene or by functional category.
Among the 1,092 unannotated genes, 488 (45%) have no overt sequence features suggesting physiological or biochemical function (that is, they have no similarity to previously characterized proteins or known functional domains; they are listed in Additional data files; and also see Materials and methods). Examination of the remaining 55% provided evidence that many of the predictions are likely to be correct. First, a handful of genes that were not annotated in our version of GO have in fact been characterized in the literature. For example, SVMs correctly predicted that phospholamban, the regulator of the Ca2+-ATPase in cardiac sarcoplasmic reticulum [27] is involved in 'muscle contraction or development'. Other genes are similar to characterized genes in other species: for example, the mouse homolog of the yeast 'Extra Spindle Poles' (ESP1) gene was predicted by SVM to function in 'mitotic cell cycle', 'cytokinesis', and 'microtubule based process', consistent with the function of its yeast counterpart [28].
A more comprehensive and objective analysis was enabled by the fact that, in an independent sequence-based analysis we conducted (see Materials and methods), known protein domain structures were encoded by 461 (42%) of these 1,092 unannotated genes (listed in the Additional data; see also the Materials and methods section) [29]. These provided further independent support for many of the predictions, since neither the primary sequences nor the domain features of the unannotated genes played a part in the predictions. In many cases, the domains also augment the predicted physiological function with a potential biochemical mechanism. For example, 3 of the 11 genes predicted in the category 'acyl-CoA/fatty acid/peroxisome' encode a short-chain dehydrogenase motif, suggesting that they are metabolic enzymes. Among the 86 unannotated genes predicted to function in 'microtubule-based process' are 4 with chromosome-segregation ATPase domains, one with an intermediate filament protein domain, one with a kinesin-motor domain, one with a myosin heavy-chain domain, and one with a tropomodulin domain, all of which are consistent with microtubule- and/or cytoskeleton-related functions. Of the four proteins predicted in 'skeletal development', one encodes a fibrillar collagen carboxy-terminal domain, and one encodes a collagen triple-helix repeat. Some of the relationships between predictions and domains are striking on the simple basis of their numbers: 7 of the 95 genes predicted in 'humoral immune response' encode an immunoglobulin domain; 13 of the 87 genes predicted in 'chromosome organization/DNA packaging' have high mobility group (HMG) domains, and 23 of the 149 genes predicted in 'RNA processing/ribosome biogenesis' encode helicase domains, RNA-binding domains, or RNA-modifying motifs. Table 1 lists a selection of statistically significant associations between the different prediction classes shown in Figure 7 and protein domains.
Comparisons among data sets for predicting gene functions
Although there was a significant correlation among the three different mouse tissue-specific data sets compared in Figure 2a, there were also many cases in which the three data sets disagreed in their assessment of relative abundance of individual genes in different tissues (Figure 2a and data not shown). We reasoned that the SVM cross-validation analysis could provide an objective measure of the quality of the different data sets: since random measurements lead to very poor predictions (Figure 6a), any errors in the data would tend to degrade the precision and recall values. While our manuscript was in preparation, an additional data set was released by Su et al., the authors of reference [15]. Their newer data [30] include measurements of 36,182 known and predicted genes over 61 tissues, measured in duplicate using custom-built Affymetrix arrays, and are thus similar in scope to our data set. Figure 6c shows a comparison between cross-validation results from running SVMs on the three data sets: ours, that of Su et al. [30], and that of Bono et al. [17], with each restricted to the 13 tissues and 1,800 genes common to all three, and the same GO-BP annotations used for all three data sets. Figure 6c shows that, although our data fare slightly better, the power of our data set and that of Su et al. [30] for predicting GO-BP categories are comparable. This confirms that distinct and coordinate regulation of many mammalian functional pathways is authentic because it is observed in two independent data sets.
Comparison of tissue-specificity with co-expression for predicting gene functions
We used two different approaches to ask how well tissue specificity can predict the functional classes of genes, in comparison to co-expression. First, from our data we compiled three sets of lists: genes that are expressed in each of the 55 individual samples; genes that are expressed highest in each of the individual 55 samples and in groups of functionally related samples (for example, treating all neural tissues as a single group); and also genes that are expressed uniquely in individual samples. All of these lists (175 in total) are compiled in the Additional data files. For each of the 992 GO-BP categories, we assessed the precision and recall for each of these lists (that is, whether these lists can distinguish genes in the category from those not in the category), and then identified the best precision value and its associated recall value for that category. Figure 6d shows a histogram of the difference between SVM precision and tissue-specificity precision, at the same recall value, for each GO-BP category. The vast majority of data points are greater than zero (P < 10-76; two-sided pairwise t test), indicating that co-expression patterns can be used (by SVMs) to predict gene functions significantly better than tissue-specificity alone.
It is possible that improved results might be obtained by other ad hoc procedures for sorting the genes in different ways, or by more automated procedures for generating large numbers of lists. However, an alternative analysis suggests that this is unlikely: when we re-ran SVMs with the matrix of 1s and 0s indicating which gene is expressed (or not) in each tissue, rather than the matrix of quantitative expression values, the resulting predictions were inferior (dotted magenta line in Figure 6a). In theory, if any combination of on/off information about gene expression in different tissues was informative for identifying genes in any category, it would have been identified by the SVMs. The result we obtained indicates that the quantitative measurements contain critical information reflecting functions of genes that is not, for the most part, contained in the binary (expressed/not expressed) information.
Validation of predictions by de novo functional analysis
Finally, we asked whether new functional predictions could be confirmed by directed experimentation. Among the genes we predicted to function in RNA processing and ribosome biogenesis was PWP1, which encodes a protein that includes WD40 repeats and which has previously been found to be up-regulated in pancreatic cancer tissue [31]. In our data, PWP1 was most highly expressed in embryonic tissues, as is characteristic of most genes annotated as 'RNA processing' by GO-BP (Figure 8a). The encoded protein Pwp1p is highly conserved across eukarya (Figure 8b) but to our knowledge it has not been functionally characterized in any species, although it has been found in the human nucleolus [32], and in yeast it is essential for cell growth [33]. We created a titratable-promoter allele of yeast PWP1, and found that cells depleted for Pwp1p displayed a striking reduction in 25S rRNA (Figure 8c), confirming the involvement of this gene in RNA processing and ribosome biogenesis. Given that WD40 repeats are thought to be protein interaction domains, we also asked whether Pwp1p physically associates with other proteins. We found that epitope-tagged yeast Pwp1 protein co-purified with known trans-acting ribosome biogenesis factors, as well as with several ribosomal protein subunits (Figure 8d), consistent with a direct role in ribosome biosynthesis.
Discussion
Simultaneous gene discovery, network mapping, and functional inference
The data presented here and the resulting inferences for the physiological roles of mammalian genes significantly extend previous microarray-based analyses of mammalian gene expression [7,15,17,21,23,24,30]. First, the data support the notion that there are thousands of mouse genes that are not represented in current cDNA databases [12,34-36]. Amongst all 21,622 confidently detected transcripts (Figure 4), 5,600 were not associated with a cDNA; 3,551 of these had EST but not cDNA support, indicating that many of them correspond to bona fide genes (Figure 2b). Moreover, inferences for the physiological roles of these transcripts can be obtained by analysis of quantitative expression levels across tissues; the 1,092 unannotated genes for which we made high-confidence predictions (Figure 7) contain 54 with no EST or cDNA support, and an additional 114 that have only EST support.
Second, our estimate that more than 58% of all transcripts are regulated together with genes in specific functional categories is much higher than previous estimates. One analysis, based on cursory analysis of early yeast and Xenopus expression data, suggested that only 5–10% of all genes fall within 'synexpression' groups [2]. Our results represent a minimum estimate of the correspondence between gene function and gene expression, because shortcomings in either the annotations or the data would tend to reduce these figures. Our results indicate that there are regulatory pathways that control many distinct biological processes in mammals, and that it is already possible to interpret the expression patterns of the majority of mammalian genes in a functionally meaningful way by comparing them to the patterns of the subset of genes that are already annotated. Moreover, it may be more straightforward than was previously anticipated to apply the same computational techniques to mammalian microarray data that are now being applied to identify 'network modules' and regulatory mechanisms in far simpler organisms, such as yeast (see for example, [37]). These potential applications represent an obvious future extension of the work presented here.
Third, while previous analyses using microarray expression data to predict gene functions [1-10] have focused on the fact that genes in large or general categories can be recognized (for example, ribosome biogenesis, translation, or proteolysis), we show quantitatively that this methodology is applicable to a much wider variety of functional categories, many of which are specific to higher organisms (for example, the category 'Pregnancy/embryo implantation' is specific to mammals; Figures 5,7).
Fourth, our analysis shows that the use of quantitative gene expression measurements to infer mammalian gene functions is more powerful than the traditional approach of using information on simple tissue specificity. Genes in many GO-BP categories were precisely identified by SVM using quantitative co-expression, but not on the basis of tissue specificity or tissue restriction (Figure 6a,6d). It appears from our results that genes in functional categories with more widespread expression (such as 'epidermal differentiation', 'regulation of cell migration', and 'apoptotic program'), categories corresponding to basic cellular functions (such as 'oxidative phosphorylation', 'RNA processing', and 'DNA replication'), and even categories that describe interactions among different cell types ('taxis', 'glycoprotein metabolism', 'cell-cell adhesion') can all be recognized and distinguished from genes in other categories on the sole basis of their coordinate expression across many tissues (Figures 5,7), even though they are expressed in many tissues (and in some cases all tissues, as in the case of mRNA splicing and other 'housekeeping' functions). PWP1 is an example of a gene that is widely expressed but has a pattern of expression that was predictive of its function (Figure 8).
A strategy and resource for mouse functional genomics
Analysis of mutant phenotypes is one of the most powerful and definitive ways to study gene functions. Many of the predicted gene functions (see Figure 7 and the Additional data files) in turn predict specific mutant phenotypes; for example, mutation of genes predicted to function in 'vision' would be expected to display defects in sight or eye morphology, while mutation of those predicted to function in 'RNA processing/ribosome biogenesis' might be lethal embryonically, but with alterations in RNA profiles or ribosome content. We have already initiated efforts to validate predicted gene functions in animals: of the XM genes on our array, 2,917 are already represented in collections of publicly available gene-trap ES cell lines [38] (indicated on the right in Figure 7). It will become increasingly straightforward to test these predictions as RNA interference methods are refined and the collection of mapped mouse mutants expands [38-40]. All of our predictions are listed in the Additional data files and will provide guidance for future efforts in mammalian functional genomics and/or support for other functional studies.
In contrast to other available mouse tissue-specific data sets (such as those described in [30]), all of our data, as well as the SVM predictions, can be downloaded anonymously without restriction from the Additional data files or from our website [16] and can be freely copied, modified, and propagated. The oligonucleotide sequences are provided, so that copies of our array design can be obtained and modified by other labs, and our expression data can be mapped to any clone collection or updated sequence annotation by batch BLAST. Many other supporting files are provided, including GO annotations, maps to gene-trap collections, and genomic locations of the probes. To facilitate perusal of the data, we have also created a web tool [19] that displays subsets of the gene expression data together with functional information and SVM predictions. This tool currently supports queries originating with a gene of interest, a functional category of interest, or a region of the chromosome of interest, which may facilitate the use of gene expression patterns and predicted gene functions in identifying genes that confer mapped traits [41]. We anticipate expanding and refining this resource to mirror both additional data and updated annotations.
Conclusion
We have created an extensive mouse expression data set and asked whether quantitative gene expression patterns correspond to functional gene categories. Our major finding is that most tissues express many functional categories, consistent with the fact that they contain many different cell types performing many different functions, but that many different functional pathways are coordinately expressed in a quantitative manner across tissues such that many categories display one or more distinctive patterns. For example, embryonic heads contain many cell types, and consequently express genes in a variety of categories including 'CNS/brain development', 'M phase', 'skeletal development', and 'microtubule-based process', yet an SVM can distinguish genes in these categories because they are differentially regulated across all 55 tissues in a way that is characteristic for each functional category (Figures 5,7). The simplest explanation for this observation is that there are discrete factors or sets of factors that control each coordinately regulated pathway. We conclude that functional genomics strategies that rely on quantitative transcriptional co-expression will be as fruitful in mammals as they have been in simpler organisms, and that transcriptional control of mammalian physiology may be more modular than is generally appreciated.
Materials and methods
Mouse mRNA isolation
Mouse tissues were isolated from the following strains: ICR (whole brain, testis, skeletal muscle, heart, lung, liver, embryo at 15 days, embryo at 12.5 days, embryo at 9.5 days, mammary gland, placenta at 9.5 days and placenta at 12.5 days); CD1 (Charles River Laboratory, Wilmington, USA; cortex, cerebellum, striatum, hindbrain, midbrain, bone marrow, knee, teeth, mandible, calvaria, femur (bone marrow flushed diaphyses), tongue surface, snout, large intestine, thyroid, aorta, brown fat, lymph node, olfactory bulb, adrenal gland, prostate, digits, trachea, trigeminal nucleus); C3H (The Jackson Laboratory, Bar Harbor, USA; salivary gland, thymus, ovary, uterus, tongue, stomach, small intestine, spleen, colon, uterus, pancreas, epididymis, eye, bladder, skin); C57BL/6 (The Jackson Laboratory, Bar Harbor, USA; spinal cord); Black Swiss (NTac:NIHBS; Embryonic heads); and R1 (ES cells). With the exception of embryonic tissues and ES cells, tissues were harvested from 3–6 month-old mice. Following recommended University of Toronto protocols, mice were euthanized by barbiturate injection and tissues were dissected as quickly as possible (within 10 minutes), snap-frozen in liquid nitrogen, and preserved at -80°C until use. RNA was extracted using homogenization and Trizol reagent (Invitrogen, Carlsbad, USA) following the instructions from the manufacturer, and mRNA was purified as described previously [3].
Microarray probe design
A FASTA file of 42,192 known and predicted mRNAs (XM sequences) was obtained from Deanna Church at NCBI on July 9, 2002 and is posted as Additional data file 1. Interspersed repeats and low complexity DNA sequences were masked with Repeat-Masker [42]. The 500 nucleotides from the 3' end of each mRNA were extracted and 10 non-overlapping Tm-balanced probes were generated using PrimerX [43] with default settings. The most unique among the 10 was identified on the basis of having the highest ΔG difference between the first (identical) and second most significant BLAST hits among the 42,192 initial XM mRNAs. Then, 41,699 probe sequences (those for which probes could be designed using this procedure) were submitted for oligonucleotide microarray production (Agilent Technologies, Palo Alto, USA). These arrays are manufactured using an ink-jet process, in which oligonucleotides are synthesized on the array by direct deposition of phosphoramidites [13]. The specificity, sensitivity, and reproducibility of these 60-mer arrays has been described elsewhere in detail [13].
Among the probes on the array, 40,822 were unique; those that were not unique can be attributed primarily to gene duplications, predominantly pseudogenes of GAPDH, ribosomal proteins, and retrovirus-like sequences. To minimize the impact of redundancy on statistical analyses, we collapsed the data from 1,928 duplicated probes and XM sequences that were in these sequence families (including 100 probes duplicated between the two array designs) into 525 groups that shared identical probe sequence and/or were both annotated and regulated in the same way. We also mapped all of the XM sequences to the current version of the mouse genome (Build 32) and to three cDNA databases (UniGene, RefSeq, and Fantom II; see below) and identified 1,991 XM sequences in which XM sequences adjacent on the chromosome also mapped to the same cDNA; these were collapsed into 904 groups. The Additional data files include a table mapping the 41,699 probes against the 39,309 presumed distinct transcripts.
Labeling and hybridization
The mRNA (1–2 μg) was reverse-transcribed with random nonamer primers (1 μg per reaction) and T18VN (0.25 μg per reaction) to synthesize cDNA. The reaction contained a 1:1 mixture of 5-(3-aminoallyl) thymidine 5'-triphosphate (Sigma, St. Louis, USA) and thymidine triphosphate (TTP) in place of TTP alone. The cDNA products were bound to QIAquick PCR Purification columns (Qiagen, Hilden, Germany) following the manufacturer's instructions, washed three times with 80% ethanol, and eluted with water. Purified cDNA was reacted with N-hydroxy succinimide esters of Cy3 or Cy5 (Amersham Pharmacia Biotech, Piscataway, USA) following the manufacturer's instructions. Hydroxylamine-quenched Cy-labeled cDNAs were separated from free dye molecules using QIAquick columns. Mixed labeled cDNAs were added to hybridization buffer containing 1 M NaCl, 0.5% sodium sarcosine, 50 mM methyl ethane sulfonate (MES), pH 6.5, 33% formamide and 40 μg herring sperm DNA (Invitrogen, Carlsbad, USA). Hybridizations were carried out in a final volume of 0.5 ml injecting into an Agilent hybridization chamber at 42°C on a rotating platform in a hybridization oven (Robbins Scientific Corporation, Sunnyvale, USA) for 16–24 h. Slides were then washed (rocking for approximately 30 seconds in 6 × SSPE, 0.005% sarcosine, then rocking for approximately 30 seconds in 0.06 × SSPE) and scanned with a 4000A microarray scanner (Axon Instruments, Union City, USA). Hybridizations were performed in duplicate with fluor reversal: that is, each mRNA sample was examined in duplicate, once in the Cy3 channel and once in the Cy5 channel, on separate arrays. Each array was hybridized with two samples simultaneously, each from an individual tissue. Essentially identical results were obtained from single-channel data from the same mRNA sample analyzed on different arrays, which were distinct from individual channels on the same arrays analyzed with a different mRNA. The organization of the hybridizations, and the data for individual channels, are given in the Additional data files.
Image processing and normalization
TIFF images were quantitated with GenePix (Axon Instruments). Individual channels were spatially detrended (that is, overall correlations between spot intensity and position on the slide were removed) by high-pass filtering (see [44]) using 10% outliers. We applied variance stabilizing normalization (VSN) [45] using 25% of the genes to normalize all single channels to each other. We manually identified and removed measurements that were inconsistent between dye-swaps, by either removing data from residual artifacts apparent on microarray images or removing the higher of the two disparate intensity measurements (in order to minimize false-positive detections). Measurements were transformed to arcsinh values (which are similar to natural log values, but are defined for negative numbers which emerge from the VSN) and for each measurement the median across all arrays was subtracted to obtain relative expression ratios for each gene in each tissue compared to all tissues. Remaining inconsistencies between dye-swaps were addressed by removing the higher of any two measurements that differed by more than two arcsinh units (in order to further minimize false-positive detections). The dye-swap arcsinh values were then averaged between replicates and among multiple probes detecting the same sequence. Clustering and manual analysis indicated that ratios below zero were generally not biologically meaningful (and probably stem largely from measurement error among low-intensity spots); hence ratios below zero were set to zero for all analyses using median-subtracted arcsinh values (Figures 1, 2, 4, 5, 6, 7 and SVM analyses). Missing values (fewer than 0.01% of all data points) were set to zero. Median-subtracted arcsinh values correspond approximately to the following ratios (arcsinh = linear): 0 = 1/1; 1 = 2.7/1; 2 = 7.5/1; 3 = 20/1, 4 = 55/1; 5 = 155/1, 6 = 405/1.
Annotations
Mouse GO-BP annotations were downloaded from the Gene Ontology website [46] and the European Bioinformatics Institute (EBI) [47] and both were mapped to XM gene sequences by sequence identity to the annotated source sequences. The full annotation database is on our website [16]. Fewer than 0.01% of these annotations were derived from gene expression (IEP code); we confirmed that removal of these genes had no appreciable impact on statistical analysis or the SVM analysis, and hence the use of these annotations to analyze gene expression is not circular. The Mouse Genome Informatics (MGI) annotations are reported to be manually compiled, whereas the EBI annotations include automated sequence-based annotations (for example, potassium channels are annotated as being in 'ion transport' and the mouse homolog of the yeast Tim8 protein, which is a translocase of the inner mitochondrial membrane, is annotated as being in 'mitochondrial translocation'). All GO-BP annotations were propagated up all possible edges of the GO graph. Redundant GO-BP categories were excluded. Categories with fewer than three genes among the 21,622 expressed genes were excluded from our analysis since they are not appropriate for the statistical tests we used, and those with more than 500 genes were excluded because they are not specific to distinct physiological processes.
False-discovery rate analysis
Each gene was associated with a co-regulated group consisting of the 50 annotated genes with the highest Pearson correlation coefficient relative to it. Annotation enrichment of this group in each GO-BP category was scored using the hypergeometric P value [48]. The minimum value of this score across all GO-BP categories was used as the measure for significant enrichment in any GO-BP category. P values were assigned to these measures using a permutation scheme on the gene labels. The statistical significance of the P values was evaluated using the Benjamini-Hochberg (BH) linear step-up procedure [22] to ensure a false discovery rate (FDR) of less than 1%. For annotated genes, a second measure was computed: the minimum among its annotated categories of the hypergeometric P values of its co-regulated group. A gene-specific permutation scheme associated P values with these scores and the FDR was also controlled at 1%.
Cluster diagonalization
Starting with an initial hierarchical clustering (agglomerative, average linkage, based on Pearson correlation coefficient), rows were divided into groups by removing a small number of links at the highest levels of the tree and grouping together all rows contained within the same disconnected subtree. Each row group was then associated with the column that contained the maximum expression value averaged over all the profiles in the group. The row groups were then sorted in increasing order of their associated column numbers.
Support vector machines
We used the SVM software package Gist [49] version 2.0.8 in Linux with parameter settings '-radial -zeromeanrow -diagfactor 0.5'. Precision was established by three-fold cross validation.
Identification of corresponding clones in cDNA and EST databases
We identified the closest corresponding mouse mRNAs in FANTOM II [50] (60,770 sequences); RefSeq [51] (16,601 sequences); UniGene [52] (87,495 sequences); and Ensembl [53] (32,911 sequences) using BLASTN with a threshold of E-60. We identified corresponding mouse mRNAs in dbEST [54] (3,939,961 sequences) using BLASTN with a threshold of E-20.
Identification of genes common to other microarray data and Spearman rank correlations
For Figure 2a, mRNA sequences were downloaded from [55] (for Su et al. data [15]) and [56] (for Bono et al. data [17]). The Su et al. [15] gene expression data were downloaded from [57] (9,977 sequences represented on the array) and the Bono et al. [17] data, from [58] (54,005 sequences represented on the array). The selected 41,699 NCBI mRNAs were used in a BLAST search against these two mRNA databases; a BLAST comparison between the two databases was also performed to retain only genes for which the closest sequence to each XM gene is also the closest sequence between the two other databases. All BLAST searches were performed with threshold E-60, and the best hit was selected for the multiple blast results. The 1,109 genes that have common hits in all the BLAST results and with gene expression data available were selected for the gene expression analysis. The 1,109 genes from all three datasets were normalized to make them comparable. To facilitate comparison, in the Bono et al. [17] dataset, each gene was median-centered in each tissue by subtracting its median expression value across all 13 common tissues. The Su et al. [15] data were arcsinh-transformed before median-centering. The data from the study described here that was used in the comparison was not zeroed, as it was in other analyses, and was median-centered using the median calculated only on the 13 common tissues, rather than all 55. The Spearman rank correlation coefficients of each pair of tissues among all three studies were transformed to Z-scores by multiplication by sqrt(1108) and then converted to P values using the cumulative probability density of a standard normal distribution.
For Figure 6c, an alternative mapping strategy was employed: our probe sequences, the Bono et al. [17] clone sequences, and the Su et al. [30] probe sequences were associated with 30,832 MGI sequences by mapping directly to corresponding MGI/GenBank sequences; 1,800 genes were identified in which a reciprocal best match between the probe sequences and the MGI sequence was identified in all three studies.
RT-PCR
Primer pairs were designed to have a matching Tm (59°C) and sequences are listed in the Additional data files. RT-PCR assays were performed using the OneStep RT-PCR Kit (Qiagen). Reactions were performed in 25 μl volumes containing 0.5 ng polyA+ mRNA, 7.5 units porcine RNAguard (Amersham) and 300 pM each of the forward and reverse primers. After 30 rounds of amplification, the reaction products were separated on 2% agarose gels stained with ethidium bromide. Inverted black-and-white images of the gels were recorded using a Syngene gel documentation system and GeneSnap software (Synopics, Frederick, USA). In total, 107 primer pairs were tested. Of the 57 XM genes tested that corresponded to a known cDNA, 42 were among those that were amplified (74%). Of the 25 tested that corresponded to an EST but not to a known cDNA, 12 were amplified (48%). However, of the 25 tested that did not correspond to a cDNA or EST, only one was amplified (4%).
Identification of genes associated with gene traps
Six different gene-trap resources were searched to identify genes associated with gene trap ES cell lines. For BayGenomics [59], Centre for Modeling Human Disease (CMHD) [60], University of California Resource of Gene Trap Insertions [61], and Fred Hutchinson Cancer Research Center (FHCRC) [62], the gene-trap sequence tags were downloaded from the website and searched against the selected 41,699 mRNA sequences using BLASTN. For the German Genetrap Consortium (GGTC) [63] and Mammalian Functional Genomics Centre (MFGC) [64], the web-based BLAST servers were used to search the 41,699 mRNA sequences against their gene-trap sequence databases. The hits with lengths equal to or larger than 50 nucleotides, and identity equal to or larger than 98%, were considered to be associated with the gene-trap ES-cell lines.
RNA extraction, northern blotting, affinity purification, and mass spectrometry
The TetO7-PWP1 and isogenic wild-type control strains were created and analyzed as previously described for other essential yeast genes [9]. Briefly, strains were exposed to 10 μg/ml doxycycline (Sigma) for a total of 24 h before harvesting for RNA extraction. RNA extraction and northern blotting were performed using standard protocols and oligonucleotide probes as described previously [9]. TAP purification of Pwp1p was performed as previously described [9] using 1.3l culture volumes; gel-purified proteins were identified by MALDI-TOF mass spectrometry.
Additional data files
There are 40 Additional data files comprising all the raw data; they are also available on our website [16]. A web tool for querying and browsing the data online is also available [19].
Supplementary Material
Additional data file 1
XM (predicted mRNA sequences) from NCBI
Click here for additional data file
Additional data file 2
XP (encoded protein sequences) from NCBI
Click here for additional data file
Additional data file 3
Array Design 1 (spot map)
Click here for additional data file
Additional data file 4
Array Design 2 (spot map)
Click here for additional data file
Additional data file 5
Array Master file - 41,699 probes; the Master file contains probe sequences, columns listing the closest sequences in RIKEN, ENSEMBL, Refseq, and Unigene; GenBank description, EST overlap, GO-BP annotations, Domain (most significant)
Click here for additional data file
Additional data file 6
Array information after removal of redundant probes: probe combinations
Click here for additional data file
Additional data file 7
Array information after removal of redundant probes: master file - 39,309 presumed distinct transcripts
Click here for additional data file
Additional data file 8
Hybridization records
Click here for additional data file
Additional data file 9
Figure 1 data
Click here for additional data file
Additional data file 10
Figure 2b data
Click here for additional data file
Additional data file 11
Figure 4a data
Click here for additional data file
Additional data file 12
Figure 4b data
Click here for additional data file
Additional data file 13
Figure 4c data
Click here for additional data file
Additional data file 14
Figure 4d data
Click here for additional data file
Additional data file 15
Figure 5 data
Click here for additional data file
Additional data file 16
Figure 7 data
Click here for additional data file
Additional data file 17
Supplementary figures S1-3
Click here for additional data file
Additional data file 18
Data: 41,699 probes - single channel, arcsinh intensities
Click here for additional data file
Additional data file 19
Data: 41,699 probes - replicates combined, arcsinh intensities
Click here for additional data file
Additional data file 20
Data: 41,699 probes - median-subtracted, zeroed
Click here for additional data file
Additional data file 21
Data: 39,309 presumed distinct transcripts - median-subtracted, zeroed
Click here for additional data file
Additional data file 22
Data: 21,622 presumed distinct transcripts - median-subtracted and zeroed, expressed above 99% of negative-control spots
Click here for additional data file
Additional data file 23
Data: binary matrix of expression above 99% of negative-control spots
Click here for additional data file
Additional data file 24
Annotations: GO annotations among 39,309 presumed distinct transcripts (12,543 annotated genes, 47,900 annotations)
Click here for additional data file
Additional data file 25
Annotations: GO annotations among 21,622 presumed distinct transcripts (9,499 annotated genes, 37,876 annotations)
Click here for additional data file
Additional data file 26
Annotations: superGO annotations
Click here for additional data file
Additional data file 27
Annotations: map between GO and superGO annotations
Click here for additional data file
Additional data file 28
SVM Predictions: GO-BP SVM Predictions, 15% precision
Click here for additional data file
Additional data file 29
SVM Predictions: GO-BP SVM Predictions, 50% precision
Click here for additional data file
Additional data file 30
SVM Predictions: superGO SVM Predictions, 15% precision
Click here for additional data file
Additional data file 31
SVM Predictions: superGO SVM Predictions, 50% precision
Click here for additional data file
Additional data file 32
RT-PCR primer sequences
Click here for additional data file
Additional data file 33
XM genes with gene trap lines
Click here for additional data file
Additional data file 34
XM motifs
Click here for additional data file
Additional data file 35
779 known and putative DNA-binding transcription factors among XM genes
Click here for additional data file
Additional data file 36
Table of accession numbers of genes common to Zhang, Su, and Bono data
Click here for additional data file
Additional data file 37
Tech report on spatial detrending
Click here for additional data file
Additional data file 38
Map locations of XM genes (BLASTed against Build 32) - retained only if top hit of both XM sequence and array probe overlap (30,387 presumed distinct transcripts)
Click here for additional data file
Additional data file 39
SVM functional predictions for 7,147 unannotated mapped transcripts (may be useful for positional cloning), also see [19]
Click here for additional data file
Additional data file 40
175 lists of genes that are expressed in individual tissues, highest in individual tissues, or specific to individual tissues
Click here for additional data file
Acknowledgements
We thank David MacLennan, Bill Stanford, and Charlie Boone for helpful conversations, Li Zhang, Richard Hill, Tony Candeliere, Dominic Falconi, and Usha Bhargava for assistance in obtaining mouse tissues, Dawn Richards and Victoria Canadien at Affinium Pharmaceuticals for MALDI-MS, Deanna Church for assistance with XM sequences, and Shawna Hiley for proofreading the manuscript. This work was supported by grants to T.R.H. and B.J.B. from CIHR, Genome Canada, and the CFI. WZ was supported by a University of Toronto Open scholarship and Q.D.M. is supported by an NSERC postdoctoral award. M.G.F. is supported by the Krembil Chair in Neural Repair and Regeneration.
Figures and Tables
Figure 1 Expression of previously characterized tissue-specific genes. Genes were identified manually by searching MEDLINE abstracts [66] and XM sequence description fields (see Additional data file 1) for keywords corresponding to the appropriate tissues. Rows and columns were ordered manually.
Figure 2 Validation of expression data by independent confirmation. (a) The P value of Spearman's Rank correlations (see Materials and methods) is shown for all possible comparisons among the 13 tissues common to all three studies (ours and those by Su et al. [15] and Bono et al. [17]) and 1,109 genes for which the same isoform is unambiguously represented on the arrays used in each of the studies (see Materials and methods). (b) Microarray data and RT-PCR results for 47 known and predicted XM genes are shown. Genes were selected to represent primarily those without GO Biological Processes (GO-BP) assignment and to encompass expression in all 18 tissues, and were biased towards those with functions predicted by support vector machines (SVMs) in categories of interest (or expressed in tissues of interest). The three columns on the far right show whether each XM gene was uncharacterized (not annotated) in GO-BP, and whether it is represented by a cDNA or EST.
Figure 3 Defining whether a gene is expressed, and how many genes are detected as expressed per sample. (a) The curves show the cumulative distribution for negative-control probes (cyan line) and for probes on the array (blue line), over all arrays, to illustrate how genes were defined as expressed. The dotted black line indicates the 99thpercentile for the negative control spots. (b) The number of genes expressed in any given number of tissues (between 1 tissue and 55 tissues; for example, there are 4,475 genes detected in only one sample, 171 genes expressed in exactly 27 samples, 1,790 genes detected in all 55 samples, and so on). The genes expressed in each of the 55 tissues were determined as in (a). (c) Number of genes defined as expressed in each of the 55 tissues, using criteria in (a).
Figure 4 Correspondence between gene expression patterns and GO-BP annotations. (a) Ratios for the 21,622 expressed genes were grouped by two-dimensional hierarchical agglomerative clustering and diagonalization, using the Pearson correlation coefficient. (b) Negative logs of P values resulting from applying the Wilcoxon-Mann-Whitney (WMW) test to each of the GO-BP categories in each of the tissues are shown. The categories (vertical axis) were clustered and ordered as in (a). (c, d) 'Density' of GO-BP annotations significantly enriched in specific points along the vertical axis at left (genes) are indicated; note that genes are in the same order in (a, b, c).
Figure 5 Expression of genes in 17 different functional categories. The categories were ordered manually. The genes within each category were clustered separately from those in other categories. The order of tissues is preserved from previous figures.
Figure 6 Predicting GO-BP categories of mouse genes using microarray data and SVMs. (a) The number of the 992 initial GO-BP categories exceeding the indicated precision value, with recall fixed for each line; for example at 40% recall (green line), around 100 categories achieve precision of 30%. To estimate the significance of the colored lines, we repeated their calculation after permuting the gene labels in the annotation database. The dotted gray line indicates the maximum number of GO categories that achieve the indicated precision, with recall of 10% or greater. The dotted magenta line indicates the result obtained using 'binary' expression data (expressed/not expressed) in each tissue. (b) The number of genes with predicted GO-BP categories (blue line) or superGO categories (red line) at varying precision values. The individual predictions are given in the Additional data files. (c) Comparison of the overall predictive capacity of three data sets, restricted to the 13 tissues and 1,800 genes shared by all three data sets. Each of the lines corresponds to the 30% recall line in (a). All of the lines are to the lower right of those in (a), since fewer genes and tissues were used. (d) A histogram comparing the precision of predictions derived from lists of tissue-specific genes with the precision of predictions from SVMs. For each category, the tissue-specific list yielding the highest precision value was identified, along with its associated recall value, and the SVM precision for the same category at the same recall value was identified. The difference between the two precision values is plotted for each category, such that instances where the SVM is superior are to the right of center.
Figure 7 Expression patterns of 1,092 unannotated genes predicted to belong to any of 117 'superGO' categories with 50% confidence or higher. The vertical axis was clustered and diagonalized as in Figure 4. The height of each predicted category has been normalized to facilitate display; the number of genes predicted in each category is indicated at the left. The gene order (vertical axis) has been clustered within each category to illustrate that some categories are characterized by multiple patterns. The proportion (%) of predicted genes in each category that have gene-trap ES cell lines available are represented at right (color scale from 0 to 100%).
Figure 8 PWP1 functions in ribosomal large-subunit biogenesis. (a) The expression pattern of mouse Pwp1 is similar to that of most known RNA-processing proteins. (b) The domain structures of Pwp1 homologs identified by BLASTP searches. Accession number and amino-acid length is given. We identified a single strong match in each of the species shown. Domains were identified by CDD search [29]. (c) A northern blot showing the accumulation of 35S rRNA precursor (blue arrow), reduction in other rRNA precursors (top panel), and reduction in 25S rRNA (red arrow) in the yeast TetO7-PWP1 mutant (strain TH_2220) in comparison to the parental wild-type strain (R1158) [9]. The U2 spliceosomal RNA is shown for comparison; its apparent abundance is increased because 5 μg RNA was loaded per lane, and the relative proportion of rRNA to snRNA is decreased in the mutant. Blotting procedures and probes were as previously described [9]. (d) Affinity-purification of yeast Pwp1p-TAP reveals association with proteins known to function in ribosomal large-subunit biogenesis (Ebp2p, Nop12p, Brx1p) as well as a subset of ribosomal proteins. The asterisks mark degradation products of Pwp1p-TAP.
Table 1 Domains associated with genes predicted to function in specific biological processes
Predicted function Enriched domain Description of domain Proportion of genes with this domain -log10 significance (P)
Chromosome organization or DNA packaging HMG HMG (high mobility group) box 13/87 10.5
Pregnancy/embryo implantation Hormone_1 Somatotropin hormone family 3/14 7.3
Acyl-CoA/fatty acid/peroxisome FabG Short-chain alcohol dehydrogenase 3/11 6.8
RNA/ribosome metabolism/processing RRM RNA recognition motif 10/149 6.6
Carboxylic acid/amine metabolism ECH Enoyl-CoA hydratase/isomerase family 3/62 6.2
Humoral immune response Sp100 The function of this domain is unknown 2/95 6.1
Vision Uteroglobin The function of this domain is unknown 3/56 5.9
RNA-nucleus import/export COG5136 U1 snRNP-specific protein C 2/13 5.7
Microtubule-based process Smc Chromosome-segregation ATPases 4/86 5.2
P values were calculated using the hypergeometric P value [48], which compares against expectation from random draws among the 15,443 XM genes with encoded domains. Domain names and descriptions are from the NCBI 'COG' database [65].
==== Refs
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns Proc Natl Acad Sci USA 1998 95 14863 14868 9843981 10.1073/pnas.95.25.14863
Niehrs C Pollet N Synexpression groups in eukaryotes Nature 1999 402 483 487 10591207 10.1038/990025
Hughes TR Marton MJ Jones AR Roberts CJ Stoughton R Armour CD Bennett HA Coffey E Dai H He YD Functional discovery via a compendium of expression profiles Cell 2000 102 109 126 10929718 10.1016/S0092-8674(00)00015-5
Kim SK Lund J Kiraly M Duke K Jiang M Stuart JM Eizinger A Wylie BN Davidson GS A gene expression map for Caenorhabditis elegans Science 2001 293 2087 2092 11557892 10.1126/science.1061603
Wu LF Hughes TR Davierwala AP Robinson MD Stoughton R Altschuler SJ Large-scale prediction of Saccharomyces cerevisiae gene function using overlapping transcriptional clusters Nat Genet 2002 31 255 265 12089522 10.1038/ng906
Chu S DeRisi J Eisen M Mulholland J Botstein D Brown PO Herskowitz I The transcriptional program of sporulation in budding yeast Science 1998 282 699 705 9784122 10.1126/science.282.5389.699
Clark EA Golub TR Lander ES Hynes RO Genomic analysis of metastasis reveals an essential role for RhoC Nature 2000 406 532 535 10952316 10.1038/35020106
Toth A Rabitsch KP Galova M Schleiffer A Buonomo SB Nasmyth K Functional genomics identifies monopolin: a kinetochore protein required for segregation of homologs during meiosis I Cell 2000 103 1155 1168 11163190 10.1016/S0092-8674(00)00217-8
Peng WT Robinson MD Mnaimneh S Krogan NJ Cagney G Morris Q Davierwala AP Grigull J Yang X Zhang W A panoramic view of yeast noncoding RNA processing Cell 2003 113 919 933 12837249 10.1016/S0092-8674(03)00466-5
Stuart JM Segal E Koller D Kim SK A gene-coexpression network for global discovery of conserved genetic modules Science 2003 302 249 255 12934013 10.1126/science.1087447
Gasch AP Spellman PT Kao CM Carmel-Harel O Eisen MB Storz G Botstein D Brown PO Genomic expression programs in the response of yeast cells to environmental changes Mol Biol Cell 2000 11 4241 4257 11102521
Waterston RH Lindblad-Toh K Birney E Rogers J Abril JF Agarwal P Agarwala R Ainscough R Alexandersson M An P Initial sequencing and comparative analysis of the mouse genome Nature 2002 420 520 562 12466850 10.1038/nature01262
Hughes TR Mao M Jones AR Burchard J Marton MJ Shannon KW Lefkowitz SM Ziman M Schelter JM Meyer MR Expression profiling using microarrays fabricated by an ink-jet oligonucleotide synthesizer Nat Biotechnol 2001 19 342 347 11283592 10.1038/86730
Yeh RF Lim LP Burge CB Computational inference of homologous gene structures in the human genome Genome Res 2001 11 803 816 11337476 10.1101/gr.175701
Su AI Cooke MP Ching KA Hakak Y Walker JR Wiltshire T Orth AP Vega RG Sapinoso LM Moqrich A Large-scale analysis of the human and mouse transcriptomes Proc Natl Acad Sci USA 2002 99 4465 4470 11904358 10.1073/pnas.012025199
The functional landscape of mouse gene expression
Bono H Yagi K Kasukawa T Nikaido I Tominaga N Miki R Mizuno Y Tomaru Y Goto H Nitanda H Systematic expression profiling of the mouse transcriptome using RIKEN cDNA microarrays Genome Res 2003 13 1318 1323 12819129 10.1101/gr.1075103
Richards M Tan SP Tan JH Chan WK Bongso A The transcriptome profile of human embryonic stem cells as defined by SAGE Stem Cells 2004 22 51 64 14688391 10.1634/stemcells.22-1-51
Mouse gene prediction database
Hossler FE Monson FC Structure and blood supply of intrinsic lymph nodes in the wall of the rabbit urinary bladder studies – with light microscopy, electron microscopy, and vascular corrosion casting Anat Rec 1998 252 477 484 9811226 10.1002/(SICI)1097-0185(199811)252:3<477::AID-AR16>3.0.CO;2-H
Miki R Kadota K Bono H Mizuno Y Tomaru Y Carninci P Itoh M Shibata K Kawai J Konno H Delineating developmental and metabolic pathways in vivo by expression profiling using the RIKEN set of 18,816 full-length enriched mouse cDNA arrays Proc Natl Acad Sci USA 2001 98 2199 2204 11226216 10.1073/pnas.041605498
Benjamini Y Hochberg Y Controlling the false discovery rate: a practical and powerful approach to multiple testing J Roy Stat Soc B 1995 57 289 300
Mootha VK Lindgren CM Eriksson KF Subramanian A Sihag S Lehar J Puigserver P Carlsson E Ridderstrale M Laurila E PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes Nat Genet 2003 34 267 273 12808457 10.1038/ng1180
Lee HK Hsu AK Sajdak J Qin J Pavlidis P Coexpression analysis of human genes across many microarray data sets Genome Res 2004 14 1085 1094 15173114 10.1101/gr.1910904
Brown MP Grundy WN Lin D Cristianini N Sugnet CW Furey TS Ares M JrHaussler D Knowledge-based analysis of microarray gene expression data by using support vector machines Proc Natl Acad Sci USA 2000 97 262 267 10618406 10.1073/pnas.97.1.262
Vapnik V The Nature of Statistical Learning Theory 1995 New York: Springer
Luo W Grupp IL Harrer J Ponniah S Grupp G Duffy JJ Doetschman T Kranias EG Targeted ablation of the phospholamban gene is associated with markedly enhanced myocardial contractility and loss of β-agonist stimulation Circ Res 1994 75 401 409 8062415
Baum P Yip C Goetsch L Byers B A yeast gene essential for regulation of spindle pole duplication Mol Cell Biol 1988 8 5386 5397 3072479
Marchler-Bauer A Anderson JB DeWeese-Scott C Fedorova ND Geer LY He S Hurwitz DI Jackson JD Jacobs AR Lanczycki CJ CDD: a curated Entrez database of conserved domain alignments Nucleic Acids Res 2003 31 383 387 12520028 10.1093/nar/gkg087
Su AI Wiltshire T Batalov S Lapp H Ching KA Block D Zhang J Soden R Hayakawa M Kreiman G A gene atlas of the mouse and human protein-encoding transcriptomes Proc Natl Acad Sci USA 2004 101 6062 6067 15075390 10.1073/pnas.0400782101
Honore B Baandrup U Nielsen S Vorum H Endonuclein is a cell cycle regulated WD-repeat protein that is up-regulated in adenocarcinoma of the pancreas Oncogene 2002 21 1123 1129 11850830 10.1038/sj.onc.1205186
Andersen JS Lyon CE Fox AH Leung AK Lam YW Steen H Mann M Lamond AI Directed proteomic analysis of the human nucleolus Curr Biol 2002 12 1 11 11790298 10.1016/S0960-9822(01)00650-9
Issel-Tarver L Christie KR Dolinski K Andrada R Balakrishnan R Ball CA Binkley G Dong S Dwight SS Fisk DG Saccharomyces Genome Database Methods Enzymol 2002 350 329 46 12073322
Guigo R Dermitzakis ET Agarwal P Ponting CP Parra G Reymond A Abril JF Keibler E Lyle R Ucla C Comparison of mouse and human genomes followed by experimental verification yields an estimated 1,019 additional genes Proc Natl Acad Sci USA 2003 100 1140 1145 12552088 10.1073/pnas.0337561100
Parra G Agarwal P Abril JF Wiehe T Fickett JW Guigo R Comparative gene prediction in human and mouse Genome Res 2003 13 108 117 12529313 10.1101/gr.871403
Penn SG Rank DR Hanzel DK Barker DL Mining the human genome using microarrays of open reading frames Nat Genet 2000 26 315 318 11062470 10.1038/81613
Bar-Joseph Z Gerber GK Lee TI Rinaldi NJ Yoo JY Robert F Gordon DB Fraenkel E Jaakkola TS Young RA Gifford DK Computational discovery of gene modules and regulatory networks Nat Biotechnol 2003 21 1337 1342 14555958 10.1038/nbt890
Stanford WL Cohn JB Cordes SP Gene-trap mutagenesis: past, present and beyond Nat Rev Genet 2001 2 756 768 11584292 10.1038/35093548
Nadeau JH Balling R Barsh G Beier D Brown SD Bucan M Camper S Carlson G Copeland N Eppig J Sequence interpretation. Functional annotation of mouse genome sequences Science 2001 291 1251 1255 11233449 10.1126/science.1058244
Kunath T Gish G Lickert H Jones N Pawson T Rossant J Transgenic RNA interference in ES cell-derived embryos recapitulates a genetic null phenotype Nat Biotechnol 2003 21 559 561 12679785 10.1038/nbt813
Walker JR Su AI Self DW Hogenesch JB Lapp H Maier R Hoyer D Bilbe G Applications of a rat multiple tissue gene expression data set Genome Res 2004 14 742 749 15060018 10.1101/gr.2161804
RepeatMasker documentation
Primer Selection
Shai O Morris Q Frey BJ Spatial bias removal in microarray images University of Toronto Technical Report PSI-2003-21 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 S96 S104 12169536
Gene Ontology download
Gene Onotology Annotation @ EBI
Tavazoie S Hughes JD Campbell MJ Cho RJ Church GM Systematic determination of genetic network architecture Nat Genet 1999 22 281 285 10391217 10.1038/10343
GIST Support vector machines
Carninci P Waki K Shiraki T Konno H Shibata K Itoh M Aizawa K Arakawa T Ishii Y Sasaki D Targeting a complex transcriptome: the construction of the mouse full-length cDNA encyclopedia Genome Res 2003 13 1273 1289 12819125 10.1101/gr.1119703
Pruitt KD Tatusova T Maglott DR NCBI Reference Sequence project: update and current status Nucleic Acids Res 2003 31 34 37 12519942 10.1093/nar/gkg111
Pontius JU Wagner L Schuler GD UniGene: a unified view of the transcriptome The NCBI Handbook 2003 Bethesda: National Center for Biotechnology Information
Clamp M Andrews D Barker D Bevan P Cameron G Chen Y Clark L Cox T Cuff J Curwen V Ensembl 2002: accommodating comparative genomics Nucleic Acids Res 2003 31 38 42 12519943 10.1093/nar/gkg083
dbEST
Affymetrix
Fantom
Gene Expression Atlas
Additional data files from Bono et al. [17]
BayGenomics
Centre for Modeling Human Disease
University of California Resource for GeneTrap Insertions
Fred Hutchinson Cancer Research Center
German GeneTrap Consortium
Mammalian Functional Genomics Centre
NCBI COGs database
Medline
| 15588312 | PMC549719 | CC BY | 2021-01-04 16:05:45 | no | J Biol. 2004 Dec 6; 3(5):21 | utf-8 | J Biol | 2,004 | 10.1186/jbiol16 | oa_comm |
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-91569137910.1186/1471-2148-5-9Research ArticlePhenotypic error threshold; additivity and epistasis in RNA evolution Takeuchi Nobuto [email protected] Petrus H [email protected] Paulien [email protected] Theoretical Biology/Bioinformatics Group, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands2005 3 2 2005 5 9 9 28 10 2004 3 2 2005 Copyright © 2005 Takeuchi et al; licensee BioMed Central Ltd.2005Takeuchi 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 error threshold puts a limit on the amount of information maintainable in Darwinian evolution. The error threshold was first formulated in terms of genotypes. However, if a genotype-phenotype map involves redundancy ("mutational neutrality"), the error threshold should be formulated in terms of phenotypes since there is no unique fittest genotype. A previous study formulated the error threshold in terms of phenotypes, and their results showed that a rather low degree of mutational neutrality can increase the error threshold unlimitedly.
Results
We obtain an analytical formulation of the phenotypic error threshold by considering the "additive assumption", in which base substitutions do not influence each other (no epistasis). Our formulation shows that an increase of the error threshold due to mutational neutrality is limited. Computer simulations of RNA evolution are conducted to verify our formulation, and the results show a good agreement between the analytical prediction and the simulations. The comparison with the previous formulation illustrates that it is important for the prediction of the error threshold to consider that the number of base substitutions per replication is rather large near the error threshold. To examine the additive assumption, a detailed analysis of additivity and epistasis in RNA folding of a particular sequence is performed. The results show a high degree of epistasis in RNA folding; furthermore, the analysis also elucidates the reason of the success of the additive assumption.
Conclusions
We conclude that an increase of the error threshold by mutational neutrality is limited, and that the additive assumption achieves a good prediction of the error threshold in spite of a high degree of epistasis in RNA folding because the average number of base substitutions of sequences retaining the phenotype per replication is sufficiently small to avoid of the effect of epistasis.
==== Body
Background
The error threshold is a limit on the permissible mutation rate for which "survival of the fittest" holds in Darwinian evolution [1]. The error threshold can be seen as a limit on the amount of information maintainable in evolutionary systems (information threshold) since an increase in sequence length results in an increase in error rate. The information threshold leads to a paradox in prebiotic evolution [2]. Suppose that to increase the maintainable amount of information, an evolving system must acquire a more complex molecular mechanism to reduce the mutation rate. However, to have such a complex molecular mechanism the system must maintain a longer sequence in the first place. Thus, the system will encounter a barrier in the evolution of complexity (cf. [3]).
The error threshold was first formulated in terms of genotypes. However if some changes in genotype do not alter the phenotype or the fitness (mutational neutrality), there is no unique genotype which can be stably maintained. Instead, the survival of a phenotype should be considered, and thus the error threshold should be formulated in terms of phenotypes [4].
In this paper, first we formulate the phenotypic error threshold analytically by employing the additive assumption, in which base substitutions do not influence each other. Under the additive assumption, we obtain the probability that a replication does not alter the phenotype (neutral replication) as a function of the number of base substitutions and the fraction of "neutral substitutions". Our results show a qualitative difference from the previous formulation [5]. Second, we analyze epistasis in RNA folding of a particular RNA sequence to examine the additive assumption.
Results and discussion
Phenotypic error threshold
Analytical formulation
The quasispecies equation describes (prebiotic) replicator dynamics in well-mixed systems [1]. We transform the equation in two ways: (1) describing the abundance of phenotypes instead of that of genotypes by denoting the population of genotypes which share the same phenotype by one variable (see [5] for mathematical details); (2) distinguishing only two classes of phenotypes, the focal phenotype (denoted by x) and the others called mutants (denoted by y). The population size is assumed to be large enough to express the abundance of the phenotypes by normalized concentration. The population dynamics of the phenotypes is described as
dx/dt = σQx + σΛ (1 - Q)x - Dx - Φx,
dy/dt = y + σ (1 - Λ)(1 - Q)x - Dy - Φy (1)
where σ (> 1) is the replication rate of x (that of y is normalized to 1); Q is the replication accuracy of x. Λ is the fraction of neutral mutants of x. D is the degradation rate (or the death rate) assumed to be uniform over the phenotypes. Φ = (σ - D)x + (1 - D)y is the excess production (or the mean fitness). The terms - Φx and - Φy induce a selection pressure. We neglect back mutation from y to x (this simplification will be discussed in the next section). From Eq. 1, we obtain the survival condition of x as Q + Λ (1 - Q) >σ-1, for which the stationary value of x is larger than zero. From this inequality, we will deduce the phenotypic error threshold, i.e., the maximum error rate of replication for which x can be stably maintained in the system.
We distinguish two classes of single base substitutions: neutral and deleterious substitutions. The class of a substitution is determined by the effect of the substitution on the phenotype when there are no other substitutions in the genotype: A substitution which retains the phenotype is a neutral substitution; otherwise it is a deleterious substitution. A beneficial substitution is not considered since the focus of the study is on the maintenance of x. [A replicator is thought of as a polymer in our study. We refer to a monomer as a base, having RNA in mind. The formulation of the phenotypic error threshold itself is independent of this terminology.]
To calculate the effective replication accuracy Qe = Q + Λ (1 - Q), we assume that a mutant is neutral iff there is no deleterious substitution: The effect of a substitution on the phenotype is independent of the other substitutions; i.e., no epistasis is assumed (the additive assumption). Let λ denote the fraction of neutral substitutions in all possible single substitutions, and let d denote the number of substitutions per replication. Then, the probability that d substitutions are all neutral substitutions (thus neutral replication) is approximated by λd by assuming that the number of neutral substitutions in d substitutions follows the binomial distribution (the binomial approximation). This approximation is valid if the probability of correct replication per base (denoted by q) is sufficiently large so that d is small. Denoting the sequence length of replicators by N, the effective replication accuracy (Qe) is obtained as
by assuming that q and λ are uniform among the genotypes in x, that q is invariable over sequence positions, and that N is the same among the populations. [A similar formula was obtained in [6] as Qe = (q + v(1 - q))N, where v is a parameter to be tuned to match the formula to the observed value of Qe. Therefore v in [6] implicitly involves both the additive effect and epistasis.]
The minimum q for which x can survive is derived from Q + Λ(1 - Q) >σ-1 and Eq. 2 as
qmin = (σ-1/N - λ)/(1 - λ). (3)
The phenotypic error threshold is 1 - qmin. As seen in Fig. 1 (the solid line), the increase in the error threshold is limited for almost all values of λ. This is because if q decreases, the number of substitutions per replication (d) increases; hence the probability of neutral replication (λd) decreases (cf. [5] and Eq. 5). At a large λ (= σ-1/N), there is a singularity such that qmin becomes zero. However, this singularity is not plausible in two ways: (1) Such a large λ is not realistic (see below); (2) q at the singularity is so small that the binomial approximation is threatened. We studied the validity of the binomial approximation, and found that the inaccuracy in the binomial approximation is largest if in some positions of the sequence all possible single substitutions are neutral, but in the rest of the positions all possible single substitutions are deleterious; i.e., the distributions of neutral and deleterious substitutions over the sequence positions are completely separated. By taking this extreme example, qmin is calculated with the additive assumption but without the binomial approximation. As Fig. 1 (the dotted line) shows, for a wide range of λ the binomial approximation is valid. qmin is underestimated by the binomial approximation only around the singularity (λ > 0.8), and thus the singularity is actually located at higher λ, which makes the singularity even less plausible. We conclude that the increase in the error threshold due to mutational neutrality is limited.
Figure 1 Error threshold The minimum permissible replication accuracy per base (qmin) is plotted against λ for three different ways of the calculation. The solid line is obtained from the additive assumption and the binomial approximation (Eq. 3). The binomial approximation is threatened at a high error rate. To examine this, the error threshold is calculated without the binomial approximation (but with the additive assumption) in the extreme example, where the binomial approximation deviates most (see the text). Let Nδ be the sum of the sequence length of the parts where all single mutations are deleterious. Then qmim is calculated as . The dotted line represents the so calculated error threshold in this extreme example. The x-axis for the dotted line (i.e., λ) is calculated as (N - Nδ)/N. The dashed line is obtained from the formulation of Reidys et al. [5] (Eq. 5). In all cases, N = 100 and σ = 10. (The same values of N and of σ as those used in [5] are chosen for a comparison purpose.)
From Eq. 3, we obtain the information threshold, i.e., the maximum permissible sequence length as
Nmax = ln(σ-1)/ln(q + (1 - q)λ). (4)
As Fig. 2 (the solid lines) shows, the increase in the information threshold is limited for plausible values of λ. (Nmax reaches infinity only when λ increases to one.) However, this result does not mean that a longer sequence in fact can have a larger λ, and thus it can be maintained. We studied the relationship between λ and the sequence length by utilizing RNA folding, which is a well-studied prototype of genotype-phenotype map, where the genotype is the primary structure of an RNA sequence and the phenotype is the minimum free energy secondary structure of the RNA sequence. We utilized Vienna RNA package [7] to fold RNA (The default parameters are used in the all occasions in the study). The average λ for different sequence length was obtained through comparing the secondary structure of randomly created RNA sequences with that of all possible mutants with only one substitution. A substitution which retains the original secondary structure is considered as neutral; otherwise deleterious. As Fig. 2 (the filled circles) shows, the average λ is a decreasing function of the sequence length. This relationship further limits an increase of information threshold due to mutational neutrality.
Figure 2 Information threshold The solid lines represent the maximum maintainable sequence length (Nmax) plotted against λ, where Nmax is calculated by using Eq. 4 where σ = 2, 10 and 100 (as indicated within the figure) and q = 0.95. The value of q was chosen to be plausible for ribozyme polymerization [14]. These solid lines show the dependence of Nmax on λ; on the other hand, the dependence of λ on N (the length of sequence) is examined by calculating the average λ in RNA folding for various values of N. The filled circles represent so obtained λ values as a function of N. In obtaining λ, a neutral substitution is defined as a single base substitution which does not alter the secondary structure of a focal sequence. For each sequence length, a hundred randomly generated RNA sequences are examined. In RNA folding program [7], the default parameters are used (in the all occasions in the study).
Comparison between the analytical prediction and computer simulations
We compare our analytical prediction with computer simulations. Our computer program simulates the evolution of RNA replicators in a well-mixed flow reactor (e.g. see [8]). In the simulations, each RNA sequence replicates and/or is diluted (be taken out from the reactor) with a certain probability in every time step. RNA folding is utilized again as a genotype-phenotype map (computed by [7]): The fitness of an RNA sequence depends on the secondary structure (i.e., the phenotype) of the RNA sequence. The fittest phenotype is set to the secondary structure of a yeast tRNAphe (the clover leaf structure, N = 76). RNA sequences which have the fittest phenotype replicate with the probability 0.01 per time step; all the other RNA sequences (mutants) replicate with the probability 0.001 per time step (thus σ = 10). The replication introduces mutations with a certain probability. Back mutations are not allowed to occur – the effect of back mutations is negligible if the sequence length is large enough [9] (this was confirmed by the simulations which were the same as the above except for allowing back mutations [data not shown]). The dilution probability (Φ) is calculated as the average probability of replication divided by the target population size. The target population size is set to 10000. All the simulations start with 10000 yeast tRNAphe sequences. The degradation of sequences is ignored (D = 0).
The results of the computer simulations showed that the "representative λ" (defined in Methods section – Non-uniform distribution of λ) of the fittest sequences increased from 0.307 to ca. 0.40 for the examined values of the error rate (data not shown). (This is also true for the population average of λ [op. cit.].) The value of the representative λ fluctuates over the time (st. dev. = 0.01 at 1 - q = 0.0475).
The equilibrium fraction of the fittest sequences of the computer simulations is compared to that of the analytical prediction over the different error rate in Fig. 3. The analytical prediction is calculated under the additive assumption from Eq. 1 and Eq. 2 by using the time averaged representative λ observed in the simulations after the evolution (λ = 0.4). As Fig. 3 shows, the calculation (the [red] solid line without error bars) closely predicts the result obtained from the computer simulations (the [black] solid line with error bars). The predicted error threshold (0.05) is slightly higher than that observed (between 0.045 and 0.048) probably due to the assumption of infinite population in Eq. 1 (see [3,10]).
Figure 3 Comparison of population structure between analytical predictions and computer simulations The fraction of the fittest phenotype population is plotted against the error rate per base (1 - q). The black line (the solid line with error bars) is obtained from computer simulations. The red line (the solid line without error bars) is calculated with the additive assumption from Eq. 1 and Eq. 2 with σ = 10 and N = 76 as in the simulations. λ is set to be 0.4 which is the time averaged representative λ value observed in the simulations after the population evolves (see the Methods section – Non-uniform distribution of λ – for the definition of the representative λ). The green line (the dotted line) is obtained from the formulation of Reidys et al. [5] with the same parameters as the above. The blue line (the dashed line) is calculated with the four λ approximation by using λ values reported in [5] (λ = 0.2489 on average) and σ = 10 and N = 76 as in our simulations.
Comparison with a previous formulation
Reidys et al. [5] derived the phenotypic error threshold as
from Qe = Q + λ (1 - Q). This equation shows an unlimited increase in the error threshold for λ ≥ σ-1 (see the dashed line in Fig. 1 and the [green] dotted line in Fig. 3). However, Qe = Q + λ (1 - Q) is valid only if either (1) a neutral set is uniformly distributed over the genotype space [a neutral set is a set of genotypes where all genotypes map to the same phenotype], or (2) q is so large that most mutants have d = 1. The uniform distribution of neutral sets in the genotype space is not applicable in RNA folding as shown later. The latter possibility is discussed next.
Studies of replicator dynamics on a neutral network often consider a very large value of q so that most mutants have d = 1 (e.g., [11]). [A neutral network is a neutral set, or its subset, where every genotype is connected to at least one genotype of the set by one or two base substitutions.] However, if the error rate (1 - q) is close to the error threshold, mutants can have on average d > 1 even if λ = 0, for which the error threshold is at the lowest error rate (see the dashed line in Fig. 4). The average d of the neutral sequences (i.e., the exact copies and the neutral mutants) per replication (this will be later called the average d per neutral replication) is lower than the average d per replication. However at the error threshold, even the average d per neutral replication is larger than one for λ > 0.32 (see Fig. 4, the solid lines). Above consideration asserts that the error threshold will be substantially overestimated if one considers only a single mutation.
Figure 4 Number of substitutions per replication in mutants The y-axis is the number of base substitutions (d) per replication (or per neutral replication) at the error threshold. The thick solid line represents the average d per neutral replication (i.e., the average d of the sequences which retain the master phenotype per replication): d = Np/(qmim + p) where p = λ (1 - qmin) and λ = (σ-1/N - qmin)/(1 - qmin). The thin solid line represents the standard deviation of it, i.e., ± (qmim + p) Np . The dashed line represents the average d per replication, which is N(1 - qmin). N is 100 and σ is 10. The lines are plotted against qmin (the lower x-axis), and the corresponding λ is shown in the upper x-axis.
Reidys et al. [5] obtained an extension of Eq. 5, the so called "four λ approximation". This extension divides a sequence in four sub-sequences in order to take into account the fact that the fraction of neutral substitutions varies over the sequence position. This extension still overestimates Qe though less so than Eq. 5 because the approximation now permits four substitutions per replication as a side effect of the subdivision. Note that this extension makes a fairly good prediction on the error threshold (see the [blue] dashed line Fig. 3) because the use of a small non-evolved λ value coincidentally cancels out the overestimation.
In conclusion, it is crucial for the calculation of the error threshold to consider that the number of substitutions per replication is large near the error threshold.
Epistasis in RNA folding
The rather impressive success of the additive assumption is counter-intuitive in view of RNA folding, in which many interactions occur between bases. In the next part of the paper, we study a particular RNA sequence, namely the yeast tRNAphe (which comprises the initial population of the previously described RNA evolution simulations), in terms of additivity and epistasis. The objective of this study is to understand how the additive assumption achieves a good prediction in spite of a high degree of nonlinearity in RNA folding [12,13].
We compare the secondary structure of randomly sampled mutants to that of the tRNAphe. Similar to the previous section, a mutant is neutral if its secondary structure is the same as that of the tRNAphe; otherwise, it is deleterious. To evaluate the deviation from the additive assumption, we categorize mutants into four classes as shown in Table 1. Negative epistasis refers to a mutant which is predicted to be neutral under the additive assumption, but turns out to be deleterious due to the interaction of the base substitutions. Positive epistasis refers to the reverse case.
Table 1 Definition of positive and negative epistasis.
mutants are
neutral deleterious
δ = 0 additive neutral negative epistasis
δ > 0 positive epistasis additive deleterious
δ is the number of deleterious base substitutions.
As Fig. 5 shows, the additive assumption underestimates the degree of mutational neutrality. The same conclusion was drawn differently in [13], where the additive neutral mutant is defined as a neutral mutant which lies in the same neutral network as that of the original sequence. Our results show that positive epistasis occurs more frequently than negative epistasis in total. What actually happens is as follows. If mutants with δ > 0 are only considered, positive epistasis occurs very rarely compared to additive deleterious case: No more than 0.5% of the mutants are neutral at d = 5 if they carry at least one deleterious substitution. If mutants with δ = 0 are only considered, negative epistasis is rather frequent relative to additive neutral case: As much as 35% of the mutants are deleterious at d = 5 even if they carry only neutral substitutions. However, replication with δ > 0 occurs far more frequently than replication with δ = 0: As much as 99.7% of the replication contains at least one deleterious substitution at d = 5 and λ = 0.307. Therefore, the relative frequency of epistasis is flipped around. Consequently, the additive assumption underestimates the degree of mutational neutrality. (Note that in Fig. 3 the additive assumption predicts the fraction of the fittest sequences always slightly smaller than that of the computer simulations.)
Figure 5 Additivity and epistasis in RNA folding The frequency of mutant classes is plotted against the number of base substitutions (d). (a) Log. plot. The patterns in the bars indicate the mutant classes: (from bottom) mesh, additive neutral; dots, positive epistasis; black, negative epistasis; stripes, additive deleterious (see Table 1 for the definition). The data were generated by RNA folding (by using [7]) with a S. cerevisiae tRNAphe sequence as a reference sequence: GCGGAUUUACCUCAGUUGGGAGAGGGCCAGACUGAACAUCUGGAGGUCCGGCGCGCGAUACGCCGAAUUCGCACCA (each non-RNA is converted to RNA). We examined all possible mutants at d = 1, 2 and the subsets of mutants for other d values (2<d<10, the portion of examined mutants is respectively, 10, 1, 0.1, 0.01, 0.3 × 10-3, 0.1 × 10-3, 0.4 × 10-5%). These observations in RNA folding are compared with the following two analytical predictions. The solid line is the probability of neutral replication estimated under the additive assumption (λd, λ = 0.307). The dashed line is the probability of neutral replication estimated with epistasis ((d), see Methods section – Probabilistic approach). (b) Linear plot. Symbols: ● the frequency of the neutral mutants (additive neutral and positive epistasis); ○ the frequency of the deleterious mutants (additive deleterious and negative epistasis).
The effect of epistasis is already noticeable when d > 2 as seen in the comparison between the probability of neutral replication under the additive assumption (λd) and that observed in RNA folding (see Fig. 6a). Since the average d per replication is more than 3 close to the error threshold in our simulations, Fig. 6a may seem to suggest that the additive assumption would substantially underestimate the effective replication accuracy (Qe) near the error threshold.
Figure 6 Comparison between additivity and epistasis in RNA folding (a) The relative probability of neutral replication under the additive assumption (λd) is plotted against the number of base substitutions (d), where the probability of neutral replication with epistasis (i.e. the fraction of neutral mutants observed in the yeast tRNAphe folding) is set to be 1 for each d as a reference. It can be seen that the effect of epistasis on the probability of neutral replication becomes larger as the number of substitution (d) increases. The same data as that of Fig. 5 are used. (b) The solid line (the left y-axis) represents the relative effective replication accuracy (Qe) under the additive assumption plotted against the error rate (1 - q), where Qe calculated with epistasis is set to be 1 as reference (see Methods section – Probabilistic approach – for details). It can be seen that the effect of epistasis on Qe increases as the error rate (1 - q) increases. The shape of the curve is in a similar manner as that of the curve in Fig 6a. Although the x-axis of Fig 6b is different from that of Fig 6a, one can relate the two graphs via the average d per neutral replication, with which the different x-axes can be transformed to each other. The average d per neutral replication is represented by the dashed line (the right y-axis plotted against 1 - q). Its value is calculated under the additive assumption as d = Np/(q + p) where p = λ (1 - q), λ = 0.4 and N = 76.
We calculate the effective replication accuracy (Qe) including the effect of epistasis in order to compare it with Qe calculated under the additive assumption. The first trial was to include a "trivial" epistasis in base paired regions (helices) as a part of the additive effect (see Methods section – Trivial epistasis). However, the analysis showed that epistasis occurs mainly in a "non-trivial" way (data not shown), and thus it is not sufficient for our sake to include a trivial epistasis. We next took a probabilistic approach to calculate Qe with epistasis (see Methods section – Probabilistic approach). The results of this method agree with the observation (see the dashed line in Fig. 5).
We compare Qe calculated under the additive assumption to that calculated with epistasis as shown in Fig. 6b (the solid line). As the comparison shows, the additive assumption indeed underestimates Qe; however, the underestimation becomes prominent only if the error rate is higher than the error threshold (1 - qmin = 0.05). As Fig. 6b (the dashed line) shows, the average d per neutral replication is ca. 1.5 at the error threshold, which is much smaller than the average d per replication (ca. 3.8). This means that the main contribution to Qe under the additive assumption is from the mutants of d = 1 or 2 at the error threshold. According to Fig. 6a the additive assumption is a good approximation at d = 1 or 2. Therefore, the additive assumption accurately estimates Qe and thus the error threshold. When the average d per neutral replication reaches 3, the additive assumption substantially underestimates Qe (see Fig. 6b), which is consistent with Fig. 6a. [However, note that this analysis does not imply that the average number of substitutions per replication is safely assumed to be near one. On the contrary, in our calculation it was more than one – actually 3.8 – at the error threshold.]
In the above examination of the additive assumption, there are two points which must be examined further: (1) The analysis of epistasis was performed on a yeast tRNAphe, which comprises the initial population of the RNA evolution simulations, but the results may differ if the analysis is done for a sequence which appears later in the RNA evolution simulations. Thus, we performed the same analysis to a sequence which was chosen from the population of the fittest sequences after the evolution in the simulations (at the 20000th time step). The results, however, did not change our conclusion (data not shown). (2) If the length of sequences is larger, the average d per neutral replication may increase, and thus the additive assumption may break down before the error threshold. However, it turns out from the analytical calculation that the average d per neutral replication at the error threshold decreases as N increases when λ is invariant (cf. the caption of Fig. 6b). Furthermore, λ decrease as N increases (see the filled circles in Fig. 2). Therefore, if the sequence length is larger, the average d per neutral replication will be actually smaller. We also conducted computer simulations of RNA evolution with a longer sequence length (200 bases). The results showed that the average d per neutral replication (calculated under the additive assumption) at the error threshold was indeed smaller (ca. 1.2 substitutions with λ ≈ 0.35) than in the previous case of the shorter sequence length (ca. 1.5 substitutions with λ ≈ 0.4), and the additive assumption still predicts the results of the simulations closely (data not shown).
Conclusions
• The phenotypic error threshold was formulated under the additive assumption. The formulation asserted that mutational neutrality increases the error threshold but the increase is limited.
• The importance of considering multiple substitutions per replication at the error threshold was illustrated.
• The comparison with the computer simulations and the analysis of epistasis showed that the additive assumption correctly estimates the effective replication accuracy (Qe) and thus the error threshold.
• The reason why the additive assumption achieves a good prediction of the error threshold in spite of a high degree of (non-trivial) epistasis in RNA folding is that the average number of substitutions per neutral replication is small enough to avoid of the effect of epistasis.
Methods
Non-uniform distribution of λ
If λ is not uniform over the genotypes sharing the same phenotype, the effective replication accuracy (Qe) depends on the distribution of the genotypes in the population. In this case, Qe is calculated under the additive assumption as
where XI denotes the population of the focal phenotype (I), and xi (resp. λi) is the population (resp. the fraction of neutral substitutions) of the genotype i. The set SI denotes the set of genotypes which have the phenotype I. If xi and λi are known, the representative λ of the phenotype can be calculated from the following equation as
The difference between the representative λ and the population average of λ was very small in the computer simulations. (The population average was always slightly smaller [ca. 99%] than the representative λ unless the distribution of λ in the fittest population is completely homogeneous [data not shown].)
Calculation of Qe with epistasis
Trivial epistasis in RNA folding
It is trivial that epistasis occurs between bases which make a pair (hydrogen bond) in the reference secondary structure. Our first trial to include epistasis in the calculation of Qe was to include this epistasis as a part of the additive effects of mutations as described in [5]. In this approach, the reference sequence is subdivided into non-paired regions and paired regions; paired regions are treated as strings of base pairs (one pair of bases is considered as one character); a substitution of a base pair is considered as an elementary step of mutations in paired regions. Following this procedure, the epistasis occurring between bases in a pair is now treated as an additive effect. [For example, two mutations – GC→GG and GC→CC – occurring in paired base must be deleterious because the bases can not make a pair any more. Given that the combined mutation – GC→CG – is neutral, it will be a case of positive epistasis in the previous procedure. However, in the new procedure it will be a case of additive neutral because the combined mutation is treated as one substitution of a base pair.] We categorized the mutants into the previously defined four groups of mutants (i.e. additive neutral, additive deleterious, positive epistasis and negative epistasis) using the same data as that of Fig. 5. However, the result did not differ much from that shown in Fig. 5 (data not shown). We conclude that epistasis occurs mainly in a non-trivial way, and thus this approach is not effective for our purpose.
Probabilistic approach
Since (non-trivial) epistasis makes it difficult to predict what happens to the phenotype given a specific change in genotype, we take the following probabilistic approach: We assume that a mutant is neutral with a certain probability (denoted by μ(v, δ)), which depends on the number of neutral base substitutions (denoted by v) and on that of deleterious base substitutions (denoted by δ). Then, the probability of neutral replication is obtained (by using the binomial approximation) as
where d = v + δ. Qe is thereupon derived as
We measured μ(v, δ) in the tRNAphe folding as shown in Fig. 7ab. When δ = 0 and v > 0, μ declines a little slower than exponentially as v increases due to negative epistasis (Fig. 7a). When v = 0 and δ > 1, μ is not zero due to positive epistasis, and μ decreases slower than exponentially as δ increases (Fig. 7b). When v > 0 and δ > 0, μ(v, δ) increases, saturates, and finally decreases as v increases (Fig. 7a): neutral substitutions can compensate deleterious substitutions. We express the above observations as follows:
Figure 7 Probabilistic approach in calculating the effective replication accuracy with epistasis (a) The probability that a mutant is neutral with v neutral substitutions and δ deleterious substitutions (i.e., μ(v, δ)) is plotted against the number of neutral substitutions (v). Symbols: ● δ = 0; □ δ = 1; ◇ δ = 2; ○ δ = 3; * δ = 4. The plots were obtained from the same data set as that of Fig. 5. The solid lines represent the results of curve fitting. We used Eq. 10 (v>0, δ = 0) to the δ = 0 data set, and Eq. 10 (v>0, δ>0) to the δ = 1 data set. The second fitting was done after we obtained α and εn from the first fitting, and εd and η from the fitting in Fig. 7b. The dotted lines are the estimation made with the obtained parameters (listed below). (b) μ(v, δ) plotted against the number of deleterious substitutions (δ). Symbols: ● v = 0; □ v = 1; ◇ v = 2; ○ v = 3; * v = 4. The solid part of the line represents the curve fitting; the dotted part is an exception, i.e., μ(0, 1) = 0. We used Eq. 10 (v = 0, δ>0) toward the v = 0 data set in the fitting. All the fitting was done after transforming both the equations and the data sets to logarithmic scale to reduce the biased importance of the points in small d. The obtained parameters are as follows: εn = 0.1190, α = 0.8483, εnd = 2.418, β = 2.333, γ = 3.996, εd = 0.02697 and η = 0.6380.
where εn, εd and εnd are the epistatic parameters of the interactions among neutral substitutions, among deleterious substitutions, and between neutral and deleterious substitutions, respectively. Note that in the additive assumption, all epistatic parameters are zero. α and η represent non-exponential decay. To express the compensation by neutral substitutions, we arbitrarily used a saturation function βv/(γ + v) where β and γ are parameters. To obtain the parameters, we fitted Eq. 10 to the data in Fig. 7ab (the solid lines) as explained in the caption. As shown in Fig. 7a (the dotted lines), the theoretical estimation turns out to be a slight underestimation. (d) was calculated from the above obtained parameters, and the calculated values match the observed ones (see the dashed line in Fig. 5).
Authors' contributions
NT contributed to the entire part of the study. PHP contributed to the computer programing of the RNA evolution simulation. PH contributed to the conceptual development of the study and the manuscript preparation as the supervisor.
==== Refs
Eigen M Selforganization of Matter and the Evolution of Biological Macromolecules Naturwissenschaften 1971 58 465 523 4942363 10.1007/BF00623322
Maynard Smith J Szathmary E Chaper 4 The Major Transitions in Evolution 1997 reprint New York: Oxford University Press 41 58
van Nimwegen E Crutchfield JP Metastable Evolutionary Dynamics: Crossing Fitness Barriers or Escaping via Neutral Paths? Bull Math Biol 2000 62 799 848 11016086 10.1006/bulm.2000.0180
Huynen MA Stadler PF Fontana W Smoothness within ruggedness: the role of neutrality in adaptation Proc Natl Acad Sci USA 1996 93 397 401 8552647 10.1073/pnas.93.1.397
Reidys C Forst CV Schuster P Replication and mutation on neutral networks Bull Math Biol 2001 63 57 94 11146884 10.1006/bulm.2000.0206
Wilke CO Selection for fitness versus selection for robustness in RNA secondary structure folding Evolution 2001 55 2412 2420 11831657
Hofacker IL Fontana W Stadler PF Bonhoeffer LS Tacker M Schuster P Fast Folding and Comparison of RNA Secondary Structures Monatsh Chem 1994 125 167 188 10.1007/BF00818163
Fontana W Schuster P A computer model of evolutionary optimization Biophys Chem 1987 26 123 47 3607225 10.1016/0301-4622(87)80017-0
Eigen M McCaskill J Schuster P The molecular quasi-species Adv Chem Phys 1989 75 149 263
Nowak M Schuster P Error Thresholds of Replication in Finite Populations Mutation Frequencies and the Onset of Muller's Ratchet J Theor Biol 1989 137 375 395 2626057
van Nimwegen E Crutchfield JP Huynen M Neutral evolution of mutational robustness Proc Natl Acad Sci USA 1999 96 9716 9720 10449760 10.1073/pnas.96.17.9716
Huynen MA Konings DA Hogeweg P Multiple coding and the evolutionary properties of RNA secondary structure J Theor Biol 1993 165 251 267 7507189 10.1006/jtbi.1993.1188
Wilke CO Lenski RE Adami C Compensatory mutations cause excess of antagonistic epistasis in RNA secondary structure folding BMC Evol Biol 2003 3 3 12590655 10.1186/1471-2148-3-3
Johnston WK Unrau PJ Lawrence MS Glasner ME Bartel DP RNA-Catalyzed RNA Polymerization: Accurate and General RNA-Templated Primer Extension Science 2001 292 1319 1325 11358999 10.1126/science.1060786
| 15691379 | PMC550645 | CC BY | 2021-01-04 16:37:17 | no | BMC Evol Biol. 2005 Feb 3; 5:9 | utf-8 | BMC Evol Biol | 2,005 | 10.1186/1471-2148-5-9 | oa_comm |
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-181571793110.1186/1471-2164-6-18Research ArticleIdentification of functional SNPs in the 5-prime flanking sequences of human genes Mottagui-Tabar Salim [email protected] Mohammad A [email protected] Yosuke [email protected]öm Pär G [email protected] Boris [email protected] Wyeth W [email protected] Claes [email protected] Center for Genomics and Bioinformatics, Karolinska Institutet, SE-17177 Stockholm, Sweden2 Centre for Molecular Medicine and Therapeutics, University of British Columbia, Vancouver, BC V5Z 4H4, Canada2005 17 2 2005 6 18 18 2 11 2004 17 2 2005 Copyright © 2005 Mottagui-Tabar et al; licensee BioMed Central Ltd.2005Mottagui-Tabar 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
Over 4 million single nucleotide polymorphisms (SNPs) are currently reported to exist within the human genome. Only a small fraction of these SNPs alter gene function or expression, and therefore might be associated with a cell phenotype. These functional SNPs are consequently important in understanding human health. Information related to functional SNPs in candidate disease genes is critical for cost effective genetic association studies, which attempt to understand the genetics of complex diseases like diabetes, Alzheimer's, etc. Robust methods for the identification of functional SNPs are therefore crucial. We report one such experimental approach.
Results
Sequence conserved between mouse and human genomes, within 5 kilobases of the 5-prime end of 176 GPCR genes, were screened for SNPs. Sequences flanking these SNPs were scored for transcription factor binding sites. Allelic pairs resulting in a significant score difference were predicted to influence the binding of transcription factors (TFs). Ten such SNPs were selected for mobility shift assays (EMSA), resulting in 7 of them exhibiting a reproducible shift. The full-length promoter regions with 4 of the 7 SNPs were cloned in a Luciferase based plasmid reporter system. Two out of the 4 SNPs exhibited differential promoter activity in several human cell lines.
Conclusions
We propose a method for effective selection of functional, regulatory SNPs that are located in evolutionary conserved 5-prime flanking regions (5'-FR) regions of human genes and influence the activity of the transcriptional regulatory region. Some SNPs behave differently in different cell types.
==== Body
Background
Single nucleotide polymorphisms (SNPs) are the most common form of genomic variations occurring on average every 1000 nucleotides. The vast majority of SNPs are neutral allelic variants, however the few that do influence a phenotype in a measurable way, are important for understanding the underlying genetics of human health. SNPs are the focus of a large number of human genetics studies attempting to understand their impact on complex diseases like Alzheimers, Parkinsons, diabetes, etc. Most SNPs, by the virtue of their location within genes (introns, 3'-UTRs, etc) or between genes, are considered most likely to be benign and not to contribute to a phenotype, whether it may be the manifestation of a disease or quicker metabolism of a drug. Among the group of SNPs located within coding regions of genes and causing a change in the peptide sequence (non-synonymous SNPs or 'nsSNPs') or among SNPs located within promoters (regulatory SNPs or rSNPs), a majority may not influence the overall activity of the protein or the gene expression. With the per-SNP validation and genotyping cost relatively high, it is increasingly important to develop strategies to predict functionally relevant SNPs in silico. The SNP databases in public domain, like NCBI/dbSNP and HGVbase, have facilitated this by highlighting all nsSNPs and also further classifying the location of the amino acid within the encoded proteins [1] to more accurately predict the detrimental effects of a change in peptide sequence. Several recent studies have attempted to focus on the subset of nsSNPs that most likely influence phenotype [2-6]. Of the approximately 4.5 Million SNPs in dbSNP [7], an estimated 10,000 nsSNP exist and approximately 10–15% of those are projected to be damaging [6]. Comparatively fewer attempts have been made to predict and validate functional promoter SNPs [8].
Transcriptional regulatory regions in the 5'-FR of human genes encode short (often < 25 bp) [9,10] sequences which serve as targets for binding of transcription factors (TFs). Understanding the conditions of binding, specificity and identity of the factors would help us understand the mechanism of regulation of human genes. Eukaryotic TFs tolerate considerable sequence variation in their target sites and recent bioinformatics works [11-13] have developed methods to model the DNA binding specificity of individual TFs [10]. Such matrices, although highly accurate [9,14], are less specific in the identification of sites with in vivo function [11], mainly due to our limited understanding of additional factors involved in TF specificity such as factor cooperative binding, protein-protein interactions, chromatin superstructures and TF concentrations. Currently the most successful approach to overcome this information gap is based on the assumption that sequences conserved between species (here human and mouse) would most likely mediate biological function [15-19].
The 7TM (7 trans-membrane domain proteins), also known as the hetero-trimeric GTP-binding protein (G protein)-coupled receptors (GPCRs) are members of a large family with an estimated 700 genes in the human genome [20]. By some estimates, nearly 60% of drugs marketed today target directly or indirectly the GPCR family members [21]. Several studies have collectively analyzed the occurrence, and importance, of coding SNPs to pharmaceutical efforts, in this family of genes [22-24]. Characterizing polymorphisms that are located in the 5'-FR of these genes and that influence expression has been reported earlier. We therefore selected a subset of clinically and pharmacologically important GPCR genes and their 5'-FR sequences to test our bioinformatics and laboratory experimentation approach for prediction of functionally important SNPs in regulatory regions. Our selection system evaluates the influence of SNPs on TF-DNA complex stability, and further investigates the influence of such SNPs on promoter activity. We present a proof-of-concept for such a strategy and identify issues and problem-areas for future developments.
Results
The lists of the full names and Ensembl ENSG numbers [25] of the 176 GPCR genes are shown in additional files [See Additional file 1]. From a total of approximately 800 SNPs in proximal 5 kb regions, less than 200 were mapped to regions of mouse-human genome conservation. Of these approximately 200 SNPs, 36 were predicted to influence TF binding, in regions of sequence conservation of over 70% in human-mouse; the alignments for two such regions are indicated in additional files [See Additional file 2]. Table 1 lists the 21 genes, along with the SNPs, TFs and TFBS sequences and positions relative to transcription start site. These 36 candidate SNPs in Table 1 were qualified by our selection criteria, as described in the methods section, and were predicted to influence the binding of TFs in a qualitative manner. The absolute binding score of the TF differed by at least 2 units between the two alleles. Ten SNPs within the 5'-FR of 7 genes were selected for EMSA tests and are shown in bold letters in Table 1. The choice of these genes was based on our understanding of their significance in human physiology and relevance to research interests within the GPCR research community. Table 2 shows the results from the EMSA experiments, where values in each column for Allele 1 and Allele 2 are the ratios of measurements from each of the 5 different concentrations of the competitor oligomers (labeled × 5 through × 25 in Table 2 and in Figure 1) divided by the measurements without competitor (labeled 'C' in Figure 1). The decrease in level of the labeled product as a consequence of increasing non-labeled oligomer concentration is an indication of the efficiency of displacement, thereby reflecting the relative stability of the DNA-protein complex. A marginal increase in level of radio-labeled complex, instead of a decrease (Table 2 rs1800508) is generally considered to be due to additional factor involvements. Table 2, column 'Ratio' shows the difference, calculated for the highest concentration of the non-labeled competitor (25-fold), between the efficiency of competition between a perfect-match competitor and the allelic mismatch competitor. Values close to 1.00 indicate no difference in rate of competition between the two alleles, and therefore no relative difference in stability of the DNA-protein complex. While 4 SNPs exhibit mobility shift with a difference of 2-fold or more (rs267412, rs509813, rs945032 and rs2528521), three SNPs exhibit a moderate, nevertheless reproducible shift (rs1799722, rs2882225 and rs1538251). Finally, three SNPs fail to show any significant shift (rs968554, rs267413 and rs1800508). In all, seven out of ten polymorphic markers indicated reproducible binding differences between the alleles. Figure 1 shows pictures of EMSA gels for two of the SNPs, rs1799722 and rs2528521.
Figure 1 Polyacrylamide gels from electromobility shift assays. Polyacrylamide gels showing the decrease in amounts of protein complex with labelled oligomer as the concentration of the competing non-labelled oligomer increases (lanes marked x5 through x25). Lane marked ’C’ has no competitor and represents the basal levels of labelled complex. The measure of displacement of the labelled oligomer is expressed as a ratio of radio-labelled product, for each lane, divided by the basal value, presented in Table 2. Comparison of allele specific DNA-protein complex stability is a ratio of the highest competitor concentration (x25) of each of the two alleles. Thus for rs1799722 (A) allele A binds proteins 1.66 times better than allele G and for rs2528521 (B) allele A binds proteins 2.62 times better than allele G (Table 2, extreme right column).
Table 1 List of positive SNP candidates for EMSA studies. The genes and the SNPs which were tested in this report are indicated in bold text. The name of the TF and the predicted consensus site from the transcription start site are indicated.
Gene Name /Ensembl ENSG TF name (distance from start site) SNP rs ID Allele 1 Allele 2
Beta-2 adrenergic receptor ENSG00000164272 Nkx(-4257) rs2082382 ttcagtg ttcggtg
Gklf(-4216) rs2082395 aagtgagaag aagtgagaaa
c-ETS(-1049) rs1432622 gatcct gatctt
P2y purinoceptor 5 (p2y5) ENSG00000139679 SAP-1(-149) rs2233571 agcggaaat agtggaaat
Anion exchange protein 2 ENSG00000164889 Yin-Yang(-3480) rs2069453 gccatg gccgtg
c-ETS(-2988) rs2069451 tttccc tgtccc
SPI-1(-2988) rs2069451 gggaaa gggaca
SPI-B(-2990) rs2069451 atgggaa atgggac
c-MYB_1(-1536) rs2069442 gggagttg gggacttg
Nkx(-1537) rs2069442 tcaagtc tcaactc
SP1(-1440) rs2069441 agggctggga agagctggga
C-c chemokine receptor type 1 ENSG00000163823 SPI-B(-117) rs3181080 acaagaa actagaa
SOX17(-118) rs3181080 tttcttgtc tttctagtc
Frizzled 6 precursor (frizzled-6) ENSG00000164930 deltaEF1(-629) rs3758096 caccta aaccta
Proteinase activated receptor 3 ENSG00000164220 c-ETS(-32) rs2069647 catcct cctcct
deltaEF1(-31) rs2069647 ctcctt atcctt
Chemokine (C-X-C) receptor 6 ENSG00000163819 c-MYB_1(-469) rs2234352 tacagatg tatagatg
Thing1-E47(-469) rs2234352 catctgtaaa catctataaa
5-hydroxytryptamine 1a receptor ENSG00000178394 ARNT(-2174) rs968554 caagtg caactg
c-MYB_1(-2174) rs968554 tccagttg tccacttg
deltaEF1(-2174) rs968554 cacttg cagttg
n-MYC(-2174) rs968554 cacttg cagttg
USF(-2174) rs968554 caagtgg caactgg
USF(-2175) rs968554 cacttgg cagttgg
ARNT(-2174) rs968554 cacttg cagttg
deltaEF1(-2174) rs968554 caactg caagtg
n-MYC(-2174) rs968554 caagtg caactg
Muscarinic acetylcholine receptor M1 ENSG00000168539 MZF_1-4(-148) rs509813 tggggg tggcgg
MZF_5-13(-149) rs509813 gtggggggag gtggcgggag
MZF_1-4(-147) rs509813 gggggg ggcggg
SP1(-147) rs509813 ggggggagga ggcgggagga
Dopamine receptor D1a ENSG00000184845 FREAC-4(-4311) rs267412 gtaaaccc gtaagccc
TCF11MafG(-4446) rs267413 actgac acagac
Follicle stimulating hormone receptor ENSG00000170820 HFH-3(-81) rs2882225 ggatgctttttt ggatgctgtttt
HFH-2(-82) rs2882225 gatgcttttttt gatgctgttttt
c-ETS(-80) rs2349718 cttctt cttttt
Gklf(-87) rs2349718 aaaaaaaaag aaaaaaaaaa
SPI-1(-80) rs2349718 aagaag aaaaag
5-hydroxytryptamine 2c receptor ENSG00000147246 HFH-1(-1736) rs3795182 ccatgtttata ccatatttata
MEF2(-1734) rs3795182 atatttataa atgtttataa
FREAC-4(-1734) rs3795182 ataaacat ataaatat
c-ETS(-271) rs3813928 tatcct taccct
MZF_1-4(-273) rs3813928 tgagga tgaggg
SPI-B(-273) rs3813928 tgaggat tgagggt
Bradykinin receptor 2 ENSG00000168398 Ahr-ARNT(-535) rs945032 tgggtg tgggta
MZF_1-4(-80) rs1800508 tgggca tgagca
AP2alpha(-79) rs1800508 gcccaggag gctcaggag
TCF11-MafG(-61) rs1799722 aatgat agtgat
Adenosine A3 receptor ENSG00000121933 AP2alpha(-4276) rs1538251 gccctctgg tccctctgg
Alpha-1a adrenergic receptor ENSG00000120907 c-ETS(-4898) rs562843 cttctt cttatt
SPI-1(-4898) rs562843 aagaag aataag
Nkx(-4900) rs562843 ataagtt agaagtt
C-c chemokine receptor type 2 ENSG00000121807 TCF11MafG(-1823) rs3092964 catgcc catacc
Ahr-ARNT(-1825) rs3092964 tgcatg tgcata
TCF11MafG(-1823) rs3749462 catgcc catacc
Ahr-ARNT(-1825) rs3749462 tgcatg tgcata
Putative chemokine receptor ENSG00000119594 SPI-B(-86) rs3825163 tcaggaa ccaggaa
FREAC-4(-80) rs3825163 gtaaccat ataaccat
TCF11-MafG(-81) rs3825163 gataac ggtaac
Nkx(-161) rs2256572 ttatttg ctatttg
S8(-163) rs2256572 tatta tacta
TCF11MafG(-232) rs590447 gctgac gccgac
deltaEF1(-230) rs590447 cagctt cggctt
Calcitonin receptor ENSG00000004948 TCF11MafG(-511) rs2528521 agtgac agtggc
Lectomedin-3 ENSG00000150471 AP2alpha(-770) rs905963 gccccgagc accccgagc
SPI-1(-763) rs905963 gcgaac gcgagc
c-ETS(-365) rs1505666 cctcct ccttct
SPI-1(-366) rs1505666 gagaag gaggag
MZF_1-4(-367) rs1505666 agagga agagaa
Glucagon-like P2 ENSG00000065325 SPI-B(-881) rs1402655 tgagaaa tgataaa
G protein-coupled receptor ENSG00000102865 Yin-Yang(-118) rs2240047 gccatg gccctg
TCF11MafG(-118) rs2240047 catggc cagggc
Gfi(-1575) rs724615 aaaatcacag aaaatgacag
Table 2 Oligonucleotide sequences used for EMSA. For every SNP, 4 oligonucleotides (2 complimentary pairs) were synthesized, one pair for each allele. One oligonucleotide sequence from each pair had additional GG dinucleotide overhangs at the 5'end for fill-in labeling reaction. Care was taken to make sure that the additional GG-dinucleotide did not influence the predicted TF binding capability. The complementary sequences lacked the GG pairs. Only the allelic sequence predicted to bind most stably was chosen for the fill-in labeling reaction (marked *) while the gel shift assays were carried out using competitor with a perfect match versus a competitor with the allelic mismatch. The polymorphic site is underlined. Column 'Ratio' shows the difference in competition between the labeled and non-labeled oligomers at 25-fold excess, by dividing Allele2 (x25) values by Allele1 (x25).
Gene name and rs ID Sequence Allele 1(competitor oligo is a perfect match) Allele 2 (competitor oligo has a mismatch) Ratio
x5 x10 x15 x20 x25 x5 x10 x15 x20 x25
Serotonin receptor (5-HT-1A) rs968554 ENST00000323865 GGAAAAGAATCCA CTTGGGCCAATG *
GGAAAAGAATCCA
GTTGGGCCAATG 1.29 - 1.27 1.23 1.00 1.16 - 1.12 1.13 0.96 0.96
Dopamine receptor DRD1 rs267412 ENST00000329144 GGAATGTAAACCC
AACACAAAAG *
GGAATGTAAGCCC
AACACAAAAG 0.72 - - 0.54 0.56 0.98 - 1.06 1.03 1.12 2.05
Dopamine receptor DRD1 rs267413 ENST00000329144 GGTATAAAAGTCA
GTGAATACAG *
GGTATAAAAGTCT
GTGAATACAG 0.96 - 0.81 0.74 0.81 0.97 - 0.98 0.97 1.01 1.24
Muscarinic acetylcholine receptor M1 rs509813 ENST00000306960 GGCTTGGGCTCCT
CCCCCCAGCCAAC *
GGCTTGGGCTCCT
CCCGCCAGCCAAC 0.21 0.11 0.08 0.05 0.09 0.70 0.60 0.46 0.46 0.37 4.11
Follicle stimulating hormone receptor. rs2882225 ENST00000304421 GGCAAGGGAGCTG
TTTTTTTT
GGCAAGGGAGCTT
TTTTTTTT * 1.10 1.00 0.86 0.71 0.75 1.96 1.73 1.67 1.36 1.35 1.80
Adenosine-A3 receptor. rs1538251 ENST00000241356 GGTGGCCACCAGA
GGGCAGCACG *
GGTGGCCACCAGA
GGGAAGCACG 1.08 1.11 0.95 0.89 0.76 1.49 1.41 1.44 1.51 1.40 1.84
Bradykinin receptor B2 rs1800508 ENST00000306005 GGGAAGTGCCCAG
GAGGC *
GGGAAGTGCTCAG
GAGGC 1.67 1.28 1.17 1.16 1.20 1.10 1.06 1.11 1.27 1.33 1.03
Bradykinin receptor B2 rs945032 ENST00000306005 GGTTCCTGGGTGC
GGG *
GGTTCCTGGGTAC
GGG 0.88 0.72 0.72 0.62 0.55 1.15 1.22 1.29 1.22 1.27 2.30
Bradykinin receptor B2 rs1799722 ENST00000306005 GGCTGGGTAGTGA
TGTCATCAGC
GGCTGGGTAATGA
TGTCATCAGC * 0.36 0.21 0.16 0.12 0.12 0.50 0.29 0.22 0.20 0.20 1.66
Calcitonin receptor precursor rs2528521 ENST00000316576, ENST00000248548 GGCTGTCCCCGGA
GTGGCGGCT
GGCTGTCCCCGGA
GTGACGGCT * 0.54 0.36 0.25 0.22 0.21 0.90 0.83 0.73 0.60 0.55 2.62
For EMSA-positive markers with no validation information at dbSNP, HGVbase or Celera Discovery Systems™ we did a validation analysis using RFLP (restriction fragment length polymorphism) on DNA samples from 25 healthy Nordic individuals. The three SNPs which failed to show positive gel-shift results (serotonin receptor 5HT-1A: rs968554; DRD1: rs267413; and BK-2: rs1800508) were not investigated any further. For dopamine receptor D1 (DRD1) polymorphism (rs267412) the genotype distribution was found to be TT = 30%, AA = 20% and AT = 50%, and for calcitonin receptor promoter (CT-R) polymorphism rs2528521, it was GG = 40%, AA = 30% and GA = 30%. While rs267412 was found to be in Hardy Weinberg Equilibrium (HWE), rs2528521 was not. A larger population sample should be genotyped to accurately measure HWE for both these loci. Bradykinin B2 (BK-2) promoter polymorphism rs1799722 is noted to be polymorphic (major allele C: 56%) at NCBI/dbSNP. The polymorphic nature of this locus (rs1799722) was also confirmed by crosschecking at Celera Discovery Systems™. Also for the second BK-2 SNP rs945032, the allele frequency information was found at dbSNP (major allele = 80%). NCBI's dbSNP provided no allele frequency information about rs1538251 adenosine-A3 receptor (ADORA3), and on sequencing 20 DNA samples this marker proved to not be polymorphic in Nordic sample population, and was eliminated from further analysis (data not shown). The genotype frequency of muscarinic acetylcholine receptor M1 (CHRM1) SNP rs509813 was documented at Celera Discovery Systems™. The contig-position of rs2882225 (follicle simulating hormone receptor; FSHR) was not in agreement between the three major public genome databases, i.e. NCBI, Ensembl and Santa Cruz Genome Assembly (UCSC). It was mapped within the transcript for FSHR by NCBI/dbSNP, and completely absent from Ensembl and UCSC [26]. Therefore of the seven SNPs exhibiting positive EMSA, five SNPs (in four genes) qualified for analysis of their influence on promoter activity.
For expression analysis in living cells, the published promoter regions, or putative regulatory 5'FR of up to 2 Kb, were cloned and basal levels of luciferase were monitored. Repeated attempts to clone the promoter region of DRD1 failed (data not shown). Furthermore the position of rs267413 is mapped at -4446 nucleotides with respect to the transcription start site, whereas the minimum length of the genomic fragment known to drive DRD1 expression is limited to 2571 nucleotides. Considering the distal position of the marker, we decided not to examine rs267413 further in this study. The promoter regions of BK-2, CHRM1 and CT-R were cloned successfully. A total of four dissimilar human cell lines (HeLa, Hep2G and SK-N-MC, HEK293) were used to monitor the influence of the four SNPs (rs945032, rs1799722, rs2528521 and rs509813) to investigate differences in expression that are possibly due to differences in TF expression in different cell types. BK-2 promoter SNPs rs945032 (genotype = GG) and rs1799722 (genotype = AA) showed approximately 40%-60% higher activity in HeLa cells as compared to their other homozygote alleles AA and GG, respectively (Figure 2). The BK-2 SNP rs945032 behaves in a reciprocal manner in two (HeLa, and HEK293) cell types. The BK-2 SNP rs1799722 allele 'C' increases expression only in HeLa while decreasing expression in the other three cell types, similar to CHRM1 marker rs509813. CT-R marker rs2528521 and CHRM1 marker rs509813 failed to show any influence on luciferase expression levels in HeLa and SK-N-MC cells. Finally, the CT-R SNP rs2528521, allele 'C', influences expression in a significant manner, but only in one cell line (Hep2G). Hence, different alleles behave differently in different cell environments.
Figure 2 Comparative promoter activity in different cell lines. Influence of four functional promoter SNPs on promoter activity is dependent on cell types. Measurements are an average of four independent experiments. A ‘T’ indicates an ‘AA’ and a ‘C’ indicates a ‘GG’ genotype.
Discussion
SNPs that are located within coding regions and result in a change in the peptide sequence may be classified as 'damaging' or 'altering' if predicted to be in structurally or functionally important sites of the three dimensional structure of the protein. It is less straightforward to predict the functional importance of SNPs within regulatory regions. TFs tolerate variation in their binding sites and all positions in a site do not contribute equally to the binding energy. Therefore, the quantitative effect a given rSNP has on gene expression depends on its position and the bases involved. Complex human diseases like Parkinson's, diabetes and obesity are polygenic diseases, where many predisposing genetic and environmental factors together, over a period of time, cause a disease state. Differences in expression of genes and cellular concentrations of proteins due to common polymorphisms in 5' regulatory regions could equally elucidate gene function as the examination of non-synonymous SNPs in coding regions. We decided to test the 5'-FR of a group of physiologically and clinically important genes, for SNPs within TFBS, which could potentially influence the kinetics of binding affinity.
A common strategy for modeling the binding preferences of a transcriptions factor is to construct a position weight matrix (PWM) from known binding sites. PWMs are probabilistic models that capture the nucleotide preference at each position of a TFBS as well as the differential contribution of positions to the overall binding energy. When a putative binding site sequence is assessed using a PWM, a score is obtained that theoretically should be proportional to the binding energy between the TF and that sequence [10]. It has been convincingly shown, and generally accepted, that by considering DNA sequence conservation between mouse and human, the over-predictive nature of TFBS modeling can be significantly remedied. Therefore we chose for this work, not to include 'negative controls', that is, SNPs from outside of mouse-human conserved regions. We do indeed think that larger studies in the future should perhaps incorporate certain number of such negative controls to validate the theoretical predictions. Thus, using PWMs it should be possible to overcome the difficulties with rSNP detection stated above. We reasoned that if the score difference between alleles is large, it should correspond to a difference in gene expression that is reproducible in living cells. Using the JASPAR database of PWMs [27] and a phylogenetic footprinting strategy previously shown to be successful [17], we developed a method to detect putative TFBS and identify rSNPs likely to affect TF binding significantly. By incorporating phylogenetic footprinting, the method reported in this study emphasizes SNPs present in genomic regions that are highly conserved between human and mouse, thereby increasing the probability of a downstream functional influence of variations within theses sequences.
Electromobility Shift Assays (EMSA) produce DNA-protein binding interactions in artificial conditions. Therefore in silico prediction methods based on other in vitro or in vivo selection technologies, like 'systematic evolution of ligands by exponential enrichment' (SELEX), may not agree with the experimental outcome of EMSA procedures. Since the construction of PWMs is often concluded from published records based on SELEX enrichment approaches, it is informative to experimentally validate the predicted binding site using methods like EMSA. Therefore, we validated a subset of our predictions with in vitro electro-mobility shift. We used a stringent selection criterion, that is, qualified only alleles demonstrating an absolute binding score difference of 2.0 or more. A stringent selection criterion would no doubt decrease excessive hits and false positives at the expense of certain loss of true positives. Our results showed that approximately 60%-70% (i.e. 7 out of the selected 10 SNPs) of predicted sites (Table 2) bind proteins from HeLa nuclear extracts.
We finally attempted to correlate the EMSA findings from the 10 SNPs with influences on actual promoter activity within living cells. Due to mapping discrepancies of one SNP and failure to clone one promoter, we tested only four out of the seven EMSA-positive SNPs identified thus far. Of the four SNPs tested in a promoter-less expression vector, two (rs945032 and rs1799722) indicated significant influence on promoter activity, while two showed convincing and reproducible, yet comparatively limited influence (rs2528521 and rs509813) on promoter activity. The influence of these polymorphisms (rs945032 and rs1799722) indicate that any given functional variation within a regulatory region might exert a measurable influence within the context of a cell type determined by the TF expression profile of the cells and perhaps competitive binding of the TF to overlapping multiple binding sites.
There are several factors in the current approach which indicate that there are far more rSNPs than currently detected using available technologies. The EMSA assays employ HeLa nuclear extract, thereby limiting our findings to the TF expression repertoire of only HeLa nuclei. The TFBS package used a limited collection of high quality PWMs, which unfortunately represent only a small part of the approximately 2,000 known human and mouse TFs. The theoretical thresholds set for selections of alleles which are predicted to differentially bind TF require further rigorous testing to ensure that the present selection is optimal.
Conclusions
From a total of approximately 200 SNPs in evolutionally conserved 5'-FR of 176 human GPCR genes, our prediction algorithm selected 36 SNPs with possible influence on TFBS. When ten of these 36 SNPs were tested for mobility shifts, seven exhibited a positive result, and four of these were further tested for influence on promoter activity using an in situ reporter system. Finally, two of the four showed significant and reproducible influences which were dependent on the cell environment. Thus starting from a large pool of potential regulatory SNPs, we successfully identified a small fraction that actually influenced promoter activity. We therefore propose a method for effective selection of functional, regulatory SNPs, in evolutionary conserved 5'-FR regions of human genes, as a means for identification of candidate SNPs for genetic association analysis studies.
Methods
Sequence alignment and TFBS detection
The GPCR genes were selected from Ensembl [25]. Human and mouse genome assemblies (versions hg12 and mm2, respectively) and mappings of GenBank and RefSeq cDNA sequences to the assemblies were retrieved from the UCSC Genome Browser Database [26]. In addition, cDNA sequences for the 176 7TM or GPCR genes (online supplement) were mapped to the human genome assembly and 50,821 mouse cDNA sequences from the RIKEN project [29] were mapped to the mouse genome assembly using the client/server version of BLAT [30] with default settings. For each of the 176 GPCR-encoding human cDNAs, we retrieved the genomic mapping with the highest number of matching bases. Orthologous mouse loci were identified by similarly retrieving mouse genomic mappings for mouse cDNAs defined as orthologs to the human cDNAs in GeneLynx [31]. To more reliably identify transcriptional start sites we searched for other cDNA mappings overlapping the retrieved mappings and indicating similar gene structures. For each gene, the cDNA mapping extending furthest 5' was then used for further analysis. We extracted human genomic sequences from -5000 to +100 relative to starts of human cDNA mappings and mouse genomic sequences from -30000 to +100 relative to starts of mouse cDNA mappings. Orthologous genomic sequences were aligned using BLASTZ [32] with default settings. Aligned regions preceding human cDNA mappings were searched for putative rSNPs as follows. SNP data for the human genomic regions was retrieved from dbSNP, build 114. For each SNP within an aligned region, two allelic versions of a 110-bp alignment slice centered around the SNP were searched for putative TFBS using the TFBS Perl modules [33] and all position-weight matrices in the JASPAR database describing vertebrate TFBS and having an information content of at least 7 [27]. Hits fulfilling the following 3 criteria were considered putative TFBS: (a) situated within regions of at least 70% sequence identity (conservation) over 50 base pairs; (b) situated at corresponding (aligned) positions in human and mouse sequences; (c) having a relative matrix score of at least 0.5 in both human and mouse sequences. Selected for further analysis were putative TFBS with a relative matrix score exceeding 0.8 in one of the human alleles and either undetected in the other allele or having a difference in absolute matrix score of at least 2 between the human alleles.
Electromobility Shift Assays (EMSA)
Table 2 lists the sequences of oligonucleotides used for EMSA tests. For setting up of EMSA experimental procedures, an earlier published positive shift assay was reproduced using a polymorphism in the gene MMP12 [34]. Method modifications were then applied as described below.
Double stranded oligonucleotides were synthesized with 5'-GG dinucleotide overhangs. The 3'-end of the complementary strand was labeled with [α32P] dCTP with fill-in reaction using Klenow flagment. The labeled oligonucleotide were passed through ProbeQuant G-50 Micro Columns (Pfizer-Pharmacia Inc) and the concentration was adjusted to 0.035 nM. A 0.8 μl volume portion was mixed with 1.6 μl HeLa Nuclear Extract (Promega™), 1.6 μl 5x Gel Shift Binding 5x Buffer (Promega™), and 3.2 μl water. After 10 min incubation at room temperature, 0.8 μl of non-labeled competitor DNA, either one allele or the other allele, was added in varying concentrations (5-, 10-, 15-, 20- and 25-folds greater than the radio-labeled oligonucleotide). After 20 min room temperature incubation, the entire 8 μl reaction was loaded on polyaclylamide gel (5% 22.5 mM Tris/22.5 mM boric acid/0.5 mM EDTA buffer in, BIO RAD™). Thereafter, electrophoresis was performed in TBE for 20 min at 200 V. Gels were placed on Whatmann 3 MM™ filters and to facilitate drying a BIO RAD™ gel-dryer was used for 30 minutes. The dried gel were exposed to intensifying screen and analyzed by Typhoon Image Analyzer 9400 (Pfizer-Pharmacia Inc™).
Sequencing and RFLP (Restriction fragment length polymorphism)
The frequency of Dopamine receptor D1 polymorphism rs267413 was determined by sequencing a 174 base-pair fragment of the promoter in 10 DNA samples from healthy Swedish individuals. The sequence of the forward primer used for PCR was 5'-GGGGTACCACTTGACCGTTCTGTTGCTTT-3' where a KpnI restriction site (GGTACC) and GG-dinucleotide was added to the 5' end. The sequence of the reverse primer was 5'-TCTTTTAAGCTCTACTGTGGGTGA-3'. Calcitonin receptor promoter polymorphism rs2528521 was analyzed by RFLP (fragment length was 334 bp, restriction enzyme Tsp45I). Forward primer sequence used for PCR was 5'-ACCCCCAAGGTGTCTCTTCT-3' and reverse primer: 5'- GAGGGACCCGAGTTAGACCT-3'. The primer sequences for Bradykinin promoter SNP rs1799722 were as follows: Forward primer 5'-CCAGGAGGCTGATGACGTCA-3'. The fourth base from 3'-end was changed from A to G from original genomic sequence to create a Tsp45I restriction site for RFLP analysis.
Reverse primer: 5'-TCAGTCGCTCCCTGGTACTG-3'. Fragment length amplified was 150 bp. PCR conditions for all RFLP and sequencing reactions were as follows: 94 C (4 min), followed by 42 cycles of 94 C (1 min), 61 C (30 sec) and 72 C (30 sec).
Luciferase expression system for promoter activity analysis
A promoter-less luciferase vector (Basic PGL3, Promega™) was used for cloning known promoter regions between restriction sites KpnI and BglII of the plasmid vector. Primers for CHRM1: Forward: 5'-GGGGTACCGCAGGACCCACATCTCTAGG-3' Reverse = 5'-GAAGATCTTCACCAGGGCACCCAAT-3'. Primers for BK-2: Forward = 5'-GGGGTACCATCTGAGACTCTGTTTCCC-3' reverse = 5'-GAAGATCTTTCAGTCGCTCCCTGGTACT-3'. Primers for CT-R: Forward = 5'-GGGGTACCCCTTGGAATCAACTTGCCT-3' reverse = 5'-TTCTCGAGCGTCCTTGGAATCAACTTGC-3'. Genomic DNA of 27 individuals of Nordic origin were amplified and sequenced to identify the genotype of sample DNA. Cloned DNA were sequenced by using primer set GLprimer2 : 5'-CTAGCAAATAGGCTGTCCC-3' and 5'-CTTTATGTTTTTGGCGTCTTCC-3'. HeLa cells were plated in 24 well plates one day before transfection in appropriate medium with serum without antibiotics. Basic PGL3 Plasmids containing the cloned promoter region (180 ng) were co-transfected with 20 ng of pRL-TK plasmid, using Lipofectamine 2000 (Invitrogen™). Luciferase activities were determined using a dual Luciferase Reporter Assay system (Promega™) according to the manufacturer's instructions.
Abbreviations
TF: Transcription Factor; TFBS: transcription factor binding site; SNP: Single Nucleotide Polymorphism; GPCR: G-protein coupled receptors; 5'-FR: 5' flanking regulatory region.
Authors' contributions
Original Concepts and supervision: CW, WWW, BL and SM-T; Bioinformatics: PGE, BL and SMT; Running costs: SMT and CW; Manus preparation: SMT, PGE, CW; Electromobility Shift Assays and RFLP assays: YM; Luciferase expression: MAF and YM.
Supplementary Material
Additional File 1
ENSG ids A list of 176 initial GPCRs considered for this study, along with the Ensembl ENSG Ids.
Click here for file
Additional File 2
Alignments Alignment information for sequence flanking rs945032 and rs1799722 in human and mouse.
Click here for file
Acknowledgments
This work was supported by Pfizer Inc. and by the Swedish National Research Foundation. We would like to express our thanks to Dr. Madis Metsis for his technical advice and unbiased analysis of the EMSA results and to Dr. Albin Sandelin for valuable discussion.
==== Refs
Brookes AJ Lehvaslaiho H Siegfried M Boehm JG Yuan YP Sarkar CM Bork P Ortigao F HGBASE: a database of SNPs and other variations in and around human genes Nucleic Acids Res 2000 28 356 360 10592273 10.1093/nar/28.1.356
Cargill M Altshuler D Ireland J Sklar P Ardlie K Patil N Shaw N Lane CR Lim EP Kalyanaraman N Nemesh J Ziaugra L Friedland L Rolfe A Warrington J Lipshutz R Daley GQ Lander ES Characterization of single-nucleotide polymorphisms in coding regions of human genes Nat Genet 1999 22 231 238 10391209 10.1038/10290
Chasman D Adams RM Predicting the Functional Consequences of Non-synonymous Single Nucleotide Polymorphisms: Structure-based Assessment of Amino Acid Variation, Journal of Molecular Biology 2001 307 683 706 11254390 10.1006/jmbi.2001.4510
Ramensky V Bork P Sunyaev S Human non-synonymous SNPs: server and survey Nucl Acids Res 2002 30 3894 12202775 10.1093/nar/gkf493
Sunyaev S Ramensky V Bork P Towards a structural basis of human non-synonymous single nucleotide polymorphisms Trends in Genetics 2000 16 198 200 10782110 10.1016/S0168-9525(00)01988-0
Sunyaev S Ramensky V Koch I Lathe III W Kondrashov AS Bork P Prediction of deleterious human alleles Hum Mol Genet 2001 10 591 11230178 10.1093/hmg/10.6.591
Sherry ST Ward MH Kholodov M Baker J Phan L Smigielski EM Sirotkin K dbSNP: the NCBI database of genetic variation Nucleic Acids Res 2001 29 308 311 11125122 10.1093/nar/29.1.308
Ponomarenko JV Merkulova TI Orlova GV Fokin ON Gorshkova EV Frolov AS Valuev VP Ponomarenko MP rSNP_Guide, a database system for analysis of transcription factor binding to DNA with variations: application to genome annotation Nucleic Acids Res 2003 31 118 121 12519962 10.1093/nar/gkg112
Pennacchio LA Rubin EM Genomic strategies to identify mammalian regulatory sequences Nat Rev Genet 2001 2 100 109 11253049 10.1038/35052548
Stormo GD DNA binding sites: representation and discovery Bioinformatics 2000 16 16 10812473 10.1093/bioinformatics/16.1.16
Fickett JW Quantitative discrimination of MEF2 sites Mol Cell Biol 1996 16 437 8524326
Fickett JW Predictive methods using nucleotide sequences Methods Biochem Anal 1998 39 231 245 9707933
Workman CT Stormo GD ANN-Spec: a method for discovering transcription factor binding sites with improved specificity Pac Symp Biocomput 2000 467 478 10902194
Tronche F Ringeisen F Blumenfeld M Yaniv M Pontoglio M Analysis of the Distribution of Binding Sites for a Tissue-specific Transcription Factor in the Vertebrate Genome, Journal of Molecular Biology 1997 266 231 245 9047360 10.1006/jmbi.1996.0760
Duret L Bucher P Searching for regulatory elements in human noncoding sequences Curr Opin Struct Biol 1997 7 399 406 9204283 10.1016/S0959-440X(97)80058-9
Krivan W Wasserman WW A Predictive Model for Regulatory Sequences Directing Liver-Specific Transcription Genome Res 2001 11 1559 11544200 10.1101/gr.180601
Lenhard B Sandelin A Mendoza L Engstrom P Jareborg N Wasserman WW Identification of conserved regulatory elements by comparative genome analysis J Biol 2003 2 13 12760745 10.1186/1475-4924-2-13
Loots GG Locksley RM Blankespoor CM Wang ZE Miller W Rubin EM Frazer KA Identification of a coordinate regulator of interleukins 4, 13, and 5 by cross-species sequence comparisons Science 2000 288 136 140 10753117 10.1126/science.288.5463.136
Shabalina SA Ogurtsov AY Kondrashov VA Kondrashov AS Selective constraint in intergenic regions of human and mouse genomes Trends Genet 2001 17 373 376 11418197 10.1016/S0168-9525(01)02344-7
Rubin GM Yandell MD Wortman JR Gabor Miklos GL Nelson CR Hariharan IK Fortini ME Li PW Apweiler R Fleischmann W Cherry JM Henikoff S Skupski MP Misra S Ashburner M Birney E Boguski MS Brody T Brokstein P Celniker SE Chervitz SA Coates D Cravchik A Gabrielian A Galle RF Gelbart WM George RA Goldstein LS Gong F Guan P Harris NL Hay BA Hoskins RA Li J Li Z Hynes RO Jones SJ Kuehl PM Lemaitre B Littleton JT Morrison DK Mungall C O'Farrell PH Pickeral OK Shue C Vosshall LB Zhang J Zhao Q Zheng XH Lewis S Comparative genomics of the eukaryotes Science 2000 287 2204 2215 10731134 10.1126/science.287.5461.2204
Muller G Towards 3D structures of G protein-coupled receptors: a multidisciplinary approach Curr Med Chem 2000 7 861 888 10911020
Rana BK Shiina T Insel PA Genetic variations and polymorphisms of G protein-coupled receptors: functional and therapeutic implications Annu Rev Pharmacol Toxicol 2001 41 593 624 11264470 10.1146/annurev.pharmtox.41.1.593
Sadee W Hoeg E Lucas J Wang D Genetic variations in human G protein-coupled receptors: implications for drug therapy AAPS PharmSci 2001 3 E22 11741273 10.1208/ps030322
Small KM Seman CA Castator A Brown KM Liggett SB False positive non-synonymous polymorphisms of G-protein coupled receptor genes FEBS Letters 2002 516 253 256 11959142 10.1016/S0014-5793(02)02564-4
Hubbard T Andrews D Caccamo M Cameron G Chen Y Clamp M Clarke L Coates G Cox T Cunningham F Curwen V Cutts T Down T Durbin R Fernandez-Suarez XM Gilbert J Hammond M Herrero J Hotz H Howe K Iyer V Jekosch K Kahari A Kasprzyk A Keefe D Keenan S Kokocinsci F London D Longden I McVicker G Melsopp C Meidl P Potter S Proctor G Rae M Rios D Schuster M Searle S Severin J Slater G Smedley D Smith J Spooner W Stabenau A Stalker J Storey R Trevanion S Ureta-Vidal A Vogel J White S Woodwark C Birney E Ensembl 2005 Nucleic Acids Res 2005 33 Database Issue D447 D453 15608235
Karolchik D Baertsch R Diekhans M Furey TS Hinrichs A Lu YT Roskin KM Schwartz M Sugnet CW Thomas DJ The UCSC Genome Browser Database Nucleic Acids Res 2003 31 51 54 12519945 10.1093/nar/gkg129
Sandelin A Alkema W Engstrom P Wasserman WW Lenhard B JASPAR: an open-access database for eukaryotic transcription factor binding profiles Nucleic Acids Res 2004 32 Database issue D91 D94 14681366 10.1093/nar/gkh012
Okazaki Y Furuno M Kasukawa T Adachi J Bono H Kondo S Nikaido I Osato N Saito R Suzuki H Analysis of the mouse transcriptome based on functional annotation of 60,770 full-length cDNAs Nature 2002 420 563 573 12466851 10.1038/nature01266
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
Lenhard B Hayes WS Wasserman WW GeneLynx: a gene-centric portal to the human genome Genome Res 2001 11 2151 2157 11731507 10.1101/gr.199801
Schwartz S Kent WJ Smit A Zhang Z Baertsch R Hardison RC Haussler D Miller W Human-mouse alignments with BLASTZ Genome Res 2003 13 103 107 12529312 10.1101/gr.809403
Lenhard B Wasserman WW TFBS: Computational framework for transcription factor binding site analysis Bioinformatics 2002 18 1135 1136 12176838 10.1093/bioinformatics/18.8.1135
Jormsjo S Whatling C Walter DH Zeiher AM Hamsten A Eriksson P Allele-specific regulation of matrix metalloproteinase-7 promoter activity is associated with coronary artery luminal dimensions among hypercholesterolemic patients Arterioscler Thromb Vasc Biol 2001 21 1834 1839 11701474
| 15717931 | PMC550646 | CC BY | 2021-01-04 16:39:32 | no | BMC Genomics. 2005 Feb 17; 6:18 | utf-8 | BMC Genomics | 2,005 | 10.1186/1471-2164-6-18 | oa_comm |
==== Front
BMC NephrolBMC Nephrology1471-2369BioMed Central London 1471-2369-6-21571591810.1186/1471-2369-6-2Research ArticleAntiglucocorticoid RU38486 reduces net protein catabolism in experimental acute renal failure Mondry Adrian [email protected] Bioinformatics Institute, 30 Biopolis Street, #07-01 Matrix Building, 138671 Singapore2005 17 2 2005 6 2 2 23 9 2004 17 2 2005 Copyright © 2005 Mondry; 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 acute renal failure, a pronounced net protein catabolism occurs that has long been associated with corticoid action. By competitively blocking the glucocorticoid receptor with the potent antiglucocorticoid RU 38486, the present study addressed the question to what extent does corticoid action specific to uremia cause the observed muscle degradation, and does inhibition of glucocorticoid action reduce the protein wasting?
Methods
RU 38486 was administered in a dose of 50 mg/kg/24 h for 48 h after operation to fasted bilaterally nephrectomized (BNX) male adult Wistar rats and sham operated (SHAM) controls. Protein turnover was evaluated by high performance liquid chromatography (HPLC) of amino acid efflux in sera from isolated perfused hindquarters of animals treated with RU 38486 versus untreated controls.
Results
Administration of RU 38486 reduces the total amino acid efflux (TAAE) by 18.6% in SHAM and 15.6% in BNX and efflux of the indicator of net protein turnover, phenylalanine (Phe) by 33.3% in SHAM and 13% in BNX animals as compared to the equally operated, but untreated animals. However, the significantly higher protein degradation observed in BNX (0.6 ± 0.2 nmol/min/g muscle) versus SHAM (0.2 ± 0.1 nmol/min/g muscle) rats, as demonstrated by the marker of myofribrillar proteolytic rate, 3-Methylhistidine (3 MH) remains unaffected by administration of RU 38486 (0.5 ± 0.1 v. 0.2 ± 0.1 nmol/min/g muscle in BNX v. SHAM).
Conclusion
RU 38486 does not act on changes of muscular protein turnover specific to uremia but reduces the effect of stress- stimulated elevated corticosterone secretion arising from surgery and fasting. A potentially beneficial effect against stress- induced catabolism in severe illness can be postulated that merits further study.
==== Body
Background
As part of the complex uremic metabolic syndrome, pronounced disturbances of carbohydrate and lipid metabolism are commonly observed, as are pathologic changes of amino acid and protein turnover [1]. An increased net protein degradation in uremia was seen as early as 1949 by Persike and Addis[2], and in the same year, Bondy et al. [3] showed that adrenal hormones are involved therein. These early findings were validated in the late eighties by Schäfer et al. [4,5] who postulated a leading role for glucocorticoids as cause of the observed changes. Schäfer et. al used the experimental approach of inhibiting activation of the glucocorticoid receptor by enteral application of the potent glucocorticoid antagonist RU 38486 in acutely uremic rats [6] and found a decrease both in the accumulation of 3-methylhistidine, an amino acid that is produced in actomyosine catabolism and is not further metabolized, and in the activity of myofibrillar protease. However, while an effect of RU38486 on liver gluconeogenesis and urea synthesis in uremia could be demonstrated [7,8], so far there is no proof of a direct action of RU38486 on muscle metabolism in uremia. To address this problem, the present study made use of the classical experimental design of the isolated perfused hindquarter of the rat [9], in which roughly 40% of the rat body's total muscle mass can be evaluated under closely defined in vitro conditions.
With this experimental design and by comparing sham-operated and bilaterally nephrectomized animals, the present study looked at the question to what extent does corticoid action specific to uremia cause the observed muscle degradation, and does inhibition of glucocorticoid action reduce the protein wasting?
Methods
Animal experimentation was carried out on male Wistar rats, aged 11–15 weeks, weighing 217–225 g, from the animal experimentation facilities of Heinrich Heine University, Düsseldorf. Permission to use animals for experimentation was given by Regierungspräsident Düsseldorf, file nr. 26.4203.1-217/87 according to German federal law. Surgery for nephrectomy, sham nephrectomy and preparation for perfusion was carried out under narcosis with hexobarbital (EVIPAN- Na: 15–20 mg/ 100 g BW).
Bilateral nephrectomy was performed using a dorsal access, ligation of renal vessel string, and excision of the kidney, leaving the adrenal glands in place. Sham operated animals underwent the same manipulations except for the ligation and excision. After surgery, animals were fasted for 48 h until perfusion. Nephrectomized animals had access to drinking water on the day of surgery for 8 hours, then were deprived of liquid to avoid lung edema. Sham animals had free access to drinking water throughout.
Animals were randomly assigned to one of four groups: bilaterally nephrectomized (BNX) and sham operated (SHAM) treated with RU38486 and untreated BNX and SHAM animals. For treatment, RU38486 was dissolved in phenylmethanol, then mixed with sesame oil to form a milky suspension which was injected into subcutaneously into the lateral abdomen in three subdoses within 24 h, adding up to a total dose of 5 mg/100 g/BW.
48 h after initial surgery, animals were narcotized and prepared for perfusion as previously described[10].
The hindquarter was linked to the recirculation system after full passage of 70 ml of pre- perfusion medium, as shown in illustration 1. The pre- perfusion medium was discarded and not used for the recirculation experiment.
The perfusion was carried out with a half- synthetic medium on the basis of Krebs- Ringer- bicarbonate buffer (KRB), pH 7.38 [11]. Oxygen carriers were calf erythrocytes prepared from fresh calf blood sampled two days before experimentation and maintained with 300 mg/l Ampicillin and 220 ml/l citric acid/ glucose stabilizer. Bovine albumine maintained the physiologically correct oncotic pressure. 10-6 mmol/l phentolamine were added to avoid vessel contractions.
During perfusion, the arterial pH and perfusate oxygenation were monitored using a pH- meter and a total oxygen content analyzer (LEX- O2- CON, Lexington Instr., Mass., USA). At the beginning and end of perfusion, plasma samples were frozen for amino acid analysis. Amino acid analysis by HPLC was carried out using 25 μl of deproteinized perfusate sample, mixed with o-phthaldialdehyde (OPA)/3-mercaptopropionic acid to form OPA- adducts that were separated on a reversed phase column and measured by fluometry. Quantification was done by comparison with a standard amino acid mix including 3-methylhistidine. Of the 20 proteinogenic amino acids, cysteine, proline, and asparic acid were not included in HPLC analysis.
Statistical analysis was done using the "Student" t- test for ungrouped, non- paired data with f = n1 + n2-2 and a significance level of p < 0.05.
Results
Loss of body weight (BW)
During the 48 h fasting period between operation and perfusion, animals had a pronounced loss of BW (table 1). In SHAM, it was 35.5 ± 5.3 g; administration of RU 38486 reduced this to 27.6 ± 5.9 g (p < 0.05). Nephrectomized animals demonstrated a much less pronounced loss of BW due to a significant increase in tissue hydratation (Table 2). RU 38486 reduces the weight loss in nephrectomized animals, too; however, this effect is much less pronounced and lacks statistical significance (BNX 15.7 ± 4.8 g, BNX + RU 12.2 ± 4.8 g).
Oxygen utilization and development of acidosis
Oxygen utilization in the perfused muscle tissues is roughly the same in all four groups (data not shown) and equal to in vivo data previously reported [12] from rats after 24 hour fasting. As expected, pH dropped significantly lower during perfusion in the nephrectomized groups (SHAM: 7.378 ± 0.033, BNX 7.321 ± 0.018, p = <0.01; SHAM + RU 7.4 ± 0.031, BNX + RU 7.312 ± 0.013, p < 0.001).
Parameters of amino acid and protein metabolism
Total amount and spectrum of amino acids released during reperfusion: During perfusion, amino acids are released in varying amounts as shown in ill. 2. Nephrectomized animals (BNX) showed a general increase in amino acid release. This, however, is significant only in a few individual amino acids. The total amino acid efflux increases by 10,4 % (p = 0.05) without qualitative change. Notable exception is glycine, which is released to a lesser amount in nephrectomized animals. Amino acid release after treatment with RU 38486: Nephrectomy equally increases the amino acid efflux in animals treated with RU 38486 by 14.4% (p < 0.05) without change in spectrum. The total efflux of amino acids, however, is significantly reduced in the comparison SHAM/ SHAM+RU (-18.6%, p < 0.05) and BNX/ BNX+RU (-15.6%, p < 0.001). Release of 3-methylhistidine: 3-methylhistidine, a derivate of histidin mainly from actin and myosin in sceletal muscle and intestinal mucosa [13], is not reutilized after proteolysis, but excreted via the kidney as 3-methylhistidine or N- acetyl- 3-methylhistidine. During reperfusion of sham- nephrectomized animals, 3-methylhistidine is released from the hindquarter to a small amount that is incresed by roughly 300% in the nephrectomized animals. Administration of RU38486 has no effect on 3-methylhistidine efflux. (table 3).
Discussion
Background
Acute renal failure is a catabolic state, and unfortunately the inherent acceleration of protein breakdown cannot be suppressed effectively by provision of exogenous nutritional substrates [14]. The situation is multicausal. Unspecific mechanisms induced by the process of acute disease, underlying illness and associated complications are just one side of the problem. On the other, one observes specific uremic effects, insulin resistance, hormonal derangements, metabolic acidosis, circulating proteases and other inflammatory mediators together with effects induced by the acute loss of renal function and the type and intensity of renal replacement therapy [1,15-17].
One factor that has for long been associated with the disturbances of protein metabolism is glucocorticoid action. The first observations date back to the nineteen-forties, when Persike and Addis [2] reported an increased urea- nitrogen production in experimental uremia, and Bondy and coworkers [3] demonstrated that adrenal hormones were involved in this dysregulation. Half a century later, it is still not fully understood to what extent steroid hormone action is responsible for the catabolic situation observed in renal insufficiency[18]. It has been shown that administration of high doses of glucocorticoids to adrenalectomized rats resulted in decreased protein synthesis, increased protein degradation, and a negative nitrogen balance [19]. In patients with chronic renal failure, a positive correlation between muscle proteolysis and the plasma cortisol level has been observed [20]. The in vivo influence of both glucocorticoids and metabolic acidosis on muscle proteolysis has been elucidated in whole- body leucine turnover studies in adrenalectomized rats [21]. These findings indicate that glucocorticoids play an important role in net protein degradation. Price formulated this so: "glucocorticoids are required but not directly responsible for the acidosis-induced increase in the mRNAs encoding proteins of this degradative pathway"[22].
Experimental approach
In order to evaluate the relative importance of glucocorticoid action on protein metabolism in acute renal failure, an experimental setting was chosen that allowed to study glucocorticoid action indirectly by selective blockade of the glucocorticoid receptor with the potent antiglucocorticoid RU38486, a substance that binds to the receptor without activating the further process of transcription [23]. Parenteral administration of a total of 50 mg/kg BW/ d of RU38486 allowed to avoid additional irritation of the animals by a gastric catheter in the postoperative phase. Relevant action of RU38486 has been observed in enteral substitution at a dose of 20 mg/kg BW/d [6]. The degree to which RU 38486 blocks the glucocorticoid receptor depends very much on the mode of application, and the target tissue. While a recent study[24] shows that 80% of glucocorticoid receptors are blocked in rat muscle following oral application of mifepristone of 50 mg/ kg BW, Kim et al.[25] demonstrated effective blocking of glucocorticoid receptors in rat brain following subcutaneous application of 80 mg/ kg over two days. Schaefer et al.[6], on whose experimental set- up the present study was modeled, had reported significant effects of an oral dose of 20 mg/ kg on muscle. In view of this, the choice for the experimental procedure seems justified as the present study uses a substantially higher dose.
The isolated perfusion procedure introduced by Ruderman[9] is well established for the representative study of muscle metabolism. In this setting, the perfused muscle mass is approximately 40% of the total muscle mass. Taking into consideration the different metabolic requirements of the perfused tissues, roughly 90% of oxydative metabolism occurs in the muscle[12], making this experimental setting truly a skeletal muscle preparation that permits the observation of even very discrete metabolic changes during reperfusion.
In the given experimental setting, it is difficult to account for the in detail contribution of protein degradation, amino acid intermediate metabolism, and protein synthesis. Factors that modify the efflux are transport systems in the cell membrane [26-28], which can be concentration- dependent (system L) or acting against the concentration gradient (system A), and the intermediary metabolism within the muscle cells[29]. Numerous previous assessments of the metabolic situation in the isolated perfused hindlimb demonstrate that these factors are relatively minor contributors, while amino acid efflux is nearly exclusively characterized by the net balance of protein metabolism both in anabolic and catabolic situations[12,30-34]. It is mostly due to changes in skeletal muscle, with only minor contributions from other tissues in this preparation [35-37].
Results
Sham- operated animals are catabolic at the time of perfusion, having lost about 36 g BW (see results), which is approximately 16% of initial BW. Rats of this age are still growing, with an increase of approx. 5 g/ day (2–3% BW)[38]. The weight loss is due to both lipolysis[39] and protein loss[40], which causes the typical increase of amino acid release in the hindquarter of fasting rats[12]. Compared to data[34] from non- operated rats fasted for 48 hours under otherwise identical conditions, the total amino acid release is increased by 30% in the sham- operated rats described here. While nephrectomy increases the amino acid release by approximately 15%, the relative decrease of amino acid release following administration of RU 38486 is similar in both nephrectomized and sham- operated animals. This indicates a stress- accentuated adaptation to fasting caused by corticosteron secretion[41] increased beyond the normal range, with a further effect of uremia. Increased amino acid serum concentrations during fasting are mostly due to an inhibition of protein synthesis[42], although proteolysis mainly of myofibrillary proteins does play a role[43]. However, corticosterone is only one of several effectors at play. RU 38486 affects neither acidosis nor lactate/ pyruvate ratio. Both factors may contribute to the continuously increased amino acid release. In the case of acidosis, this may be due to an action on acid inhibitable transporters such as system A which reduce the supply of nutrients to the cells[28]. Another possible mechanism is through inhibition of leptin by acidosis[44], which in neutral pH might counteract muscle wasting[45]. Balancing acidosis in chronically uremic rats with increased corticosterone secretion inhibited protein degradation, but had no effect on the defective protein synthesis[21]. More recently, RU 38486 was shown to be ineffective in blocking acid- mediated protein degradation as its action is only an indirect one, mediated via insuline- like growth factor I (IGF- I)[24,46]. These findings indicate that RU 38486 acts through an inhibition of the corticosterone- mediated decrease of protein- synthesis without affecting other factors that act predominantly on the level of protein degradation. While all these and more factors contribute to the muscle degradation seen in excess glucocorticoid situations, the mechanisms responsible in ultima causa remain still unclear[18]. Nephrectomy enhances the catabolic situation: the total amino acid efflux is increased by roughly 15% compared to sham- operated animals. The relative increase of amino acids that are not metabolized, such as phenylalanin and tyrosin, indicates that this effect is due to the acute and complex metabolic situation of uremia, without differenciating between inhibition of protein synthesis and stimulation of protein degradation. A multitude of effectors partake in this metabolic turmoil [15], of which glucocorticoids have been accused of playing a leading role [47]. At first sight, this opinion is supported by the finding that isolated hindquarters of animals treated with RU 38486 show a significant reduction of total amino acid efflux (16–19%, p < o.05- p <0.001) compared to untreated animals that underwent the same surgical procedure, indicating that RU38486 inhibits some common degrading influence on protein metabolism.
By contrast, in the comparison of the two groups treated with RU38486, the amino acid release remains increased in nephrectomized animals, albeit to a lesser extent, stressing the very point that glucocorticoids are only one of several factors that contribute to the net protein wasting. Chronically uremic rats with increased corticosterone- secretion [48] showed a less pronounced increase in protein degradation when acidosis was balanced while the defective protein synthesis remained unchanged. In the present experimental setting, acidosis evolving during perfusion was not corrected for. Acidosis and glucocorticoid action are seen as concomitant factors in the activation of the ubiquitin- proteasome pathway of muscle proteolysis [49], and a pH- responsive element in the promoter region for the ubiquitin- proteasome pathway has been reported [47]. In the situation of uncorrected acidosis, it therefore seems likely RU 38486 may have had an inhibitory effect on the corticoid- induced decrease in protein synthesis without influencing the proteolytogenic effects of other putative agents.
This presumption is supported by the finding that RU 38486 had no effect on the efflux of 3-methylhistidin. While this result is in contrast to Schäfer et al. [6,8], Lowell et al. found no reduction of the efflux of 3-methylhistidin after adrenalectomy in the perfused hindquarter of fasted animals [50], and in rats with chronic uremia, RU38486- resistant protein catabolism with unchanged release of 3-methylhistidin has been demonstrated in vivo by Teschner [51]. As responsiveness of protein synthesis and degradation to amino acid availability seem to be regulated differentially [52] and activation of glucocorticoid- mediated proteolysis occurs only at relatively elevated hormone levels [53] compared to the inhibition of protein synthesis [54], it seems possible to speculate that RU38486 may have a more pronounced effect on net protein catabolism at substantially higher doses.
The presented data fail to show that RU38486 inhibits glucocorticoid action in the specific uremic setting while it clearly reduces the elevated net protein catabolism compared to non- operated animals. This suggests that glucocorticoid mediated protein wasting in acute uremia is rather a by- product of the overall stress, in the present experimental setting caused by surgery and fasting, than due to an independent action specific to uremia.
While this finding abolishes hopes to counteract muscle wasting in uremia by administration of an anticorticoid drug and indirectly rather stresses the well described[55] clinical importance of a balanced acid- base status, it may open speculation about the usefulness of RU486 in post- traumatic states and severe illness.
Conclusion
Both sham- operated and nephrectomized animals show an increase in net protein catabolism. RU38486 clearly reduces the net protein wasting in both groups, but the increase in net protein catabolism observed over sham- operated animals remains unchanged in nephrectomized rats. The effect of antiglucocorticoid RU38486 may be attributed to an inhibition of fasting and operative stress- induced cortisol action which, even when within the physiological range, promotes increased protein turnover [56], and to a protective effect against the inhibition of protein synthesis. While RU38486 had no effect on net protein catabolism that could be specifically attributed to uremia, the demonstrated anticatabolic effectiveness in a stress accentuated metabolic situation should be studied more closely. Possible targets for therapeutic application under this aspect include post- traumatic states and severe illness.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
A.M. carried out animal experimentation, sample analysis, statistical analysis and wrote the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Professor Martin Schwenen, Düsseldorf, helped with the experimental design, the surgery and was always available for discussion. He has the author's heartfelt gratitude for his generous support.
Figures and Tables
Figure 1 A: Isolated rat hindquarter preparation. The perfused muscle mass is below the dotted line. The tail is clamped off. B: Recirculating perfusion system. Heart piece is the oxygenator (1), wherein the medium is gassed (95% O2, 5% CO2) and from where it is pumped (3) up via a filter (nylon mash, 20 μm width) (2) and pH and pO2 probes (4) to attain hydrostatic pressure of 80 mm Hg (5) before entering the hindquarter (6). Intermediate sampling and pH adjustment can be carried out if necessary (7).
Figure 2 Release rate of amino acids (nmol/ min/ g muscle) from isolated perfused rat hindquarter. Error bars indicate standard deviation. The figure is based on tables 4 and 5 which give the p-values for the comparison between sham- operated and nephrectomized animals (table 4) and RU 38486 untreated and treated groups (table 5), respectively.
Table 1 Body weight (BW) at day of surgical intervention (OP) and perfusion (EX). Values are gram ± standard deviation. No significant intergroup differences.
BW [g] at OP BW [g] at EX Weight loss [g]
SHAM (n = 6) 223.67 ± 8.02 188.17 ± 7.28 35.5 ± 5.3
BNX (n = 6) 217.33 ± 13.81 201.67 ± 11.91 15.7 ± 4.8
SHAM + RU (n = 5) 223.60 ± 9.21 196.00 ± 9.70 27.6 ± 5.9
BNX + RU (n = 5) 225.20 ± 5.36 213.00 ± 7.35 12.2 ± 4.8
Table 2 Dry weight (DW) in % of total tissue weight and hydration (Hy) at day of perfusion in ml/ g of total tissue weight. Average values ± standard deviation. Intergroup differences with significance levels of: *p < 0,001; #p < 0,05.
SHAM* BNX* SHAM + RU# BNX + RU#
n 6 6 5 5
DW [%] 24.41 ± 1.46 21.07 ± 0.98 24.26 ± 1.14 22.10 ± 1.46
Hy [ml/g] 0.761 ± 0.015 0.795 ± 0.010 0.762 ± 0.012 0.784 ± 0.015
Table 3 Total amino acid efflux (TAAE) and release of 3-methyl histidine (3 MH) in serum from isolated perfused rat hindquarter at 60 min. perfusion. Values given in nmol/min/g muscle + standard-deviation. Intergroup differences with significance of: #p < 0,01; *p < 0,005; ~, $p < 0,001.
SHAM BNX SHAM + RU BNX + RU
n 6 6 5 5
TAAE 65.05 ± 5.88* 71.83 ± 5.02# 52.98 ± 2.96* 60.61 ± 5.62#
3 MH 0.2 ± 0.09~ 0.55 ± 0.15~ 0.18 ± 0.06$ 0.53 ± 0.07$
Table 4 Amino acid release from the isolated perfused hindlimb. P- values indicate significant differences in the comparison between sham- operated and nephrectomized animals, either without or with RU 38486 treatment. A: p- value for SHAM vs. SHAM + RU; B: p- value for BNX vs. BNX + RU; C: p- value for SHAM vs. BNX; D: p- value for SHAM + RU vs. BNX + RU.
SHAM SHAM+RU BNX BNX+RU A: p-value B: p-value C: p-value D: p-value
Ala 13.26 10.08 15.61 13.23 <0.02 n.s <0.05 <0.05
Glu 15 13.69 17.39 15.37 n.s n.s <0.05 <0.05
Gly 7.56 7.11 6.53 5.26 n.s <0.005 n.s <0.001
Ser 3.27 2.75 3.12 2.6 n.s <0.025 n.s n.s
Thr 3.58 2.98 3.62 3.11 n.s ~0.05 n.s n.s
Val 2.2 1.49 2.32 1.7 <0.005 <0.001 n.s n.s
Leu 2.08 1.43 2.04 1.66 <0.005 ~0.05 n.s n.s
Ile 0.95 0.63 0.97 0.63 <0.025 <0.005 n.s n.s
Phe 1.78 1.23 2.29 2.02 <0.05 n.s <0.05 <0.001
Tyr 1.41 1.09 1.68 1.48 n.s ~0.05 n.s <0.01
Lys 6.53 5.08 6.91 5.54 <0.05 <0.005 n.s n.s
Arg 2.5 1.9 3.16 2.57 <0.05 <0.05 <0.02 <0.05
His 1.63 1.22 2.33 2.12 n.s n.s <0.025 <0.001
Glu 0.31 0.28 0.32 0.26 n.s n.s n.s n.s
Asn 1.49 1.05 1.7 1.57 n.s n.s n.s <0.02
Met 1.1 0.65 1.17 1.03 <0.02 n.s n.s <0.001
Trp 0.43 0.32 0.47 0.46 n.s n.s n.s <0.01
==== Refs
Druml W Protein metabolism in acute renal failure Miner Electrolyte Metab 1998 24 47 54 9397417 10.1159/000057350
Persike EC Addis T Increased rate of urea formation following removal of renal tissue. Am J Physiol 1949 158 149 156
Bondy PK Engel FL Farrar B The metabolism of amino acids and protein in the adrenalectomized nephrectomized rat. Endocrinoly 1949 44 476 484
Schaefer RM Moser M Kulzer P Peter G Heidland A Horl WH Massry SG Hormonal regulation of muscle protein catabolism in acutely uremic rats: effect of adrenalectomy and parathyroidectomy Adv Exp Med Biol 1988 240 257 266 3072844
Schaefer RM Weipert J Moser M Peter G Heidbreder E Horl WH Heidland A Reduction of urea generation and muscle protein degradation by adrenalectomy in acutely uremic rats Nephron 1988 48 149 153 3344055
Schaefer RM Teschner M Kulzer P Leibold J Peter G Heidland A Evidence for reduced catabolism by the antiglucocorticoid RU 38486 in acutely uremic rats Am J Nephrol 1987 7 127 131 3300336
Schaefer RM Riegel W Stephan E Keller H Horl WH Heidland A Normalization of enhanced hepatic gluconeogenesis by the antiglucocorticoid RU 38486 in acutely uraemic rats Eur J Clin Invest 1990 20 35 40 2108035
Schaefer RM Teschner M Riegel W Heidland A Reduced protein catabolism by the antiglucocorticoid RU 38486 in acutely uremic rats Kidney Int Suppl 1989 27 S208 11 2636660
Ruderman NB Houghton CR Hems R Evaluation of the isolated perfused rat hindquarter for the study of muscle metabolism Biochem J 1971 124 639 651 5135248
Altman KI Schwenen M Increased catabolism of muscle proteins as a manifestation of radiation myopathy Radiat Environ Biophys 1987 26 171 180 3659268
Cohen PP Umbreit , Burris and Stauffer Suspending media for animal tissues Manometric techniques 1957 Minneapolis, Burgess Publishing Co. 147 150
Schwenen M Skeletmuskulatur und metabolische Homöostase: Physiologische und Pathophysiologische Aspekte der Glukokortikoidwirkung auf den Muskelstoffwechsel Institut für physiologische Chemie II 1981 Düsseldorf, Universität Düsseldorf
Haverberg LN Omstedt PT Munro HN Young VR Ntau-methylhistidine content of mixed proteins in various rat tissues Biochim Biophys Acta 1975 405 67 71 1174569
Cianciaruso B Bellizzi V Napoli R Sacca L Kopple JD Hepatic uptake and release of glucose, lactate, and amino acids in acutely uremic dogs Metabolism 1991 40 261 269 2000038 10.1016/0026-0495(91)90107-8
Newby FD Price SR Determinants of protein turnover in health and disease Miner Electrolyte Metab 1998 24 6 12 9397411 10.1159/000057344
Bailey JL Mitch WE Mechanisms of protein degradation: what do the rat studies tell us J Nephrol 2000 13 89 95 10858969
Price SR Du JD Bailey JL Mitch WE Molecular mechanisms regulating protein turnover in muscle Am J Kidney Dis 2001 37 S112 4 11158874
Ma K Mallidis C Bhasin S Mahabadi V Artaza J Gonzalez-Cadavid N Arias J Salehian B Glucocorticoid-induced skeletal muscle atrophy is associated with upregulation of myostatin gene expression Am J Physiol Endocrinol Metab 2003 285 E363 71 12721153
Quan ZY Walser M Effect of corticosterone administration at varying levels on leucine oxidation and whole body protein synthesis and breakdown in adrenalectomized rats Metabolism 1991 40 1263 1267 1961118 10.1016/0026-0495(91)90026-S
Garibotto G Russo R Sofia A Sala MR Robaudo C Moscatelli P Deferrari G Tizianello A Skeletal muscle protein synthesis and degradation in patients with chronic renal failure Kidney Int 1994 45 1432 1439 8072256
May RC Bailey JL Mitch WE Masud T England BK Glucocorticoids and acidosis stimulate protein and amino acid catabolism in vivo Kidney Int 1996 49 679 683 8648908
Price SR Bailey JL England BK Necessary but not sufficient: the role of glucocorticoids in the acidosis-induced increase in levels of mRNAs encoding proteins of the ATP-dependent proteolytic pathway in rat muscle Miner Electrolyte Metab 1996 22 72 75 8676830
Baulieu EE RU 486 (mifepristone). A short overview of its mechanisms of action and clinical uses at the end of 1996 Ann N Y Acad Sci 1997 828 47 58 9329823
Pickering WP Baker FE Brown J Butler HL Govindji S Parsons JM Pawluczyk IZ Walls J Bevington A Glucocorticoid antagonist RU38486 fails to block acid-induced muscle wasting in vivo or in vitro Nephrol Dial Transplant 2003 18 1475 1484 12897084 10.1093/ndt/gfg203
Kim YM Lee JY Choi SH Kim DG Jahng JW RU486 blocks fasting-induced decrease of neuronal nitric oxide synthase in the rat paraventricular nucleus Brain Res 2004 1018 221 226 15276881 10.1016/j.brainres.2004.05.068
Shotwell MA Kilberg MS Oxender DL The regulation of neutral amino acid transport in mammalian cells Biochim Biophys Acta 1983 737 267 284 6303424
Rennie MJ Muscle protein turnover and the wasting due to injury and disease Br Med Bull 1985 41 257 264 3896381
Bevington A Brown J Butler H Govindji S K MK Sheridan K Walls J Impaired system A amino acid transport mimics the catabolic effects of acid in L6 cells Eur J Clin Invest 2002 32 590 602 12190959 10.1046/j.1365-2362.2002.01038.x
Goldberg AL Chang TW Regulation and significance of amino acid metabolism in skeletal muscle Fed Proc 1978 37 2301 2307 350636
Ruderman NB Berger M The formation of glutamine and alanine in skeletal muscle J Biol Chem 1974 249 5500 5506 4278315
Thienhaus R Tharandt L Zais U Staib W [Effect of glucocorticoides on the release of amino acids in the perfused rat hindquarter (author's transl)] Hoppe Seylers Z Physiol Chem 1975 356 811 817 1181274
Jefferson LS Li JB Rannels SR Regulation by insulin of amino acid release and protein turnover in the perfused rat hemicorpus J Biol Chem 1977 252 1476 1483 838725
Caldwell MD Lacy WW Exton JH Effects of adrenalectomy on the amino acid and glucose metabolism of perfused rat hindlimbs J Biol Chem 1978 253 6837 6844 690127
Schwenen M Altman KI Schroder W Radiation-induced increase in the release of amino acids by isolated, perfused skeletal muscle Int J Radiat Biol 1989 55 257 269 2563398
Clark AS Kelly RA Mitch WE Systemic response to thermal injury in rats. Accelerated protein degradation and altered glucose utilization in muscle J Clin Invest 1984 74 888 897 6470144
Clark AS Mitch WE Comparison of protein synthesis and degradation in incubated and perfused muscle Biochem J 1983 212 649 653 6349623
Goldstein L Perlman DF McLaughlin PM King PA Cha CJ Muscle glutamine production in diabetic ketoacidotic rats Biochem J 1983 214 757 767 6414461
Odedra BR Bates PC Millward DJ Time course of the effect of catabolic doses of corticosterone on protein turnover in rat skeletal muscle and liver Biochem J 1983 214 617 627 6193785
Li JB Wassner SJ Protein synthesis and degradation in skeletal muscle of chronically uremic rats Kidney Int 1986 29 1136 1143 3747330
Goodman MN Lowell B Belur E Ruderman NB Sites of protein conservation and loss during starvation: influence of adiposity Am J Physiol 1984 246 E383 90 6720943
Millward DJ Odedra B Bates PC The role of insulin, corticosterone and other factors in the acute recovery of muscle protein synthesis on refeeding food-deprived rats Biochem J 1983 216 583 587 6365077
Li JB Goldberg AL Effects of food deprivation on protein synthesis and degradation in rat skeletal muscles Am J Physiol 1976 231 441 448 961895
Li JB Wassner SJ Effects of food deprivation and refeeding on total protein and actomyosin degradation Am J Physiol 1984 246 E32 7 6364831
Teta D Bevington A Brown J Pawluczyk I Harris K Walls J Acidosis downregulates leptin production from cultured adipocytes through a glucose transport-dependent post-transcriptional mechanism J Am Soc Nephrol 2003 14 2248 2254 12937300 10.1097/01.ASN.0000083903.18724.93
Gat-Yablonski G Ben-Ari T Shtaif B Potievsky O Moran O Eshet R Maor G Segev Y Phillip M Leptin reverses the inhibitory effect of caloric restriction on longitudinal growth Endocrinology 2004 145 343 350 14525912 10.1210/en.2003-0910
Kritsch KR Murali S Adamo ML Ney DM Dexamethasone decreases serum and liver IGF-I and maintains liver IGF-I mRNA in parenterally fed rats Am J Physiol Regul Integr Comp Physiol 2002 282 R528 36 11792663
Wang X Jurkowitz C England BK Mechanisms for low levels of branched chain amino acids in uremia: Dual regulation of branched chain a ketoacid dehydrogenase by pH and glucocorticoids. J Am Soc Nephrol 1996 7 1867
May RC Kelly RA Mitch WE Mechanisms for defects in muscle protein metabolism in rats with chronic uremia. Influence of metabolic acidosis J Clin Invest 1987 79 1099 1103 3549778
Bailey JL Wang X England BK Price SR Ding X Mitch WE The acidosis of chronic renal failure activates muscle proteolysis in rats by augmenting transcription of genes encoding proteins of the ATP-dependent ubiquitin-proteasome pathway J Clin Invest 1996 97 1447 1453 8617877
Lowell BB Ruderman NB Goodman MN Evidence that lysosomes are not involved in the degradation of myofibrillar proteins in rat skeletal muscle Biochem J 1986 234 237 240 3707546
Teschner M Schaefer RM Rudolf C Peter G Heidland A Independence of enhanced protein catabolism from glucocorticoids in chronically uremic rats Res Exp Med (Berl) 1989 189 339 345 2813969
Giordano M Castellino P DeFronzo RA Differential responsiveness of protein synthesis and degradation to amino acid availability in humans Diabetes 1996 45 393 399 8603758
Shoji S Pennington RJ The effect of cortisone on protein breakdown and synthesis in rat skeletal muscle Mol Cell Endocrinol 1977 6 159 169 832761 10.1016/0303-7207(77)90082-X
Tomas FM Munro HN Young VR Effect of glucocorticoid administration on the rate of muscle protein breakdown in vivo in rats, as measured by urinary excretion of N tau-methylhistidine Biochem J 1979 178 139 146 435272
Bailey JL Mitch WE Twice-told tales of metabolic acidosis, glucocorticoids, and protein wasting: what do results from rats tell us about patients with kidney disease? Semin Dial 2000 13 227 231 10923349 10.1046/j.1525-139x.2000.00063.x
Quan ZY Walser M Effects of corticosterone administration on nitrogen excretion and nitrogen balance in adrenalectomized rats Am J Clin Nutr 1992 55 695 700 1550045
| 15715918 | PMC550647 | CC BY | 2021-01-04 16:03:43 | no | BMC Nephrol. 2005 Feb 17; 6:2 | utf-8 | BMC Nephrol | 2,005 | 10.1186/1471-2369-6-2 | oa_comm |
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-161571323010.1186/1471-2458-5-16Research ArticleLeisure-time versus full-day energy expenditure: a cross-sectional study of sedentarism in a Portuguese urban population Gal Diane L [email protected] Ana-Cristina [email protected] Henrique [email protected] Department of Hygiene and Epidemiology, University of Porto Medical School, Porto, Portugal2005 15 2 2005 5 16 16 5 8 2004 15 2 2005 Copyright © 2005 Gal et al; licensee BioMed Central Ltd.2005Gal 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
Low physical activity is known to be a potential risk factor for cardiovascular disease. With high prevalence of cardiovascular diseases in the Portuguese urban population, little is known about how sedentary this population is and what factors are associated to sedentary lifestyles. This study's objective was to examine sedentary lifestyles and their determinants through a cross-sectional study.
Methods
2134 adults (18 years and older) were interviewed using a standard questionnaire, comprising of social, behavioural and clinical information. Time spent in a variety of activities per day, including: work, household chores, sports, sedentary leisure time and sleep, were self-reported. Energy expenditure was estimated based on the related metabolic equivalent (MET) and time spent in each activity (min/day). Those with less than 10% of energy expenditure at a moderate intensity of 4 METs or higher were categorised as sedentary. The proportion of sedentary people and 95% Confidence Intervals (CI) were calculated, and the magnitude of associations, between sedentary lifestyles and the population characteristics, were computed as age-adjusted odds ratios using logistic regression.
Results
Sedentarism in both genders during leisure time is high at 84%, however in full day energy expenditure, which includes physical activity at work, sleeping hours and household chores, 79% of males and 86% of females are found to be sedentary. In leisure-time only, increased age is associated with higher odds of being sedentary in both genders, as well as in women with increased BMI. In comparison, in full-day energy expenditure, sedentarism is more likely to occur in those with higher levels of education and in white-collar workers.
Conclusions
A high prevalence of sedentarism is found in the study participants when measuring leisure-time and full-day energy expenditure. The Portuguese population may therefore benefit from additional promotion of physical activity.
==== Body
Background
Physical activity has been defined by World Health Organization [1] as comprising of all movements in everyday life, including work, recreation, and sports activities and has been categorised in levels of intensity from light to moderate to vigorous. Health benefits, including decreasing risk of coronary heart disease [2-5] have mostly been associated with moderate to vigorous activities [6]. Practicing 30 minutes of moderate physical activity at least five days a week is widely promoted to achieve health benefits and is also felt to be an achievable lifestyle change for sedentary adults [1].
Physical activity tends to be associated with lower cardiovascular morbidity and mortality and an overall improved quality of life. Some studies [6-9] have defined or examined sedentary lifestyles and associated factors at the population level, however few have approached the subject in the context of Southern Europe. This may be due in part to the fact that it is difficult to classify a person as sedentary since no universally accepted classification is currently available. Previous studies have used a simple question or an evaluation of whether adults perform at least 30 minutes of moderate activity five times a week [10], or did not take part in leisure time physical activity at all [4,11], to classify participants as inactive or sendentary.
Portugal has the highest stroke mortality rates in Western Europe and cardiovascular diseases cause approximately 40% of deaths [12,13]. The current study's objective was to examine sedentarism in leisure time and throughout a full day in a Portuguese urban population and to cross-sectionally assess the associations between sedentarism and demographic, social, behavioural, clinical and anthropometric factors.
Methods
Data was obtained as part of an ongoing cross-sectional health survey of adults living in the city of Porto, Portugal. Random digit dialling was used to select a single person over 17 years old from each household, without allowing for substitution of refusals. A participation rate of 70% was achieved [14]. Using a structured questionnaire, trained interviewers collected data from 2134 adults on demographic, personal and family medical history, and behavioural characteristics (physical activity, smoking, alcohol intake and diet) [15]. Sixty-seven participants who scored less than 24 on the Folstein mini-mental state examination [16] were considered probably unable to provide reliable information due to cognitive impairment and were excluded from the analysis. An additional 16 participants who did not fit the survey criteria (did not live in Porto or had severe disabilities and diseases) were also excluded. As well, participants missing relevant data were not included in the study analyses, leaving 2004 participants (1226 women; 778 men) in the analysis.
As reported in an earlier publication [17], education was recorded as completed years of schooling and divided into three broad categories: less than 5, 5–11, and more than 11 years. Body weight was measured to the nearest 0.1 kg using a digital scale, and height was measured to the nearest centimetre in the standing position using a wall standiometer. Body mass index (BMI) was calculated as weight in kilograms divided by square height in meters. The distribution of BMI is reported by standard WHO categories and nomenclature [18]: underweight to normal (<25.0 kg/m2), overweight (25.0–29.9 kg/m2), and obese (> = 30 kg/m2). The number of self-reported medical visits occurring in the last 12 months was grouped based on tertiles (0–1, 2–3, >3). Current occupation was self-reported and divided into the three usual categories of white collar, blue-collar and retired or unemployed. White collar work included all non-manual and superior professionals such as teachers, health professionals, secretaries etc. Blue-collar work included all manual professionals including agriculturers, taxi drivers, cooks, factory workers and sewers. Women who stated that they performed domestic work in their own home and had no other employment were classified as unemployed. Each participant was also asked about chronic diseases requiring continued medical care. Energy intake was estimated based on a semi-quantitative food frequency questionnaire validated for the Portuguese population and results were presented in tertiles, separately for each gender. Alcohol intake was self-reported and classified into three categories: current drinkers (daily alcohol intake), ex-drinkers (no alcohol for more than 6 months), and never or occasional drinkers. Smoking was self-reported and classified based on the WHO categories [19]: current smoker included both daily and occasional smokers, ex-smokers were those who had not smoked a cigarette in the last 6 months, and non-smokers were those who never smoked at all.
Participants completed a physical activity questionnaire designed to estimate usual individual daily energy expenditure, focused on the activity in the past year. Time spent in a variety of activities per day, including: work, transport to or from work, household chores, sports, sedentary leisure time and sleep, was self-reported and activity intensity categorised as very light, light, moderate and heavy with a corresponding average of 1.5, 2.5, 5.0 and 7.0 METs respectively, where one MET is equal to the energy expended at the basal metabolic rate or at rest [20]. Due to the manner in which these questions were presented during the face-to-face interviews, a large variation resulted in the number of hours reported per day (Average = 19 hours/day (minimum = 6.5 to maximum = 32.5 hours) with 46 (2.3%) participants reporting activity resulting in more than 24 hours per day). Energy expenditure was estimated by multiplying the related metabolic equivalent (MET) to the self-reported time spent in each activity (min/day). Participants with less than 10% of daily energy expenditure at a moderate or high intensity level (>4 METs) during leisure-time or throughout the day were categorised as sedentary, the remaining being considered active [21]. Proportions of sedentary individuals and 95% CI were calculated for both leisure-time and full-day energy expenditure, the latter including energy expended at work, during sleep and in household chores. The magnitude of associations, between sedentary lifestyles and the factors studied, were computed as age-adjusted odds ratios using logistic regression. In the analyses of the percentages of sedentarism and its associations with leisure-time energy expenditure, 15 people were excluded since they did not report any leisure-time activities. Analyses were conducted using Stata 7.0.
Results
In the exploration of the population studied it was found that a significant difference between men and women was noted in most baseline characteristics other than in the age distribution and the hours of sleep (mean hours of sleep per night equalled 8; 95%CI 7.9–8.1). A higher proportion of household work was undertaken by women (95.5% versus 55.5% of men), a higher proportion of men were married (84.3% versus 61.3% of women), 68.3% of women and 55.7% of men reported no chronic disease. 31% of all participants worked between 20 and 40 hours a week, with a higher percentage of men working greater than 40 hours (31.1% versus 15.2% of women). It is also worthy to note that a high percentage of both genders reported not undertaking regular leisure-time sports and exercise (69.3% of women and 58.9% of men). All subsequent analyses were performed separately for males and females.
Overall sedentary lifestyle percentages (Table 1 and 2) are high, 83.7% (95%CI: 80.9–86.2) for males and 84.4% (95%CI: 82.2–86.3) for females during leisure time. A lower percentage was found for males with 78.8% (95%CI: 75.7–81.6) when the full day energy expenditure is calculated including physical activity at work, hours of sleep and household chores. The full-day sedentary lifestyle percentage for women, however, increased slightly to 86.1% (95%CI: 84.0–88.0).
Table 1 Sedentarism in the Female Population
Characteristics Leisure-time Energy Expenditure * Full-day Energy Expenditure **
N (%) % (95% CI) OR† (95% CI) % (95% CI) OR† (95% CI)
1226 (61.2) 84.4 (82.2–86.3) 86.1 (84.0–88.0)
Age (years)
18–29 88 (7.2) 68.2 (57.4–77.7) 75.0 (64.6–83.6)
30–39 125 (10.2) 71.4 (62.4–79.3) 1.2 (0.6–2.1) 77.6 (69.3–84.6) 1.2 (0.6–2.2)
40–49 294 (24.0) 84.6 (80.0–88.6) 2.6 (1.5–4.5) 83.3 (78.6–87.4) 1.7 (0.9–3.0)
50–59 311 (25.4) 85.1 (80.5–88.8) 2.7 (1.5–4.6) 85.2 (80.7–88.9) 1.9 (1.1–3.4)
60–69 238 (19.4) 91.5 (87.2–94.7) 5.0 (2.7–9.6) 90.8 (86.3–94.1) 3.3 (1.7–6.3)
70+ 170 (13.9) 90.0 (84.5–94.1) 4.2 (2.1–8.2) 98.2 (94.9–99.6) 18.6 (5.4–64.1)
Marital Status
Married 751 (61.3) 85.3 (82.5–87.7) 86.4 (83.7–88.7)
Not married 475 (38.7) 82.9 (79.1–86.1) 0.8 (0.6–1.2) 85.7 (82.1–88.6) 0.8 (0.6–1.2)
Education (years)
<5 years 530 (43.2) 93.7 (91.2–95.6) 87.0 (83.7–89.7)
5–11 years 330 (26.9) 84.5 (80.0–88.1) 0.4 (0.2–0.6) 83.9 (79.4–87.6) 1.2 (0.8–1.8)
>11 years 366 (29.9) 70.6 (65.6–75.2) 0.2 (0.1–0.3) 86.9 (82.9–90.1) 2.2 (1.4–3.6)
BMI (kg/m2)
<25 455 (37.1) 77.5 (73.2–81.2) 85.3 (81.6–88.3)
25–30 452 (36.9) 85.1 (81.4–88.2) 1.3 (0.9–1.9) 85.2 (81.5–88.3) 0.7 (0.5–1.0)
>30 319 (26.0) 93.1 (89.6–95.5) 2.9 (1.8–4.9) 88.7 (84.6–91.9) 0.9 (0.6–1.4)
Physician visits in last year (n)
0–1 422 (34.4) 84.4 (80.5–87.7) 84.4 (80.5–87.6)
2–3 visits 381 (31.1) 79.1 (74.6–83.0) 0.6 (0.4–0.8) 84.8 (80.7–88.2) 0.8 (0.6–1.2)
>3 423 (34.5) 89.0 (85.6–91.8) 1.2 (0.8–1.8) 89.1 (85.7–91.9) 1.1 (0.7–1.7)
Occupation
White collar worker 425 (34.7) 77.1 (72.7–81.0) 88.2 (84.7–91.1)
Blue collar worker 193 (15.7) 92.1 (87.3–95.5) 3.0 (1.7–5.4) 61.7 (54.4–68.5) 0.2 (0.1–0.3)
Unemployed or retired 608 (49.6) 87.0 (84.0–89.5) 1.4 (0.9–2.1) 92.4 (90.0–94.4) 1.0 (0.6–1.6)
Energy Intake (kcal/day)
<1800 400 (32.7) 83.3 (79.3–86.9) 1.0 (0.7–1.5) 86.8 (83.0–89.9) 1.0 (0.7–1.5)
1800–2300 461 (37.7) 82.7 (78.8–86.0) 85.7 (82.1–88.7)
>2300 362 (29.6) 87.5 (83.6–90.7) 1.8 (1.2–2.7) 85.9 (81.9–89.3) 1.3 (0.9–1.9)
Alcohol Use
Non/Occasional-drinkers 576 (47.0) 80.7 (77.2–83.8) 83.7 (80.3–86.6)
Ex-drinkers 103 (8.4) 94.2 (87.8–97.8) 2.6 (1.1–6.2) 93.2 (86.5–97.2) 1.7 (0.8–4.0)
Drinkers 547 (44.6) 86.3 (83.1–89.0) 1.3 (0.9–1.3) 87.4 (84.2–90.0) 1.2 (0.9–1.8)
Tobacco Use
Non-smokers 893 (72.8) 84.9 (82.3–87.2) 86.6 (84.1–88.7)
Ex-smokers 119 (9.7) 81.7 (73.5–88.3) 0.9 (0.6–1.6) 89.1 (82.0–94.1) 1.6 (0.9–3.0)
Smokers 214 (17.5) 83.6 (77.9–88.3) 1.5 (0.9–2.3) 82.7 (77.0–87.5) 1.2 (0.8–1.8)
* Leisure time energy expenditure encompasses the energy expended in all leisure activities (not including sleep, work and household chores) where being sedentary is defined as spending less than 10% of their time in activities requiring ≥ 4 metabolic equivalents (MET).
** Full-day energy expenditure encompasses the energy expended in all activities in a day where being sedentary is defined as above.
† Age-adjusted Odds ratios
Table 2 Sedentarism in the Male Population
Characteristics Leisure-time Energy Expenditure * Full-day Energy Expenditure **
N (%) % (95% CI) OR† (95% CI) % (95% CI) OR† (95% CI)
778 (38.8) 83.7 (80.9–86.2) 78.8 (75.7–81.6)
Age (years)
18–29 49 (6.3) 57.1 (42.2–71.2) 69.4 (54.6–81.7)
30–39 68 (8.7) 74.6 (62.5–84.5) 2.2 (1.0–4.9) 73.5 (61.4–83.5) 1.2 (0.5–2.8)
40–49 173 (22.2) 79.2 (72.4–85.0) 2.9 (1.5–5.6) 78.0 (71.1–84.0) 1.6 (0.8–3.2)
50–59 180 (23.1) 85.6 (79.6–90.3) 4.4 (2.2–9.0) 76.7 (69.8–82.6) 1.4 (0.7–2.9)
60–69 174 (22.4) 89.5 (84.0–93.7) 6.4 (3.0–13.5) 80.5 (73.8–86.1) 1.8 (0.9–3.7)
70+ 134 (17.2) 94.0 (88.5–97.4) 11.7 (4.7–29.2) 86.6 (79.6–91.8) 2.8 (1.3–6.2)
Marital Status
Married 656 (84.3) 85.1 (82.1–87.7) 78.7 (75.3–81.7)
Not married 122 (15.7) 76.2 (67.7–83.5) 1.1 (0.6–2.0) 79.5 (71.3–86.3) 1.4 (0.8–2.5)
Education (years)
<5 years 262 (33.7) 93.4 (89.7–96.1) 74.8 (69.1–79.9)
5–11 years 282 (36.2) 84.0 (79.2–88.1) 0.5 (0.3–0.9) 78.7 (73.5–83.4) 1.6 (1.0–2.4)
>11 years 234 (30.1) 72.5 (66.3–78.2) 0.3 (0.2–0.5) 83.3 (77.9–87.9) 2.5 (1.5–4.2)
BMI (kg/m2)
<25 279 (35.9) 79.9 (74.7–84.4) 77.4 (72.1–82.2)
25–30 375 (48.2) 84.9 (80.8–88.3) 1.2 (0.8–1.9) 78.7 (74.1–82.6) 1.1 (0.7–1.6)
>30 124 (15.9) 88.7 (81.8–93.7) 1.8 (0.9–3.4) 82.3 (74.4–88.5) 1.3 (0.8–2.3)
Physician visits in last year (n)
0–1 354 (45.5) 80.9 (76.3–84.8) 76.0 (71.1–80.3)
2–3 visits 236 (30.3) 84.7 (79.4–89.0) 1.1 (0.7–1.7) 79.2 (73.5–84.2) 1.1 (0.7–1.7)
>3 188 (24.2) 87.8 (82.2–92.1) 1.1 (0.6–1.9) 83.5 (77.4–88.5) 1.4 (0.9–2.3)
Occupation
White collar worker 336 (43.2) 83.1 (79.5–86.2) 85.4 (82.0–88.3)
Blue collar worker 133 (17.1) 86.6 (81.8–90.6) 1.6 (0.9–2.9) 65.6 (59.4–71.5) 0.2 (0.1–0.3)
Unemployed or retired 309 (39.7) 53.8 (25.1–80.8) 0.8 (0.5–1.4) 69.2 (38.6–90.9) 0.6 (0.4–1.0)
Energy Intake (kcal/day)
<2300 261 (33.7) 88.1 (83.6–91.8) 1.1 (0.7–1.8) 86.6 (81.8–90.5) 1.6 (1.0–2.5)
2300–2900 283 (36.5) 84.3 (79.6–88.4) 79.5 (74.3–84.1)
>2900 231 (29.8) 77.7 (71.8–82.9) 0.8 (0.5–1.3) 68.8 (62.4–74.7) 0.6 (0.4–0.9)
Alcohol Use
Non/Occasional drinkers 87 (11.2) 80.2 (70.2–88.0) 82.8 (73.2–90.0)
Ex-drinkers 48 (6.2) 83.3 (69.8–92.5) 0.7 (0.2–1.8) 79.2 (65.0–89.5) 0.6 (0.2–1.5)
Drinkers 643 (82.7) 84.2 (81.1–86.9) 1.0 (0.5–1.8) 78.2 (74.8–81.3) 0.7 (0.4–1.2)
Tobacco Use
Non-smokers 218 (28.0) 82.6 (76.9–87.4) 78.9 (72.9–84.1)
Ex-smokers 296 (38.1) 86.1 (81.6–89.8) 1.0 (0.6–1.6) 77.0 (71.8–81.7) 0.8 (0.5–1.3)
Smokers 264 (33.9) 82.1 (76.9–86.5) 1.2 (0.7–1.9) 80.7 (75.4–85.3) 1.3 (0.8–2.0)
* Leisure time energy expenditure encompasses the energy expended in all leisure activities (not including sleep, work and household chores) where being sedentary is defined as spending less than 10% of their time in activities requiring ≥ 4 metabolic equivalents (MET).
** Full-day energy expenditure encompasses the energy expended in all activities in a day where being sedentary is defined as above.
†Age-adjusted Odds ratios
Few differences were found in the level of sedentarism in adults when considering differences in population characteristics. Younger participants tend to have lower percentages of sedentarism compared to older participants. In the leisure-time only estimation unmarried men (76.2%; 95%CI: 67.7–83.5) and female white-collar workers (77.1%; 95%CI: 72.7–81.0) tend to be more active. Women and men tend to be more active in leisure-time with increasing years of education, changing from 94% sedentarism in those with less than five years of education to 84% for those with five to eleven years of education and 72.5% for men and 70.6% for women with greater than 11 years of schooling. However, when the full-day energy expenditure is used as the estimate, these differences are no longer found and the trends with education and occupation are reversed. The lowest levels of sedentarism are found in males (74.8%; 95%CI: 69.1–79.9) with less education, and men who have a high energy intake (68%; 62.4–74.7) as well as, both male (65.6%; 95%CI: 59.4–71.5) and female (61.7%; 95%CI: 54.4–68.5) blue-collar workers.
When further examining the results of the age-adjusted associations between the population characteristics and sedentarism it is found that few factors were associated with an increased proportion of sedentarism. Marital status, physician visits in the last year and tobacco consumption, once adjusted for age, were not associated with differences in energy expenditure and sedentarism. In the leisure-time only estimation, increased age was associated with higher odds of being sedentary in both males and females. Sedentarism increased with increased BMI in women (BMI 25–30 = OR 1.3 95% CI: 0.9–1.9 ; BMI >30 = OR 2.9 95%CI:1.8–4.9), as well as in women with a high energy intake (>2300 Kcal/day = OR 1.8 95%CI:1.2–2.7). Sedentarism also increased in males who were ex-drinkers when compared to non or occasional drinkers (OR 2.6 95%CI: 1.1–6.2). Following what was noted earlier, higher levels of education were associated with higher levels of activity in leisure-time in both males (5–11 years education = OR 0.5 95%CI: 0.3–0.9; >11 years education = OR 0.3 95%CI:0.2–0.5) and females (OR 0.4 95%CI: 0.2–0.6 and OR 0.2 95%CI: 0.1–0.3, respectively) and blue-collar workers were more likely to be sedentary (males: OR 1.6 95%CI: 0.9–2.9; females: OR 3.0 95%CI: 1.7–5.4).
In comparison, in the full-day energy expenditure estimation, increased odds in age were not as strong for men and evidence of an association was not apparent with increased BMI or calorie intake for women. However, the reverse association was identified where those with higher amounts of education tended to be more sedentary in both males (5–11 years education = OR 1.6 95%CI: 1.0–2.4; >11 years education = OR 2.5 95%CI:1.5–4.2) and females (OR 1.2 95%CI: 0.8–1.8 and OR 2.2 95%CI: 1.4–3.6, respectively) and blue-collar workers were found to be significantly less sedentary (OR 0.2 95%CI:0.1–0.3). Relationships with energy intake also were identified in men with those consuming less than 2300 Kcal/day on average having a higher odds of being sedentary (OR 1.6 95%CI: 1.0–2.5) and those consuming greater than 2900 Kcal/day having a lower odds of being sedentary (OR 0.6 95%CI:0.4–0.9).
Discussion
Our results highlight the primarily sedentary nature of this adult urban population, with 70% of women and 60% of men not undertaking any regular physical activity or sports during leisure time. Similar studies, which have only evaluated leisure time physical activity, have identified comparable levels of sedentarism, as well as associations between sedentarism and certain population factors. In a European Union study [22] conducted in 1997, it was reported that the Portuguese population was one of the most sedentary among the 15 countries studied, with 85.2% of men and 90.0% of women being classified as sedentary compared to 83.7% of men and 84.4% of women in this study. It would be expected that, since the sampling in the study was meant to be representative of the whole country, a greater difference in the overall levels of sedentarism would be found mainly due to the differences in the levels and types of activities undertaken by rural and city dwellers. The small sample number in the European study (1007 participants) may also not capture the full extent of activities undertaken by the population in general. Similar associations were also noted for leisure-time energy expenditure, where the prevalence of sedentarism was higher with age and higher in the less educated in the European and in a Swiss [23] study. Other associations found in the European study, that obese individuals had higher prevalence of sedentarism was only found to be true in women in our study and no association was found between sedentarism and current smoking as identified in the European study. Lower levels of physical activity have also been associated with those who were female, older and with lower socio-economic status in a New Zealand study [10]. The differences between other studies results and ours may be due to true differences between the study sample baseline characteristics, or possible due to the study methods utilised.
The questionnaire, which was used to collect data on physical activity, was developed according to the European Prospective Investigation into Cancer and Nutrition study questionnaire, which showed acceptable repeatability and validity [24]. Formal validation of the questionnaire was undertaken using four seven days records (data not published). Participant recall may limit accurate capturing, through the questionnaire, of time and intensity spent undertaking various activities [25] during an average day or week in the last year. Although a variation in the hours of activities reported in a day was found, the percentage of participants reporting over 24 hours of activity was small (2.2%) and would not substantially affect the results of the study. As well, the description of types of activities provided in the questionnaire allow for METs to be estimated based on the Compendium of physical activity[20]. Variation is also present between studies in the categorisation of metabolic equivalents for activities with moderate intensity, with the US Surgeon General reporting moderate intensity exercise as being equal to 3–5 times the basal metabolic rate [26], while other studies use a cut off for moderate activity being more than 4, [2,21,22] or even greater than 5 METs [8]. Thirty minutes of activity at 4 METs, in an adult with 75 kg, will lead to an approximate energy expenditure of 150 Kcal per day or 1050 Kcal per week, which is a minimum level of moderate intensity daily activity recommended in the US Surgeon General report [26]. As energy expenditure varies from person to person, previous studies [21,22] have measured energy expenditure and have defined someone as being sedentary if they expend less than 10% of their daily energy in the performance of moderate-intensity activities (at least 4 times the basal metabolism rate) and therefore, on average, expend less than the recommended150 Kcal per day.
The above mentioned studies only recorded and based results on leisure-time energy expenditure, excluding the potential input of physical activity that is undertaken at work or on household tasks. As presented in the results of this study, the inclusion of work-time energy expenditure shows that those less educated and those with manual occupations are less sedentary, which reverses the association seen in the leisure-time only estimation. The differences between the associations of the two separate measurements, leisure-time energy expenditure versus full-day energy expenditure, therefore demonstrate the potential for work-related and household-related physical activity to significantly affect the proportion of sedentarism, and the associations between sedentarism and the factors studied. Efforts, therefore, need to be made to include all components of daily physical activity and energy expenditure and to study the effects of this energy expenditure as a whole, on cardiovascular disease and other health benefits of moderate and high-intensity energy expenditure, which has also been highlighted by Salmon et al [27]. However, the different psychosocial aspects expectedly associated with the decision of engaging in leisure time physical activity or related to hard work as part of occupational tasks might result in different effects on health for the same amount of energy expenditure.
The European Society of Cardiology has outlined, in a recent position paper, the need for physical activity to be prescribed in primary and secondary prevention and to implement successful strategies to reduce cardiovascular risks [28]. It has been observed for centuries that physical activity maintains and improves health and well-being, however health-systems have done little to promote and support appropriate levels of physical activity, especially in groups with elevated cardiovascular risk [29]. The lack of knowledge of the determinants of, and health problems related to, sedentarism and of the best interventions for behavioural change and long-term adherence to physical activity may play a part in low prescription of physical activity. Interventions to decrease sedentarism through primary health care [11] and in workplace settings [30] have had positive results, however all interventions may not affect change, such as was found with a population-wide print-media intervention [31]. Lessons can be learned from these interventions, and appropriate public health interventions prepared, in order to reduce the high levels of sedentarism, which acts as a main factor in high cardiovascular risk.
Conclusions
The urban Portuguese population has a very high prevalence of reported sedentarism potentially contributing to the high levels of cardiovascular disease in the country. Caution, however needs to be taken in the classification of individuals as sedentary when considering leisure-time versus full-day energy expenditure, as work and household-related activities can account for a large portion of the energy spent. Including these measures may also affect the overall associations found between sedentarism and the population characteristics.
Abbreviations
BMI Body Mass Index
MET Metabolic Equivalent
RMR Resting Metabolic Rate
WHO World Health Organisation
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DG designed the study, performed the statistical analysis and drafted the manuscript. ACS participated in the design of the study and in the interpretation of the results. HB conceived the study, and participated in its design and coordination. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was funded by the Fundação para a Ciência e Tecnologia, Praxis 2/2.1/SAU/1332/95, POCTI/ESP/35767/99, and POCTI/ESP/42361/2001.
==== Refs
Pan American Health Organisation Physical activity: How much is needed? 2002 Washington, USA, PAHO 2pp
Sesso HD Paffenbarger RSJ Lee IM Physical activity and other risk factors in men: The Harvard Alumni Health Study Circulation 2000 102 975 980 10961960
Stampfer MJ Hu FB Manson JE Rimm EB Willett WC Primary prevention of coronary heart disease in women through diet and lifestyle N Engl J Med 2000 343 16 22 10882764 10.1056/NEJM200007063430103
Kaprio J Kujala UM Koskenvuo M Sarna S Physical activity and other risk factors in male twin pairs discordant for coronary heart disease Atherosclerosis 2000 150 193 200 10781651 10.1016/S0021-9150(99)00368-8
Manson JE Hu FB Edwards RJW Colditz GA Stampfer MJ Willett WC Speizer FE Hennekens CH A prospective study of walking as compared with vigorous excercise in the prevention of coronary heart disease in women N Engl J Med 1999 341 650 658 10460816 10.1056/NEJM199908263410904
Dong L Block G Mandel S Activities contributing to total energy expenditure in the United States: Results from the NHAPS study Int J Behav Nutr Physical Activity 2004 1 4 10.1186/1479-5868-1-4
Schmitz KH Lytle LA Phillips GA Murray DM Birnbaum AS Kubik MY Psychosocial correlates of physical activity and sedentary leisure habits in young adolescents: The teens eating for energy and nutrition at school study Prev Med 2002 34 266 278 11817924 10.1006/pmed.2001.0982
Sternfeld B Ainsworth BE Quesenberry CP Physical activity patterns in a diverse population of women Prev Med 1999 28 313 323 10072751 10.1006/pmed.1998.0470
Ham SA Yore MM Fulton JE Kohl III HW Prevalence of no leisure time physical activity 35 States and the District of Columbia, 1988-2002. MMWR Weekly 2004 53 82 86
Elley CR Kerse NM Arroll B Why target sedentary adults in primary health care? Baseline results from the Waikato Heart, Health, and Activity Study Prev Med 2003 37 342 348 14507491 10.1016/S0091-7435(03)00142-7
Titze S Martin BW Seiler R Stronegger W Marti B Effects of a lifestyle physical activity intervention on stages of change and energy expenditure in sedentary employees Psychology Sport Exerc 2001 2 103 116 10.1016/S1469-0292(00)00016-9
World Health Organisation: Regional Office for Europe Highlights on Health in Portugal 1997 Copenhagen, DK, WHO 38pp
Instituto Nacional de Estatística Resultados definitivos: As causas de morte em Portugal 2000 2002 Portugal, INE 2pp
Ramos E Lopes C Barros H Investigating the effect of nonparticipation using a population based case control study on myocardial infarction Ann Epidemiol 2004 14 437 441 15246333 10.1016/j.annepidem.2003.09.013
Lopes CMM Alimentação e enfarte agudo do miocárdio 2000 Portugal, University of Porto 269pp
Folstein MF Folstein SE Mchush PR " Mini mental state ". A practical method for grading the cognitive state of patients for the clinician J Psychiatr Res 1975 12 189 198 1202204 10.1016/0022-3956(75)90026-6
Santos AC Barros H Prevalence and determinants of obesity in an urban sample of Portuguese adults Public Health 2003 117 430 437 14522159 10.1016/S0033-3506(03)00139-2
Expert panel on the identification and evaluation and treatment of overweight adults Clinical guidelines on the identification, evaluation and treatment of overweight and obesity in adults: executive summary Am J Clin Nutr 1998 68 899 917 9771869
World Health Organisation Guidelines for controlling and monitoring the tobacco epidemic 1997 Geneva, Sw, WHO 190pp
Ainsworth BE Haskell WL Leon AS Jacobs DR Montoye HJ Sallis JF Paffenbarger RSJ Compendium of Physical Activities: classification of energy costs of human physical activities Med Sci Sports Exerc 1993 71 79 8292105
Bernstein MS Morabia A Sloutskis D Definition and prevalence of sedentarism in an urban population Am J Public Health 1999 89 862 867 10358676
Varo JJ Martínez González AM de Irala Estévez J Kearney J Gibney M Martínez JA Distribution and determinants of sedentary lifestyles in the European Union Int J Epidemiol 2003 32 138 146 12690026 10.1093/ije/dyg116
Bernstein MS Costanza M Morabia A Physical activity of urban adults: a general population survey in Geneva Soz Präventivemed 2001 46 49 59
Pols MA Peeters PH Ocke MC Bueno-de-Mesquita HB Slimani N Kemper HC Collette HJ Relative validity and repeatability of a new questionnaire on physical activity Prev Med 1997 26 37 43 9010896 10.1006/pmed.1996.9995
Duncan GE Sydeman SJ Perri MG Limacher MC Martin AD Can sedentary adults accurately recall the intensity of their physical activity? Prev Med 2001 33 18 26 11482992 10.1006/pmed.2001.0847
Physical activity and health. A report of the Surgeon General 1996 Atlanta, Georgia, US Dept of Health and Human Services 278pp
Salmon J Owen N Bauman A Schmitz KH Booth M Leisure-time, occupational and household physical activity among professional, skilled and less-skilled workers and homemakers Prev Med 2000 30 191 199 10684742 10.1006/pmed.1999.0619
Giannuzzi P Mezzani A Saner H Björnstad H Fioretti P Mendes M al. Physical activity for primary and secondary prevention: Position paper of the Working Group on Cardiac Rehabilitation and Exercise Physiology of the European Society of Cardiology Eur J Cardiovasc Prev Rehabil 2003 10 319 327 14663293 10.1097/01.hjr.0000086303.28200.50
Oldridge N Physical activity in primary and secondary prevention there is a treatment gap Eur J Cardiovasc Prev Rehabil 2003 10 317 318 14663292 10.1097/01.hjr.0000099030.73419.67
Steptoe A Rink E Kerry S Psychosocial predictors of changes in physical activity in overweight sedentary adults following counselling in primary care Prev Med 2000 31 183 194 10938220 10.1006/pmed.2000.0688
Marshall AL Bauman AE Owen N Booth ML Crawford D Marcus BH Reaching out to promote physical activity in Australia: A state wide randomised controlled trial of a stage targeted intervention Health Promotion 2004 18 283 287 15011926
| 15713230 | PMC550648 | CC BY | 2021-01-04 16:28:55 | no | BMC Public Health. 2005 Feb 15; 5:16 | utf-8 | BMC Public Health | 2,005 | 10.1186/1471-2458-5-16 | oa_comm |
==== Front
BMC PhysiolBMC Physiology1472-6793BioMed Central London 1472-6793-5-41571003610.1186/1472-6793-5-4Research ArticleComparative analysis of mouse skeletal muscle fibre type composition and contractile responses to calcium channel blocker Mänttäri Satu [email protected]ärvilehto Matti [email protected] Department of Biology, University of Oulu, Oulu, FIN-90014, Finland2005 14 2 2005 5 4 4 24 5 2004 14 2 2005 Copyright © 2005 Mänttäri and Järvilehto; 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 this study, we examined the correlation between excitation-contraction coupling characteristics and skeletal muscle fibre type by (1) localizing the distribution of dihydropyridine receptor (DHPR) protein and (2) comparing the effect of DHPR blocker on muscles with different fibre type composition, in order to better understand the differences between contractile phenotypes of fibres and to explain the contradictory reports to date on the interaction of dihydropyridines with skeletal muscle isoform of DHPR.
Results
Histochemical experiments revealed that fluorophore conjugated dihydropyridines stain selectively the membranes of muscle fibres. The staining was most evident in type IIA fibres. The major fibre type in gluteus and femoris, revealed by mATPase staining, was IIA (45.0 and 38.1 %, respectively). In gastrocnemius the content of IIA fibres was 22.7 %. Contraction forces before and after the addition of blocker for the three muscles investigated were: gluteus 0.075 ± 0.017 N vs. 0.052 ± 0.011 N, femoris 0.045 ± 0.005 N vs. 0.033 ± 0.005 N and gastrocnemius 0.089 ± 0.016 N vs. 0.075 ± 0.014 N, respectively. The attenuation of contraction force proportional to the cross-sectional area of the muscle was significantly (P = 0.023) higher in gluteus (28.3 ± 3.5 %) and femoris (27.6 ± 3.2 %) as compared to gastrocnemius (16.1 ± 2.5 %). However, no significant change in the control measurements was observed ruling out the possibility of fatigue.
Conclusion
The results indicate that the attenuation of the contraction force was largest in muscles with a high percentage of type IIA fibres. This supports our finding that the abundance of dihydropyridine receptors of IIA fibres outnumbers that in the other fibre types. The present data show that the correlation of density of dihydropyridine receptors can be one of the important factors influencing the overall contractile properties of the muscle and for its part explain the contradictory results of previous studies on coupling process.
==== Body
Background
One of the features characterizing mammalian skeletal muscle tissue is the structural variability of muscle fibres. The inclusive properties of skeletal muscle lead to functional diversity of muscle fibres, which has been related to differences in relative proportions of membrane structures and to different amounts of contractile proteins [1]. Although the interfibre-type differences are well recognized, very little is known about the characteristics of the process leading from electrical membrane excitation to contraction (E-C coupling) related to different cell types. The first indication of variation in E-C coupling between representatives of different fibre types came from studies using vaseline-gap technique in order to record slow calcium currents and asymmetric charge movement in fast- and slow-twitch muscles [2]. The results suggested functional and structural differences between fast- and slow-twitch mammalian muscles with respect to dihydropyridine receptor (DHPR) density. Recently Goodman et al [3] provided the first evidence about E-C coupling characteristics being related to myosin heavy chain based fibre type and the muscle from which the fibre originated. Our finding about the density of DHPR varying between different fibre types [4] supports the hypothesis that optimum contractile function of skeletal muscle is related to its fibre type composition via differences in E-C coupling.
Voltage-dependent dihydropyridine receptors play an important role in the function of skeletal muscle. As well known, contraction is initiated by a depolarization of the transverse tubular (T-tubular) membrane, which in turn causes calcium to be released from the stores of sarcoplasmic reticulum (SR). The DHPR is located in the T-tubule membrane acting as a voltage sensor. Linked to the ryanodine receptor located in the sarcoplasmic reticulum membrane, this channel protein triggers the intracellular Ca2+-release via the ryanodine receptor [5]. The DHPR itself is a heteromultimer composed of 5 subunits; α1 (170 kDa), α2 (140 kDa), β (55 kDa), γ (33 kDa) and δ (24–33 kDa) [6,7]. The pore forming α1 subunit contains the receptor for dihydropyridines (DHPs) and other calcium antagonists. DHPs, such as nifedipine, nitrendipine, and nimodipine, bind to DHP receptor in a specific, saturable and reversible fashion. In addition, receptors show high affinity for DHP ligands. The dissociation constant (KD) for skeletal muscle membranes is usually in the range of 1–10 nM [8]. Dihydropyridines are used clinically as blockers in the treatment of hypertension and coronary heart disease [9].
Besides being the voltage sensor for Ca2+ release needed in E-C coupling, the α1-subunit of the molecule functions also as a calcium channel [10] giving rise to a slow inward calcium current [11]. However, DHPR appears to be unimportant as an ion-conducting channel [12]. Firstly, the calcium current is activated too slowly to generate contraction of the fibre. Second, the contraction continues even if the extracellular calcium is removed, and finally, blocking the slow calcium current through the channel does not prevent contraction. On the other hand, Beam et al [5] reported a notable increase in the magnitude of inward calcium current in developing muscle thus implicating an important role of the current in maturation.
In the present study we examined the E-C coupling characteristics of muscles with different dihydropyridine receptor expression defined by fibre type composition. To test the hypothesis that muscles with different cell type compositions, i.e. DHPR quantities, react differently to calcium channel blocker, we measured contraction forces of three muscles before and after adding the blocker solution. We found variation in the activities of muscles depending on the cell type composition of the muscles. The findings confirm the previously observed pattern and elucidate the role of dihydropyridine receptor in muscle cell contraction.
Results
Myosin heavy chain analysis and fibre typing
In order to analyse the distribution of cell types in the gluteus maximus (GLU), rectus femoris (RF) and gastrocnemius (GAS) muscles, tissue samples were collected and subjected to SDS-PAGE. The different muscles of mouse contained four myosin heavy chain (MHC) isoforms in the increasing electrophoretic mobility, MHCIIa, MHCIId, MHCIIb and MHCI (Fig. 1) as compared to those of rat gastrocnemius muscle (in this report MHC isoform expressed in fibre is identified by roman numeral and lower case letter, while the fibre type is identified by a roman numeral and a corresponding capital letter according to Hämäläinen and Pette [13]). The distribution of these myosin heavy chain isoforms differed among the three muscles studied (Table 1). In GAS, the major isoform was MHCIId (35.6 %), expressed together with smaller amounts of IIb (28.1 %). On the other hand, the quantity of MHCIIa was smaller (20.6 %). In contrast to GAS, both GLU and RF were characterized by higher contents of MHCIIa (26.6 % and 27.7 %, respectively), together with MHCd (34.3 %; 23.7 %) and lower amount of MHCIIb (15.9 %; 24.3 %). All the muscles contained also portions of MHCI (GAS: 15.7 %, GLU: 23.2 %, RF: 24.3 %). A statistically significant difference (P < 0.05) was found in the portion of IIb between GAS and GLU as well as in IId between GAS and RF. Four fibre types were also identified in muscle sections stained with mATPase (Table 1). The major fibre type in GAS was IID (51.5 %). However, in contrast to the results from SDS-PAGE, IIA fibre type (22.7 %) outnumbered the IIB (18.2 %). The most frequent fibre type in both GLU and RF was IIA (45.0 %; 38.1 %), followed by IIB (31.1 %) in GLU and IID (21.6 %) in RF. There was a statistically significant difference (P < 0.05) in the portion of IIA between GLU and RF. In addition, between GAS and RF a significant difference was found in the proportion of I (P < 0.05) and IID (P < 0.01). Between GAS and GLU a significant difference was found in proportion of IIA (P < 0.05) and IID (P < 0.01). To summarise, the significant difference in proportion of IIA between GAS and RF/GLU is particularly relevant to the present study.
DHPR expression and correlation to fibre types
We have previously shown that the density of dihydropyridine receptors (L-type calcium channels) in T-tubule membranes increases markedly during the postnatal development of mouse skeletal muscle [4]. Furthermore, the findings showed that the fast oxidative glycolytic (FOG) fibre type has the most evident appearance of DHPRs. Fig. 2A shows localization of DHPR in mouse RF muscle. DHPR is strongly expressed in some muscle fibres whereas weakly expressed at the membrane of others. The corresponding control slide, preincubated with nifedipine resulted in a loss of staining (Fig. 2C). To correlate the DHPR expression with muscle fibre typing, histochemical experiments were performed as described. According to the staining intensity, IIA fibre type corresponds to the muscle fibres that strongly express DHPRs (Fig. 2B).
Force measurements
GLU, RF and GAS muscles were weighed and the length and diameter were measured. Table 2 shows some morphological details for the muscles studied. To reveal the function of L-type calcium channels in muscles with different kinds of cell type distributions, the single twitch forces of the muscles were measured before and after the addition of a specific channel blocker nifedipine. Maximum contraction forces in proportion to the cross-sectional area of the muscle, and the average responses to 1 μM nifedipine solution are shown in Fig. 3. The force production of GAS (0.089 ± 0.016 N) was considerably higher in comparison to the ones of GLU (0.075 ± 0.017 N) and RF (0.045 ± 0.005 N). The effect of nifedipine on contractile force was inhibiting in all the muscles studied. The in vitro nifedipine effect on the muscle force production was determined as the percentage of the remaining channel mediated contraction force. The weakening of contraction force proportional to the cross-sectional areas of the muscles was significantly (P = 0.023) higher in GLU (28.3 ± 3.5 %) and RF (27.6 ± 3.2 %) in comparison to GAS (16.1 ± 2.5 %). Experiments performed in increasing concentrations of nifedipine resulted in a clear dose-response curve (Fig. 4) and revealed a decreasing force output as the blocker concentration increased. Furthermore, the decrease was significantly smaller in GAS (P < 0.05 at nifedipine concentration of 5 μM), in comparison to the decrease in GLU.
Discussion
The cell type composition of a muscle is one of the features characterizing the function of the muscle. Type I fibres are most prevalent in muscles involved in the maintenance of posture, whereas type II fibres are used for movements which require high power output. Many vertebrate locomotor muscles are composed of a mixture of fibre types. Additionally, one muscle fibre may contain several myosin heavy chain types indicating heterogeneity at fibril level [14]. The m. gastrocnemius (whole) used in this study has a high type IIB content whereas both m. rectus femoris and m. gluteus maximus are predominantly composed of types IIA and IID. The results from myosin heavy chain analysis and fibre typing are somewhat different. The identification of four different fibre types using histochemical methods is difficult because of the presence of several types of myosin heavy chains within one single fibre. Therefore, the MHC analysis is more qualitative. The correlation of high density of DHPRs with IIA fibres was stated on the basis of ATPase activity. This was previously also shown by statistical analysis where fibre size was additionally taken into account [4].
Nifedipine was used as an antagonist in order to specifically block the dihydropyridine receptors. Nifedipine is a dihydropyridine derivative that binds in a specific, stereoselective and saturable fashion to DHPR. The selective calcium channel inhibitor prevents the influx of extracellular calcium through the L-type calcium channel [15] and also has a clear effect on the contraction activity of the skeletal muscle fibre. The effects of nifedipine on depolarization-induced force responses are inhibitory and dose dependent [16]. As the half life of the blocker is 2 – 6 h, there is enough time to perform reliable measurements.
By blocking the channel with nifedipine, a clear selective decrease in the contraction force of the different muscles studied was observed (Fig. 3). The inhibition percentages of RF and GLU were, however, significantly higher as compared to that of GAS. This data indicates that the same concentration of blocker causes a different response between different muscles. A similar variability is noted when compared to the cell type composition of the muscles. The muscles with high IIA fibre type content have strongly reduced contraction force as a result of addition of a calcium channel blocker. On the other hand, the effect of a blocker on the muscle with a lower IIA type content is much weaker. The findings are attributable to the varying densities of dihydropyridine receptors in muscles. In GAS, the amount of DHPRs is reduced, likely due to the low IIA fibre type content. Thus the blocking effect of a receptor antagonist is weaker as compared to the muscles with a higher IIA content. Furthermore, the response of GAS and GLU to the gradual saturation of DHPRs with nifedipine molecules is different (Fig. 4). Consistently, the inhibition percentage of the contraction force of GAS decreased less as compared to GLU. In addition, a constant level of the contraction force was reached sooner in GAS indicating a complete saturation of the receptors available.
There are several examples of the divergent behaviour of the dihydropyridine receptor in skeletal muscle. First, dihydropyridines are shown to have both stimulatory [17] and inhibitory [18] effects on excitation-contraction coupling. Furthermore, unlike in cardiac muscle, calcium release from SR does not require inward current, and as yet is induced by blockade of dihydropyridine receptors [19]. However, it is clear that DHPRs are essential in excitation-contraction coupling since animals with mdg/mdg mutation, which results in a lack of receptor proteins, die at birth because of paralysis of the respiratory muscles [20]. On the other hand, when the α1-subunit is introduced into the nuclei of the dysgenic myotubes, some cells contract upon an electrical stimulation. Despite of this, the recovered influx of calcium ions is not necessary for the contraction since the cells are contracting also even if the current is blocked with cadmium. Hence calcium antagonist drugs seem to have very few pharmacologically relevant actions on skeletal muscle as observed also by Walsh et al [21]. The role of inward current is still an open question although it has been suggested that the current is needed to maintain the calcium stores inside the cell [5] or for the conditions of activeness of the voltage sensor [22].
In the present study we report one plausible explanation for the large variation of the results observed in the behaviour of the L-type calcium channels in skeletal muscle. The uneven density distribution of dihydropyridine receptors indicates a difference in E-C coupling machinery between muscle fibre types. In this study, we showed that the difference is also detectable in the contraction forces between muscles of different cell type composition. Although nifedipine specifically blocks the dihydropyridine receptor, there could also be other differences in addition to the density of receptors between the muscles causing the difference in the contraction forces. A larger number of different muscles and developmental stages might clarify the differences in the present results.
Goodman et al [3] provided the first evidence that E-C coupling characteristics are related to fibre types based on myosin heavy chain. By measuring depolarization-induced force responses of skinned single fibres of rat, they concluded that the optimum force production of a skeletal muscle is related to its MHC isoform composition. Although we found no published data on the E-C coupling phenotype of MHC IIa isoform, the results from previous studies suggest that the parameters describing force and velocity properties of a single muscle fibre are significantly higher in the fast, type IIA and IIB, fibres than in slow, type I fibres.
Conclusions
Taken together, our results obtained from three different muscles confirmed that the expression of dihydropyridine receptor is a fibre type specific character. In addition, the present data indicate that the density of DHPRs is one important factor influencing the overall contractile properties of the muscle. Furthermore, the results of our experiments point out that when determining the physiological relevance of DHPRs, it is necessary to compare histochemistry and protein analysis to relevant functional properties such as contraction of the muscle. Moreover, it is noted, that differences between developing and adult type of muscle cells may emerge due to the differentiation of phenotypes of muscle fibres. In the future it might be useful to examine the correlation from a larger number of different muscles, and from muscles of different developmental stage.
Methods
Electrophoretic separation of myosin heavy chain isoforms
M. rectus femoris (N = 6), m. gastrocnemius (N = 6) and m. gluteus maximus (N = 6) from adult mice (strain CD-1) were removed after the animals were killed by paracervical dislocation in accordance with the Animal Ethics Committee of the University of Oulu (licence no. 046/03). The muscles were homogenized in 6 vol of homogenization buffer [62.5 mM Tris-HCl, pH 6.8] and boiled in sample buffer as previously described for 5 min at a final protein concentration of 0.5 mg ml-1 [23]. Total protein was assayed according to the method of Bradford [24]. Myosin heavy chain isoforms were separated by gradient (5–8 %) sodiumdodecylsulfate polyacrylamide gel electrophoresis performed at 5°C for 23 h (120 V constant voltage). The gels were stained with Coomassie Brilliant Blue. The stained gels were scanned and the separated protein bands were analysed with FluorS MultiImager program (Bio-Rad, USA).
Fibre typing
Serial cross-sections, 8 μm thick, were cut on a cryostat microtome at -25°C, mounted on cover slips and stained for myosinATPase (mATPase) with acid preincubation according to Hämäläinen and Pette [25]. The sections were preincubated at room temperature for 7 min in sodium acetate (54.3 mM) – sodium barbital (32.6 mM) solution adjusted with HCl to pH 4.6. After washing with CaCl2 (18 mM) and Tris-HCl (199 mM) the sections were incubated in substrate solution (4.5 mM ATP, 19.5 mM CaCl2, 116 mM 2-amino-2-methyl-1-propanol; pH 9.4) at room temperature for 45 min. After incubations in 11 mM CaCl2, 2% CoCl2, and 10 mM sodium barbital, the colour was developed in 2% (v/v) ammonium sulphide for 45 s. After washing with distilled water, the sections were dehydrated in ethanol, cleared in xylene and mounted with DPX. An alkaline preincubation in a solution containing 34 mM 2-amino-2-methyl-1-propanol, 120 mM CaCl2, adjusted to pH 10.3 with HCl was also used as previously described by Guth and Samaha [26]. After preincubation the sections were processed as above. The percentage of each fibre type in different muscles was calculated with the use of the LSM 5 PASCAL software 3.2 (Leo, Germany) for analyzing images.
Fluorescence staining
For fluorescence labelling, RF muscle from mouse was cut into 8 μm cryostat sections, dried for 30 min and treated as previously described by Mänttäri et al [4]. The concentration of the high affinity (-)-enantiomer of dihydropyridine was 20 nM. Control samples were preincubated with 10 μM nifedipine in the phosphate buffer for 10 min prior to the addition of dihydropyridine conjugate.
Force measurements
Adult mice (weighing 23 – 45 g) were killed by paracervical dislocation, and the pelvic region and right hind limb were skinned. The muscle of interest (m. rectus femoris, m. gastrocnemius or m. gluteus maximus) was dissected out and suspended between a fixed clamp at the base of an organ bath and a Grass FTO3C (USA) force-displacement transducer. Muscles were maintained in an oxygenated (95% O2 – 5% CO2) physiological saline solution (+37°C, pH 7.4) containing 137.0 mM NaCl, 2.7 mM KCl, 1.0 mM MgSO4· 7H2O, 1.8 mM CaCl2, 0.4 mM NaH2PO4, 12 mM NaHCO3 and 5.5 mM glucose. After an equilibration period of 2 min, the muscles were supramaximally stimulated using steel electrode, pulses of 1 ms duration at 80 V (model S44, Grass Instrument Co.) and muscle length selected to elicit maximal twitch force. The maximum force of the muscle was measured in series of three stimulus impulses with one minute equilibration time between the stimulations. Recordings of transducer output were A/D converted and collected on a computer at a sample frequency of 1000 Hz.
In order to examine the role of a specific, L-type calcium channel blocker, the muscles first measured in saline solution were mounted in a bath containing 1 μM nifedipine [1,4-dihydro-2,6-dimethyl-4-(2-nitrophenyl)-3,5-pyridinedicarboxylic acid dimethyl ester], equilibrated for 2 min, and assessed as described above i.e. the muscle was a control in itself. In addition, series of measurements were made with m. gastrocnemius (N = 6) and m. gluteus (N = 6) using nifedipine concentration range of 2 – 30 μM. The concentration of the drug was progressively increased by adding the nifedipine solution to the saline solution bath. During the experiment, the nifedipine solutions were maintained in dark to prevent photo-bleaching. To exclude the possibility of fatigue caused by successive stimulations, the protocol control measurement was performed similarly devoid of nifedipine.
Statistical analysis
The statistical significance of differences between means of contraction forces before and after the addition of calcium channel blocker was evaluated by paired samples t-test. The difference in parameters between the different types of muscle were analysed with independent samples t-test for equality of means. A P value of < 0.05 was accepted as indicative of a significant difference between the two sets of observations. The significance of differences between the muscles was evaluated with one way analysis of variance (normality test passed). The significant difference was stated as mentioned above. The values are presented as the means ± SE. All the statistical analyses were performed with the SPSS for Windows software.
List of abbreviations used
DHP; dihydropyridine
DHPR; dihydropyridine receptor
E-C coupling; excitation-contraction coupling
FOG; fast oxidative glycolytic
GAS; musculus gastrocnemius
GLU; musculus gluteus maximus
mATPase; myosin adenosine triphosphatase
MHC; myosin heavy chain
RF; musculus rectus femoris
SDS-PAGE; sodium dodecyl sulphate polyacrylamide gel electrophoresis
SR; sarcoplasmatic reticulum
T-tubule; transverse tubule
IIa, IIb, IId; myosin heavy chain type
IIA, IIB, IID; fibre type
Authors' contributions
SM carried out most of the experiments, assisted in interpretation of results, and participated in writing the manuscript. MJ conceived of the study, participated in its design and coordination, and participated in writing the manuscript.
Acknowledgements
The authors thank MSc Tuula Korhonen for her excellent technical assistance. We would also like to thank Drs Esa Hohtola and Ahti Pyörnilä for their technical support and valuable comments on the manuscript.
Figures and Tables
Figure 1 Electrophoretic separation of myosin heavy chain isoforms. Myosin isoforms of mouse muscles were compared to those of rat gastrocnemius muscle (lane 1). Densitogram of each lane is shown above the gel. The muscles analysed were m. gastrocnemius (lane 2), m. gluteus (lane 3) and m. rectus femoris (lane 4).
Figure 2 Characterization of fibre type specificity of dihydropyridine receptor Two subsequent sections of mouse musculus gastrocnemius were assayed for DHPR localization by fluorescence staining (A) and for fibre type analysis by staining for mATPase (B). After preincubation in pH 4.6, type IIA displays a relatively low acid stability and is stained most lightly (*). The corresponding fibre in (A) is also recognized by fluorophore conjugated DHP blocker. The fibre on the right of the asterisked fibre shows also fluorescence. The intensity is, however, somewhat lower than in the fibre next to it and originates probably from the special fibre type (hybrid, type IIDA). (C) A control sample preincubated with nifedipine. Bar 20 μm.
Figure 3 Effect of nifedipine on contraction force. Maximum contraction forces of GAS, RF and GLU in proportion to the cross-sectional area of the muscle (white), and the average responses to 1 μM nifedipine solution (grey) are shown. The fatigue control was performed with GAS, RF and GLU in physiological salt solution devoid of nifedipine in order to control the procedure and to rule out the possibility of fatigue. The statistical significance of differences between means of contraction forces before and after the addition of nifedipine solution is shown by levels (*, P < 0.05; ** P < 0.01, paired samples t-test).
Figure 4 A force-concentration curve for GAS (Δ) and GLU (×) GLU was also used as a saline standard (○) in order to rule out the possibility of fatigue. The x-axis displays the increase of nifedipine concentration and the y-axis shows the relative force response of the muscles in % of the maximum initial value. Values are given as means ± SE (dashed lines for GAS). The significance of the decrement of contraction force in proportion to the cross-sectional area of the muscle is shown by levels (*, P < 0.05; ** P < 0.01, paired samples t-test) in GLU at nifedipine concentration of 5 μM onwards. N = 6 in each measurement.
Table 1 The proportion of fibre types in different muscles. The fibre type composition of m. rectus femoris, m. gastrocnemius and m. gluteus maximus were determined by SDS-PAGE based on electrophoretic separation of myosin heavy chain isoforms, and staining for myofibrillar ATPase after acid and alkaline preincubation. IIa, IIb, IId; myosin heavy chain types, IIA, IIB, IID; fibre types. Statistical significance of the differences between the corresponding groups (I vs. I, IIa vs. IIA etc.) from the two methods used: NS non significant; * P < 0.05; ** P < 0.01; *** P < 0.001.
muscle fibre populations % (SE)
SDS-PAGE m. rectus femoris m. gastrocnemius m. gluteus maximus
I 24.3 (3.0) 15.7 (4.6) 23.2 (3.0)
IIa 27.7 (1.1) 20.6 (3.2) 26.6 (2.9)
IIb 24.3 (2.3) 28.1 (3.1) 15.9 (1.4)
IId 23.7 (1.7) 35.6 (2.2) 34.3 (1.3)
total 100.0 100.0 100.0
mATPase
I 21.1 (2.3) NS 7.6 (1.3) ** 8.0 (2.6) **
IIA 38.1 (1.5) *** 22.7 (3.8) NS 45.0 (0.8) ***
IIB 19.2 (2.3) * 18.2 (1.8) ** 31.1 (3.6) **
IID 21.6 (1.1) NS 51.5 (3.1) *** 15.9 (1.6) ***
total 100.0 100.0 100.0
Table 2 A selection of morphological features of different muscles.
mass size
muscle N body (g) muscle (g) length (mm) diameter (mm)
m. rectus femoris (SE) 8 32.13 (1.64) 0.12 (0.01) 11.10 (0.23) 5.20 (0.20)
m. gastrocnemius (SE) 12 31.25 (2.18) 0.15 (0.01) 11.41 (0.31) 5.95 (0.19)
m. gluteus maximus (SE) 13 27.50 (1.51) 0.11 (0.01) 12.70 (0.30) 6.61 (0.24)
==== Refs
Punkt K Fibre types in skeletal muscles Adv Anat Embryol Cell Biol 2002 162 1 109
Lamb GD Walsh T Calcium currents, charge movement and dihydropyridine binding in fast- and slow-twitch muscles of rat and rabbit J Physiol 1987 393 595 617 2451745
Goodman C Patterson M Stephenson G MHC-based fiber type and E-C coupling characteristics in mechanically skinned muscle fibers of the rat Am J Physiol Cell Physiol 2003 284 C1448 C1459 12734106
Mänttäri S Pyörnilä A Harjula R Järvilehto M Expression of L-type calcium channels associated with postnatal development of skeletal muscle function in mouse J Muscle Res Cell Mot 2001 22 61 67 10.1023/A:1010305421661
Beam KG Tanabe T Numa S Structure, function and regulation of skeletal muscle dihydropyridine receptor Ann NY Acad Sci 1989 560 127 137 2545129
Curtis BM Catterall WA Purification of the calcium antagonist receptor of the voltage-sensitive calcium channel from skeletal muscle transverse tubules Biochemistry 1984 23 2113 2118 6329263
Leung AT Imagawa T Campbell KP Structural characterization of the 1,4-dihydropyridine receptor of the voltage dependent Ca2+-channel from rabbit skeletal muscle. Evidence for two distinct high molecular weight subunits J Biol Chem 1987 626 7943 7946
Gould RJ Murphy KMM Snyder SH Tissue heterogeneity of calcium channel antagonist binding sites labelled by [3H] nitrendipine Mol Pharmacol 1984 25 235 241 6422256
Triggle DJ Calcium-channel antagonists: mechanisms of action, vascular selectivities, and clinical relevance Cleve Clin J Med 1992 59 617 627 1330367
Catterall W Structure and modulation of Na+- and Ca2+-channels Ann NY Acad Sci 1993 707 1 19 9137538
Sanches JA Stefani E Inward calcium current in twitch muscle fibres of the frog J Physiol 1978 283 197 209 309941
Gonzales-Serratos H Valle-Aguilera R Slow inward calcium currents have no obvious role in muscle excitation-contraction coupling Nature 1982 298 292 294 6806669 10.1038/298292a0
Hämäläinen N Pette D Patterns of myosin isoforms in mammalian skeletal muscle fibres Microsc Res Tech 1995 30 381 389 7787237
Pette D Staron R Transitions of muscle fiber phenotypic profiles Histochem Cell Biol 2001 115 359 372 11449884
Rios E Brum G Involvement of dihydropyridine receptors in excitation-contraction coupling in skeletal muscle Nature 1987 325 717 720 2434854 10.1038/325717a0
Posterino GS Lamb GD Effect of nifedipine on depolarization-induced force responses in skinned skeletal muscle fibres of rat and toad J Muscle Res Cell Mot 1998 19 53 65
Kitamura N Ohta T Ito S Nakazato Y Effects of nifedipine and Bay K 8644 on contractile activities in single skeletal muscle fibers of the frog Eur J Pharmacol 1994 256 169 176 7519558 10.1016/0014-2999(94)90242-9
Neuhaus R Rosenthal R Lüttgau HCh The effects of dihydropyridine derivatives on force and Ca2+ current in frog skeletal muscle fibres J Physiol 1990 427 187 209 2170635
Weigl LG Hohenegger M Kress HG Dihydropyridine-induced Ca2+ release from ryanodinesensitive Ca2+ pools in human skeletal muscle cells J Physiol 2000 525 461 469 10835047 10.1111/j.1469-7793.2000.t01-1-00461.x
Tanabe T Beam K Powell JA Numa S Restoration of excitation-contraction coupling and slow calcium current in dysgenic muscle by dihydropyridine receptor complementary DNA Nature 1988 336 134 139 2903448 10.1038/336134a0
Walsh KB Bryant SH Schwartz A Effect of calcium antagonist drugs on calcium currents in mammalian skeletal muscle fibers J Pharmacol Exp Ther 1986 236 403 407 2418195
Melzer W Herrmann-Frank A Lüttgau HCh The role of Ca2+ ions in excitation-contraction coupling of skeletal muscle fibres Biochim Biophys Acta 1994 1241 59 116
Laemmli UK Cleavage of structural proteins during assembly of the head of bacteriophage T4 Nature 1970 227 680 685 5432063
Bradford M A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal Biochem 1976 72 248 254 942051
Hämäläinen N Pette D The histochemical profiles of fast fiber types IIB, IID, and IIA in skeletal muscles of mouse, rat, and rabbit J Histochem Cytochem 1993 41 733 743 8468455
Guth L Samaha FJ Procedure for the histochemical demonstration of actomyosin ATPase Exp Neurol 1970 28 365 367 4248172 10.1016/0014-4886(70)90244-X
| 15710036 | PMC550649 | CC BY | 2021-01-04 16:03:51 | no | BMC Physiol. 2005 Feb 14; 5:4 | utf-8 | BMC Physiol | 2,005 | 10.1186/1472-6793-5-4 | oa_comm |
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-151570116610.1186/1471-2458-5-15Research ArticleCardiovascular comorbidities among public health clinic patients with diabetes: the Urban Diabetics Study Robbins Jessica M [email protected] David A [email protected] Christopher N [email protected] Philadelphia Department of Public Health, 500 South Broad Street, Philadelphia, PA, USA2 Jefferson Medical College, Philadelphia, PA, USA2005 8 2 2005 5 15 15 2 9 2004 8 2 2005 Copyright © 2005 Robbins et al; licensee BioMed Central Ltd.2005Robbins 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
We sought to determine the frequency and distribution of cardiovascular comorbidities in a large cohort of low-income patients with diabetes who had received primary care for diabetes at municipal health clinics.
Methods
Outpatient data from the Philadelphia Health Care Centers was linked with hospital discharge data from all Pennsylvania hospitals and death certificates.
Results
Among 10,095 primary care patients with diabetes, with a mean observation period of 4.6 years (2.8 after diabetes diagnosis), 2,693 (14.3%) were diagnosed with heart disease, including 270 (1.4%) with myocardial infarction and 912 (4.8%) with congestive heart failure. Cerebrovascular disease was diagnosed in 588 patients (3.1%). Over 77% of diabetic patients were diagnosed with hypertension. Incidence rates of new complications ranged from 0.6 per 100 person years for myocardial infarction to 26.5 per 100 person years for hypertension. Non-Hispanic whites had higher rates of myocardial infarction, and Hispanics and Asians had fewer comorbid conditions than African Americans and non-Hispanic whites.
Conclusion
Cardiovascular comorbidities were common both before and after diabetes diagnosis in this low-income cohort, but not substantially different from mixed-income managed care populations, perhaps as a consequence of access to primary care and pharmacy services.
==== Body
Background
Patients with diabetes are at increased risk of a wide range of complications and comorbidities, which adversely affect quality of life, mortality, and health care services utilization. Cardiovascular diseases are the primary causes of morbidity and mortality among patients with diabetes, yet data on cardiovascular disease among patients with diabetes are limited [1]. Although microvascular pathologies, including retinopathy and nephropathy, have been shown to be strongly associated with glycemic control, macrovascular complications, including heart disease and cerebrovascular disease, appear to be less responsive to glycemic control, and strategies to reduce them are focussed on controlling other risk factors such as hypertension, hyperlipidemia, smoking, and obesity [2]. This complex array of risk factors and outcomes makes diabetes care exceptionally demanding both for patients and for health care providers.
Data on the prevalence and incidence of macrovascular comorbidities are largely derived either from clinical trials with selected patient populations, self-reported cross-sectional survey data, or managed-care studies of insured patients of higher socioeconomic status. The rising prevalence of diabetes in low-income inner-city populations underscores the need to understand the frequency with which these comorbidities occur in such populations and to assess possible disparities in their occurrence.
The Philadelphia Health Care Centers (HCCs) provide primary care and pharmacy services to approximately 100,000 patients annually through eight neighborhood centers and one sexually transmitted disease clinic. Individual patients are not billed for any services provided. Although the HCCs are available to all Philadelphia residents, their patients are almost universally low-income; over 60% have no health insurance. A majority of HCC patients are African American, but there are also substantial numbers of non-Hispanic white, Hispanic, and Asian patients. Diabetes is one of the most common diagnoses among adult HCC patients. Since to our knowledge no previous study has reported the prevalence or incidence of cardiovascular diseases in a comparable, ethnically diverse, low-income diabetic population, we sought to determine the frequency and distribution of cardiovascular comorbidities among HCC patients with diabetes.
Methods
The Urban Diabetics Study cohort consists of 10,095 patients diagnosed with diabetes in the HCCs between January 1, 1996 and December 31, 2001. All diabetic patients were included except those for whom Social Security numbers were unavailable (less than 6% of all diabetic patients). All HCC visits between March 1, 1993 and December 31, 2001, all hospital discharges from any Pennsylvania hospital between January 1, 1993 and December 31, 2001, and death records have been linked for these patients. The data includes outpatient visits and hospitalizations both before and after the initial diabetes diagnosis, within the study time period. Race/ethnicity was classified based on the outpatient records. Diabetes incidence was calculated based on the admission date of the first hospitalization for which a diabetes diagnosis was recorded or the date of the first HCC outpatient visit for which a diabetes diagnosis was recorded, whichever came first.
The cardiovascular comorbidities assessed included any heart disease (ICD9 codes 402 and 410–429, ICD10 codes I5-I9, I11, I13, I20-I27, and I30-I52), myocardial infarction (MI) (ICD9 code 410, ICD10 codes I21-I23), congestive heart failure (CHF) (ICD9 code 428, ICD10 code I50), cerebrovascular disease (ICD9 codes 430–438, ICD10 code I6), and hypertension (ICD9 codes 401–405, ICD10 codes I10-I15). For each comorbid condition we assessed baseline prevalence at or before first diabetes diagnosis, incidence rate per 100 person-years of follow-up time following first diabetes diagnosis among those free of the condition at baseline, and cumulative prevalence, i.e., any occurrence over the course of the observation period. Logistic regression and proportional hazards regression were used to simultaneously assess multiple demographic risk factors, including sex, race/ethnicity, exact age, and year of diabetes incidence. Patients with unknown race/ethnicity (n = 7) were excluded from multivariable analyses.
Results
Demographic characteristics of the 10,095 patients included are shown in Table 1. Mean age at diabetes diagnosis was 49.2 years, and mean observation period was 4.59 years, including 1.74 years before and 2.85 years after first known diabetes diagnosis. Among the patients for whom income was ascertained, median income was $7,092; 86% had annual incomes below $15,000. Table 2 shows which of the linked data sources provided diagnoses of each comorbidity.
Table 1 Demographic characteristics of Urban Diabetics Study cohort
Total African American Hispanic Non-Hispanic White Asian Other Unknown
N N N N N N N
Total 10095 7243 1119 1056 394 276 7
Females 5671 4230 528 540 224 145 4
Age
0–4 17 12 2 2 1 0 0
5–14 38 28 2 2 4 2 0
15–24 213 149 25 28 8 3 0
25–34 477 359 52 41 11 14 0
35–44 1086 826 89 97 50 23 1
45–54 1614 1214 161 151 50 35 3
55–64 1542 1139 142 146 72 41 2
65–74 508 362 44 55 21 25 1
75–84 149 114 11 16 6 2 0
85+ 30 27 0 2 1 0 0
Year of Diabetes Diagnosis
1996 948 720 79 90 45 14 0
1997 969 714 78 116 40 20 1
1998 817 604 77 88 23 24 1
1999 854 637 92 74 26 25 0
2000 926 699 84 75 35 33 0
2001 1157 856 118 97 55 29 2
Males 4424 3013 591 516 170 131 3
Age
0–4 28 26 0 1 1 0 0
5–14 78 68 7 2 1 0 0
15–24 419 371 20 22 3 3 0
25–34 1174 1015 81 55 15 7 1
35–44 2567 2243 155 113 29 26 1
45–54 3183 2797 163 132 52 38 1
55–64 2807 2467 119 134 45 42 0
65–74 902 800 36 32 20 14 0
75–84 290 254 9 23 3 1 0
85+ 58 54 1 2 1 0 0
Year of Diabetes Diagnosis
1996 732 509 92 101 17 13 0
1997 743 496 96 101 27 22 1
1998 692 486 95 74 21 16 0
1999 638 423 80 84 24 27 0
2000 694 457 103 75 33 25 1
2001 925 642 125 81 48 28 1
Table 2 Data sources for cardiovascular comorbid conditions
Outpatient only Hospital only Death Certificate only Outpatient And Hospital Outpatient and Death Certificate Hospital and Death Certificate Outpatient, Hospital, and Death Certificate
Heart Disease 1125 700 38 731 9 31 59
Congestive Heart Failure 330 314 6 246 2 7 7
Myocardial Infarction 5 240 13 3 0 9 0
Cerebrovascular Disease 109 333 9 123 0 11 3
Hypertension 5248 218 0 2296 9 3 30
Heart disease
Baseline prevalence of any heart disease was 14.8 %, including 4.1% with CHF and 1.0% with MI (Table 3). In multiple regression analyses, non-Hispanic white race/ethnicity (compared to the African-American reference group) was associated with substantially higher baseline prevalence of MI (Table 4). Hispanic race/ethnicity was associated with lower baseline prevalence of CHF, and Asian race/ethnicity was associated with lower baseline prevalences of MI, CHF, and any heart disease. Age and male sex were positively associated with baseline prevalence of MI, CHF, and any heart disease. Year of diabetes diagnosis was negatively associated with baseline prevalence of MI, CHF, and any heart disease, indicating that baseline prevalences declined between 1996 and 2001.
Table 3 Baseline prevalence, incidence, and cumulative prevalence of cardiovascular comorbid conditions
Comorbidities Total Cases Cumulative Prevalence Baseline Prevalence At or Before Initial Diabetes Diagnosis Incident Cases After Diabetes Diagnosis Incidence Rate
N % N % N % of those at risk per 100 Person-Years
Heart Disease 2,693 26.7% 1,495 14.8% 1,198 13.9% 5.36
Congestive Heart Failure 912 9.0% 417 4.1% 495 5.1% 1.85
Myocardial Infarction 270 2.7% 104 1.0% 166 1.7% 0.59
Cerebrovascular Disease 588 5.8% 294 2.9% 294 3.0% 1.07
Hypertension 7,804 77.3% 5,226 51.8% 2,578 52.9% 26.46
Table 4 Associations with baseline prevalence of comorbid conditions
Odds Ratio, (95% Confidence Interval)
Male White Hispanic Asian Other Age (per 10 years) Year of Diabetes Incidence
Heart Disease 1.24 (1.10–1.39) 1.32 (1.11–1.57) 0.59 (0.48–0.73) 0.45 (0.31–0.65) 0.49 (0.32–0.74) 1.73 (1.66–1.81) 1.02 (0.99–1.05)
Congestive Heart Failure 1.33 (1.08–1.63) 0.86 (0.63–1.19) 0.28 (0.17–0.48) 0.37 (0.18–0.75) 0.32 (0.13–0.79) 1.76 (1.63–1.89) 1.06 (1.00–1.12)
Myocardial Infarction 1.36 (0.92–2.01) 1.94 (1.19–3.14) 0.52 (0.23–1.21) 0.00 (0.00->9.99) 0.31 (0.04–2.26) 1.47 (1.28–1.70) 1.03 (0.92–1.15)
Cerebrovascular Disease 1.46 (1.15–1.85) 0.90 (0.61–1.32) 0.64 (0.41–1.00) 0.56 (0.27–1.14) 0.10 (0.01–0.70) 1.89 (1.73–2.06) 1.01 (0.94–1.08)
Hypertension 0.75 (0.69–0.82) 0.69 (0.60–0.79) 0.39 (0.34–0.45) 0.32 (0.26–0.41) 0.30 (0.23–0.40) 1.85 (1.78–1.91) 1.85 (1.78–1.91)
Odds ratios are adjusted for other variables shown in the table: sex, race/ethnicity, exact year of age, and year of diabetes incidence. Reference groups for odds ratios are: Females (for Males); African Americans (for White, Hispanic, Asian, and Other); lower age; and earlier year of diabetes diagnosis.
Incidence of heart disease, among those who had not been diagnosed with it at the time of their diabetes diagnosis, was 5.36 cases per 100 person-years, with 13.9% of those at risk being diagnosed before the end of the study period (Table 3). Incidence of MI was 0.59 per 100 person-years. Incidence of CHF was 1.85 per 100 person-years. In multiple regression analyses (Table 5), Non-Hispanic whites, as compared to African-Americans, were at higher risk of MI and all heart disease. Hispanic and Asian patients were at lower risk of CHF. Incidence of heart disease increased slightly with later year of diabetes diagnosis. Age at diabetes diagnosis was associated with higher incidence of all heart disease outcomes, as was male sex.
Table 5 Risk factors for incident comorbid conditions
Hazard Ratio, (95% Confidence Interval)
Male White Hispanic Asian Other Age (per 10 years) Year of Diabetes Incidence
Heart Disease 1.34 (1.19–1.50) 1.26 (1.06–1.50) 1.00 (0.84–1.20) 0.83 (0.62–1.13) 0.63 (0.42–0.95) 1.48 (1.42–1.54) 1.07 (1.03–1.12)
Congestive Heart Failure 1.57 (1.31–1.87) 1.10 (0.85–1.43) 0.57 (0.40–0.81) 0.55 (0.31–0.98) 0.99 (0.59–1.67) 1.56 (1.46–1.67) 1.06 (0.99–1.13)
Myocardial Infarction 1.71 (1.25–2.32) 1.81 (1.23–2.67) 0.95 (0.56–1.61) 0.47 (0.15–1.49) 0.45 (0.11–1.84) 1.56 (1.40–1.75) 1.01 (0.90–1.15)
Cerebrovascular Disease 1.20 (0.96–1.52) 1.08 (0.77–1.51) 0.61 (0.39–0.97) 0.47 (0.21–1.05) 0.56 (0.23–1.35) 1.66 (1.53–1.82) 0.99 (0.90–1.09)
Hypertension 1.04 (0.96–1.13) 0.71 (0.62–0.81) 0.67 (0.59–0.75) 0.78 (0.65–0.94) 0.61 (0.49–0.76) 1.32 (1.28–1.36) 1.29 (1.25–1.33)
Hazard ratios are adjusted for other variables shown in the table: sex, race/ethnicity, exact year of age, and year of diabetes incidence. Reference groups for hazard ratios are: Females (for Males); African Americans (for White, Hispanic, Asian, and Other); lower age; and earlier year of diabetes diagnosis.
Through the entire study period, 2,693 patients (26.7%) were diagnosed with heart disease (Table 3). MI was diagnosed in 270 patients (2.7%) and CHF in 912 (9.0%). In multiple regression analyses (Table 6), Non-Hispanic white race/ethnicity was associated with higher cumulative prevalence of MI (odds ratio [OR] 1.90) and any heart disease (OR 1.32). Hispanic race/ethnicity was associated with lower cumulative prevalence of CHF and any heart disease (OR 0.78, 95% CI 0.66–0.91). Asian race/ethnicity was associated with lower cumulative prevalences of all heart disease outcomes. Age at baseline and male sex were positively associated with cumulative prevalence of MI, CHF, and any heart disease. Year of diabetes diagnosis was negatively associated with cumulative prevalence of each of the heart disease outcomes.
Table 6 Associations with cumulative prevalence of comorbid conditions
Associations with Cumulative Prevalence of Comorbidities Odds Ratio, (95% Confidence Interval)
Male White Hispanic Asian Other Age (per 10 years) Year of Diabetes Incidence
Heart Disease 1.33 (1.21–1.46) 1.32 (1.14–1.53) 0.78 (0.66–0.91) 0.58 (0.45–0.76) 0.52 (0.38–0.72) 1.71 (1.65–1.78) 0.85 (0.82–0.87)
Congestive Heart Failure 1.48 (1.28–1.71) 0.99 (0.79–1.23) 0.43 (0.31–0.58) 0.44 (0.28–0.70) 0.66 (0.41–1.05) 1.68 (1.59–1.78) 0.85 (0.82–0.88)
Myocardial Infarction 1.56 (1.22–2.00) 1.90 (1.39–2.59) 0.78 (0.50–1.23) 0.27 (0.09–0.85) 0.40 (0.13–1.26) 1.53 (1.39–1.67) 0.81 (0.75–0.87)
Cerebrovascular Disease 1.32 (1.11–1.57) 0.99 (0.76–1.29) 0.62 (0.44–0.86) 0.49 (0.28–0.86) 0.31 (0.13–0.70) 1.80 (1.69–1.93) 0.82 (0.78–0.86)
Hypertension 0.87 (0.79–0.97) 0.58 (0.49–0.68) 0.41 (0.36–0.48) 0.39 (0.30–0.49) 0.33 (0.25–0.43) 1.85 (1.78–1.93) 0.91 (0.88–0.93)
Odds ratios are adjusted for other variables shown in the table: sex, race/ethnicity, exact year of age, and year of diabetes incidence. Reference groups for hazard ratios are: Females (for Males); African Americans (for White, Hispanic, Asian, and Other); lower age; and earlier year of diabetes diagnosis.
Cerebrovascular disease
Cerebrovascular disease was diagnosed in 588 patients, evenly split between cases diagnosed at baseline and incident cases subsequent to diabetes incidence (Table 3). Baseline prevalence of cerebrovascular disease was 2.9%. In multiple regression analyses (Table 4), Hispanic, Asian, and "other" race/ethnicities were associated with lower baseline prevalences of cerebrovascular disease. Age at baseline and male sex were positively associated with baseline prevalence of cerebrovascular disease, while year of diabetes diagnosis was negatively associated with baseline prevalence of cerebrovascular disease (OR 0.78, 95% CI 0.74–0.83).
The incidence rate for cerebrovascular disease, among those free of it at baseline diabetes diagnosis, was 1.07 per 100 person-years. In multiple regression analyses, only Hispanic race/ethnicity was associated with lower risk. Age was the only variable associated with a higher incidence of cerebrovascular disease.
Cumulative prevalence of cerebrovascular disease was 5.8%. Hispanic and Asian race/ethnicities were associated with lower cumulative prevalences of cerebrovascular disease, as was year of diabetes diagnosis. Age at baseline and male sex were positively associated with cumulative prevalence of cerebrovascular disease.
Hypertension
A majority of patients (51.8%) had a diagnosis of hypertension at baseline (Table 3). Non-Hispanic white, Hispanic, Asian, and "other" race/ethnicities were all associated with lower baseline prevalence of hypertension, as was male sex (Table 4). Age and year of diabetes diagnosis were associated with higher baseline prevalence.
An additional 2,578 patients were diagnosed with hypertension after diabetes diagnosis, for a cumulative prevalence of 7,804 (77.3%). The incidence rate was 26.46 cases per 100 person-years. Non-Hispanic white, Hispanic, Asian, and "other" race/ethnicities were all associated with lower incidence of hypertension than the African-American reference group, while age and later date of diabetes diagnosis were associated with higher incidence.
Cumulative prevalence of hypertension was negatively associated with non-Hispanic white, Hispanic, Asian, and "other" race/ethnicities, male sex and year of diabetes diagnosis, and positively associated with age.
Discussion
Patients in the Urban Diabetics Study cohort, like other diabetic patient populations, faced a substantial burden of cardiovascular comorbidity. Many were affected by comorbid conditions before they were diagnosed with diabetes, including 14.8% with preexisting heart disease, 2.9% with preexisting cerebrovascular disease, and 51.8% with preexisting hypertension. Among those free of these comorbidities at the time of diabetes diagnosis, similar proportions went on to develop them during followup.
The diabetic patients served by the Philadelphia HCCs are almost uniformly low-income and uninsured or underinsured. A large majority are African American. Nonetheless, the prevalence and incidence of major cardiovascular complications does not, in general, appear to exceed the rates of these complications in other patient populations, whether assessed in nationally-representative surveys or in insured, predominantly middle class managed care populations. Disparities associated with race/ethnicity were in varying directions. Non-Hispanic whites had lower rates of hypertension than African Americans but higher rates of heart disease, especially MI, while Asians and Hispanics had more favorable outcomes on several measures.
As expected, age was positively associated with all cardiovascular comorbidities. Male sex was associated with higher incidence and prevalence of all comorbidities except hypertension.
These findings are comparable to those for other diabetic populations not limited to low-income or uninsured patients. The cumulative prevalence of any heart disease in this study population (26.7%) was similar to the 24.5% prevalence of self-reported coronary heart disease among diabetic patients in the nationally representative 1999–2001 National Health Interview Surveys (NHIS), while the cumulative prevalence of cerebrovascular disease in our study (5.8%) was substantially lower than self-reported stroke in the NHIS sample (9.3%) [3]. However, the age distribution of diabetics in the NHIS sample was substantially older than in this population. If the comparison is restricted to NHIS diabetic patients age 35–64, national prevalences were 31% lower than in our population for heart disease (18.4% vs. 26.7%) and almost identical for cerebrovascular disease (5.9% vs. 5.8%). The overall incidence rate of MI (0.59 per 100 person-years) in the Urban Diabetics cohort was lower than that in the Kaiser Permanente managed care population studied by Karter, et al. [4]. Again, this may reflect the fact that the Urban Diabetics cohort was younger (mean age 49.2 years vs. 59.8 years). When the overall incidence rate in our population is multiplied by the hazard ratio for 10 years additional age in our study, the result (0.92 per 100 person-years) is almost identical to the age- and sex-adjusted rate for African Americans reported by Karter, et al. (0.91 per 100 person-years). The incidence rate of CHF, on the other hand, was higher in our population than in studies of Kaiser Permanente patients [5], while that of cerebrovascular disease was similar to the incidence of stroke reported by Karter, et al. Another study conducted with the Kaiser Permanente diabetes patients found evidence of hypertension for 74% of those diagnosed with diabetes [6], similar to the 77% cumulative prevalence in the Urban Diabetics study.
Other studies have reported somewhat divergent findings for heart disease [7,8] and cerebrovascular disease [9-11], but these studies involved either younger type 1 diabetic patients [7], earlier time periods [8], an older population [9], or analyses that excluded some patients with prior histories of cardiovascular disease [10,11]. Several studies have reported high rates of hypertension among diabetic patients, although none as high as the cumulative prevalence in this cohort [9,11-14].
Many of our findings on racial/ethnic differences in the incidence of MI, CHF, and cerebrovascular disease among diabetic patients are also similar to those found by Karter, et al. Like them, we found higher risks of MI for non-Hispanic whites than for other groups, and lower risks of CHF and cerebrovascular disease for Hispanics and Asians than for African Americans and non-Hispanic whites. The magnitudes of the differences between African-American and non-Hispanic white rates were broadly similar between the two studies, while rates for Asians and Hispanics were more variable. The 1999–2001 NHIS also found the prevalence of coronary and other heart disease among patients with diabetes (based on self-reported data) highest among non-Hispanic whites and lowest among Hispanics (rates for Asians were not reported) [3]. Unlike our study, the NHIS analyses found rates of cerebrovascular disease higher among African Americans with diabetes than among either Hispanics or non-Hispanic whites. Among diabetic patients in the Veterans Health Administration, rates of cardiovascular disease were lower for African Americans, Hispanics, and Asians than for non-Hispanic whites [9]. However, that study was restricted to patients with at least 4 outpatient clinic visits in a 12-month period, and race/ethnicity was not ascertained for 21.4% of the patients in that study, leaving considerable potential for bias in these results.
In our study, male sex was associated with higher baseline prevalence of all comorbidities, higher incidence of the heart disease comorbidities, and higher cumulative prevalence of all comorbidities except hypertension. Our findings are broadly consistent with those in other diabetic populations [3,4,15].
The similarities of our findings to those from the Kaiser Permanente studies are striking, given that the latter assessed an insured, West Coast cohort in which most Hispanics were Mexican American [4], while our study population was an overwhelmingly poor, uninsured or underinsured, East Coast cohort in which most Hispanics were Puerto Rican. The national origins of the Asian patients in the two studies probably also differ, as the Philadelphia Asian community includes larger proportions of South and Southeast Asians and smaller proportions of Japanese and Filipino individuals than the Asian communities of northern California [16]. Among both Hispanics and Asian Americans, national groups vary widely in socioeconomic status and health outcomes [17-19].
Our study is based on administrative records from several sources, each of which may contain errors, and is unlikely to capture as many cardiovascular endpoints than studies with study-specific, patient-level data collection [20,21]. Because the outpatient encounter forms allowed a maximum of four diagnostic codes, there may have been underreporting of some comorbidities, although cardiovascular disease was unlikely to go unrecorded. Outpatient care outside of the eight Philadelphia HCCs and hospitalizations outside of Pennsylvania were not ascertained. The design of the study could therefore undercount comorbid conditions and cardiovascular endpoints for patients who acquired health insurance or moved out of the city and therefore changed outpatient providers. As a test of the sensitivity of the results to migration in and out of the health system, we repeated the analyses, restricting the follow-up time from the dates of the first to the last to outpatient visit; the results were not materially changed.
Among the strengths of the study are its inclusion of a wide range of complications, 8+ year longitudinal design, large sample size, and the combination of data sources to enhance endpoint ascertainment. Our design captured diabetes diagnoses and cardiovascular comorbidity and endpoint data from three complementary sources: outpatient visits in any of the eight Philadelphia HCCs, inpatient visits within any hospital in Pennsylvania and death certificate records. Table 2 supports the observation of Kashner and colleagues that the addition of outpatient administrative data resulted in significantly higher estimates of comorbidity prevalence, strongly suggesting that it has value as an additional data source [22].
By limiting the study to patients with an initial diabetes diagnosis after the first 34 months for which we have data, we were able to ascertain the prevalence of cardiovascular conditions before and at the time of diabetes diagnosis as well as incident conditions diagnosed after diabetes incidence. We are unaware of any study that has presented similar data except for one limited to elderly African Americans and whites in North Carolina, which included only 653 patients with diabetes [23]. The fact that patients included in the study had no record of a diabetes diagnosis either in the Philadelphia Health Care Centers or in a hospital discharge record between March 1993 and the date of their diabetes diagnosis, which was no earlier than January 1996, should ensure that the great majority of patients included here were incident diabetes cases.
There is a strong, well-established association between socioeconomic disadvantage and cardiovascular disease in industrialized populations [24], including diabetic populations [25]. The absence of excess cardiovascular disease beyond that experienced in other diabetic patient populations in this disadvantaged cohort suggests that some factor has offset this disadvantage for this cohort. One possibility is that the provision of primary care, other outpatient services to which patients are referred by their primary care physicians, and prescription medications without out-of-pocket costs to all patients in the Philadelphia HCCs removes a significant barrier to effective care. Restricted use of medications due to costs is common [26] and has been shown to lead to poorer outcomes for several chronic diseases, including cardiovascular diseases [27,28].
This is the only study of which we are aware to report on cardiovascular comorbidity in a large, ethnically diverse cohort of low-income patients with diabetes with near-universal ascertainment of race/ethnicity. It includes substantial populations of low-income Asians, Hispanics, and non-Hispanic whites as well as African Americans, and allows comparisons of the experiences of these groups. The data is longitudinal and includes both outpatient and hospital discharges over an extended period. By looking at comorbid conditions both before and after diabetes incidence, we have provided both a picture of the health status of these patients at the initial presentation of diabetes and estimates of the incidence of additional comorbidities while being treated for diabetes.
Conclusions
Cardiovascular comorbidities were common both before and after diabetes diagnosis in this low-income cohort, but not substantially different from mixed-income managed care populations. It is possible that access to primary care and pharmacy services without fees to individuals in this public health clinic system prevented the poorer outcomes usually seen for disadvantaged patients.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JMR and DAW conceived the study and obtained the data. JMR designed the study, performed the analyses, and drafted the manuscript. DAW and CNS 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
This research was supported by grant #R21DK064201-01 from the National Institute of Diabetes, Digestive and Kidney Diseases.
==== Refs
Engelgau MM Geiss LS Saaddine JB Boyle JP Benjamin SM Gregg EW Tierney EF Rios-Burrows N Mokdad AH Ford ES Imperatore G Narayan V The evolving diabetes burden in the United States Ann Intern Med 2004 140 945 950 15172919
Goldberg RB Cardiovascular disease in patients who have diabetes Cardiology Clinics 2003 21 399 413 14621454
Centers for Disease Control and Prevention Self-reported heart disease and stroke among adults with and without diabetes – United States, 1999–2001 MMWR Morbidity & Mortality Weekly Report 2003 52 1065 70 14603181
Karter AJ Ferrara A Liu JY Moffet HH Ackerson LM Selby JV Ethnic disparities in diabetic complications in an insured population JAMA 2002 287 2519 2527 12020332 10.1001/jama.287.19.2519
Iribarren C Karter AJ Go AS Ferrara A Liu JY Sidney S Selby JV Glycemic control and heart failure among adult patients with diabetes Circulation 2001 103 2668 2673 11390335
Selby JV Peng T Karter AJ Alexander M Sidney S Lian J Arnold A Pettitt D High rates of co-occurrence of hypertension, elevated low-density lipoprotein cholesterol, and diabetes mellitus in a large managed care population Am J Manag Care 2004 10 163 170 15005509
Zgibor JC Songer TJ Kelsey SF Drash AL Orchard TJ Influence of health care providers on the development of diabetes complications: long-term follow-up from the Pittsburgh Epidemiology of Diabetes Complications Study Diabetes Care 2002 25 1584 1590 12196431
Wingard DL Barrett-Connor E National Diabetes Data Group Heart disease and diabetes Diabetes in America 1995 2 Washington, DC: US Govt Printing Office (NIH publ no 95-1468) 429 448
Young BA Maynard C Boyko EJ Racial differences in diabetic nephropathy, cardiovascular disease, and mortality in a national population of veterans Diabetes Care 2003 26 2392 2399 12882868
Kothari V Stevens RJ Adler AI Stratton IM Manley SE Neil A Holman RR Risk of stroke in type 2 diabetes estimated by the UK Prospective Diabetes Study risk engine Stroke 2002 33 1776 1781 12105351 10.1161/01.STR.0000020091.07144.C7
Lee CD Folsom AR Pankow JS Brancati FL Cardiovascular events in diabetic and nondiabetic adults with or without history of myocardial infarction Circulation 2004 109 855 860 14757692 10.1161/01.CIR.0000116389.61864.DE
Baumann LC Chang MW Hoebeke R Clinical outcomes for low-income adults with hypertension and diabetes Nursing Research 2002 51 191 198 12063418 10.1097/00006199-200205000-00008
Spijkerman AMW Adriaanse MC Dekker JM Nijpels G Stehouwer CDA Bouter LM Heine RJ Diabetic patients detected by population-based stepwise screening already have a diabetic cardiovascular risk profile Diabetes Care 2002 25 1784 1789 12351478
Klein BEK Klein R Lee KE Components of the metabolic syndrome and risk of cardiovascular disease and diabetes in Beaver Dam Diabetes Care 2002 25 1790 1794 12351479
Barzilay JI Spiekerman CF Kuller LH Burke GL Bittner V Gottdiener JS Brancati FL Orchard TJ O'Leary DH Savage PJ Prevalence of clinical and isolated subclinical cardiovascular disease in older adults with glucose disorders: the Cardiovascular Health Study Diabetes Care 2001 24 1233 1239 11423508
Census 2000 Summary File 3
Arias E Anderson RN Hsiang-Ching K Murphy SL Kochanek KD Deaths:Final data for 2001 Natl Vital Stat Rep 2003 52 1 115 14570230
Anand SS Yusuf S Vuksan V Devanesen S Teo KK Montague PA Kelemen L Yi C Lonn E Gerstein H Hegele RA McQueen M Differences in risk factors, atherosclerosis and cardiovascular disease between ethnic groups in Canada: the study of health assessment and risk in ethnic groups (SHARE) Indian Heart J 2000 52 S35 43 11339439
Klatsky A The risk of hospitalization for ischemic heart disease among Asian Americans in Northern California Am J Public Health 1994 84 1672 1675 7943495
Ballantyne CM Hoogeveen RC Bang H Coresh J Folsom AR Heiss G Sharrett AR Lipoprotein-associated phospholipase A2, high-sensitivity C-reactive protein, and risk for incident coronary heart disease in middle-aged men and women in the Atherosclerosis Risk in Communities (ARIC) study Circulation 2004 109 837 842 14757686 10.1161/01.CIR.0000116763.91992.F1
Burt VL Cutler JA Higgins M Horan MJ Labarthe D Whelton P Brown C Roccella EJ Trends in the prevalence, awareness, treatment, and control of hypertension in the adult US population. Data from the health examination surveys, 1960 to 1991 Hypertension 1995 26 60 69 7607734
Kashner TM Agreement between administrative files and written medical records: a case of the Department of Veterans Affairs Med Care 1998 36 1324 1336 9749656 10.1097/00005650-199809000-00005
Fillenbaum GG Pieper CF Cohen HD Cornoni-Huntley JC Guralnik J Comorbidity of five chronic health conditions in elderly community residents: determinants and impact on mortality J Gerontol A Biol Sci Med Sci 2000 55 M84 M89 10737690 10737690
Kaplan GA Keil JE Socioeconomic factors and cardiovascular disease: a review of the literature Circulation 1993 88 1973 1998 8403348
Bachmann MO Eachus J Hopper CD Davey Smith G Propper C Pearson NJ Williams S Tallon D Frankel S Socio-economic inequalities in diabetes complications, control, attitudes and health service use: a cross-sectional study Diabet Med 2003 20 921 929 14632718 10.1046/j.1464-5491.2003.01050.x
Artz MB Hadsall RS Schondelmeyer SW Impact of generosity level of outpatient prescription drug coverage on prescription drug events and expenditure among older persons Am J Public Health 2002 92 1257 1263 12144981
Tamblyn R Laprise R Hanley JA Abrahamowicz M Scott S Mayo N Hurley J Grad R Latimer E Perreault R McLeod P Huang A Larochelle P Mallet L Adverse events associated with prescription drug cost-sharing among poor and elderly persons JAMA 2001 285 421 429 11242426 10.1001/jama.285.4.421
Heisler M Langa KM Eby EL Fendrick AM Kabeto MU Piette JD The health effects of restricting prescription medication use because of cost Med Care 2004 42 626 34 15213486 10.1097/01.mlr.0000129352.36733.cc
| 15701166 | PMC550650 | CC BY | 2021-01-04 16:28:55 | no | BMC Public Health. 2005 Feb 8; 5:15 | utf-8 | BMC Public Health | 2,005 | 10.1186/1471-2458-5-15 | oa_comm |
==== Front
BMC Fam PractBMC Family Practice1471-2296BioMed Central London 1471-2296-6-61569847110.1186/1471-2296-6-6Research ArticleInfluence of patient symptoms and physical findings on general practitioners' treatment of respiratory tract infections: a direct observation study Fischer Thomas [email protected] Susanne [email protected] Michael M [email protected] Eva [email protected] Department of General Practice, Georg-August-University, Goettingen, Germany2005 7 2 2005 6 6 6 13 8 2004 7 2 2005 Copyright © 2005 Fischer et al; licensee BioMed Central Ltd.2005Fischer 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 high rate of antibiotic prescriptions general practitioners (GPs) make for respiratory tract infections (RTI) are often explained by non-medical reasons e.g. an effort to meet patient expectations. Additionally, it is known that GPs to some extent believe in the necessity of antibiotic treatment in patients with assumed bacterial infections and therefore attempt to distinguish between viral and bacterial infections by history taking and physical examination. The influence of patient complaints and physical examination findings on GPs' prescribing behaviour was mostly investigated by indirect methods such as questionnaires.
Methods
Direct, structured observation during a winter "cough an cold period" in 30 (single handed) general practices. All 273 patients with symptoms of RTI (age above 14, median 37 years, 51% female) were included.
Results
The most frequent diagnoses were 'uncomplicated upper RTI/common cold' (43%) followed by 'bronchitis' (26%). On average, 1.8 (95%-confidence interval (CI): 1.7–2.0) medicines per patient were prescribed (cough-and-cold preparations in 88% of the patients, antibiotics in 49%). Medical predictors of antibiotic prescribing were pathological findings in physical examination such as coated tonsils (odds ratio (OR) 15.4, 95%-CI: 3.6–66.2) and unspecific symptoms like fatigue (OR 3.1, 95%-CI 1.4–6.7), fever (OR 2.2, 95%-CI: 1.1–4.5) and yellow sputum (OR 2.1, 95%-CI: 1.1–4.1). Analysed predictors explained 70% of the variance of antibiotic prescribing (R2 = 0,696). Efforts to reduce antibiotic prescribing, e.g. recommendations for self-medication, counselling on home remedies or delayed antibiotic prescribing were rare.
Conclusions
Patient complaints and pathological results in physical examination were strong predictors of antibiotic prescribing. Efforts to reduce antibiotic prescribing should account for GPs' beliefs in those (non evidence based) predictors. The method of direct observation was shown to be accepted both by patients and GPs and offered detailed insights into the GP-patient-interaction.
==== Body
Background
Over-prescribing in respiratory tract infections (RTI) has been the topic of numerous studies in general practice. Explanations for the overuse of antibiotics despite weak scientific evidence are multifaceted and vary from efforts to protect patients from "complicated" courses of disease to assumed patient expectations and to inadequate knowledge of physicians [1-4]. Recent investigations demonstrated that patient symptoms and physical findings were also associated with antibiotic prescribing [5,6]. A qualitative study using focus groups reported that general practitioners (GPs) tried to distinguish between viral and bacterial infection by history taking and physical examination [7]. This is in line with a questionnaire-based self-registration study showing a positive correlation between antibiotic prescribing and diagnoses/symptoms assumed to be associated with bacterial infection (e.g. 'sinusitis', 'tonsillitis' or 'yellow-green sputum') [8]. The antibiotic prescription rate in this study, however, was rather low (28%) and only few predictive items could be detected. Because self-registration studies are methodically limited (varying documentation quality, predetermining questionnaire items and completion of the forms in the sense of "scientific acceptability" [9], we designed a direct observation study to gain detailed insights into the influence of patient symptoms and physical examination on GPs' therapeutic decision. Additionally, we expected to obtain information about the doctor-patient-dialogue (e.g. about recommended self-medication or behaviour advice).
Methods
We performed a structured, direct observation-based study. This concept is borrowed from social and cultural anthropology developed in the early 1960s [10]. Starting from an unstructured (qualitative) approach, which enabled open, unbiased data collection, a more systematic, structured concept using checklists (quantitative approach) was developed over time, particularly in nursing research [11-13]. We developed a checklist to record information on the interaction between GPs and patients with suspected RTI. Items included were based on history taking and physical examination protocols and the checklist was evaluated and adapted in a pilot study.
Data collection focused on patient complaints, results of physical examinations, further diagnostic procedures and diagnoses. Physical examination findings were coded according to the level of precision obtained (from 'pathological finding' to 'coated tonsils'). The registration of medication included prescriptions as well as recommendations for over-the-counter medicines (OTC) and the dispensing of drug samples (previously brought in by pharmaceutical representatives). Data were mostly acquired through observation during the consultation.
Potentially participating GPs were randomly selected from a GP register and addressed with a form letter. Of 62 GPs addressed, 30 participated this study (16 located in a medium-sized town in Lower-Saxony and 14 in rural areas of North Rhine-Westphalia). To avoid biasing GPs' behaviour no specific hypotheses were shared with the participants.
GPs were visited for one day by S. F. (medical student at time of observation) during a "cough and cold" period in winter; and all consecutive patients (age above 14 years) with symptoms consistent with RTI were included. All patients were informed that a medical student wanted to participate in the consultation and all of them accepted her attendance.
SAS software (Version 8.1) was used to analyse the data [14]. Respiratory tract infections were classified according to the International Classification for Primary Care (ICPC) [15]: 'upper respiratory tract infection (URTI)/common cold' (R74), 'sinusitis' (R75), 'tonsillitis' (R76), 'laryngitis' (R77) and 'bronchitis' (R78). Since exacerbations of chronic lung diseases are not defined in ICPC, we constructed a dummy variable (including R91, R95 and R96). Multiple diagnoses were accepted and all diagnoses were recorded as stated after the consultation by the GPs. Multiple logistic regression models were used to test for associations between patient characteristics, symptoms, diagnoses and recommended treatment (stepwise backward elimination, α = 0,05). Degree of effect is reported as odds ratios (OR) with 95% confidence intervals (CI). The most frequent patient complaints (cough, sneezing, sore throat, headache, fever, fatigue, hoarseness, myalgia, earache and facial pain) and pathological findings in physical examination as well as age, smoking status and duration of symptoms were included as predictive variables in the multivariate analyses. Drugs were classified using the Anatomic-Therapeutic-Chemical Classification (ATC) codes [16].
Results
Doctors and patients
GPs' median age was 48 years, their median experience in general practice was 12 years and 17% of the GPs were female. A total of 273 patients (51% women) were included (representing 21.4% of all patients visiting their GP at the time of observation). The median number of included patients per GP was 9. No patient refused participation. Patients' median age was 37 years (range 14 to 88).
Diagnoses
Upper respiratory tract infections/common cold (URTI), bronchitis, tonsillitis and sinusitis were the dominating diagnoses and 88% of all patients got at least one of these diagnoses( Table 1). In total 319 diagnoses were made (40 patients were assigned 2 diagnoses and 3 patients 3 diagnoses).
Table 1 Frequency of diagnoses
Diagnoses No. of patients (% of 273 patients)
URTI/common cold 117 (42.9)
Bronchitis 70 (25.6)
Sinusitis 33 (12.1)
Tonsillitis 30 (11.0)
Acute exacerbation chronic lung diseases 24 (8.8)
Otitis media 17 (6.3)
Laryngitis 12 (4.4)
Other 16 (5.9)
(multiple diagnoses possible, URTI = upper respiratory tract infection)
Self-medication and non-medical therapy
49 patients (18%) were asked by their GPs about self-medication and 83% of them acknowledged the previous use of OTC- drugs (predominantly symptomatic cough and cold drugs, particularly mucolytics). In 40% of these patients, GPs prescribed an OTC-preparation of the same ATC-classification. The use of household remedies was addressed in 12% of the patient encounters and three-quarter of the patients confirmed that they tried them before the consultation (inhalation in 36%, gargling in 27% and drinking of tea in 15%).
Drug treatment
On average 1.8 (95%-CI: 1.68–1.95) drugs were prescribed per patient (Table 2). Only 12% of the patients left the consultation without a drug prescription, 27% received prescriptions for 3 and more drugs. The most frequently used drugs were acetylcysteine (89 prescriptions) and ambroxole (43 prescriptions). Only 8 patients asked for a specific prescription by themselves and 4 of them received the favoured drug. In total, 17 pharmaceutical samples were handed out to 14 patients (mostly cough and cold preparations). The recommendation to buy an OTC-preparation was given to 7 patients (mostly paracetamol).
Table 2 Drug Treatment (according to ATC-code)
Main groups Patients (%)
Cough and cold preparations (R05) 208 (76.2)
Expectorants, excl. combinations with cough suppressants (R05C) 163 (59.7)
Other cold combination preparations (R05X) 57 (20.9)
Cough suppressants, excl. combinations with expectorants (R05D) 41 (15.0)
Cough suppressants and expectorants, combinations (R05F) 14 (5.1)
Antibacterials for systemic use (J01) 134 (49.1)
Macrolides and lincosamides (J01F) 52 (19.0)
Beta-lactam antibacterials, penicillins (J01C) 27 (9.9)
Cough suppressants and expectorants, combinations with antibacterials(*) (R05G) 25 (9.2)
Other Beta-lactam antibacterials (J01D) 19 (7.0)
Tetracyclines (J01A) 7 (2.9)
Quinolone antibacterials (J01M) 4 (1.5)
Nasal preparations (R01) 43 (15.8)
Anti-Asthmatics (R03) 21 (7.7)
Throat preparations (R02) 10 (3.7)
Other (Breast unctions (R04), Echinacea-preparations (L03), otologicals (S02)) 18 (6.6)
((*)mostly tetracyclines, ATC = Anatomic, Therapeutic, Chemical Classification)
Predictors of prescriptions
As shown in table 3 an association between specific diagnoses and antibiotic prescription rates could be demonstrated. Antibiotic prescription rate was highest in patients with the diagnosis 'tonsillitis' (90%) and relatively low in patients with the diagnosis 'common cold' (18%). The prescription rate of cough and cold preparations was above 90% in all diagnoses except 'tonsillitis'.
Table 3 Prescription rates of antibiotics and cough and cold preparations
Diagnoses Prescription rates (in %) of
Antibiotics [95% CI] Cough and cold preparations [95%CI]
Tonsillitis 89.7 [78.6–99.3] 55.2 [37.1–73.3]
Bronchitis 77.1 [67.3–87.0] 94.3 [85.3–95.9]
Laryngitis 75.0 [50.5–99.5] 91.7 [76.0–100.0]
Sinusitis 63.6 [47.2–80.0] 90.9 [81.1–100.0]
Acute exacerbation chronic lung disease* 50.0 [30.0–70.0] 95.8 [87.8–100.0]
URTI/ common cold 18.0 [11.0–24.9] 90.6 [85.3–95.9]
all patients 48.5 [42.6–54.5] 87.9 [84.0–91.7]
(URTI = upper respiratory tract infection, *see methods)
Multiple logistic regression analyses demonstrated associations between patient complaints, physical examination results and the prescription of antibiotics (Table 4). The calculated model explained 70% of the variance of antibiotic prescribing (R2 = 0,695). Patients' age, smoking status and symptoms such as cough, sore throat, hoarseness, sputum, sneezing, headache and earache had no influence. Also, GPs' characteristics influenced the prescription rate: Although patients' age was uniformly distributed between younger and older GPs, younger GPs (<50 years) prescribed fewer antibiotics than older ones did (37.3 versus 54.4% of the patients, p < 0.05). Neither the number of treated patients (measured as number of individual patients treated per quarter of a year) nor the GPs' experience in general practice (measured as time of practicing) had influence on antibiotic prescription rates.
Table 4 Influence of patient complaints and physical examination results on antibiotic prescription
Complaints/Physical examination results OR 95%-CI
Pathologically altered tonsils in mouth and throat inspection 15.41 3.6–66.16
Pathological otoscopy findings 8.85 1.16–67.58
Pathological cervical lymph node palpation findings 6.24 1.97–19.71
Rales in lung auscultation 4.29 2.09–8.83
Pathological results in paranasal sinus palpation (sinus tenderness) 3.20 1.38–7.42
Fatigue 3.09 1.42–6.72
Wheezing in lung auscultation 2.91 1.17–7.23
Fever 2.19 1.06–4.54
Yellow sputum 2.10 1.07–4.14
(multiple logistic regression, OR = Odds ratio, CI = Confidence interval, for included variables see methods)
Cough and cold preparations were associated with symptoms such as cough (OR 5.09, 95%-CI: 2.43–10.69), headache (OR 2.43, 95%-CI: 1.02–5.83), sneezing (OR 2.39, 95%-CI: 1.2–4.76), and abnormal findings in throat examination (OR 2.16, 95%-CI: 1.11–4.2). The calculated model explained 19% of the variance of cough and cold preparation prescriptions (R2 = 0.186).
Non-pharmaceutical treatment
Recommendations for non-pharmaceutical treatment were given to 35% patients (increased fluid intake in 19%, inhalation 18%, bed rest 8%, multiple items possible). The advice of increased fluid intake was given to 27% of patients with the prescription of acetylcysteine (which is recommended in the package insert). One quarter of smokers were advised to stop smoking.
Sickness certification
A sickness certification was issued to 57% of the professionally active patients for a median duration of 4 days. 18% declined the proposal of a sick note.
Patient rescheduling
43% of the patients were asked to return for a "control visit" and 21% were informed to revisit if they became worse. Patients with the diagnoses 'tonsillitis' (in 63%) and 'bronchitis' (61%) were more often asked for a "control visit" than patients with the diagnoses 'URTI/ common cold' (23%).
Discussion
Our observational study confirmed the overuse of antibiotics and cough and cold preparations despite the lack of scientific evidence. The analysis of predictive indicators for the prescription showed that antibiotic prescribing was associated with specific patient symptoms and physical examination results whereas the prescription of cough and cold preparations was mostly performed indiscriminately. Efforts to reduce antibiotic prescribing, e.g. recommendations for self-medication, counselling on home remedies or delayed antibiotic prescribing were rare.
Consultation rates for RTI
RTI consultation rates of up to one fifth of all patients seen a day may appear high when compared internationally [17]. However, contacting a GP with symptoms of RTI without appointment is a usual and frequent behaviour of German patients, who tend to frequently consult even for minor complaints. Both cultural factors and a fee-for-service oriented healthcare system are likely to influence consultation rates [18]: On average, German patients visit their GP two times often than e.g. people from the Netherlands or France and even 3 times often than in Sweden [17]. Consequently German GPs see many more patients per day than most of their foreign colleagues. As to our study, one-day visits proved to be sufficient to acquire a representative number of patients.
Antibiotic prescription
In our study, the demonstrated antibiotic prescription rate was high in comparison to other European investigations and rather approaches US-American levels [2,8]. Parallel to this observation study, a documentation study was performed in Germany using questionnaires to record GPs' behaviour with patients also displaying symptoms of RTI [8]. The results of these two studies differ although the patient characteristics and diagnoses were comparable. One explanation for the lower prescription rates in the documentation study could be a selection bias because all of the GPs in our observation study were randomly recruited, whereas all of the GPs participating in the documentation study were recruited at an EBM-workshop. In addition, the completion of the documentation sheets in the sense of "scientific acceptability" is a known confounder [8,9]. However, it can be argued that the direct observation method also influenced the prescription rate ("observer effect" or "Hawthorne effect") [19]. It appears likely that the intense public discussion in the last years about the overuse of antibiotics in primary care would rather lower the rates.
In our observation, patient complaints and physical findings such as 'pathologically altered tonsils' or 'wheezing/rales' in lung auscultation have been shown to be strong predictors of GPs prescribing behaviour comparable to recent investigations [5,6]. Besides clear evidence-based medical facts, more unspecific factors such as fever or fatigue caused GPs to prescribe antibiotics, as well. Furthermore, 'yellow sputum' was positively associated with antibiotic prescription as demonstrated in other studies before [8], although this symptom has shown to be unspecific in predicting bacterial infections (in acutely ill patients) [20,21]. It appears that GPs' behaviour is influenced by empirical "internal evidence" (to detect bacterial infections), which – at least partially -, contradicts actual scientific "external evidence" [4,5,22]. The extensive impact of those predictors (explaining 70% of the variance of antibiotic prescribing) demonstrates that efforts to reduce high antibiotic prescribing rates in RTI must focus on GPs' beliefs in pathogenesis and the misattributing of clinical signs and patient symptoms to bacterial infection.
Diagnoses made by GPs were strong predictors of prescribing behaviour. Patients with the diagnosis 'URTI/ common cold' received significantly fewer antibiotics than those with 'tonsillitis', 'sinusitis' or 'bronchitis'. Despite the controversial scientific evidence for antibiotic therapy in those conditions, the majority of participating GPs seemed to be convinced that antibiotics were necessary or likely to be helpful [4,5,22]. Furthermore, patients with these diagnoses were frequently rescheduled for a "control visit" probably reflecting GPs' beliefs about "potentially dangerous diseases".
Complementary to quantitative analysis direct observation offered interesting insights in the GP-patient-consultation. Only one patient asked for an antibiotic prescription, which is in contrast to the common belief about patient expectations [3,23]. Surprisingly, the observer noticed only few efforts of GPs to reduce antibiotic prescription rates. The questionable benefit of antibiotics was discussed in very few consultations only and no doctor used delayed prescribing which has been demonstrated to be a useful tool to reduce antibiotic intake [24].
The impact of non-medical reasons of prescribing should not be disregarded [25]. Our observational study survey could not measure the impact of tacit or implicit beliefs and relational aspects on the very complex issue of motivation for prescribing. However, the importance of non-medical reasons (e.g. patients' expectations or GPs perception of those expectations) as an independent predictor of antibiotic prescribing has already been shown in several studies [3,5,7,23,26]. Additional influences of the healthcare system seem probable: fee-for-service remuneration, liberal access to GPs and specialist may result in GPs fearing to lose patients (and money) when denying a prescription.
Prescription of cough and cold preparations
Prescription of cough and cold preparations in patients with acute RTI is not supported by scientific evidence [27-29], in particular for the expectorants acetylcysteine and ambroxole, which are the most commonly used cough and cold preparations in Germany. Although some investigations have demonstrated an effect of those substances on chronic bronchitis, evidence in acute RTI is so far absent because well-performed control trials of ambroxole and acetylcysteine are rare [30].
International comparisons of OTC-prescription rates are restricted because of different health insurance systems (and refunding policies) in different countries. Recent data from Belgium – a country with an at least partially comparable health system – showed high prescription rates for cough and cold preparations, too [31]. Interestingly, Belgium GPs preferred prescribing of cough suppressants rather than expectorants, which is the leading group in Germany (Table 2).
The fact that only 19% of the variance of cough and cold preparation prescriptions was explained by clinical signs and patient symptoms supports the assumption that non-medical reasons dominated the decision to prescribe. Frequently, the term "pseudo-placebo" is used in this context to describe the function of those preparations [32] and assumed patient expectations were mainly held responsible for the prescription [3,23].
Eighty-three percent of patients asked confirmed the use of OTC-preparations before the consultation, which were often identical with the subsequently prescribed preparations. GPs did not appreciate the opportunity to encourage this self-medication and cost reducing patient behaviour, perhaps because they were afraid that patients would not consult them with similar complaints at another time [33]. In this context, it must be noted that, in contrast to the UK, German GPs are paid only for patients actually consulting them.
Conclusions
The method of direct observation allowed for detailed and sensitive insights in GP-patient-consultations. Nevertheless, investigators must be aware of the specific disadvantages of the method (e.g. the "observer effect").
Furthermore, we could demonstrate that patient complaints and physical examination results had a strong impact on GPs prescribing behaviour, especially on antibiotic prescription. Although this GP behaviour is not in accordance with actual scientific evidence, GPs' understanding of pathogenesis and the value of clinical signs should be strongly considered when efforts are made to reduce the overuse of antibiotics in primary care.
List of abbreviations
ATC code = "Anatomic, Therapeutic, Chemical Classification" code
CI = Confidence interval
GP = General practitioner
OR = Odds ratio
RTI = Respiratory tract infection
URTI = Upper respiratory tract infection
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
TF participated in the study design, analysis, interpretation of data and drafted the manuscript. SF participated in the study design, analysis, interpretation of data, draft of manuscript and performed the practice observations. MMK participated in the study design and draft of manuscript. EHP participated in the study design, analysis, interpretation of data and draft of the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Mainous AG Hueston WJ Clark JR Antibiotics and upper respiratory infection: do some folks think there is a cure for common cold? J Fam Pract 1996 42 357 361 8627203
Gonzales R Steiner JF Sande MA Antibiotic prescribing for adults with colds, upper respiratory tract infections, and bronchitis by ambulatory care physicians JAMA 1997 278 901 904 9302241 10.1001/jama.278.11.901
Himmel W Lippert-Urbanke E Kochen MM Are patients more satisfied when they receive a prescription? The effect of patient expectations in general practice Scand J Prim Health Care 1997 15 118 122 9323777
Butler CC Rollnick S Pill R Maggs-Rapport F Stott N Understanding the culture of prescribing: qualitative study of general practitioners' and patients' perceptions of antibiotics for sore throats BMJ 1998 317 637 642 9727992
Dosh SA Hickner JM Mainous AG Ebell MH Predictors of antibiotic prescribing for nonspecific upper respiratory infections, acute bronchitis, and sinusitis J Fam Pract 2000 49 407 414 10836770
Fagnan LJ Prescribing antibiotics for upper respiratory infections J Fam Pract 2000 5 415 417 10836771
Coenen S van Royen P Vermeire E Hermann I Denekens J Antibiotics for cough in general practice: a qualitative decisions analysis Fam Pract 2000 17 380 385 11021895 10.1093/fampra/17.5.380
Hummers-Pradier E Pelz J Himmel W Treatment of respiratory tract infection – a study in 18 general practices in Germany Eur J Gen Pract 1999 1 15 20
Rethans JJ Westin S Hays R Methods for quality assessment in general practice Fam Pract 1996 13 468 476 8902517
Gold RL Roles in sociological field observations Soc Forces 1958 36 217 223
Pretzlik U Observational methods and strategies Nurs Res 1994 2 13 21
Turnock C Gibson V Validity in action research: a discussion on theoretical and practice issues encountered whilst using observation to collect data J Adv Nurs 2001 36 471 477 11686762 10.1046/j.1365-2648.2001.01995.x
Mulhall A In the field: notes on observation in qualitative research J Adv Nurs 2003 41 306 313 12581118 10.1046/j.1365-2648.2003.02514.x
SAS Institute Inc SAS/STAT. User's Guide Version 8 Cary, NC 1999
Lamberts H Wood M Hofmans-Okkes ed The international classification of primary care in the European community 1993 Oxford University Press, Oxford-New York-Toronto
Schwabe U ATC-Code: anatomic-therapeutic chemical classification for the German drug market (in German) Bonn: GKV-Arzneimittelindex, WIdO 1995
German Federal Ministry of Health and Social Security (in German)
Turkie P "French lessons" BMJ 2004 329 1393 10.1136/bmj.329.7479.1393
Mangione-Smith R Elliott MN McDonald L McGlynn EA An observational study of antibiotic prescribing behaviour and the Hawthorne effect Health Serv Res 2002 37 1603 1623 12546288 10.1111/1475-6773.10482
Winther B Brofeldt S Gronborg H Study of bacteria in the nasal cavity and nasopharynx during naturally acquired common colds Acta Otolaryngol 1984 98 315 320 6388229
Gonzales R Barret PH Steiner JF The relation between purulent manifestations and antibiotic treatment of upper respiratory tract infections J Gen Intern Med 1999 14 151 156 10203620 10.1046/j.1525-1497.1999.00306.x
Vinson D Lutz I The effect of parental expectation on treatment of children with cough J Fam Pract 1993 37 23 27 8345335
Cockburn J Pit S Prescribing behaviour in clinical practice: patients' expectations and doctors' perceptions of patients' expectations – a questionnaire study BMJ 1997 315 520 523 9329308
Dowell J Pitkethly M Bain J Martin S A randomised controlled trial of delayed antibiotic prescribing as a strategy for managing uncomplicated respiratory tract infection in primary care Br J Gen Pract 2001 51 200 205 11255901
Butler CC Rollnick S Kinnersley P Jones A Stott N Reducing antibiotics for respiratory tract symptoms in primary care: consolidating 'why' and considering 'how' Br J Gen Pract 1998 48 1865 1870 10198512
Coenen S Michiels B Van Royen P Van der Auwera JC Denekens J Antibiotics for coughing in general practice: a questionnaire study to quantify and condense the reasons for prescribing BMC Family Practice 2002 3 16 12217080 10.1186/1471-2296-3-16
Del Mar C Glasziou P Upper respiratory tract infection Clin Evidence 2003 9 1701 1711
Poole PJ Black PN Mucolytic agents for chronic bronchitis or chronic obstructive pulmonary disease The Cochrane Database of Systematic Reviews 2003 1 Art. No. CD001287
Schroeder K Fahey T Over-the-counter medications for acute cough in children and adults in ambulatory settings The Cochrane Database of Systematic Reviews 2004 4 Art. No. CD001831.pub2
Matthys H de Mey C Carls C Rys A Geib A Wittig T Efficacy and tolerability of Myrtol standardized in acute bronchitis. A multi-centre, randomised, double-bilnd, placebo-controlled parallel group clinical trials vs. cefuroxime and ambroxole Arzneimittelforschung 1999 50 700 11 10994153
Coenen S Van Royen P Michiels B Denekens J Optimizing antibiotic prescribing for acute cough in general practice: a cluster-randomized controlled trial J Antimicrobial Chemotherapy 2004 54 661 672 10.1093/jac/dkh374
Kochen MM General practice characteristics of pharmacotherapy (in German) Z Arztl Fortbild 1994 88 647 654
Himmel W Self-medication – economic, sociopharmacologic and toxicologic aspects (in German) Dtsch Med Wschr 2000 125 401 407 10778402
| 15698471 | PMC550651 | CC BY | 2021-01-04 16:29:12 | no | BMC Fam Pract. 2005 Feb 7; 6:6 | utf-8 | BMC Fam Pract | 2,005 | 10.1186/1471-2296-6-6 | oa_comm |
==== Front
BMC ImmunolBMC Immunology1471-2172BioMed Central London 1471-2172-6-41570116810.1186/1471-2172-6-4Research ArticleThe mannose receptor is expressed by subsets of APC in non-lymphoid organs Linehan Sheena A [email protected] Sir William Dunn School of Pathology, South Parks Road, Oxford, OX1 3RE, UK2 Department of Infectious Diseases, Centre for Molecular Microbiology and Infection, Imperial College School of Medicine, Armstrong Rd., London, SW7 2AZ, UK2005 8 2 2005 6 4 4 14 10 2004 8 2 2005 Copyright © 2005 Linehan; licensee BioMed Central Ltd.2005Linehan; 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 mannose receptor (MR) is an endocytic receptor of Mφ and endothelial cell subsets whose natural ligands include both self glycoproteins and microbial glycans. It is also expressed by immature cultured dendritic cells (DC), where it mediates high efficiency uptake of glycosylated antigens, yet its role in antigen handling in vivo is unknown. Knowledge of which APC subsets express MR will assist the design of experiments to address its immunological functions. Here the expression of MR by MHC class II positive APC in non-lymphoid organs of the mouse is described.
Results
MR positive APC were identified in several peripheral organs: skin, liver, cardiac and skeletal muscle and tongue. MR positive cells in salivary gland, thyroid and pancreas coexpressed MHC class II and the myeloid markers macrosialin and sialoadhesin, but not the dendritic cell markers CD11c or DEC-205. MR and MHC class II colocalised in confocal microscope images, implying that antigen capture may be the primary role of MR in these cells. Distinct ligands of MR were found in salivary gland and pancreas tissue lysates that are candidate physiological ligands of MR positive APC in these organs.
Conclusions
The tissue and subcellular distribution of MR suggest it is appropriately located to serve as a high efficiency antigen uptake receptor of APC.
==== Body
Background
Dendritic cells (DC), APC specialised for the efficient stimulation of naïve T cells, are of fundamental importance in the control of antigen-specific immune responses (reviewed in [1]). Immature DC are sparsely distributed in peripheral organs, where they act as sentinels, continuously sampling the antigenic environment. They undergo maturation in response to stimuli that include microbial components and tissue damage, and migrate to T dependent areas of lymphoid organs. Here, they upregulate expression of costimulatory molecules and peptide-loaded surface MHC class I and II molecules and develop the capacity to stimulate antigen-specific T cells restricted by MHC class I and II.
Immature DC capture antigens by receptor-mediated endocytosis, in addition to macropinocytosis and phagocytosis (reviewed in [2]). DC are phenotypically and functionally heterogenous (reviewed in [3]), so the ability to target antigens via specific receptors to different subsets of DC in vivo may help to reveal distinctive features of their roles. One candidate receptor for endocytosis in DC is the mannose receptor (MR), or CD206, which recognises glycoconjugates bearing terminal mannose, fucose and N-acetylglucosamine by interaction with its carbohydrate recognition domains (CRDs). Natural ligands include microbial polysaccharides, glycoproteins and glycolipids, and mammalian glycoproteins with N-linked high-mannose (reviewed in [4]). MR is expressed mainly by subsets of Mφ and endothelial cells in vivo [5,6], but it is also expressed by immature cultured DC, where it endocytoses mannosylated ligands for processing and presentation to T cells by MHC class II [7]. Mannosylation of antigen confers a greatly enhanced efficiency of presentation to T cells of the order of 100 [8], and 200 to 10 000-fold [9]. However, MR is not expressed by immature splenic DC or epidermal Langerhans cells in situ in naïve mice [6], and its contribution to T cell immunity remains unknown.
Notwithstanding their role in stimulating immune responses, it is becoming increasingly apparent that DC play a role in maintaining T cell tolerance to self antigens in the periphery (reviewed in [10,11]). Direct evidence that DC can induce T cell unresponsiveness under non-inflammatory conditions came from an elegant study in which a model MHC class II peptide was targeted to DC in situ [12]. The peptide was engineered into a mAb specific for DEC-205 [12], an endocytic receptor of DC that is structurally related to MR [13]. Although T cells initially proliferated in vivo in response to DC targeting, the response was short-lived, and mice were rendered unresponsive to subsequent challenge with antigen in adjuvant [12]. Antigen targeting via DEC-205 also led to CD8+ T cell tolerance in the steady state [14], and the generation of regulatory T cells [15]. The latter suppressed proliferation of conventional CD4+ T cells in vitro, and appeared to exert immunosuppressive effects in both CD4+ and CD8+ T cell driven immunopathologies [15].
We recently found MR positive interstitial cells in the thyroid, pancreas and salivary gland: secretory organs which bear endogenous ligands of the CRDs of MR [16]. Thyroglobulin was identified as the major MR ligand in the thyroid. Since this glycoprotein is an autoantigen, we postulated that APC in the thyroid may use MR for antigen capture. As a basis for experiments to determine the role of MR in antigen handling in vivo, especially in secretory organs, we surveyed naïve murine non-lymphoid organs for MR positive APC and report here on their phenotypic characterisation. We also provide further evidence for the existence of distinct ligands of MR in pancreas and salivary gland. In light of the immunosuppressive function of immature DC in non-inflammatory conditions, a role for MR in antigen capture by APC for the purposes of maintenance of T cell tolerance to its ligands is credible.
Results
We previously characterised the expression pattern of MR in the adult mouse by immunohistochemistry using a polyclonal Ab raised against MR, and by in situ hybridisation [6]. MR positive cells detected by immunohistochemistry in lymphoid organs were distinguished from the most well characterised DC subsets by their location and morphology, and by comparison with expression of DC markers in adjacent sections. Here, the MR-specific mAb MR6F3 [17] has been used in immunofluorescence and confocal microscopy to identify and characterise MR positive APC in non-lymphoid organs.
MR expression by MHC class II positive cells in non-lymphoid organs
The specificity of MR6F3A488, was verified by observation of the expected expression pattern of MR in the spleen. Expression was restricted to the red pulp and did not overlap in dual-labeled specimens with the CD11c positive immature DC located at the border of the white and red pulps (figure 1). No non-specific binding of the isotype matched control antibody, IgG2aA488 was detected. The lack of expression of MR by DEC-205 positive DC in the thymus was also confirmed (not shown).
Figure 1 Mannose receptor positive and negative APC subsets identified by immunofluorescence microscopy. Tissues were examined by immunofluorescence microscopy for MR expression and other APC phenotypic markers in dual-labeled specimens as indicated. MR was detected using MR6F3A488 mAb. Images of each antigen labelling are shown separately in gray-scale, and merged in colour, in some cases with nuclear counterstaining shown in blue. Colocalisation of red and green is indicated by yellow. Examples of MHC class II-positive APC which coexppress MR are indicated with arrows. These constitute a subset of APC in the skin, liver, tongue and heart. No background was detected with an irrelevant IgG2aA488 negative control (shown in the spleen inset of MR-labelling only). Background labelling due to secondary Abs alone was also not detected in control sections (shown for anti-hamster IgG in the spleen inset of CD11c labelling only). Scale bars are indicated in MR-labelled panels only, and are 100 μm.
To detect the scattered interstitial DC of non-lymphoid organs, sections were labeled for MHC class II (figure 1). Although other APC such as Mφ can express MHC class II, constitutive expression by non-DC is absent in most organs of naïve mice. In the skin, almost complete coexpression of MR and the pan-Mφ marker macrosialin (FA.11) was found, but only a minor subset of MHC class II positive cells in the dermis coexpressed MR. DC of the hepatic portal triads are known to express MHC class II (reviewed in [18]). A proportion of such cells in the liver were found to coexpress MR. MHC class II positive cells in the tongue and heart also expressed MR, as did those in skeletal muscle (Table 1). In summary, expression of MR by MHC class II positive cells in non-lymphoid organs of naïve mice is not uncommon.
Table 1 Expression of Mannose receptor by DC and APC differentiated in vivo
Organ or Tissue Cell type Expression of MR Species Reference
Lymphoid organs
Spleen Immature and mature DC - Mouse [6]
Lymph node Interdigitating cells - Mouse [6]
Lymph node T cell area APC + Human [23]
Lymph node Follicular border APC + Mouse [6]
Peyer's patch APC - Mouse [6]
Thymus Interdigitating cells - Mouse [6]
Non-lymphoid organs
Skin Langerhan's cells - Mouse, Human [6, 22-25]
Skin (atopic dermatitis and psoriasis) Inflammatory dendritic epidermal cells + Human [25]
Skin Dermal APC Minor subset + Mouse This study
Lung Immature DC Endocytose MR ligand Human [31]
Peritoneum DC - Mouse [44]
Blood DC - Human [45]
Thyroid Immature DC / APC + Pig, Mouse [27]; This study
Pancreas APC + Mouse This study
Salivary gland APC + Mouse This study
Liver Hepatic portal triad APC Subset + Mouse This study
Muscle (cardiac and skeletal) APC Subset + Mouse This study
Tongue APC Subset + Mouse This study
MR expression by novel APC in secretory organs
We previously showed that secretory cells in the salivary gland, thyroid and exocrine pancreas were rich in ligands of the CRDs of MR, and expression of MR by adjacent interstitial cells was particularly intense [16]. In adjacent sections, we noted expression of MHC class II by interstitial cells. Here a further examination of the phenotype of MR positive interstitial cells in these organs was made. Essentially all MR positive cells coexpressed macrosialin (FA.11) (figure 2). Many if not most also expressed MHC class II, although the relative fluorescence intensities of MR and MHC class II in individual cells were often very different. In the submandibular salivary gland, MHC class II was also expressed by other interstitial cells that were probably epithelial cells. Further phenotypic analysis revealed that MR positive cells coexpressed sialoadhesin (not shown), a myeloid marker expressed by some Mφ subsets, and also some DC. Expression of DEC-205 was not detected on any MR positive cells in any of the secretory organs (not shown). Although there were CD11c positive cells in the salivary gland and thyroid, MR positive APC were negative for this marker. It was not possible to assess whether the interstitial cells of the exocrine pancreas expressed CD11c as unexpected and abundant expression of this integrin by adjacent secretory cells was found.
Figure 2 Mannose receptor positive APC in secretory organs co-express macrosialin and MHC class II. Secretory organ sections were probed for MR and either macrosialin (FA.11) or MHC class II as indicated. They are shown separately and as merged images that include a nuclear counterstain in blue. In merged images of MR and FA.11 labelling, there are many yellow cells, indicating colocalisation of these markers. In merged images of MR and MHC class II labelling, there are fewer yellow cells, because the fluorescence intensities of the two labels are usually very different in individual cells. However, visual comparison of singly labelled images indicates that many cells do express both markers. Examples in each tissue are indicated with arrows. The scale bar is indicated in the first panel only and is 100 μm.
Since the MR positive cells of secretory organs exhibited a strongly myeloid phenotype but lacked expression of traditional DC markers, the possibility that they may be activated Mφ was considered. IFNγ is the most potent activator of Mφ, leading to expression of MHC class II, so the phenotype of interstitial cells in the secretory organs of IFNγ -/- mice was examined. Constitutive expression of MR and MHC class II was retained (not shown). In summary, MR positive interstitial cells displayed a novel APC phenotype, and constitutively expressed MHC class II in the absence of stimulation with IFNγ.
Confocal microscopic analysis of MR positive APC in secretory organs
The coexpression of MR and MHC class II in secretory organ APC implied that these cells may have immunologic functions. The distribution of these markers were examined by confocal immunofluorescence microscopy, in cells that expressed comparable levels of the two proteins (figure 3). In some cells, considerable pixel to pixel overlap of MR and MHC class II, which was not restricted to individual optical sections within z-series, was observed. The resolution of the microscope was inadequate to resolve individual vesicles, but regions of higher and lower detection of one marker were frequently paralleled by similar variations in fluorescence intensity of the other. In other cells, MR and MHC class II were not seen to colocalise, demonstrating that the resolution of the microscope and preservation of the tissues were adequate to allow different intracellular compartments to be distinguished in these cells (not shown). Furthermore, MR did not colocalise with the endosomal marker macrosialin in confocal microscope images (not shown). The results are therefore highly suggestive that MR and MHC class II are located in the same subcellular compartments within a proportion of APC.
Figure 3 Mannose receptor and MHC class II colocalise in confocal microscope images of secretory organ APC. The subcellular distribution of MR and MHC class II in secretory organ APC was examined by confocal microscopy as indicated. A nuclear counterstain in blue is included in merged images. Single optical planes showing a high degree of colocalisation of MR and MHC class II were selected from z-series in which colocalisation was also observed in adjacent optical sections. Boxed regions of each image are shown at three-fold higher magnification in insets. The scale bar is indicated in the first panel only and is 5 μm.
Characterisation of MR ligands in salivary gland and pancreas by lectin blotting
We previously identified thyroglobulin as an abundant ligand of MR present in the thyroid, and showed that CRD4-7Fc, an Fc-fusion protein bearing four of the CRDs of MR, which retains the same ligand specificity as MR, bound specifically to sections of salivary gland and pancreas [16]. Here, further evidence for the existence of distinct ligands of CRD4-7Fc in the salivary gland and pancreas is shown, and estimates of their Mr. Tissue lysates were separated by SDS-PAGE, transferred to nitrocellulose membranes and probed for ligands of CRD4-7Fc by lectin blotting (figure 4). Bands of approximate Mr of 55, 100, 120, 130 and >250 kD were identified from salivary gland, and bands of 60, 145, 215 and 250 kD from pancreas lysates. Binding specificity was confirmed by probing with CRD4-7Fc in the presence of mannose. Under these conditions, all binding was lost. The distribution of bands in CRD4-7Fc probed blots was compared with gels stained for protein with Coomassie. The 55 kD ligand in salivary gland may correspond to a particularly abundant fraction in the Coomassie stained gel, but other ligand bands from either organ did not appear to be associated with major protein fractions. Human salivary amylase has been shown to be a ligand of rat MR [19], so a comparison was made between the size of ligand bands and bands in blots probed for alpha amylase. Amylase was a minor component of salivary gland, although a major component of pancreas, giving rise to a doublet of bands in both with Mr of 49 and 50 kD, distinct from all bands detected in lectin blots with CRD4-7Fc. The identities of the ligands of MR in salivary gland and pancreas therefore remain unknown.
Figure 4 A Mannose receptor-fusion protein recognises distinct bands in blots of salivary gland and pancreas tissue lysates. Tissue lysates were separated by SDS-PAGE and examined by Coomassie protein staining and western and lectin blotting as indicated. The doublet of bands recognised by the antibody directed against the putative MR ligand, salivary amylase, did not correspond to the bands recognised by the MR-fusion protein, CRD4-7Fc. The specificity of this probe was confirmed by loss of all bands in blots probed with CRD4-7Fc in the presence of the competitor, mannose. Binding to mannose-BSA is shown as a positive control. Protein loaded per lane was 10 μg of salivary gland and pancreas lysates in each application except western blotting of pancreas in which 200 ng was used. 5 ng mannose-BSA was used per lane in lectin blotting. The migration pattern of molecular weight markers is indicated on the left of the image.
As a control, equivalent quantities of lung, thymus and spleen tissue lysates were subjected to lectin blotting with CRD4-7Fc as described above, but defined bands were not detected (not shown). We previously showed that Mφ within lung and thymus tissue sections contained ligands of CRD4-7Fc, as did Mφ and other cells within spleen [16]. The absence of defined bands in lectin blots of lysates of these tissues indicates that ligands generally present within tissue are heterogeneous in size, and that no single species is sufficiently abundant to be detectable after separation by SDS-PAGE. This is in stark contrast to the defined tissue-specific bands detected in lectin blots of salivary gland and pancreas tissue lysates.
Discussion
The MR was the first non-opsonic C-type lectin receptor shown to be expressed and mediate adsorptive endocytosis in DC [7,20], properties it shares with a growing family of such receptors (reviewed in [21]). Expression of MR by DC appears to be tightly regulated. In human peripheral blood monocyte-derived DC, MR expression is down-regulated by inflammatory stimuli, and for this reason it has been considered a prototype marker of immature DC [7]. However, human epidermal Langerhans cells which are functionally immature, do not express MR in situ [22-25], although have been found to be positive for MR after a short isolation procedure [26]. In fact, many DC subsets that have differentiated in vivo do not express MR; a summary of MR expression by APC is shown in table 1. Here, the expression of MR by APC in non-lymphoid organs of the naive mouse is described. Cells were analysed in situ, circumventing the problem that isolation or culture of cells for ex vivo analysis may affect expression of MR.
In contrast to our earlier observation that MR is not expressed by well-characterised subsets of DC in lymphoid organs [6], many APC in peripheral tissues expressed MR, including a proportion of MHC class II positive cells in skin, liver, tongue and heart. Whilst a limited phenotypic analysis is not sufficient to prove that these cells are DC, in the case of MR positive cells in the thyroid, pancreas and salivary glands, this designation appears to be a strong possibility. DC are normally distinguished from monocytes and macrophages by phenotype, morphology and function, properties which are most amenable to study after cell isolation. MR is expressed by interstitial cells in situ in pig thyroids, and MR positive cells have been successfully isolated from this source, and maintained in co-culture with thyrocytes in the presence of thyroid stimulating hormone [27]. Isolated cells expressed MHC class II, S100 protein and had a high endocytic capacity, characteristic of DC. They also exhibited morphological features of DC including a plasma membrane with numerous processes, a reniform nucleus and intracellular annular structures. In response to a stimulator of DC maturation, TNFα, the cells detached from the culture substratum and lost the ability to endocytose the MR ligand, FITC-dextran. By extension, the MR positive APC detected in the murine thyroid may also be immature DC. MR positive APC in the pancreas and salivary gland were phenotypically indistinguishable from the thyroid cells, according to the markers used in this study. APC in all three secretory organs exhibited a strongly myeloid phenotype, expressing macrosialin and sialoadhesin, but were unlikely to be activated Mφ since they were also present in IFNγ -/- mice. MR positive immature DC or other APC at peripheral sites are likely to down-regulate MR when they mature and migrate to lymph nodes, in accordance with in vitro models [7,27], since the mature interdigitating cells of tissue-draining lymph nodes do not express MR [6].
The function of MR in relation to antigen handling in vivo is poorly understood, in part because most studies have used glycosylated Ag that are now known to be ligands of other receptors in addition to MR (reviewed in [28]). However, a specific role for MR in enhancing uptake and presentation to both CD4+ and CD8+ T cells by DC has recently been shown using Ag fused to mAb specific for MR [29,30]. Surprisingly, reports of any ability of MR to enhance primary immune responses in vivo have not been forthcoming, but since MR is not expressed by splenic DC or epidermal Langerhans cells (figure 1; [6]), MR may not play a role in DC capture of antigen administered intravenously or by skin absorption. Isolated human lung DC exhibit a high endocytic capacity for the MR ligand, FITC-dextran [31]. Therefore, MR may play a role in handling of inhaled antigens, such as the house dust mite allergen, Der p 1, whose endocytosis by DC is dependent on MR [32]. A link between MR and allergy has been suggested based on the finding that monocyte-derived DC from allergic patients expressed more MR and endocytosed Der p 1 more efficiently than did DC derived from healthy donors [32].
Antigen uptake by MR on cultured DC promotes presentation to T cells of antigen-derived peptides on MHC class II [7]. The colocalisation of MR and MHC class II in confocal microscope images of APC in thyroid, pancreas and salivary gland described here implies a role in antigen uptake for presentation. Interestingly, in mouse bone marrow-derived DC and a MHC class II-expressing fibroblast cell line, MR did not colocalise with MHC class II [33]. However, the structurally similar receptor, DEC-205, did target MHC class II compartments, and this was found to depend on the presence of an EDE triad within the receptor's cytoplasmic tail, which MR lacks. When fused to the extracellular domain of the Fc receptor, the cytoplasmic tail of DEC-205 directed endocytosed human IgG to a compartment from which MHC class II loading and subsequent presentation of IgG-derived peptides was 100-fold more efficient than that driven by the cytoplasmic tail of MR. In another study, efficiencies of processing and presentation of ribonucleases A and B by a fibroblast cell line expressing MR and MHC class II IAK were compared [34]. These model Ag differ only by the presence of a high-mannose oligosaccharide on RNAse B. Although RNAse B, a ligand of MR, was endocytosed more efficiently than the non-ligand RNAse A, efficiencies of presentation of the two Ag were indistinguishable [34]. Therefore, MR does not constitutively enhance Ag presentation of its ligands. To do so, it may need to target MHC class II compartments. The apparent colocalisation of MR and MHC class II in secretory organ APC implies that the intracellular locations of MR or MHC class II molecules in these cells may be distinct from those in bone marrow-derived DC and transfected fibroblasts, perhaps permitting MR to target Ag to these compartments with greater efficiency. Unusually for immature DC, those cultured from pig thyroid were characterised by plasma membrane expression of MHC class II [35].
If MR is poised to direct Ag to the MHC class II compartments of secretory organ APC, the question of which Ag are captured in these environments must be raised. We previously showed that thyroglobulin, a major auto-antigen in murine autoimmune thyroid diseases, is a ligand of MR, and suggested that MR expressed by interstitial cells in the thyroid may be involved in maintaining tolerance to this Ag in normal mice [16]. Here, further evidence is presented that the pancreas and salivary glands also bear abundant MR ligands. A high molecular weight fraction in salivary gland lysate (>250 kD) may correspond to secretory IgA. This Ab inhibits adhesion of microbes to the gut wall, but is a non-inflammatory isotype, to avoid unwanted immune reactions against indigenous microflora and dietary Ags (reviewed in [36]). Uptake of secretory IgA by cultured human DC could be partially blocked by mAb against MR and this pathway has been suggested to allow modulation of mucosal immune responses [37]. The MR ligands in pancreas lysate were of different Mr to those in salivary gland, and remain undefined. A role for MR in maintenance of immune tolerance to its ligands in this organ is credible since pancreatic DC of naïve mice are believed to be tolerogenic [38]. DC isolated from the lymph node draining the pancreas, but not from other lymph nodes, were able to confer protection against diabetes when transferred to the nonobese diabetic (NOD) mouse strain [38].
In light of the location of MR positive APC in secretory organs, the abundance of endogenous ligands, and the positive identification of one ligand as thyroglobulin, the available data could suggest a tolerogenic role of MR is more likely than an immunostimulatory role in these organs in the steady-state. Such a mechanism may represent a physiological correlate of the experimentally induced immune tolerance achieved by targeting Ag to DEC-205 in naïve mice [12,14,15]. Further studies will be required to address the significance of MR to immune homeostasis.
Conclusions
MR positive APC are present in non-lymphoid organs of the naïve mouse, but in most organs examined, only a subset of MHC class II positive cells express MR, in contrast to the ubiquitous expression of MR described in cultured DC. This information will inform the design of experiments to test the function of MR in antigen handling in vivo. An immuno-regulatory role for MR in relation to its endogenous ligands in salivary gland, pancreas and thyroid is suggested by the finding that MR positive APC are abundant in these organs, they co-express MHC class II and appear to have an overlapping subcellular distribution of MR and MHC class II.
Methods
Mice
IFNγ -/- mice [39] were backcrossed for 14 generations onto the C57BL/6 background at the Sir William Dunn School of Pathology, Oxford, UK. IFNγ -/- and WT C57BL/6 mice were bred and used at 8 to 12 weeks of age in accordance with Home Office legislation. No differences in phenotype or abundance of MR positive APC were noted between male and female mice.
Antibodies and Fc-fusion protein
Rat mAb were prepared in our laboratory and used at optimal concentrations for immunolabelling. They were directed against MHC class II (clone TIB120; ATCC); macrosialin (clone FA.11; [40]) and sialoadhesin (clone 3D6; [41]). Rat mAb against DEC-205 (clone NLDC-145; [42]) and hamster mAb against CD11c (clone N418; [43]) were purchased from Serotec (Kidlington, UK). MR6F3 directed against MR [17], and rat IgG2a were conjugated to Alexa488 (Molecular Probes, Leiden, The Netherlands) according to the manufacturer's instructions. Goat anti-amylase was from Santa Cruz Biotechnology Inc (Santa Cruz, CA). CRD4-7Fc was prepared as described [16]. Secondary Abs were goat F(ab')2 anti-rat IgG, and goat F(ab')2 anti-Armenian hamster IgG, both cy3 conjugated, and horseradish peroxidase conjugated mouse F(ab')2 anti-human IgG Fc (Jackson ImmunoResearch Labs; West Grove, PA). Horseradish peroxidase conjugated donkey F(ab')2 anti-goat IgG was from Chemicon (Harrow, UK).
Immunofluorescence labelling and microscopy
For immunofluorescence microscopy, 6 μm cryosections were prepared from unfixed frozen tissues. They were permeabilised in an incubation buffer consisting of 0.5% BSA and 0.05% saponin in PBS, then blocked with 5% normal rabbit serum in incubation buffer. Sections were subsequently incubated with rat or hamster mAb specific for APC antigens for 1 hr, then the appropriate cy3-conjugated secondary Ab for 30 min. Sections were blocked with 100 μg/ml rat IgG for 30 min, before being probed with Alexa488 conjugated MR6F3 mAb or isotype matched control. Nuclei were labeled with Hoechst 33342 dye (Molecular Probes). Sections were mounted in Aqua Polymount (Polysciences, Inc., Warrington, PA). Slides were examined by fluorescence microscopy and 12-bit digital images captured using a CCD camera attached to a Zeiss Axioplan photomicroscope. Slides were prepared in the same way for confocal microscopy, except that cryosections were cut at 20 μm. Confocal microscope z-stack images were collected using a Bio-Rad Radiance 2000 MP laser scanning confocal microscope, with lasers exciting at 367, 488 and 543 nm. Images corresponding to each fluorophore were collected using individual lasers sequentially to eliminate bleed-through. Immunofluorescence and confocal images were processed using MetaMorph version 4.5 software.
Western and lectin blotting
Mouse tissue lysates were prepared in lysis buffer (2% v/v Triton X-100, 10 mM Tris pH 8.0, 10 mM NaN3, 150 mM NaCl, 10 mM EDTA, 5 mM iodoacetamide, 1 mM PMSF, 1 mg/ml pepstatin, 1μM leupeptin). Proteins were quantified in duplicate using Bicinchoninic acid protein assay kit (Pierce Chemical Company, Chester, UK). Samples were electrophoresed by SDS-PAGE, transferred to nitrocellulose membranes and subjected to western or lectin blotting using standard methods [16]. CRD4-7Fc was used at 1 μg/ml, with or without 100 mM D-mannose as a specificity control. Membranes were probed with the appropriate horseradish peroxidase-conjugated secondary Ab, and signal was developed using enhanced chemiluminescence (Amersham Life Science Ltd., Bucks., UK). A calcium containing buffer was used in lectin blotting throughout (150 mM NaCl, 10 mM Tris pH 7.4, 10 mM CaCl2) and mannose-BSA (Sigma, St. Louis, MO) was used as positive control. Gels were also stained with Coomassie R-250 to detect protein.
Abbreviations
CRD, carbohydrate recognition domain; DC, dendritic cell; MR, mannose receptor
Acknowledgements
This work was supported by a Junior Research Fellowship from St Cross College, Oxford, to the author. Prof Siamon Gordon and Dr Luisa Martinez-Pomares are thanked for useful discussions, and the generous provision of materials. I thank Dr Nick White for expert assistance in microscopy and Mrs. Liz Darley for preparing cryosections.
==== Refs
Steinman RM The dendritic cell system and its role in immunogenicity Annu Rev Immunol 1991 9 271 296 1910679 10.1146/annurev.iy.09.040191.001415
Mellman I Steinman RM Dendritic cells: specialized and regulated antigen processing machines Cell 2001 106 255 258 11509172 10.1016/S0092-8674(01)00449-4
Shortman K Liu YJ Mouse and human dendritic cell subtypes Nat Rev Immunol 2002 2 151 161 11913066 10.1038/nri746
Linehan SA Martínez-Pomares L Gordon S Macrophage lectins in host defence Microbes Infect 2000 2 279 288 10758404 10.1016/S1286-4579(00)00300-2
Takahashi K Donovan MJ Rogers RA Ezekowitz RA Distribution of murine mannose receptor expression from early embryogenesis through to adulthood Cell Tissue Res 1998 292 311 323 9560474 10.1007/s004410051062
Linehan SA Martínez-Pomares L Stahl PD Gordon S Mannose receptor and its putative ligands in normal murine lymphoid and non-lymphoid organs. In situ expression of mannose receptor by selected macrophages, endothelial cells, perivascular microglia and mesangial cells, but not dendritic cells J Exp Med 1999 189 1961 1972 10377192 10.1084/jem.189.12.1961
Sallusto F Cella M Danielli C Lanzavecchia A Dendritic cells use macropinocytosis and the mannose receptor to concentrate macromolecules in the major histocompatibility complex class II compartment: Down regulation by cytokines and bacterial products J Exp Med 1995 182 389 400 7629501 10.1084/jem.182.2.389
Engering AJ Cella M Fluitsma D Brockhaus M Hoefsmit EC Lanzavecchia A Pieters J The mannose receptor functions as a high capacity and broad specificity antigen receptor in human dendritic cells Eur J Immunol 1997 27 2417 2425 9341788
Tan MC Mommaas AM Drijfhout JW Jordens R Onderwater JJ Verwoerd D Mulder AA Van der Heiden AN Scheidegger D Oomen LC Ottenhoff TH Tulp A Neefjes JJ Koning F Mannose receptor-mediated uptake of antigens strongly enhances HLA class II-restricted antigen presentation by cultured dendritic cells Eur J Immunol 1997 27 2426 2435 9341789
Steinman RM Nussenzweig MC Avoiding horror autotoxicus: the importance of dendritic cells in peripheral T cell tolerance Proc Natl Acad Sci USA 2002 99 351 358 11773639 10.1073/pnas.231606698
Mahnke K Schmitt E Bonifaz L Enk AH Jonuleit H Immature, but not inactive: the tolerogenic function of immature dendritic cells Immunol Cell Biol 2002 80 477 483 12225384 10.1046/j.1440-1711.2002.01115.x
Hawiger D Inaba K Dorsett Y Guo M Mahnke K Rivera M Ravetch JV Steinman RM Nussenzweig MC Dendritic cells induce peripheral T cell unresponsiveness under steady state conditions in vivo J Exp Med 2001 194 769 779 11560993 10.1084/jem.194.6.769
Jiang W Swiggard WJ Heufler C Peng M Mirza A Steinman RM Nussenzweig MC The receptor DEC-205 expressed by dendritic cells and thymic epithelial cells is involved in antigen processing Nature 1995 375 151 155 7753172 10.1038/375151a0
Bonifaz L Bonnyay D Mahnke K Rivera M Nussenzweig MC Steinman RM Efficient targeting of protein antigen to the dendritic cell receptor DEC-205 in the steady state leads to antigen presentation on major histocompatibility complex class I products and peripheral CD8+ T cell tolerance J Exp Med 2002 196 1627 1638 12486105 10.1084/jem.20021598
Mahnke K Qian Y Knop J Enk AH Induction of CD4+/CD25+ regulatory T cells by targeting of antigens to immature dendritic cells Blood 2003 101 4862 4869 12543858 10.1182/blood-2002-10-3229
Linehan SA Martínez-Pomares L da Silva RP Gordon S Endogenous ligands of the carbohydrate recognition domains of the mannose receptor in murine macrophages, endothelial cells and secretory cells; potential relevance to inflammation and immunity Eur J Immunol 2001 31 1857 1866 11433382 10.1002/1521-4141(200106)31:6<1857::AID-IMMU1857>3.0.CO;2-D
Martinez-Pomares L Reid DM Brown GD Taylor PR Stillion RJ Linehan SA Zamze S Gordon S Wong SY Analysis of mannose receptor regulation by IL-4, IL-10 and proteolytic processing using novel monoclonal antibodies J Leukoc Biol 2003 73 604 613 12714575 10.1189/jlb.0902450
Thomson AW O'Connell PJ Steptoe RJ Lu L Immunobiology of liver dendritic cells Immunol Cell Biol 2002 80 65 73 11881616 10.1046/j.0818-9641.2001.01058.x
Niesen TE Alpers DH Stahl PD Rosenblum JL Metabolism of glycosylated human salivary amylase: in vivo plasma clearance by rat hepatic endothelial cells and in vitro receptor mediated pinocytosis by rat macrophages J Leukoc Biol 1984 36 307 320 6207253
Sallusto F Lanzavecchia A Efficient presentation of soluble antigen by cultured human dendritic cells is maintained by granulocyte/macrophage colony-stimulating factor plus interleukin 4 and downregulated by tumor necrosis factor alpha J Exp Med 1994 179 1109 1114 8145033 10.1084/jem.179.4.1109
Figdor CG van Kooyk Y Adema GJ C-type lectin receptors on dendritic cells and Langerhans cells Nat Rev Immunol 2002 2 77 84 11910898 10.1038/nri723
Uccini S Sirianni MC Vincenzi L Topino S Stoppacciaro A Lesnoni La Parola I Capuano M Masini C Cerimele D Cella M Lanzavecchia A Allavena P Manotvani A Baroni CD Ruco LP Kaposi's sarcoma cells express the macrophage-associated antigen mannose receptor and develop in peripheral blood cultures of Kaposi's sarcoma patients Am J Path 1997 150 929 938 9060831
Noorman F Braat EA Barrett-Bergshoeff M Barbé E van Leeuwen A Lindeman J Rijken DC Monoclonal antibodies against the human mannose receptor as a specific marker in flow cytometry and immunohistochemistry for macrophages J Leukoc Biol 1997 61 63 72 9000538
Mommaas AM Mulder AA Jordens R Out C Tan MC Cresswell P Kluin PM Koning F Human epidermal Langerhans cells lack functional mannose receptors and a fully developed endosomal/lysosomal compartment for loading of HLA class II molecules Eur J Immunol 1999 29 571 580 10064073 10.1002/(SICI)1521-4141(199902)29:02<571::AID-IMMU571>3.0.CO;2-E
Wollenberg A Mommaas M Oppel T Schottdorf EM Gunther S Moderer M Expression and function of the mannose receptor CD206 on epidermal dendritic cells in inflammatory skin diseases J Invest Dermatol 2002 118 327 334 11841552 10.1046/j.0022-202x.2001.01665.x
Condaminet B Péguet-Navarro J Stahl PD Dalbiez-Gauthier C Schmitt D Berthier-Vergnes O Human epidermal Langerhans cells express the mannose-fucose binding receptor Eur J Immunol 1998 28 3541 3551 9842897 10.1002/(SICI)1521-4141(199811)28:11<3541::AID-IMMU3541>3.3.CO;2-W
Croizet K Rabilloud R Kostrouch Z Nicolas JF Rousset B Culture of dendritic cells from a nonlymphoid organ, the thyroid gland: evidence for TNFalpha-dependent phenotypic changes of thyroid-derived dendritic cells Lab Invest 2000 80 1215 1225 10950112
Keler T Ramakrishna V Fanger MW Mannose receptor-targeted vaccines Expert Opin Biol Ther 2004 4 1953 1962 15571457 10.1517/14712598.4.12.1953
Ramakrishna V Treml JF Vitale L Connolly JE O'Neill T Smith PA Jones CL He LZ Goldstein J Wallace PK Keler T Endres MJ Mannose receptor targeting of tumor antigen pmel17 to human dendritic cells directs anti-melanoma T cell responses via multiple HLA molecules J Immunol 2004 172 2845 2852 14978085
He LZ Ramakrishna V Connolly JE Wang XT Smith PA Jones CL Valkova-Valchanova M Arunakumari A Treml JF Goldstein J Wallace PK Keler T Endres MJ A novel human cancer vaccine elicits cellular responses to tumor-associated antigen, human chorionic gonadotropin β Clin Cancer Res 2004 10 1920 1927 15041707
Cochand L Isler P Songeon F Nicod LP Human lung dendritic cells have an immature phenotype with efficient mannose receptors Am J Respir Cell Mol Biol 1999 21 547 554 10536111
Deslee G Hammad H Angyalosi G Tillie-Leblond I Mantovani A Tonnel AB Pestel J Involvement of the mannose receptor in the uptake of Der p 1, a major mite allergen, by human dendritic cells J Allergy Clin Immunol 2002 110 763 770 12417886 10.1067/mai.2002.129121
Mahnke K Guo M Lee S Sepulveda H Swain SL Nussenzweig M Steinman RM The dendritic cell receptor for endocytosis, DEC-205, can recycle and enhance antigen presentation via major histocompatibility complex class II-positive lysosomal compartments J Cell Biol 2000 151 673 684 11062267 10.1083/jcb.151.3.673
Napper CE Taylor ME The mannose receptor fails to enhance processing and presentation of a glycoprotein antigen in transfected fibroblasts Glycobiol 2004 14 7C 12C 10.1093/glycob/cwh109
Croizet K Trouttet-Masson S Rabilloud R Nicolas JF Bernier-Valentin F Rousset B Signaling from epithelial to dendritic cells of the thyroid gland: evidence for thyrocyte-derived factors controlling the survival, multiplication, and endocytic activity of dendritic cells Lab Invest 2001 81 1601 1613 11742031
van Egmond M Damen CA van Spriel AB Vidarsson G van Garderen E van de Winkel JG IgA and the IgA Fc receptor Trends Immunol 2001 22 205 211 11274926 10.1016/S1471-4906(01)01873-7
Heystek HC Moulon C Woltman AM Garonne P van Kooten C Human immature dendritic cells efficiently bind and take up secretory IgA without the induction of maturation J Immunol 2002 168 102 107 11751952
Clare-Salzler MJ Brooks J Chai A Van Herle K Anderson C Prevention of diabetes in nonobese diabetic mice by dendritic cell transfer J Clin Invest 1992 90 741 748 1522229
Dalton DK Pitts-Meek S Keshav S Figari IS Bradley A Stewart TA Multiple defects of immune cell function in mice with disrupted interferon-γ genes Science 1993 259 1739 1742 8456300
Smith MJ Koch GLE Differential expression of murine macrophage surface glycoprotein antigens in intra-cellular membranes J Cell Sci 1987 87 113 3312248
Crocker PR Gordon S Properties and distribution of a lectin-like haemagglutinin differentially expressed by stromal tissue macrophages J Exp Med 1986 164 1862 1875 3783087 10.1084/jem.164.6.1862
Kraal G Breel M Janse M Bruin G Langerhans cells, veiled cells and interdigitating cells in the mouse recognized by a monoclonal antibody J Exp Med 1986 163 981 997 3950549 10.1084/jem.163.4.981
Metlay JP Witmer-Pack MD Agger R Crowley MT Lawless D Steinman RM The distinct leukocyte integrins of mouse spleen dendritic cells as identified with new hamster monoclonal antibodies J Exp Med 1990 171 1753 1771 2185332 10.1084/jem.171.5.1753
Apostolopoulos V Barnes N Pietersz GA McKenzie IF Ex vivo targeting of the macrophage mannose receptor generates anti-tumor CTL responses Vaccine 2000 18 3174 3184 10856797 10.1016/S0264-410X(00)00090-6
Kato M Neil TK Fearnley DB McLellan AD Vuckovic S Hart DN Expression of multilectin receptors and comparative FITC-dextran uptake by human dendritic cells Int Immunol 2000 12 1511 1519 11058570 10.1093/intimm/12.11.1511
| 15701168 | PMC550652 | CC BY | 2021-01-04 16:36:44 | no | BMC Immunol. 2005 Feb 8; 6:4 | utf-8 | BMC Immunol | 2,005 | 10.1186/1471-2172-6-4 | oa_comm |
==== Front
BMC Med GenetBMC Medical Genetics1471-2350BioMed Central London 1471-2350-6-81571003810.1186/1471-2350-6-8Research ArticleSNP genotyping to screen for a common deletion in CHARGE Syndrome Lalani Seema R [email protected] Arsalan M [email protected] Susan D [email protected] Michael [email protected] Carlos A [email protected] Laura M [email protected] Nancy L [email protected] Jeffrey A [email protected] William J [email protected] John W [email protected] Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, Texas, USA2 Genome Quebec and McGill University Innovation Centre, McGill University, Montreal, Quebec, Canada3 Department of Anesthesiology, Baylor College of Medicine, Houston, Texas, USA4 Department of Pediatrics, Baylor College of Medicine, Houston, Texas, USA5 Department of Pediatrics (Cardiology), Baylor College of Medicine, Houston, Texas, USA2005 14 2 2005 6 8 8 16 7 2004 14 2 2005 Copyright © 2005 Lalani 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
CHARGE syndrome is a complex of birth defects including coloboma, choanal atresia, ear malformations and deafness, cardiac defects, and growth delay. We have previously hypothesized that CHARGE syndrome could be caused by unidentified genomic microdeletion, but no such deletion was detected using short tandem repeat (STR) markers spaced an average of 5 cM apart. Recently, microdeletion at 8q12 locus was reported in two patients with CHARGE, although point mutation in CHD7 on chromosome 8 was the underlying etiology in most of the affected patients.
Methods
We have extended our previous study by employing a much higher density of SNP markers (3258) with an average spacing of approximately 800 kb. These SNP markers are diallelic and, therefore, have much different properties for detection of deletions than STRs.
Results
A global error rate estimate was produced based on Mendelian inconsistency. One marker, rs431722 exceeded the expected frequency of inconsistencies, but no deletion could be demonstrated after retesting the 4 inconsistent pedigrees with local flanking markers or by FISH with the corresponding BAC clone. Expected deletion detection (EDD) was used to assess the coverage of specific intervals over the genome by deriving the probability of detecting a common loss of heterozygosity event over each genomic interval. This analysis estimated the fraction of unobserved deletions, taking into account the allele frequencies at the SNPs, the known marker spacing and sample size.
Conclusions
The results of our genotyping indicate that more than 35% of the genome is included in regions with very low probability of a deletion of at least 2 Mb.
==== Body
Background
CHARGE Association is characterized by ocular coloboma, cranial nerve abnormalities, common outflow tract heart defects, choanal atresia, cupped-shaped pinnae, Mondini dysplasia of the inner ear [1,2] and growth delay. The embryology and mechanisms of maldevelopment in CHARGE are not well understood. CHARGE Association may be genetically heterogeneous, a possibility supported by the rare and variable chromosomal aberrations observed in a few affected individuals. We have identified a de novo mutation in a semaphorin gene, SEMA3E, in an affected patient, identified upon mapping the translocation breakpoints in an unrelated individual with a de novo balanced translocation involving chromosomes 2 and 7: karyotype 46, XY, t(2;7)(p14;q21.11) [3]. Recently, Vissers et al. have reported mutations in CHD7 gene, a member of the chromodomain family in a substantial number of patients [4]. In a large number of children, however, the genetic mechanism of this complex birth defect remains unidentified. Graham has suggested that a subgroup of children with CHARGE Association have a recognizable syndrome [5] and may have a common pathologic and molecular basis. We have focused our study in this subset of patients and hypothesized that within this group, there is a common chromosomal region where recombination events lead to frequent microdeletion. In a previous study we used short tandem repeat (STR) markers spaced at an average of 5 cM to examine ten CHARGE case-parent trios for a large common deletion [6]. STR markers, because they are multi-allelic, generally are highly informative in each trio and there is little ambiguity regarding Mendelian transmission of parental alleles. The study did not identify any common deletion, but it was limited because of the relatively broad marker spacing. We have now extended the analysis by genotyping 3258 diallelic SNP markers with an average spacing of approximately 800 kb, in the same nuclear families. If microdeletion is the underlying basis of this disorder, then genotyping using this dense set of SNPs would be expected to uncover loci with loss of expected heterozygosity in the probands. We have taken into account the major and minor allele frequencies of each of the 3258 SNP markers and used a metric called the expected deletion detection, EDD [Belmont et al., personal communication] to evaluate specific chromosomal intervals for the probability of detecting a deletion in the sample set. Although no deletion was detected in the CHARGE study sample, we conclude that this method could be generally useful in other studies in which small deletions occur as part of the allelic spectrum of disease.
Methods
Patients
The diagnosis of CHARGE syndrome was established by examination by a participating clinical genetics specialist (CB, JWB, SRL). A medical history questionnaire was completed by the parents or by direct interview. For the core genotyping, 8 Caucasian and 1 Hispanic case/parent trios were selected based on the presence of 4 major criteria or three major and three minor criteria for the syndrome [7]. There were five affected males and four affected females, with ages ranging from five to twenty years. Blood was collected and transformed cell lines were established for these families. This research protocol was reviewed and approved by the Baylor College of Medicine Institutional Review Board.
Genotyping
Genotyping for this study was carried out on an Orchid BioSciences SNPstream UHT platform (Princeton, NJ) as previously described [8]. For this study, an initial set of ~4,200 T/C SNPs were identified and selected from the public databases for incorporation into a genome wide SNP panel. After selection, the complete set of SNPs was arranged into ~350 unique 12-plex reactions for the purpose of performing the assays on the UHT platform. The complete set of markers was then validated on a set of 5 CEPH pedigrees (40 individuals) and in 3 independent populations. A final set of 3258 markers was chosen from these combined datasets for analysis after eliminating SNPs that performed poorly through all populations, SNPs that failed both Hardy-Weinberg and Mendelian error calculations and any SNPs that were not polymorphic in the evaluated populations. An average genome wide spacing of ~800 kb between markers was achieved for this panel.
This genotyping method utilized Orchid's single base primer extension chemistry (SBE) to identify which bases were present at the site of interrogation. Multiplexed reactions (12-plex) were performed in a single tube that incorporated labeled chain terminating nucleotides onto the ends of the SBE oligonucleotides. These reactions were then hybridized onto a microarray format, that facilitates the solid-phase sorting of the labeled extension-primers to a set of universal tagged primers arrayed on the surface of the plate. The universal tags were arranged on the surface of the microarray plate in a 384-well microwell layout. This microarray format created a generic design consisting of 384 4 × 4 arrays that contained 12 oligonucleotides that corresponded to 12 unique universal capture tags. The four additional oligonucleotides, plotted in each array, were used for positive and negative controls. Genotyping calls were determined by the presence or absence of incorporated dyes that appeared at each spot on the printed arrays.
TaqMan polymerase chain reaction
Two Assays-on-demand SNPs, rs422951 and ss1309424, flanking rs431722 were obtained from Applied Biosystems and genotyping was performed in 384 well-plates, using the TaqMan polymerase chain reaction-based method. The final volume reaction was 5 μl using 12 ng of genomic DNA, 2.5 μl of Taqman Master mix and 0.25 μl of 20X Assays-on-Demand SNP Genotyping Assay Mix. The plate was heated at 95° for 10 minutes, followed by forty cycles of denaturation at 92° for 15 seconds and annealing/extension at 60° for 1 minute. PCR plates were read on ABI PRISM 7900HT instrument with SDS v2.0 software. Individual genotypes that were ambiguous were excluded.
Fish
Bacterial Artificial chromosomes (BACs) were selected from the public database [9] and obtained from Children's Hospital Oakland Research Institute. DNA extraction was performed according to the standard protocol. Fluorescence in situ hybridization was performed as described elsewhere [6]. Detection of the digoxigenin labeled probe was performed with anti-digoxigenin conjugated to rhodamine, giving a red signal. Biotin labeled control probe was detected with FITC (fluorescein isothiocyanate), giving a green signal. The chromosomes were counterstained with DAPI and analyzed with a Zeiss Axioskop fluorescence microscope equipped with appropriate filter combinations. Approximately 10 metaphase preparations were scored for each hybridization.
Data analysis
The genotype error rate was estimated using the method of Gordon et al. [10] and as implemented in CUE [11]. Expected deletion detection (EDD) is a new method designed for this study, which uses the allele frequency, the marker spacing and the number of pedigrees sampled to estimate the probability that a common deletion would be missed because of ambiguous genotype outcomes. Qualitatively, the information available in a single SNP marker for the purposes of detecting a deletion by lack of expected heterozygosity in a case-parent trio is limited by the many genotype configurations that could appear consistent with Mendelian inheritance, but actually harbor a deletion. Inclusion of 2 or more SNP markers in a deletion interval decreases the likelihood that a common deletion goes undetected.
Results
SNP genotyping data were subjected to analysis for Hardy-Weinberg equilibrium. As an additional test of marker integrity, a transmission disequilibrium test was also performed for each of the SNP markers to examine for distortions in allele transmission in the trios. Unequal transmission of alleles from the heterozygote parents to the affected offspring was not determined by this analysis. Non-paternity was excluded in core pedigrees. Using the method of Gordon [12], we used the genotyping data to estimate the genotyping error rate at 0.02%. Analysis of the data showed transmission inconsistent with Mendelian inheritance for 22 markers on different chromosomes (20 with 1 inconsistency, 1 with 2 inconsistencies, and 1 with 4 inconsistencies). Given the underlying genotyping error rate, we could predict that markers with >2 inconsistencies would be highly unlikely to occur. One marker, rs431722 with overall call rate of 95%, showed Mendelian inconsistency in 4 trios. This SNP was found to lie within the intron 2 of the NOTCH4 gene on chromosome 6p21.32. Human DNA sequence from clone XXbac-300A18 (GenBank accession number AL662884) on chromosome 6p21 was used for FISH. This BAC clone encompasses the NOTCH4 gene, and was confirmed by PCR amplification of the clone sequence using NOTCH4 specific primers (data not shown). The analysis of the metaphase chromosomes after staining with DAPI showed two bright hybridization signals, indicating the presence of both alleles (Figure 1). Two additional SNPs, rs422951 and ss1309424, flanking rs431722 within 780 bp, were genotyped using TaqMan chemistry. The results showed inheritance of biparental alleles in all four pedigrees. The CHD7 locus on 8q12 was investigated additionally with FISH using RP11-33I11 (GenBank accession number AC113143) and RP11-414L17 (GenBank accession number AC023102). Using these BAC clones, microdeletion of this region was excluded in all the affected patients in this study sample.
Because none of the pedigrees were consanguineous, we used parental data to estimate the allele frequencies for each of the 3258 SNP markers [13]. The EDD was then calculated for each chromosome. The percent coverage ranged from 51% on chromosome 20 to 20% on chromosome 15, with a mean of 36% for an autosomal deletion of 2 Mb (Figure 2).
Discussion
Despite various efforts to understand the molecular basis of CHARGE syndrome, with candidate genes sequencing [14,15], comparative genomic hybridization [16], and genome-wide scan for microdeletion(s) using microsatellite markers [6], the underlying molecular mechanism in many patients remained unknown until recently [4]. Based on the complex phenotype and clinical overlap with Velocardiofacial syndrome, it is a plausible hypothesis that in a subset of CHARGE patients with a homogenous phenotype, the underlying genetic mechanism is a cryptic submicroscopic deletion involving highly pleiotropic gene(s). To address this hypothesis, we had previously used microsatellite markers to ascertain loci with loss of expected heterozygosity in case-parent trios. SNPs are far more abundant than microsatellite markers but have not yet been used extensively in linkage and loss of heterozygosity (LOH) studies. The present study represents the application of SNPs to scan for potential submicroscopic deletions across the autosomes. Amos et al. have previously shown that SNP genotyping of child-parent trios provides valuable information about the presence of de novo microdeletion when sufficient families are studied [17]. However, this method is most appropriate when linkage disequilibrium is accounted for because of high SNP marker density. They have provided a general analytical framework and point out the effects of non-paternity, sample mix-up, and genotyping error in the interpretation of Mendelian inconsistency in case-parent trio data with biallelic markers. As expected, increases in rates of such phenomenon in the data decrease power to detect a microdeletion. In addition, they point out that heterogeneity in the position of a putative deletion also has a large impact on power. Their analysis is particularly apt given the probable future availability of extremely high density SNP marker maps and the technical capability to genotype hundreds of thousands of markers per research subject. However, they do not explore the effect of intermarker distance in the ability to detect deletions of various sizes.
Assuming the genotyping error rate of 0.5%, the probability of observing >2 inconsistencies per marker is very low. The SNP marker rs431722 showed Mendelian inconsistency in four of nine pedigrees, with apparent loss of a parental allele in each case i.e. a frequency much higher than expected for the genotyping error rate. Interestingly, this SNP is located within the NOTCH4 gene on chromosome 6p21.32. The Notch gene family encodes highly conserved transmembrane receptors that are involved in intercellular signaling. The Notch signaling pathway plays an essential role in regulating embryonic vascular morphogenesis and remodeling [18]. Moreover, disruption of Notch signaling via mutation in the Notch ligand JAG1 is known to result in Alagille syndrome [19], nonsyndromic Tetralogy of Fallot (TOF) [20] and possibly nonsyndromic biliary extrahepatic atresia [21]. Since TOF is a heart defect commonly seen in CHARGE Syndrome, NOTCH4 was further studied with FISH as well as flanking SNP markers for microdeletion. The results, however conclusively excluded a discernible microdeletion at this locus.
This screen is expected to detect deletions of about 1–2 Mb depending on the overlap of the SNP markers with the deletion interval. Variable coverage for each chromosome was determined for approximately 2 Mb microdeletion in this study. Almost 50% of chromosomes 7, 19 and 20 were excluded for any microdeletion greater than 2 Mb. The least coverage was observed for chromosome 15 and 18, with exclusion of 20% of the chromosome for the presence of a similar genetic aberration. Overall, we can estimate that approximately 36% of the genome had >80% chance for detecting a common 2 Mb deletion in at least 2 patients with CHARGE Syndrome.
There are several limitations to this approach in studying the genetics of CHARGE syndrome. Although the marker density is high, the reduced amount of information per marker means that only some of the trios give the possibility of a conclusive result. Denser marker sets would be predicted to fill most of the gaps, but the regions around the centromeres are likely to be difficult with any currently available technique.
The strategy out lined in this paper would work equally well for conventional Mendelian traits in which the mutant alleles included at least some deletions. Using a much denser SNP map of 1 marker every 5–10 kb, as is anticipated for whole genome association analyses, it would be possible to detect most deletions given a sufficient representation of deletions within the spectrum of gene mutations.
Conclusions
In this report we show that a SNP genotyping screen has excluded moderate length submicroscopic deletions in a subset of patients with CHARGE syndrome. Further analysis by microarray comparative genome hybridization methods or denser SNPs will allow a comprehensive assessment of the role, if any, of microdeletions in CHARGE syndrome.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SRL carried out the data analysis, FISH and TaqMan assays and drafted the manuscript. AMS assisted in the data analysis. MP carried out the SNP genotyping, JWB conceived of the study, and participated in its design and coordination. He performed the statistical analysis and drafted the manuscript. SDF, CAB, and NLG participated in patient enrollment and LMM assisted in DNA extraction. WJC and JAT assisted in manuscript preparation. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank the families for participating in this research and the CHARGE Syndrome Foundation for its support. We also acknowledge Kim McBride for providing the reagents for the TaqMan assay. This work was supported by NIH HD39056 (JWB) and the Doris Duke Charitable Foundation (SRL). Dr. Towbin is funded by the Texas Children's Foundation Chair in Pediatric Cardiac Research.
Figures and Tables
Figure 1 FISH analysis of NOTCH4 gene. Two copies of the locus on 6p21.32 using XXbac-300A18 clone.
Figure 2 Deletion coverage. Chromosome-specific maps indicating the Expected Deletion Detection over each interval.
==== Refs
Collins WO Buchman CA Bilateral semicircular canal aplasia: a characteristic of the CHARGE association Otol Neurotol 2002 23 233 4 11875356 10.1097/00129492-200203000-00022
Satar B Mukherji SK Telian SA Congenital aplasia of the semicircular canals Otol Neurotol 2003 24 437 46 12806296 10.1097/00129492-200305000-00014
Lalani SR Safiullah AM Molinari LM Fernbach SD Martin DM Belmont JW SEMA3E mutation in a patient with CHARGE syndrome J Med Genet 2004 41 E94 15235037 10.1136/jmg.2003.017640
Vissers LE van Ravenswaaij CM Admiraal R Hurst JA de Vries BB Janssen IM van der Vliet WA Huys EH de Jong PJ Hamel BC Schoenmakers EF Brunner HG Veltman JA van Kessel AG Mutations in a new member of the chromodomain gene family cause CHARGE syndrome Nat Genet 2004 36 955 7 15300250 10.1038/ng1407
Graham JM Jr A recognizable syndrome within CHARGE association: Hall-Hittner syndrome Am J Med Genet 2001 99 120 3 11241469 10.1002/1096-8628(2000)9999:999<00::AID-AJMG1132>3.0.CO;2-J
Lalani SR Stockton DW Bacino C Molinari LM Glass NL Fernbach SD Towbin JA Craigen WJ Graham JM JrHefner MA Lin AE McBride KL Davenport SL Belmont JW Toward a genetic etiology of CHARGE syndrome: I. A systematic scan for submicroscopic deletions Am J Med Genet 2003 118A 260 6 10.1002/ajmg.a.20002
Blake KD Davenport SL Hall BD Hefner MA Pagon RA Williams MS Lin AE Graham JM Jr CHARGE association: an update and review for the primary pediatrician Clin Pediatr (Phila) 1998 37 159 73 9545604
Bell PA Chaturvedi S Gelfand CA Huang CY Kochersperger M Kopla R Modica F Pohl M Varde S Zhao R Zhao X Boyce-Jacino MT Yassen A SNPstream UHT: ultra-high throughput SNP genotyping for pharmacogenomics and drug discovery Biotechniques 2002 70-2 74,76 7 12083401
National Center for Biotechnology Information
Gordon D Leal SM Heath SC Ott J An analytic solution to single nucleotide polymorphism error-detection rates in nuclear families: implications for study design Pac Symp Biocomput 2000 663 74 10902214
Calculating Undetected Errors CUE
Gordon D Heath SC Ott J True pedigree errors more frequent than apparent errors for single nucleotide polymorphisms Hum Hered 1999 49 65 70 10077724 10.1159/000022846
SNP Genotype and Allele Frequency
Martin DM Probst FJ Fox SE Schimmenti LA Semina EV Hefner MA Belmont JW Camper SA Exclusion of PITX2 mutations as a major cause of CHARGE association Am J Med Genet 2002 111 27 30 12124729 10.1002/ajmg.10473
Tellier AL Amiel J Delezoide AL Audollent S Auge J Esnault D Encha-Razavi F Munnich A Lyonnet S Vekemans M Attie-Bitach T Expression of the PAX2 gene in human embryos and exclusion in the CHARGE syndrome Am J Med Genet 2000 93 85 8 10869107 10.1002/1096-8628(20000717)93:2<85::AID-AJMG1>3.0.CO;2-B
Sanlaville D Romana SP Lapierre JM Amiel J Genevieve D Ozilou C Le Lorch M Brisset S Gosset P Baumann C Turleau C Lyonnet S Vekemans M A CGH study of 27 patients with CHARGE association Clin Genet 2002 61 135 8 11940088 10.1034/j.1399-0004.2002.610208.x
Amos CI Shete S Chen J Yu RK Positional Identification of Microdeletions with Genetic Markers Hum Hered 2003 56 107 118 14614244 10.1159/000073738
Krebs LT Xue Y Norton CR Shutter JR Maguire M Sundberg JP Gallahan D Closson V Kitajewski J Callahan R Smith GH Stark KL Gridley T Notch signaling is essential for vascular morphogenesis in mice Genes Dev 2000 14 1343 52 10837027
Li L Krantz ID Deng Y Genin A Banta AB Collins CC Qi M Trask BJ Kuo WL Cochran J Costa T Pierpont ME Rand EB Piccoli DA Hood L Spinner NB Alagille syndrome is caused by mutations in human Jagged1, which encodes a ligand for Notch1 Nat Genet 1997 16 243 51 9207788 10.1038/ng0797-243
Eldadah ZA Hamosh A Biery NJ Montgomery RA Duke M Elkins R Dietz HC Familial Tetralogy of Fallot caused by mutation in the jagged1 gene Hum Mol Genet 2001 10 163 9 11152664 10.1093/hmg/10.2.163
Kohsaka T Yuan ZR Guo SX Tagawa M Nakamura A Nakano M Kawasasaki H Inomata Y Tanaka K Miyauchi J The significance of human jagged 1 mutations detected in severe cases of extrahepatic biliary atresia Hepatology 2002 36 904 12 12297837
| 15710038 | PMC550653 | CC BY | 2021-01-04 16:03:33 | no | BMC Med Genet. 2005 Feb 14; 6:8 | utf-8 | BMC Med Genet | 2,005 | 10.1186/1471-2350-6-8 | oa_comm |
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-91571792210.1186/1471-2474-6-9Case ReportBilateral recurrent discloation of the patella associated with below knee amputation: A case report Batra Sumit [email protected] Ratnesh [email protected] Prasanna [email protected] Department of Orthopaedics, National Institute for the Orthopaedically Handiccaped, B.T. Road, Bon Hooghly, Kolkata- 700090. India2 Director, National Institute for the Orthopaedically Handiccaped, B.T. Road, Bon Hooghly, Kolkata- 700090. India3 Department of Prosthetics & Orthotics, National Institute for the Orthopaedically Handiccaped, B.T. Road, Bon Hooghly, Kolkata- 700090. India2005 17 2 2005 6 9 9 22 7 2004 17 2 2005 Copyright © 2005 Batra et al; licensee BioMed Central Ltd.2005Batra 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
Recurrent dislocation of the patella in patients with below knee amputation is a known entity. Abnormally high-riding patella (patella alta) and medial patellofemoral ligament insufficiency in these patients predisposes them to patellar instability. The established treatment of this problem is surgical realignment.
Case presentation
A 25 year old male patient with bilateral below knee amputation presented with bilateral recurrent dislocation of the patella while walking on knees on uneven ground. Clinical and radiographic studies showed patella alta. A simple shoe modification was used to treat this patient.
Conclusions
A simple shoe modification can be used to treat such a condition which is otherwise treated surgically.
==== Body
Background
Recurrent dislocation of the patella can follow a violent initial dislocation, but occur more often in knees with one or more underlying anatomic abnormalities that predispose the patella to dislocation or subluxation. In these knees, less trauma is needed for dislocation to occur. The underlying pathologic condition causes an abnormal excursion of the extensor mechanism over the femoral condyles. High-riding patella (patella alta) and a damaged medial patellofemoral ligament at the time of first episode of dislocation leads to such an abnormality and leads to recurrent dislocation of the patella [1-4]. Patella alta has been reported in patients with below knee amputation using patellar tendon bearing prosthesis. The usual treatment in these cases is surgical reconstruction. We present a case of bilateral recurrent dislocation of the patella with below knee amputation which was managed conservatively.
Case presentation
A 25 years old bilateral below knee amputee presented with recurrent dislocation of the patella while walking on knees in emergent situations on uneven ground without the prosthesis. Amputation was performed at the age of 15 years as a result of train accident. Since then he has been using patellar tendon bearing (PTB) type below knee prosthesis on both sides. First episode of dislocation occurred after 5 years of amputation. The patient used to walk on his knees without using the prosthesis for in-house activities on uneven ground. His patella used to dislocate whenever there was an unnoticed pressure on the medial side of the knee. On clinical examination, patella alta and positive apprehension test were noted on both sides. The ratio of patellar length to patellar tendon length was 0.8 on both sides demonstrating relative elongation of the patellar tendon. The normal ratio is 1.0 [1]. Modified shoes were given to the patient which were moulded in the inner surface around the patellar tendon and femoral condyles to provide uniform distribution of weight over a wider area. Another moulding was done on lateral side that prevented excessive movement of patella laterally [Fig. 1]. The patient was allowed to walk on knees after wearing these shoes [Fig. 2]. At 6 months follow up the patient is doing well with no recurrence.
Figure 1 Close-up photograph of the shoe (arrow = moulding on the lateral side providing protective force against dislocation)
Figure 2 Patient wearing modified shoes
Discussion
High-riding patella (known as Patella alta) leads to patellar instability [1]. In the presence of such instability, sudden laterally directed forces can lead to dislocation of the patella. The first episode of lateral dislocation of patella invariably damages the medial patellofemoral ligament (MPFL) which is the primary passive restraint to lateral patellar displacement[4,5]. An injured MPFL leads to further instability and both the factors combined predispose to recurrent dislocation of the patella. In below knee amputees using patellar tendon bearing (PTB) prosthesis, the prolonged, upwardly directed force against the patellar tendon gradually elongate the tendon and produce patella alta [2,3]. It takes a long time for such an instability to develop. The first incidence of patellar dislocation occurred after 5 years of amputation in our patient. Similar time lag was described by Mowery et.al. in his patients where the time lag was from 5–13 years [3].
The patient described here used to walk on knees for certain emergent and short mobility in-house activities e.g., going to toilet especially at night, going to kitchen etc. As most of rural houses in developing countries do not have cemented floors, walking on uneven ground cannot be prevented. Walking on knees in such conditions produces eccentric forces on the patella. Sudden laterally directed forces which might result when the knee strikes some elevated surface on medial side can lead to dislocation of patella in patients with patellar instability.
Walking on knees is usually discouraged in below knee amputees as it can lead to flexion contractures of the knee joint. In cases of bilateral amputees it is not always possible for the patients to wear prosthesis on both sides in emergency situations as discussed above. Considering these points, the occasional walking on knees could not be avoided in this patient. Surgical correction of instability would not have helped much as walking on knees on uneven ground could have nullified the results of surgery. A modified well moulded shoe was given to the patient for walking on knees that prevented any eccentric forces and protected the patella against dislocation.
6 months follow-up results were excellent in this patient with no episode of dislocation on either side. The patient was very comfortable in using the shoes as it takes him only 15 seconds to wear them. Hence a simple low cost shoe modification has been used to treat a condition which is mostly treated surgically.
Conclusions
Recurrent dislocation of the patella in cases of below knee amputees using patellar tendon bearing prosthesis is very rare and the usual treatment is surgical realignment. To our knowledge no case of bilateral recurrent dislocation of the patella in below knee amputees has been described in the literature. A simple shoe modification can be used in difficult situations to treat such a condition.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SB was the principle surgeon who planned the treatment protocol of this patient, in addition to conceptualizing and drafting the article.
RK guided the designing of the shoe.
PL was the prosthetist who designed and made the shoe.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Written consent was obtained from the patient for publication of study.
==== Refs
Insall J Goldberg V Salvati E Recurrent dislocation and the high riding patella Clin Orthop 1972 88 67 9 5086583
McIvor J Gillespie R Patellar instability in juvenile amputees J Pediatr Orthop 1987 7 553 6 3624466
Mowery CA Herring JA Jackson D Dislocated patella associated with below-knee amputation in adolescent patient J Pediatr Orthop 1986 6 299 301 3711321
Arendt EA Fithian DC Cohen E Current concepts of lateral patella dislocation Clin Sports Med 2002 21 499 519 12365240
Amis AA Firer P Mountney J Senavongse W Thomas NP Anatomy and biomechanics of the medial patellofemoral ligament Knee 2003 10 215 20 12893142 10.1016/S0968-0160(03)00006-1
| 15717922 | PMC550654 | CC BY | 2021-01-04 16:32:04 | no | BMC Musculoskelet Disord. 2005 Feb 17; 6:9 | utf-8 | BMC Musculoskelet Disord | 2,005 | 10.1186/1471-2474-6-9 | oa_comm |
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-191571793210.1186/1471-2164-6-19Research ArticleThe PECACE domain: a new family of enzymes with potential peptidoglycan cleavage activity in Gram-positive bacteria Pagliero Estelle [email protected] Otto [email protected] Thierry [email protected] Guilmi Anne Marie [email protected] Laboratoire d'Ingénierie des Macromolécules Institute de Biologie Structurale Jean-Pierre Ebel (CEA-CNRS UMR 5075-UJF), 41 Rue Jules Horowitz 38027 Grenoble cedex 1, France2 Laboratoire de Cristallographie Macromoléculaire Institut de Biologie Structurale Jean-Pierre Ebel (CEA-CNRS UMR 5075-UJF), 41 Rue Jules Horowitz 38027 Grenoble cedex 1, France2005 17 2 2005 6 19 19 22 10 2004 17 2 2005 Copyright © 2005 Pagliero et al; licensee BioMed Central Ltd.2005Pagliero 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 bacterial peptidoglycan is a dynamic process, synthases and cleavage enzymes are functionally coordinated. Lytic Transglycosylase enzymes (LT) are part of multienzyme complexes which regulate bacterial division and elongation. LTs are also involved in peptidoglycan turnover and in macromolecular transport systems. Despite their central importance, no LTs have been identified in the human pathogen Streptococcus pneumoniae. We report the identification of the first putative LT enzyme in S. pneumoniae and discuss its role in pneumococcal peptidoglycan metabolism.
Results
Homology searches of the pneumococcal genome allowed the identification of a new domain putatively involved in peptidoglycan cleavage (PECACE, PEptidoglycan CArbohydrate Cleavage Enzyme). This sequence has been found exclusively in Gram-positive bacteria and gene clusters containing pecace are conserved among Streptococcal species. The PECACE domain is, in some instances, found in association with other domains known to catalyze peptidoglycan hydrolysis.
Conclusions
A new domain, PECACE, putatively involved in peptidoglycan hydrolysis has been identified in S. pneumoniae. The probable enzymatic activity deduced from the detailed analysis of the amino acid sequence suggests that the PECACE domain may proceed through a LT-type or goose lyzosyme-type cleavage mechanism. The PECACE function may differ largely from the other hydrolases already identified in the pneumococcus: LytA, LytB, LytC, CBPD and PcsB. The multimodular architecture of proteins containing the PECACE domain is another example of the many activities harbored by peptidoglycan hydrolases, which is probably required for the regulation of peptidoglycan metabolism. The release of new bacterial genomes sequences will probably add new members to the five groups identified so far in this work, and new groups could also emerge. Conversely, the functional characterization of the unknown domains mentioned in this work can now become easier, since bacterial peptidoglycan is proposed to be the substrate.
==== Body
Background
The bacterial cell wall resists intracellular pressure and gives the bacterium its particular shape. Cell wall reinforcement is brought about by a strong scaffolding structure, the peptidoglycan, which is formed by glycan strands and peptide chains held together by covalent bonds, resulting in a mono- or multilayered network. The glycan strands are composed of N-acetylglucosamine (GlcNAc) and N-acetylmuramyl (MurNAc) residues linked together by β-1,4 glycosidic bonds. Peptides are covalently attached to the lactyl group of the muramic acid and their cross-linking results in the net structure of the peptidoglycan (Fig. 1a).
Figure 1 Schematic representation of peptidoglycan and of cleavage enzymes in S. pneumoniae. (a) Scheme of the pneumococcal peptidoglycan, indicating the chemical bonds cleaved by identified hydrolases in blue. The MurNAc residue containing the 1, 6-anhydro bond resulting from LT reaction is in a green circle. The putative LT pneumococcal enzyme appears in red, while enzymes CBPD and PcsB for which no enzymatic specificity is yet characterized are in black. (b) Topological representation of the glycan strand hydrolases described in S. pneumoniae. Black and hatched boxes indicate the signal peptide and the transmembrane anchor, respectively. The blue boxes illustrate the respective enzymatic active domains. Purple rectangles correspond to the Choline-Binding repeats. Green and orange boxes correspond to SH3b and coiled-coil regions, respectively. The topology was designed with the help of SMART server [39].
Peptidoglycan is synthesized in a multi-stage process. The first steps occur in the cytoplasm, where a set of enzymatic reactions gives rise to the assembly of the MurNAc-pentapeptide. This unit is in turn linked to the carrier undecaprenol lipid via a pyrophosphate group; afterwards the GlcNAc group is added, generating the lipid II precursor. The saccharidic and peptidic moieties of lipid II are subsequently exposed to the periplasmic space. At this stage, peptidoglycan biosynthesis involves polymerization of the glycan chains, catalyzed by glycosyltransferases [1] as well as interpeptide bridge formation performed by transpeptidases [2]. These two enzymatic reactions are resident on the extracellular domains of Penicillin-Binding Proteins (PBPs) which are membrane-associated molecules, present in all eubacteria [2].
Peptidoglycan metabolism is a dynamic process since this structure grows and divides in perfect synchronization with cell growth and division. Furthermore, it is well established that peptidoglycan is subject to maturation, turnover and recycling in Gram-negative bacteria [3]. To fullfil these processes, it is expected that peptidoglycan cleavage enzymes must exert their functions in coordinated action with PBPs. Indeed, a large range of different peptidoglycan hydrolases have been identified in numerous bacterial species and specific peptidoglycan hydrolases exist for almost each covalent bond [3] (Fig. 1a). The polysaccharidic component of peptidoglycan is the target of several hydrolases: the β-1,4 glycosidic bond between MurNAc and GlcNAc residues is cleaved by lyzosyme and by lytic transglycosylases (LT), the β-1,4 glycosidic bond between GlcNAc and MurNAc is hydrolyzed by glucosaminidases and amidases are responsible for the cleavage of the MurNAc-L-alanine bond (Fig. 1a).
Lyzosyme and LT enzymes cleave the same β-1,4-MurNAc-GlcNAc bond but generate different reaction products: while lyzosymes catalyze a hydrolytic reaction, LTs cleave the β-glycosidic linkage with the concomitant formation of 1,6-anhydromuramyl residues, blocking the reducing end of the glycan strands. The significance of the ring structure is not known but it has been speculated that the bond energy may be utilized for glycan strand rearrangements. In addition, the 1,6-anhydro ring may also be considered as a specific product of peptidoglycan turnover. Despite the lack of understanding of the physiological function of anhydromuropeptide product, LT enzymes must play a significant cellular role. Indeed, it has been observed that deletions of genes encoding LT proteins lead to E. coli and Neisseria meningitidis with altered cell separation phenotypes, indicating that LTs cleave septal peptidoglycan [4,5]. Macromolecular transport systems (secretion types II, III, IV and IV pilus synthesis) of Gram-negative bacteria contain LT enzymes, suggesting that peptidoglycan hole formation (essential for transport functions) is specifically performed by this enzyme family [6]. As mentioned above, the enlargement of the bacterial stress-bearing peptidoglycan structure requires the well coordinated action of synthases (PBPs) and hydrolase enzymes. The "three-for-one" growth mechanism described by Höltje proposes that a triplet of glycan strands cross-linked to each other (resulting from PBPs synthesis) is attached to the peptidoglycan layer. Subsequently, the docking strand is removed by hydrolases resulting in the insertion of the peptidoglycan triplet. The hydrolases involved in such multienzyme complexes are endopeptidases and LT enzymes [3]. This hypothesis is supported by experimental data as LT and PBPs could be co-purified from E. coli extracts [7-9]. In conclusion, LT enzymes play an important cellular role in diverse aspects of cell biology as expected from their presence in a very wide range of eubacteria as well as archaebacteria [3,10,11].
Surprinsingly, no such LT enzyme has been identified to date in the human pathogen Streptococcus pneumoniae, the causative agent of ear infections in children, as well as meningitis and pneumonia. The pattern of peptidoglycan hydrolases in this Gram-positive bacteria includes, besides a D, D-carboxypeptidase, five glycan strand cleaving enzymes (Fig 1b). Four of these are surface-exposed proteins harboring Choline-Binding Domains which are non-covalently bound to choline residues present on cell wall pneumococcal teichoic and lipoteichoic acids [12-14]. The Choline-Binding Proteins (CBPs) catalyzing peptidoglycan hydrolysis are LytA, LytB, LytC and potentially CBPD (Fig 1b). LytA is an amidase and also appears as an autolytic enzyme, causing bacteriolysis when acting in an uncontrolled manner [15]. LytB is a glucosaminidase involved in cell separation as lytB mutants form very long chains of over 100 cells [16]. LytC is a lysozyme with an autolytic behavior at 30°C [17]. Finally, CBPD and PcsB contain a CHAP domain (Cysteine, Histidine-dependent amidohydrolase/peptidase) predicted to hydrolyse the peptidoglycan in pneumococcus, but definitive biochemical data are still lacking [18-20].
Our interest in the biology of S. pneumoniae led us to investigate the presence of LT enzymes in this bacteria. Homology searches of enzyme sequences within the pneumococcus genome using bioinformatics tools allowed the identification of a new domain harboring motifs that infer potential peptidoglycan cleavage activity. For this reason we named this domain PECACE (PEptidoglycan CArbohydrate Cleavage Enzyme). This domain sequence was found exclusively in Gram-positive bacterial species, suggesting a significant cellular role. Finally, the PECACE domain is in some instances found in association with other domains, known to catalyze peptidoglycan hydrolysis: this observation reinforces the predicted function of PECACE as participating in peptidoglycan cleavage and represents another example of multifunctional proteins involved in peptidoglycan metabolism.
Results and discussion
Identification of a protein harboring the PECACE domain in S. pneumoniae
The C-terminal domain of Escherichia coli Slt70 (Soluble Lytic Transglycosylase) has a lysozyme-like fold and its amino acid sequence was employed in a search of Bacilli genomes within the NCBI Conserved Domain Search server [11,21-23]. Thirty-four Slt70-homologue sequences were retrieved using an inclusion threshold of 0.01. None of these sequence originated from the S. pneumoniae translated genome. Subsequently, each of these 34 sequences was compared with the non-redundant protein database using PSI-BLAST with a E-value threshold of 0.005 and 5 sequences showed significant matches with a unique protein in S. pneumoniae. This sequence (accession numbers NP358524, gi:15902974) contains 204 amino acids: the first 21 amino acids are predicted to form a transmembrane anchor and the subsequent 192-residue region is putatively exposed to the extracellular space (Fig. 1b). This S. pneumoniae NP358524 sequence has been tested as a pneumococcal vaccine antigen on the basis of preliminary screens for novel vaccine candidates [24].
A three-dimensional fold prediction of the S. pneumoniae NP358524 protein was performed with the 3D-PSSM server [25] which identified two matches: E. coli Slt70 (d1qsaa2, E-value:10-7) and LysG (G-type goose lyzozyme, d1531, E-value:10-3). The sequence alignment between NP358524 and Slt70 is shown in Fig. 2, defining the PECACE domain in the pneumococcal protein. The secondary structures are also reported, based on three-dimensional structures of Slt70 and on computational predictions for PECACE and suggest that the latter is highly α-helical (Fig. 2). It is of note that both Slt70 and LysG are highly similar, and both lack the catalytic aspartate residue commonly found in the active site of lysozymes [10,11,21,22]. Therefore, the PECACE domain of the NP358524 sequence appears to belong to this group of bacterial lysozymes, characterized by the absence of an aspartate residue in the catalytic site and is part of the Glycoside Hydrolase family 23 based upon CAZy classification . The catalytic acid residue in the PECACE domain is most probably Glu61 since it aligns with the catalytic Glu478 residue in the Slt70 sequence (Fig. 2). The serine residue following the catalytic glutamate and the GLMQI/V motif are essential for active-site architecture and are conserved between Slt70 and LysG. In the PECACE sequence, a threonine residue follows the catalytic glutamate and the GLMQI/V motif differs since the corresponding sequence is D(68)VMQS (Fig. 2). Finally, the second motif AYNxG which has been shown to be involved in the interaction with the substrate for Slt70 (A551YNxG) is well conserved in the PECACE sequence (A117YNxG).
Figure 2 Alignment of the PECACE domain with Slt70. Protein fold recognition was performed with the 3D-PSSM server. The NP358524 sequence (residues 31–145) from S. pneumoniae (PECACE domain) is aligned with Slt70 from E. coli (P03810, residues 478–616). Amino acids of Slt70 involved in the catalytic reaction and in ligand recognition are underlined while residues conserved in each alignment are highlighted in red. The structural prediction for S. pneumoniae PECACE domain was determined (H = helix, C = coil) while Slt70 secondary-structure information was obtained from PDB file 1QSA.
Based on this sequence analysis, we infer that the S. pneumoniae NP358524 protein, through its PECACE domain, probably catalyzes the peptidoglycan cleavage of the β-1,4-MurNAc-GlcNAc bond by employing Glu61 as the catalytic residue.
Identification of the PECACE domain in Gram-positive bacteria
The 204 amino acid sequence from S. pneumoniae NP358524, containing the PECACE domain, was used as a PSI-BLAST search query. In total, 29 distinct proteins, all from Gram-positive bacteria, were identified (E-value: 10-5) and no sequences from Gram-negative bacteria were retrieved. These sequences were aligned with ClustalW and manually edited. A conserved pattern could be extracted from this alignment: E- [ST]-X-G-X(1,16)-D-X-M-Q- [SA]- [SA]-E- [SG] which was used to search for additional sequences, but no new sequence could be detected from databases, even with a degenerated pattern. PSI-BLAST performed through the GOLD server led to the identification of 10 new sequences from Gram-positive bacteria [26]. In summary, out of the about 50 Gram-positive bacteria for which the whole genome sequence is available, 34 of them contain at least one protein harboring the PECACE domain. The final alignment of these sequences with the S. pneumoniae PECACE domain is shown in Fig. 3. The putative catalytic glutamate residue, Glu61 in the S. pneumoniae PECACE domain, is conserved in all sequences and the following residue is a Ser or Thr in accordance with Slt70 and LysG patterns. In addition, the D(68)VMQS motif in the S. pneumoniae PECACE domain is also well represented in the large majority of sequences with the consensus sequence DI/VMQSSES. Finally, the second motif AYNxG is also conserved while the Ala residue is often replaced by a Ser. In conclusion, the features identified in the S. pneumoniae PECACE domain regarding the potential enzymatic properties of peptidoglycan polysaccharide cleavage are also shared by the similar PECACE domains in Gram-positive bacteria.
Figure 3 Sequence alignment of PECACE domains identified in Gram-positive bacteria. Multiple sequence alignment was constructed using ClustalW. The lengths of the insertions in the sequences are shown in parentheses. The sequences are denoted by their GenBank Identifier (gi). The domain limits are indicated by the residue positions (first-end). The amino acids identified as catalytic or involved in ligand recognition are marked with asterisks under PECACE sequence. Alignments are coloured using the CHROMA tool using default parameters [40]. Full sequence details, group (i): Streptococcus pneumoniae R6 (gi:15902974), Streptococcus mitis NCTC 12261 (§SMT1418), Streptococcus sanguinis SK36 (&:SS_A352_G10), Streptococcus gordonii (gi:18389219), Streptococcus suis P1/7 (suis166b12), Streptococcus uberis 0140J (sub49a04), Streptococcus equi (equi324d3), Streptococcus equi subsp. Zooepidemicus (zoo26g07), Streptococcus pyogenes M1 GAS (gi:15675124), Streptococcus agalactiae 2603V/R (gi:22537230), Lactococcus lactis subsp. Cremoris SK11 (scaffold18), Streptococcus mutans UA159 (gi:24379517), Streptococcus thermophilus LMD-9 (scaffold3), Lactococcus lactis subsp. Lactis (gi:15672584), Enterococcus faecium DO (2351355_Cont543), Enterococcus faecalis V583 (gi:29376084), Bacillus subtilis subsp. subtilis str. 168 (gi:16078973), Bacillus cereus ATCC 14579 (gi:30020591), Oceanobacillus iheyensis HTE831 (gi:23100516), group (ii): Bacillus anthracis: (pXO2-08) (gi:10956398), Enterococcus faecalis: (pRE25) (gi:12957015), Enterococcus faecium (gi:22992993), Enterococcus faecalis V583 (gi:29376781), Clostridium difficile 630 (Cd81d2), Enterococcus faecalis V583 (gi:29376405), Clostridium perfringens (gi:13274506), Staphylococcus aureus subsp. aureus Mu50 (gi:15923390), Listeria monocytogenes EGD-e (gi:16803144), Streptococcus agalactiae 2603V/R (gi:22537089), Enterococcus faecium (gi:22993467), Bacillus subtilis subsp. subtilis str. 168 (gi:16077564, group (iii): Bacillus cereus ATCC 14579 (gi:30021796), group (iv): Enterococcus faecalis BM4518 (gi:33355845), group (v): Bacillus anthracis str. A2012: (pXO1) (gi:21392795), Bacillus cereus ATCC 10987: (pBc10987) (gi:44004362).
Genomic organization of pecace genes
The genomic organization of pecace genes has been analyzed in a variety of Gram-positive bacteria and a conserved distribution was observed in various streptococci species (Fig. 4). This feature indicates that genetic transfer of the whole cluster may have occured within the streptococci family, providing further evidence regarding the significant importance of the PECACE domains in bacterial physiology. However, the pneumococcal cluster is more related to the S. mitis one than to S. mutans, S. agalactiae and S. pyogenes ones, while clusters of the latter three species are related to each other. Genes located upstream and downstream of pecace are in some instances well characterized but the function of the corresponding proteins could not bring any clues about the role of PECACE, nor any evidence on pecace gene transcription. However, pecace is in all cases found in association with the same gene (whose locus name in S. pneumoniae is spr0929) but no information about the function of the protein encoded by this locus is available in databases. Transcriptional analysis of these two genes may bring informations about their potential co-regulation, a first stage in deciphering cellular function.
Figure 4 Schematic representation of the gene cluster containing pecace in Streptococci species. The coding regions and their direction of transcription are indicated by arrows. Gene names are given on top of the corresponding region.
Domain organization of proteins containing the PECACE domain
The PECACE domain is found in a large range of protein architectures, commonly associated with other peptidoglycan hydrolases, suggesting that these proteins have multiple peptidoglycan cleavages activities (Fig. 5). The identification of proteins displaying the PECACE domain was carried out using NCBI Conserved Domain Search and Pfam servers. In addition, prediction of membrane anchoring was performed with the DAS-Transmembrane Prediction server while extracellular secretion of the protein was deduced from the identification of a signal peptide.
Figure 5 Domain architecture of PECACE proteins. The domain architecture of the proteins containing the PECACE domain was organized according to searches with NCBI Conserved Domain Search server against Pfam database: CHAP/NlpC-P60 (Pfam: PF05257/PF00877), M37 peptidase (Pfam: PF01551), unknown domain 1 (gi: 33355845) and unknown domain 2 (gi: 30021796). The size of the domains is not respected in these representations.
PECACE-containing proteins appear to fall into 5 main categories (Fig. 5): (i) those which display no additional domain, (ii) CHAP-Nlpc/P60 as the associated group, (iii) CHAP-Nlpc/P60 and an unknown domain as associated groups, (iv) domains with no ascribed functions and finally (v) CHAP-Nlpc/P60 and M37 peptidase as associated groups.
The 19 proteins which contain only the PECACE domain belong to group (i) and harbor either a signal peptide or a transmembrane helix (as for the S. pneumoniae protein), leading in both cases to cell surface expression.
The CHAP-Nlpc/P60 domain is commonly associated with the PECACE domain in different modular organizations, namely in groups (ii), (iii), and (v) [18,19,27]. The CHAP domain has been recently described as a Cysteine, Histidine-dependent Amidohydrolase/Peptidase and it has been proposed to hydrolyse peptidoglycan containing γ-glutamyl [18,19]. Indeed, proteins such as N-acetylmuramyl-L-alanine amidase and D-alanyl-glycyl endopeptidase have been described as CHAP-containing enzymes [18,19]. However, while the substrate and the reaction mechanism have not been yet experimentally characterized for the CHAP domain, its role in peptidoglycan hydrolysis is inferred from its presence in multifunctional proteins recognizing peptidoglycan as substrate. Recently, hydrolytic activity of peptidoglycan has been attributed to the CHAP-containing protein PcsB in S. pneumoniae due to abnormal and uncontrolled cell wall synthesis at misplaced septa and formation of long cells in pcsB deleted mutant strains [20]. Proteins from group (ii) are expressed at the cell surface through a transmembrane anchor or are secreted, 12 members have been identified with this topology. Only one sequence (AAQ16265, gi:33355845) from Enterococcus faecalis BM4518 is part of the group (iii), and no function could be identified for the N-terminus domain preceding the PECACE domain. However the former domain is Lys-rich (14%) suggesting an electrostatic interaction with the peptidoglycan as proposed for B. subtilis endopeptidase [28]. Group (iv) is composed of an unique sequence from B. cereus ATCC 14579 (NP 833427, gi:30021796). Neither a signal peptide nor a transmembrane anchor have been detected. Furthermore, the domain of unknown function, which is different from the ones identified in groups (iii) and (v) is present in other multimodular proteins of B. cereus, in association with peptidoglycan hydrolysis enzymes. Finally, two sequences share the architecture defining the group (v) which harbor CHAP-Nlpc/P60 and Peptidase M37 domains [29]. Members of the Peptidase M37 family are generally glycylglycine endo-metallopeptidases; the archetypal member is the lysostaphin enzyme from Staphylococcus species which cleaves the pentaglycine bridge in the peptidoglycan [30]. One group (v) protein (NP 652875, gi:21392795) is encoded by Bacillus anthracis plasmid pXO1 and is required for synthesis of various anthrax toxin proteins [31]; this sequence has neither a signal peptide nor a transmembrane region. The second sequence of group (v) is located on Bacillus cereus ATCC 10987 plasmid pBc10987 (NP 982030, gi:44004362) and contains, in addition to CHAP-Nlpc/P60, Peptidase M37 and PECACE domain as well as an extra sequence to which no function has been attributed but with significant similarity with a B. anthracis plasmid pXO1 sequence (NP 652874, gi:21392794) [32].
Conclusions
In summary, a new domain named PECACE, putatively involved in peptidoglycan cleavage has been identified in S. pneumoniae. The probable enzymatic activity deduced from the detailed analysis of the amino acid sequence suggests a LT-type or goose lyzosyme-type mechanism; we are currently characterising the enzymatic properties and cellular role of the PECACE domain from S. pneumoniae. This new putative pneumococcal peptidoglycan cleavage enzyme differs largely from the other hydrolases already identified in this bacteria. Indeed, LytA, LytB, LytC and CBPD proteins are all bound to the cell wall choline residues and thus expressed at the cell surface. The presence of a signal peptide within the amino acid sequence of PcsB suggests that it is either exposed on the cell surface or secreted. On the contrary, the pneumococcal NP358524 protein displaying the PECACE domain is embeded in the cytoplasmic membrane by a hydrophobic helix. The physiological role of this membranous peptidoglycan cleavage enzyme might differ from the other peptidoglycan hydrolysing enzymes.
Interestingly, the PECACE domain has only been found in Gram-positive bacteria. It is tempting to speculate that the multilayered structure of Gram-positive peptidoglycan relates to the PECACE putative activity.
The architecture of multimodular proteins containing the PECACE domain is another example of the pattern of multiple activities harbored by many peptidoglycan hydrolases, probably needed for the regulation of peptidoglycan metabolism. The release of new bacterial genomes sequences will probably add new members that will complete the five groups identified so far in this work and new groups could also emerge. Conversely, the functional characterization of the unknown domains mentioned in this work should now be easier, as their substrate, the peptidoglycan, is now identified.
Methods
The non-redudant database of protein sequences (National center for Biotechnology Information, NIH, Bethesda) and whole bacterial genomes sequences [26] was searched using BLASP and PSI-BLAST programs with (E) value threshold of 0.005 [33]. Multiple alignments were constructed with ClustalW program [34] followed by manual correction based on PSI-BLAST results. Protein fold recognition through 3D-profiles was searched using 3D-PSSM server [25]. Conserved (and degenerated) amino acid patterns was designed and searched against non-redudant database of protein sequences . Identification of domains associated with PECACE proteins was realized using NCBI Conserved Domain Search [35] and Pfam servers [36]. Finally, prediction of transmembrane anchor and secretory signal peptide were performed with DAS server and SignalP-2.0 servers respectively [37,38].
Authors' contributions
OD conceived of the study and participated in the sequences alignment. EP carried out the sequences analysis and the writing of the manuscript with AMDG. TV coordinated the study. All authors read and approved the final manuscript.
Acknowledgments
This work was supported by the European Commission grant LSMH-CT-COBRA 2003-503335. EP is a recipient of a CEA CFR fellowship. We are grateful to Dr Andréa Dessen for constructive comments and critical review of the manuscript.
==== Refs
van Heijenoort J Formation of the glycan chains in the synthesis of bacterial peptidoglycan Glycobiology 2001 11 25R 36R 11320055 10.1093/glycob/11.3.25R
Goffin C Ghuysen JM Biochemistry and Comparative Genomics of SxxK Superfamily Acyltransferases Offer a Clue to the Mycobacterial Paradox: Presence of Penicillin-Susceptible Target Proteins versus Lack of Efficiency of Penicillin as Therapeutic Agent Microbiol Mol Biol Rev 2002 66 702 738 12456788 10.1128/MMBR.66.4.702-738.2002
Höltje JV Growth of the stress-bearing and shape-maintaining murein sacculus of Escherichia coli Microbiol Mol Biol Rev 1998 62 181 203 9529891
Adu-Bobie J Lupetti P Brunelli B Granoff D Norais N Ferrari G Grandi G Rappuoli R Pizza M GNA33 of Neisseria meningitidis is a lipoprotein required for cell separation, membrane architecture, and virulence Infect Immun 2004 72 1914 1919 15039310 10.1128/IAI.72.4.1914-1919.2004
Heidrich C Ursinus A Berger J Schwarz H Höltje JV Effects of multiple deletions of murein hydrolases on viability, septum cleavage, and sensitivity to large toxic molecules in Escherichia coli J Bacteriol 2002 184 6093 6099 12399477 10.1128/JB.184.22.6093-6099.2002
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
Vollmer W von Rechenberg M Höltje JV 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
Schiffer G Höltje JV Cloning and characterization of PBP1c, a third member of the multimodular class A penicillin-binding proteins of Escherichia coli J Biol Chem 1999 274 32031 32039 10542235 10.1074/jbc.274.45.32031
Romeis T Höltje JV Specific interaction of penicillin-binding proteins 3 and 7/8 with soluble lytic transglycosylase in Escherichia coli J Biol Chem 1994 269 21603 21607 8063800
Blackburn NT Clarke AJ Identification of four families of peptidoglycan lytic transglycosylases J Mol Evol 2001 52 78 84 11139297
Dijkstra BW Thunnissen AM 'Holy' proteins. II: The soluble lytic transglycosylase Curr Opin Struct Biol 1994 4 810 813 7712284 10.1016/0959-440X(94)90261-5
Hoskins JA Arnold WEJr Arnold J Blaszczak LC Burgett S DeHoff BS Estrem ST Fritz L Fu DJ Fuller W Geringer C Gilmour R Glass JS Khoja H Kraft AR Lagace RE LeBlanc DJ Lee LN Lefkowitz EJ Lu J Matsushima P McAhren SM McHenney M McLeaster K Mundy CW Nicas TI Norris FH O'Gara M Peery RB Robertson GT Rockey P Sun PM Winkler ME Yang Y Young-Bellido M Zhao G Zook CA Baltz RH Jaskunas SR Rosteck PR JrSkatrud PL Glass JI Genome of the bacterium Streptococcus pneumoniae strain R6 J Bacteriol 2001 183 5709 5717 11544234 10.1128/JB.183.19.5709-5717.2001
Tettelin H Nelson K Paulsen IT Eisen JA Read TD Peterson S Heidelberg J DeBoy RT Haft DH Dodson RJ Durkin AS Gwinn M Kolonay JF Nelson WC Peterson JD Umayam LA White O Salzberg SL Lewis MR Radune D Holtzapple E Khouri H Wolf AM Utterback TR Hansen CL McDonald LA Feldblyum TV Angiuoli S Dickinson T Hickey EK Holt IE Loftus BJ Yang F Smith HO Venter JC Dougherty BA Morrison DA Hollingshead SK Fraser CM Complete genome sequence of a virulent isolate of Streptococcus pneumoniae Science 2001 293 498 506 11463916 10.1126/science.1061217
Fernandez-Tornero C Lopez R Garcia E Gimenez-Gallego G Romero A A novel solenoid fold in the cell wall anchoring domain of the pneumococcal virulence factor LytA Nat Struct Biol 2001 8 1020 1024 11694890 10.1038/nsb724
Ronda C Garcia JL Garcia E Sanchez-Puelles JM Lopez R Biological role of the pneumococcal amidase. Cloning of the lytA gene in Streptococcus pneumoniae Eur J Biochem 1987 164 621 624 3569279
De Las Rivas B Garcia JL Lopez R Garcia P Purification and polar localization of pneumococcal LytB, a putative endo-beta-N-acetylglucosaminidase: the chain-dispersing murein hydrolase J Bacteriol 2002 184 4988 5000 12193614 10.1128/JB.184.18.4988-5000.2002
Garcia P Paz Gonzalez M Garcia E Garcia JL Lopez R The molecular characterization of the first autolytic lysozyme of Streptococcus pneumoniae reveals evolutionary mobile domains Mol Microbiol 1999 33 128 138 10411730 10.1046/j.1365-2958.1999.01455.x
Rigden DJ Jedrzejas MJ Galperin MY Amidase domains from bacterial and phage autolysins define a family of gamma-D, L-glutamate-specific amidohydrolases Trends Biochem Sci 2003 28 230 234 12765833 10.1016/S0968-0004(03)00062-8
Bateman A Rawlings ND The CHAP domain: a large family of amidases including GSP amidase and peptidoglycan hydrolases Trends Biochem Sci 2003 28 234 237 12765834 10.1016/S0968-0004(03)00061-6
Ng WL Kazmierczak KM Winkler ME Defective cell wall synthesis in Streptococcus pneumoniae R6 depleted for the essential PcsB putative murein hydrolase or the VicR (YycF) response regulator Mol Microbiol 2004 53 1161 1175 15306019 10.1111/j.1365-2958.2004.04196.x
Blackburn NT Clarke AJ Assay for lytic transglycosylases: a family of peptidoglycan lyases Anal Biochem 2000 284 388 393 10964424 10.1006/abio.2000.4707
Thunnissen AM Isaacs NW Dijkstra BW The catalytic domain of a bacterial lytic transglycosylase defines a novel class of lysozymes Proteins 1995 22 245 258 7479697
Thunnissen AM Dijkstra AJ Kalk KH Rozeboom HJ Engel H Keck W Dijkstra BW Doughnut-shaped structure of a bacterial muramidase revealed by X-ray crystallography Nature 1994 367 750 753 8107871 10.1038/367750a0
Wizemann TM Heinrichs JH Adamou JE Erwin AL Kunsch C Choi GH Barash SC Rosen CA Masure HR Tuomanen E Gayle A Brewah YA Walsh W Barren P Lathigra R Hanson M Langermann S Johnson S Koenig S Use of a whole genome approach to identify vaccine molecules affording protection against Streptococcus pneumoniae infection Infect Immun 2001 69 1593 1598 11179332 10.1128/IAI.69.3.1593-1598.2001
Kelley L MacCallum R Sternberg MJE Enhanced Genome Annotation using Structural Profiles in the Program 3D-PSSM J Mol Biol 2000 299 499 520 10860755 10.1006/jmbi.2000.3741
Bernal A Ear U Kyrpides N Genomes OnLine Database (GOLD): a monitor of genome projects world-wide Nucleic Acid Res 2001 29 126 127 11125068 10.1093/nar/29.1.126
Anantharaman V Aravind L Evolutionary history, structural features and biochemical diversity of the NlpC/P60 superfamily of enzymes Genome Biol 2003 4 R11 12620121 10.1186/gb-2003-4-2-r11
Margot P Wahlen M Gholamhoseinian A Piggot P Karamata D Gholamhuseinian A The lytE gene of Bacillus subtilis 168 encodes a cell wall hydrolase J Bacteriol 1998 180 749 752 9457885
Ichimura T Yamazoe M Maeda M Wada C Hiraga S Proteolytic activity of YibP protein in Escherichia coli J Bacteriol 2002 184 2595 2602 11976287 10.1128/JB.184.10.2595-2602.2002
Lange R Hengge-Aronis R The nlpD gene is located in an operon with rpoS on the Escherichia coli chromosome and encodes a novel lipoprotein with a potential function in cell wall formation Mol Microbiol 1994 13 733 743 7997184
Xie L Chatterjee C Balsara R Okeley NM van der Donk WA Heterologous expression and purification of SpaB involved in subtilin biosynthesis Biochem Biophys Res Commun 2002 295 952 957 12127987 10.1016/S0006-291X(02)00783-0
Rasko DA Ravel J Okstad OA Helgason E Cer RZ Jiang L Shores KA Fouts DE Tourasse NJ Angiuoli SV Kolonay J Nelson WC Kolsto AB Fraser CM Read TD The genome sequence of Bacillus cereus ATCC 10987 reveals metabolic adaptations and a large plasmid related to Bacillus anthracis pXO1 Nucleic Acids Res 2004 32 977 988 14960714 10.1093/nar/gkh258
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 Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Higgins D Thompson J Gibson T Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressivemultiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 1994 22 4673 4680 7984417
Marchler-Bauer A Anderson JB DeWeese-Scott C Fedorova ND Geer LY He S Hurwitz DI Jackson JD Jacobs AR Lanczycki CJ Liebert CA Liu C Madej T Marchler GH Mazumder R Nikolskaya AN Panchenko AR Rao BS Shoemaker BA Simonyan V Song JS Thiessen PA Vasudevan S Wang Y Yamashita RA Yin J Bryant SH CDD: a curated Entrez database of conserved domain alignments Nucleic Acids Res 2003 31 383 387 12520028 10.1093/nar/gkg087
Bateman A Coin L Durbin R Finn RD Hollich V Griffiths-Jones S Khanna A Marshall M Moxon S Sonnhammer ELL Studholme DJ Yeats C Eddy SR The Pfam Protein Families Database Nucleic Acids Research 2004 32 D138 D141 14681378 10.1093/nar/gkh121
Cserzo M Wallin E Simon I von Heijne G Elofsson A Prediction of transmembrane alpha-helices in procariotic membrane proteins: the Dense Alignment Surface method Protein Eng 1997 10 673 676 9278280 10.1093/protein/10.6.673
Nielsen H Engelbrecht J Brunak S von Heijne G Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites Protein Engineering 1997 10 1 6 9051728 10.1093/protein/10.1.1
Schultz J Milpetz F Bork P Ponting C SMART, a simple modular architecture research tool: identification of signaling domains Proc Natl Acad Sci U S A 1998 95 5857 5864 9600884 10.1073/pnas.95.11.5857
Goodstadt L Ponting C CHROMA: consensus-based colouring of multiple alignments for publication Bioinformatics 2001 17 845 846 11590103 10.1093/bioinformatics/17.9.845
| 15717932 | PMC550655 | CC BY | 2021-01-04 16:39:32 | no | BMC Genomics. 2005 Feb 17; 6:19 | utf-8 | BMC Genomics | 2,005 | 10.1186/1471-2164-6-19 | oa_comm |
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-191571793210.1186/1471-2164-6-19Research ArticleThe PECACE domain: a new family of enzymes with potential peptidoglycan cleavage activity in Gram-positive bacteria Pagliero Estelle [email protected] Otto [email protected] Thierry [email protected] Guilmi Anne Marie [email protected] Laboratoire d'Ingénierie des Macromolécules Institute de Biologie Structurale Jean-Pierre Ebel (CEA-CNRS UMR 5075-UJF), 41 Rue Jules Horowitz 38027 Grenoble cedex 1, France2 Laboratoire de Cristallographie Macromoléculaire Institut de Biologie Structurale Jean-Pierre Ebel (CEA-CNRS UMR 5075-UJF), 41 Rue Jules Horowitz 38027 Grenoble cedex 1, France2005 17 2 2005 6 19 19 22 10 2004 17 2 2005 Copyright © 2005 Pagliero et al; licensee BioMed Central Ltd.2005Pagliero 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 bacterial peptidoglycan is a dynamic process, synthases and cleavage enzymes are functionally coordinated. Lytic Transglycosylase enzymes (LT) are part of multienzyme complexes which regulate bacterial division and elongation. LTs are also involved in peptidoglycan turnover and in macromolecular transport systems. Despite their central importance, no LTs have been identified in the human pathogen Streptococcus pneumoniae. We report the identification of the first putative LT enzyme in S. pneumoniae and discuss its role in pneumococcal peptidoglycan metabolism.
Results
Homology searches of the pneumococcal genome allowed the identification of a new domain putatively involved in peptidoglycan cleavage (PECACE, PEptidoglycan CArbohydrate Cleavage Enzyme). This sequence has been found exclusively in Gram-positive bacteria and gene clusters containing pecace are conserved among Streptococcal species. The PECACE domain is, in some instances, found in association with other domains known to catalyze peptidoglycan hydrolysis.
Conclusions
A new domain, PECACE, putatively involved in peptidoglycan hydrolysis has been identified in S. pneumoniae. The probable enzymatic activity deduced from the detailed analysis of the amino acid sequence suggests that the PECACE domain may proceed through a LT-type or goose lyzosyme-type cleavage mechanism. The PECACE function may differ largely from the other hydrolases already identified in the pneumococcus: LytA, LytB, LytC, CBPD and PcsB. The multimodular architecture of proteins containing the PECACE domain is another example of the many activities harbored by peptidoglycan hydrolases, which is probably required for the regulation of peptidoglycan metabolism. The release of new bacterial genomes sequences will probably add new members to the five groups identified so far in this work, and new groups could also emerge. Conversely, the functional characterization of the unknown domains mentioned in this work can now become easier, since bacterial peptidoglycan is proposed to be the substrate.
==== Body
Background
The bacterial cell wall resists intracellular pressure and gives the bacterium its particular shape. Cell wall reinforcement is brought about by a strong scaffolding structure, the peptidoglycan, which is formed by glycan strands and peptide chains held together by covalent bonds, resulting in a mono- or multilayered network. The glycan strands are composed of N-acetylglucosamine (GlcNAc) and N-acetylmuramyl (MurNAc) residues linked together by β-1,4 glycosidic bonds. Peptides are covalently attached to the lactyl group of the muramic acid and their cross-linking results in the net structure of the peptidoglycan (Fig. 1a).
Figure 1 Schematic representation of peptidoglycan and of cleavage enzymes in S. pneumoniae. (a) Scheme of the pneumococcal peptidoglycan, indicating the chemical bonds cleaved by identified hydrolases in blue. The MurNAc residue containing the 1, 6-anhydro bond resulting from LT reaction is in a green circle. The putative LT pneumococcal enzyme appears in red, while enzymes CBPD and PcsB for which no enzymatic specificity is yet characterized are in black. (b) Topological representation of the glycan strand hydrolases described in S. pneumoniae. Black and hatched boxes indicate the signal peptide and the transmembrane anchor, respectively. The blue boxes illustrate the respective enzymatic active domains. Purple rectangles correspond to the Choline-Binding repeats. Green and orange boxes correspond to SH3b and coiled-coil regions, respectively. The topology was designed with the help of SMART server [39].
Peptidoglycan is synthesized in a multi-stage process. The first steps occur in the cytoplasm, where a set of enzymatic reactions gives rise to the assembly of the MurNAc-pentapeptide. This unit is in turn linked to the carrier undecaprenol lipid via a pyrophosphate group; afterwards the GlcNAc group is added, generating the lipid II precursor. The saccharidic and peptidic moieties of lipid II are subsequently exposed to the periplasmic space. At this stage, peptidoglycan biosynthesis involves polymerization of the glycan chains, catalyzed by glycosyltransferases [1] as well as interpeptide bridge formation performed by transpeptidases [2]. These two enzymatic reactions are resident on the extracellular domains of Penicillin-Binding Proteins (PBPs) which are membrane-associated molecules, present in all eubacteria [2].
Peptidoglycan metabolism is a dynamic process since this structure grows and divides in perfect synchronization with cell growth and division. Furthermore, it is well established that peptidoglycan is subject to maturation, turnover and recycling in Gram-negative bacteria [3]. To fullfil these processes, it is expected that peptidoglycan cleavage enzymes must exert their functions in coordinated action with PBPs. Indeed, a large range of different peptidoglycan hydrolases have been identified in numerous bacterial species and specific peptidoglycan hydrolases exist for almost each covalent bond [3] (Fig. 1a). The polysaccharidic component of peptidoglycan is the target of several hydrolases: the β-1,4 glycosidic bond between MurNAc and GlcNAc residues is cleaved by lyzosyme and by lytic transglycosylases (LT), the β-1,4 glycosidic bond between GlcNAc and MurNAc is hydrolyzed by glucosaminidases and amidases are responsible for the cleavage of the MurNAc-L-alanine bond (Fig. 1a).
Lyzosyme and LT enzymes cleave the same β-1,4-MurNAc-GlcNAc bond but generate different reaction products: while lyzosymes catalyze a hydrolytic reaction, LTs cleave the β-glycosidic linkage with the concomitant formation of 1,6-anhydromuramyl residues, blocking the reducing end of the glycan strands. The significance of the ring structure is not known but it has been speculated that the bond energy may be utilized for glycan strand rearrangements. In addition, the 1,6-anhydro ring may also be considered as a specific product of peptidoglycan turnover. Despite the lack of understanding of the physiological function of anhydromuropeptide product, LT enzymes must play a significant cellular role. Indeed, it has been observed that deletions of genes encoding LT proteins lead to E. coli and Neisseria meningitidis with altered cell separation phenotypes, indicating that LTs cleave septal peptidoglycan [4,5]. Macromolecular transport systems (secretion types II, III, IV and IV pilus synthesis) of Gram-negative bacteria contain LT enzymes, suggesting that peptidoglycan hole formation (essential for transport functions) is specifically performed by this enzyme family [6]. As mentioned above, the enlargement of the bacterial stress-bearing peptidoglycan structure requires the well coordinated action of synthases (PBPs) and hydrolase enzymes. The "three-for-one" growth mechanism described by Höltje proposes that a triplet of glycan strands cross-linked to each other (resulting from PBPs synthesis) is attached to the peptidoglycan layer. Subsequently, the docking strand is removed by hydrolases resulting in the insertion of the peptidoglycan triplet. The hydrolases involved in such multienzyme complexes are endopeptidases and LT enzymes [3]. This hypothesis is supported by experimental data as LT and PBPs could be co-purified from E. coli extracts [7-9]. In conclusion, LT enzymes play an important cellular role in diverse aspects of cell biology as expected from their presence in a very wide range of eubacteria as well as archaebacteria [3,10,11].
Surprinsingly, no such LT enzyme has been identified to date in the human pathogen Streptococcus pneumoniae, the causative agent of ear infections in children, as well as meningitis and pneumonia. The pattern of peptidoglycan hydrolases in this Gram-positive bacteria includes, besides a D, D-carboxypeptidase, five glycan strand cleaving enzymes (Fig 1b). Four of these are surface-exposed proteins harboring Choline-Binding Domains which are non-covalently bound to choline residues present on cell wall pneumococcal teichoic and lipoteichoic acids [12-14]. The Choline-Binding Proteins (CBPs) catalyzing peptidoglycan hydrolysis are LytA, LytB, LytC and potentially CBPD (Fig 1b). LytA is an amidase and also appears as an autolytic enzyme, causing bacteriolysis when acting in an uncontrolled manner [15]. LytB is a glucosaminidase involved in cell separation as lytB mutants form very long chains of over 100 cells [16]. LytC is a lysozyme with an autolytic behavior at 30°C [17]. Finally, CBPD and PcsB contain a CHAP domain (Cysteine, Histidine-dependent amidohydrolase/peptidase) predicted to hydrolyse the peptidoglycan in pneumococcus, but definitive biochemical data are still lacking [18-20].
Our interest in the biology of S. pneumoniae led us to investigate the presence of LT enzymes in this bacteria. Homology searches of enzyme sequences within the pneumococcus genome using bioinformatics tools allowed the identification of a new domain harboring motifs that infer potential peptidoglycan cleavage activity. For this reason we named this domain PECACE (PEptidoglycan CArbohydrate Cleavage Enzyme). This domain sequence was found exclusively in Gram-positive bacterial species, suggesting a significant cellular role. Finally, the PECACE domain is in some instances found in association with other domains, known to catalyze peptidoglycan hydrolysis: this observation reinforces the predicted function of PECACE as participating in peptidoglycan cleavage and represents another example of multifunctional proteins involved in peptidoglycan metabolism.
Results and discussion
Identification of a protein harboring the PECACE domain in S. pneumoniae
The C-terminal domain of Escherichia coli Slt70 (Soluble Lytic Transglycosylase) has a lysozyme-like fold and its amino acid sequence was employed in a search of Bacilli genomes within the NCBI Conserved Domain Search server [11,21-23]. Thirty-four Slt70-homologue sequences were retrieved using an inclusion threshold of 0.01. None of these sequence originated from the S. pneumoniae translated genome. Subsequently, each of these 34 sequences was compared with the non-redundant protein database using PSI-BLAST with a E-value threshold of 0.005 and 5 sequences showed significant matches with a unique protein in S. pneumoniae. This sequence (accession numbers NP358524, gi:15902974) contains 204 amino acids: the first 21 amino acids are predicted to form a transmembrane anchor and the subsequent 192-residue region is putatively exposed to the extracellular space (Fig. 1b). This S. pneumoniae NP358524 sequence has been tested as a pneumococcal vaccine antigen on the basis of preliminary screens for novel vaccine candidates [24].
A three-dimensional fold prediction of the S. pneumoniae NP358524 protein was performed with the 3D-PSSM server [25] which identified two matches: E. coli Slt70 (d1qsaa2, E-value:10-7) and LysG (G-type goose lyzozyme, d1531, E-value:10-3). The sequence alignment between NP358524 and Slt70 is shown in Fig. 2, defining the PECACE domain in the pneumococcal protein. The secondary structures are also reported, based on three-dimensional structures of Slt70 and on computational predictions for PECACE and suggest that the latter is highly α-helical (Fig. 2). It is of note that both Slt70 and LysG are highly similar, and both lack the catalytic aspartate residue commonly found in the active site of lysozymes [10,11,21,22]. Therefore, the PECACE domain of the NP358524 sequence appears to belong to this group of bacterial lysozymes, characterized by the absence of an aspartate residue in the catalytic site and is part of the Glycoside Hydrolase family 23 based upon CAZy classification . The catalytic acid residue in the PECACE domain is most probably Glu61 since it aligns with the catalytic Glu478 residue in the Slt70 sequence (Fig. 2). The serine residue following the catalytic glutamate and the GLMQI/V motif are essential for active-site architecture and are conserved between Slt70 and LysG. In the PECACE sequence, a threonine residue follows the catalytic glutamate and the GLMQI/V motif differs since the corresponding sequence is D(68)VMQS (Fig. 2). Finally, the second motif AYNxG which has been shown to be involved in the interaction with the substrate for Slt70 (A551YNxG) is well conserved in the PECACE sequence (A117YNxG).
Figure 2 Alignment of the PECACE domain with Slt70. Protein fold recognition was performed with the 3D-PSSM server. The NP358524 sequence (residues 31–145) from S. pneumoniae (PECACE domain) is aligned with Slt70 from E. coli (P03810, residues 478–616). Amino acids of Slt70 involved in the catalytic reaction and in ligand recognition are underlined while residues conserved in each alignment are highlighted in red. The structural prediction for S. pneumoniae PECACE domain was determined (H = helix, C = coil) while Slt70 secondary-structure information was obtained from PDB file 1QSA.
Based on this sequence analysis, we infer that the S. pneumoniae NP358524 protein, through its PECACE domain, probably catalyzes the peptidoglycan cleavage of the β-1,4-MurNAc-GlcNAc bond by employing Glu61 as the catalytic residue.
Identification of the PECACE domain in Gram-positive bacteria
The 204 amino acid sequence from S. pneumoniae NP358524, containing the PECACE domain, was used as a PSI-BLAST search query. In total, 29 distinct proteins, all from Gram-positive bacteria, were identified (E-value: 10-5) and no sequences from Gram-negative bacteria were retrieved. These sequences were aligned with ClustalW and manually edited. A conserved pattern could be extracted from this alignment: E- [ST]-X-G-X(1,16)-D-X-M-Q- [SA]- [SA]-E- [SG] which was used to search for additional sequences, but no new sequence could be detected from databases, even with a degenerated pattern. PSI-BLAST performed through the GOLD server led to the identification of 10 new sequences from Gram-positive bacteria [26]. In summary, out of the about 50 Gram-positive bacteria for which the whole genome sequence is available, 34 of them contain at least one protein harboring the PECACE domain. The final alignment of these sequences with the S. pneumoniae PECACE domain is shown in Fig. 3. The putative catalytic glutamate residue, Glu61 in the S. pneumoniae PECACE domain, is conserved in all sequences and the following residue is a Ser or Thr in accordance with Slt70 and LysG patterns. In addition, the D(68)VMQS motif in the S. pneumoniae PECACE domain is also well represented in the large majority of sequences with the consensus sequence DI/VMQSSES. Finally, the second motif AYNxG is also conserved while the Ala residue is often replaced by a Ser. In conclusion, the features identified in the S. pneumoniae PECACE domain regarding the potential enzymatic properties of peptidoglycan polysaccharide cleavage are also shared by the similar PECACE domains in Gram-positive bacteria.
Figure 3 Sequence alignment of PECACE domains identified in Gram-positive bacteria. Multiple sequence alignment was constructed using ClustalW. The lengths of the insertions in the sequences are shown in parentheses. The sequences are denoted by their GenBank Identifier (gi). The domain limits are indicated by the residue positions (first-end). The amino acids identified as catalytic or involved in ligand recognition are marked with asterisks under PECACE sequence. Alignments are coloured using the CHROMA tool using default parameters [40]. Full sequence details, group (i): Streptococcus pneumoniae R6 (gi:15902974), Streptococcus mitis NCTC 12261 (§SMT1418), Streptococcus sanguinis SK36 (&:SS_A352_G10), Streptococcus gordonii (gi:18389219), Streptococcus suis P1/7 (suis166b12), Streptococcus uberis 0140J (sub49a04), Streptococcus equi (equi324d3), Streptococcus equi subsp. Zooepidemicus (zoo26g07), Streptococcus pyogenes M1 GAS (gi:15675124), Streptococcus agalactiae 2603V/R (gi:22537230), Lactococcus lactis subsp. Cremoris SK11 (scaffold18), Streptococcus mutans UA159 (gi:24379517), Streptococcus thermophilus LMD-9 (scaffold3), Lactococcus lactis subsp. Lactis (gi:15672584), Enterococcus faecium DO (2351355_Cont543), Enterococcus faecalis V583 (gi:29376084), Bacillus subtilis subsp. subtilis str. 168 (gi:16078973), Bacillus cereus ATCC 14579 (gi:30020591), Oceanobacillus iheyensis HTE831 (gi:23100516), group (ii): Bacillus anthracis: (pXO2-08) (gi:10956398), Enterococcus faecalis: (pRE25) (gi:12957015), Enterococcus faecium (gi:22992993), Enterococcus faecalis V583 (gi:29376781), Clostridium difficile 630 (Cd81d2), Enterococcus faecalis V583 (gi:29376405), Clostridium perfringens (gi:13274506), Staphylococcus aureus subsp. aureus Mu50 (gi:15923390), Listeria monocytogenes EGD-e (gi:16803144), Streptococcus agalactiae 2603V/R (gi:22537089), Enterococcus faecium (gi:22993467), Bacillus subtilis subsp. subtilis str. 168 (gi:16077564, group (iii): Bacillus cereus ATCC 14579 (gi:30021796), group (iv): Enterococcus faecalis BM4518 (gi:33355845), group (v): Bacillus anthracis str. A2012: (pXO1) (gi:21392795), Bacillus cereus ATCC 10987: (pBc10987) (gi:44004362).
Genomic organization of pecace genes
The genomic organization of pecace genes has been analyzed in a variety of Gram-positive bacteria and a conserved distribution was observed in various streptococci species (Fig. 4). This feature indicates that genetic transfer of the whole cluster may have occured within the streptococci family, providing further evidence regarding the significant importance of the PECACE domains in bacterial physiology. However, the pneumococcal cluster is more related to the S. mitis one than to S. mutans, S. agalactiae and S. pyogenes ones, while clusters of the latter three species are related to each other. Genes located upstream and downstream of pecace are in some instances well characterized but the function of the corresponding proteins could not bring any clues about the role of PECACE, nor any evidence on pecace gene transcription. However, pecace is in all cases found in association with the same gene (whose locus name in S. pneumoniae is spr0929) but no information about the function of the protein encoded by this locus is available in databases. Transcriptional analysis of these two genes may bring informations about their potential co-regulation, a first stage in deciphering cellular function.
Figure 4 Schematic representation of the gene cluster containing pecace in Streptococci species. The coding regions and their direction of transcription are indicated by arrows. Gene names are given on top of the corresponding region.
Domain organization of proteins containing the PECACE domain
The PECACE domain is found in a large range of protein architectures, commonly associated with other peptidoglycan hydrolases, suggesting that these proteins have multiple peptidoglycan cleavages activities (Fig. 5). The identification of proteins displaying the PECACE domain was carried out using NCBI Conserved Domain Search and Pfam servers. In addition, prediction of membrane anchoring was performed with the DAS-Transmembrane Prediction server while extracellular secretion of the protein was deduced from the identification of a signal peptide.
Figure 5 Domain architecture of PECACE proteins. The domain architecture of the proteins containing the PECACE domain was organized according to searches with NCBI Conserved Domain Search server against Pfam database: CHAP/NlpC-P60 (Pfam: PF05257/PF00877), M37 peptidase (Pfam: PF01551), unknown domain 1 (gi: 33355845) and unknown domain 2 (gi: 30021796). The size of the domains is not respected in these representations.
PECACE-containing proteins appear to fall into 5 main categories (Fig. 5): (i) those which display no additional domain, (ii) CHAP-Nlpc/P60 as the associated group, (iii) CHAP-Nlpc/P60 and an unknown domain as associated groups, (iv) domains with no ascribed functions and finally (v) CHAP-Nlpc/P60 and M37 peptidase as associated groups.
The 19 proteins which contain only the PECACE domain belong to group (i) and harbor either a signal peptide or a transmembrane helix (as for the S. pneumoniae protein), leading in both cases to cell surface expression.
The CHAP-Nlpc/P60 domain is commonly associated with the PECACE domain in different modular organizations, namely in groups (ii), (iii), and (v) [18,19,27]. The CHAP domain has been recently described as a Cysteine, Histidine-dependent Amidohydrolase/Peptidase and it has been proposed to hydrolyse peptidoglycan containing γ-glutamyl [18,19]. Indeed, proteins such as N-acetylmuramyl-L-alanine amidase and D-alanyl-glycyl endopeptidase have been described as CHAP-containing enzymes [18,19]. However, while the substrate and the reaction mechanism have not been yet experimentally characterized for the CHAP domain, its role in peptidoglycan hydrolysis is inferred from its presence in multifunctional proteins recognizing peptidoglycan as substrate. Recently, hydrolytic activity of peptidoglycan has been attributed to the CHAP-containing protein PcsB in S. pneumoniae due to abnormal and uncontrolled cell wall synthesis at misplaced septa and formation of long cells in pcsB deleted mutant strains [20]. Proteins from group (ii) are expressed at the cell surface through a transmembrane anchor or are secreted, 12 members have been identified with this topology. Only one sequence (AAQ16265, gi:33355845) from Enterococcus faecalis BM4518 is part of the group (iii), and no function could be identified for the N-terminus domain preceding the PECACE domain. However the former domain is Lys-rich (14%) suggesting an electrostatic interaction with the peptidoglycan as proposed for B. subtilis endopeptidase [28]. Group (iv) is composed of an unique sequence from B. cereus ATCC 14579 (NP 833427, gi:30021796). Neither a signal peptide nor a transmembrane anchor have been detected. Furthermore, the domain of unknown function, which is different from the ones identified in groups (iii) and (v) is present in other multimodular proteins of B. cereus, in association with peptidoglycan hydrolysis enzymes. Finally, two sequences share the architecture defining the group (v) which harbor CHAP-Nlpc/P60 and Peptidase M37 domains [29]. Members of the Peptidase M37 family are generally glycylglycine endo-metallopeptidases; the archetypal member is the lysostaphin enzyme from Staphylococcus species which cleaves the pentaglycine bridge in the peptidoglycan [30]. One group (v) protein (NP 652875, gi:21392795) is encoded by Bacillus anthracis plasmid pXO1 and is required for synthesis of various anthrax toxin proteins [31]; this sequence has neither a signal peptide nor a transmembrane region. The second sequence of group (v) is located on Bacillus cereus ATCC 10987 plasmid pBc10987 (NP 982030, gi:44004362) and contains, in addition to CHAP-Nlpc/P60, Peptidase M37 and PECACE domain as well as an extra sequence to which no function has been attributed but with significant similarity with a B. anthracis plasmid pXO1 sequence (NP 652874, gi:21392794) [32].
Conclusions
In summary, a new domain named PECACE, putatively involved in peptidoglycan cleavage has been identified in S. pneumoniae. The probable enzymatic activity deduced from the detailed analysis of the amino acid sequence suggests a LT-type or goose lyzosyme-type mechanism; we are currently characterising the enzymatic properties and cellular role of the PECACE domain from S. pneumoniae. This new putative pneumococcal peptidoglycan cleavage enzyme differs largely from the other hydrolases already identified in this bacteria. Indeed, LytA, LytB, LytC and CBPD proteins are all bound to the cell wall choline residues and thus expressed at the cell surface. The presence of a signal peptide within the amino acid sequence of PcsB suggests that it is either exposed on the cell surface or secreted. On the contrary, the pneumococcal NP358524 protein displaying the PECACE domain is embeded in the cytoplasmic membrane by a hydrophobic helix. The physiological role of this membranous peptidoglycan cleavage enzyme might differ from the other peptidoglycan hydrolysing enzymes.
Interestingly, the PECACE domain has only been found in Gram-positive bacteria. It is tempting to speculate that the multilayered structure of Gram-positive peptidoglycan relates to the PECACE putative activity.
The architecture of multimodular proteins containing the PECACE domain is another example of the pattern of multiple activities harbored by many peptidoglycan hydrolases, probably needed for the regulation of peptidoglycan metabolism. The release of new bacterial genomes sequences will probably add new members that will complete the five groups identified so far in this work and new groups could also emerge. Conversely, the functional characterization of the unknown domains mentioned in this work should now be easier, as their substrate, the peptidoglycan, is now identified.
Methods
The non-redudant database of protein sequences (National center for Biotechnology Information, NIH, Bethesda) and whole bacterial genomes sequences [26] was searched using BLASP and PSI-BLAST programs with (E) value threshold of 0.005 [33]. Multiple alignments were constructed with ClustalW program [34] followed by manual correction based on PSI-BLAST results. Protein fold recognition through 3D-profiles was searched using 3D-PSSM server [25]. Conserved (and degenerated) amino acid patterns was designed and searched against non-redudant database of protein sequences . Identification of domains associated with PECACE proteins was realized using NCBI Conserved Domain Search [35] and Pfam servers [36]. Finally, prediction of transmembrane anchor and secretory signal peptide were performed with DAS server and SignalP-2.0 servers respectively [37,38].
Authors' contributions
OD conceived of the study and participated in the sequences alignment. EP carried out the sequences analysis and the writing of the manuscript with AMDG. TV coordinated the study. All authors read and approved the final manuscript.
Acknowledgments
This work was supported by the European Commission grant LSMH-CT-COBRA 2003-503335. EP is a recipient of a CEA CFR fellowship. We are grateful to Dr Andréa Dessen for constructive comments and critical review of the manuscript.
==== Refs
van Heijenoort J Formation of the glycan chains in the synthesis of bacterial peptidoglycan Glycobiology 2001 11 25R 36R 11320055 10.1093/glycob/11.3.25R
Goffin C Ghuysen JM Biochemistry and Comparative Genomics of SxxK Superfamily Acyltransferases Offer a Clue to the Mycobacterial Paradox: Presence of Penicillin-Susceptible Target Proteins versus Lack of Efficiency of Penicillin as Therapeutic Agent Microbiol Mol Biol Rev 2002 66 702 738 12456788 10.1128/MMBR.66.4.702-738.2002
Höltje JV Growth of the stress-bearing and shape-maintaining murein sacculus of Escherichia coli Microbiol Mol Biol Rev 1998 62 181 203 9529891
Adu-Bobie J Lupetti P Brunelli B Granoff D Norais N Ferrari G Grandi G Rappuoli R Pizza M GNA33 of Neisseria meningitidis is a lipoprotein required for cell separation, membrane architecture, and virulence Infect Immun 2004 72 1914 1919 15039310 10.1128/IAI.72.4.1914-1919.2004
Heidrich C Ursinus A Berger J Schwarz H Höltje JV Effects of multiple deletions of murein hydrolases on viability, septum cleavage, and sensitivity to large toxic molecules in Escherichia coli J Bacteriol 2002 184 6093 6099 12399477 10.1128/JB.184.22.6093-6099.2002
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
Vollmer W von Rechenberg M Höltje JV 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
Schiffer G Höltje JV Cloning and characterization of PBP1c, a third member of the multimodular class A penicillin-binding proteins of Escherichia coli J Biol Chem 1999 274 32031 32039 10542235 10.1074/jbc.274.45.32031
Romeis T Höltje JV Specific interaction of penicillin-binding proteins 3 and 7/8 with soluble lytic transglycosylase in Escherichia coli J Biol Chem 1994 269 21603 21607 8063800
Blackburn NT Clarke AJ Identification of four families of peptidoglycan lytic transglycosylases J Mol Evol 2001 52 78 84 11139297
Dijkstra BW Thunnissen AM 'Holy' proteins. II: The soluble lytic transglycosylase Curr Opin Struct Biol 1994 4 810 813 7712284 10.1016/0959-440X(94)90261-5
Hoskins JA Arnold WEJr Arnold J Blaszczak LC Burgett S DeHoff BS Estrem ST Fritz L Fu DJ Fuller W Geringer C Gilmour R Glass JS Khoja H Kraft AR Lagace RE LeBlanc DJ Lee LN Lefkowitz EJ Lu J Matsushima P McAhren SM McHenney M McLeaster K Mundy CW Nicas TI Norris FH O'Gara M Peery RB Robertson GT Rockey P Sun PM Winkler ME Yang Y Young-Bellido M Zhao G Zook CA Baltz RH Jaskunas SR Rosteck PR JrSkatrud PL Glass JI Genome of the bacterium Streptococcus pneumoniae strain R6 J Bacteriol 2001 183 5709 5717 11544234 10.1128/JB.183.19.5709-5717.2001
Tettelin H Nelson K Paulsen IT Eisen JA Read TD Peterson S Heidelberg J DeBoy RT Haft DH Dodson RJ Durkin AS Gwinn M Kolonay JF Nelson WC Peterson JD Umayam LA White O Salzberg SL Lewis MR Radune D Holtzapple E Khouri H Wolf AM Utterback TR Hansen CL McDonald LA Feldblyum TV Angiuoli S Dickinson T Hickey EK Holt IE Loftus BJ Yang F Smith HO Venter JC Dougherty BA Morrison DA Hollingshead SK Fraser CM Complete genome sequence of a virulent isolate of Streptococcus pneumoniae Science 2001 293 498 506 11463916 10.1126/science.1061217
Fernandez-Tornero C Lopez R Garcia E Gimenez-Gallego G Romero A A novel solenoid fold in the cell wall anchoring domain of the pneumococcal virulence factor LytA Nat Struct Biol 2001 8 1020 1024 11694890 10.1038/nsb724
Ronda C Garcia JL Garcia E Sanchez-Puelles JM Lopez R Biological role of the pneumococcal amidase. Cloning of the lytA gene in Streptococcus pneumoniae Eur J Biochem 1987 164 621 624 3569279
De Las Rivas B Garcia JL Lopez R Garcia P Purification and polar localization of pneumococcal LytB, a putative endo-beta-N-acetylglucosaminidase: the chain-dispersing murein hydrolase J Bacteriol 2002 184 4988 5000 12193614 10.1128/JB.184.18.4988-5000.2002
Garcia P Paz Gonzalez M Garcia E Garcia JL Lopez R The molecular characterization of the first autolytic lysozyme of Streptococcus pneumoniae reveals evolutionary mobile domains Mol Microbiol 1999 33 128 138 10411730 10.1046/j.1365-2958.1999.01455.x
Rigden DJ Jedrzejas MJ Galperin MY Amidase domains from bacterial and phage autolysins define a family of gamma-D, L-glutamate-specific amidohydrolases Trends Biochem Sci 2003 28 230 234 12765833 10.1016/S0968-0004(03)00062-8
Bateman A Rawlings ND The CHAP domain: a large family of amidases including GSP amidase and peptidoglycan hydrolases Trends Biochem Sci 2003 28 234 237 12765834 10.1016/S0968-0004(03)00061-6
Ng WL Kazmierczak KM Winkler ME Defective cell wall synthesis in Streptococcus pneumoniae R6 depleted for the essential PcsB putative murein hydrolase or the VicR (YycF) response regulator Mol Microbiol 2004 53 1161 1175 15306019 10.1111/j.1365-2958.2004.04196.x
Blackburn NT Clarke AJ Assay for lytic transglycosylases: a family of peptidoglycan lyases Anal Biochem 2000 284 388 393 10964424 10.1006/abio.2000.4707
Thunnissen AM Isaacs NW Dijkstra BW The catalytic domain of a bacterial lytic transglycosylase defines a novel class of lysozymes Proteins 1995 22 245 258 7479697
Thunnissen AM Dijkstra AJ Kalk KH Rozeboom HJ Engel H Keck W Dijkstra BW Doughnut-shaped structure of a bacterial muramidase revealed by X-ray crystallography Nature 1994 367 750 753 8107871 10.1038/367750a0
Wizemann TM Heinrichs JH Adamou JE Erwin AL Kunsch C Choi GH Barash SC Rosen CA Masure HR Tuomanen E Gayle A Brewah YA Walsh W Barren P Lathigra R Hanson M Langermann S Johnson S Koenig S Use of a whole genome approach to identify vaccine molecules affording protection against Streptococcus pneumoniae infection Infect Immun 2001 69 1593 1598 11179332 10.1128/IAI.69.3.1593-1598.2001
Kelley L MacCallum R Sternberg MJE Enhanced Genome Annotation using Structural Profiles in the Program 3D-PSSM J Mol Biol 2000 299 499 520 10860755 10.1006/jmbi.2000.3741
Bernal A Ear U Kyrpides N Genomes OnLine Database (GOLD): a monitor of genome projects world-wide Nucleic Acid Res 2001 29 126 127 11125068 10.1093/nar/29.1.126
Anantharaman V Aravind L Evolutionary history, structural features and biochemical diversity of the NlpC/P60 superfamily of enzymes Genome Biol 2003 4 R11 12620121 10.1186/gb-2003-4-2-r11
Margot P Wahlen M Gholamhoseinian A Piggot P Karamata D Gholamhuseinian A The lytE gene of Bacillus subtilis 168 encodes a cell wall hydrolase J Bacteriol 1998 180 749 752 9457885
Ichimura T Yamazoe M Maeda M Wada C Hiraga S Proteolytic activity of YibP protein in Escherichia coli J Bacteriol 2002 184 2595 2602 11976287 10.1128/JB.184.10.2595-2602.2002
Lange R Hengge-Aronis R The nlpD gene is located in an operon with rpoS on the Escherichia coli chromosome and encodes a novel lipoprotein with a potential function in cell wall formation Mol Microbiol 1994 13 733 743 7997184
Xie L Chatterjee C Balsara R Okeley NM van der Donk WA Heterologous expression and purification of SpaB involved in subtilin biosynthesis Biochem Biophys Res Commun 2002 295 952 957 12127987 10.1016/S0006-291X(02)00783-0
Rasko DA Ravel J Okstad OA Helgason E Cer RZ Jiang L Shores KA Fouts DE Tourasse NJ Angiuoli SV Kolonay J Nelson WC Kolsto AB Fraser CM Read TD The genome sequence of Bacillus cereus ATCC 10987 reveals metabolic adaptations and a large plasmid related to Bacillus anthracis pXO1 Nucleic Acids Res 2004 32 977 988 14960714 10.1093/nar/gkh258
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 Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Higgins D Thompson J Gibson T Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressivemultiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 1994 22 4673 4680 7984417
Marchler-Bauer A Anderson JB DeWeese-Scott C Fedorova ND Geer LY He S Hurwitz DI Jackson JD Jacobs AR Lanczycki CJ Liebert CA Liu C Madej T Marchler GH Mazumder R Nikolskaya AN Panchenko AR Rao BS Shoemaker BA Simonyan V Song JS Thiessen PA Vasudevan S Wang Y Yamashita RA Yin J Bryant SH CDD: a curated Entrez database of conserved domain alignments Nucleic Acids Res 2003 31 383 387 12520028 10.1093/nar/gkg087
Bateman A Coin L Durbin R Finn RD Hollich V Griffiths-Jones S Khanna A Marshall M Moxon S Sonnhammer ELL Studholme DJ Yeats C Eddy SR The Pfam Protein Families Database Nucleic Acids Research 2004 32 D138 D141 14681378 10.1093/nar/gkh121
Cserzo M Wallin E Simon I von Heijne G Elofsson A Prediction of transmembrane alpha-helices in procariotic membrane proteins: the Dense Alignment Surface method Protein Eng 1997 10 673 676 9278280 10.1093/protein/10.6.673
Nielsen H Engelbrecht J Brunak S von Heijne G Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites Protein Engineering 1997 10 1 6 9051728 10.1093/protein/10.1.1
Schultz J Milpetz F Bork P Ponting C SMART, a simple modular architecture research tool: identification of signaling domains Proc Natl Acad Sci U S A 1998 95 5857 5864 9600884 10.1073/pnas.95.11.5857
Goodstadt L Ponting C CHROMA: consensus-based colouring of multiple alignments for publication Bioinformatics 2001 17 845 846 11590103 10.1093/bioinformatics/17.9.845
| 15707487 | PMC550656 | CC BY | 2021-01-04 16:02:51 | no | BMC Bioinformatics. 2005 Feb 11; 6:29 | latin-1 | BMC Bioinformatics | 2,005 | 10.1186/1471-2105-6-29 | oa_comm |
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-251570518910.1186/1471-2105-6-25Research ArticleIntegrating alternative splicing detection into gene prediction Foissac Sylvain [email protected] Thomas [email protected] Unité de Biométrie et Intelligence Artificielle, INRA, 31326 Castanet Tolosan, France2005 10 2 2005 6 25 25 27 7 2004 10 2 2005 Copyright © 2005 Foissac and Schiex; 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
Alternative splicing (AS) is now considered as a major actor in transcriptome/proteome diversity and it cannot be neglected in the annotation process of a new genome. Despite considerable progresses in term of accuracy in computational gene prediction, the ability to reliably predict AS variants when there is local experimental evidence of it remains an open challenge for gene finders.
Results
We have used a new integrative approach that allows to incorporate AS detection into ab initio gene prediction. This method relies on the analysis of genomically aligned transcript sequences (ESTs and/or cDNAs), and has been implemented in the dynamic programming algorithm of the graph-based gene finder EuGÈNE. Given a genomic sequence and a set of aligned transcripts, this new version identifies the set of transcripts carrying evidence of alternative splicing events, and provides, in addition to the classical optimal gene prediction, alternative optimal predictions (among those which are consistent with the AS events detected). This allows for multiple annotations of a single gene in a way such that each predicted variant is supported by a transcript evidence (but not necessarily with a full-length coverage).
Conclusions
This automatic combination of experimental data analysis and ab initio gene finding offers an ideal integration of alternatively spliced gene prediction inside a single annotation pipeline.
==== Body
Background
Alternative splicing (AS) is a biological process that occurs during the maturation step of a pre-mRNA, allowing the production of different mature mRNA variants from a unique transcription unit. AS is known to play a key role in the regulation of gene expression and transcriptome/proteome diversity [1]. First considered as an exceptional event, AS is now thought to involve the majority of the human multi-exon genes, from 50% to 74% [1-3]. This observation raises new issues for genome annotation, especially concerning the computational gene finding process that generally provides only one exon-intron structure per sequence.
In the context of structural gene prediction, two classes of approaches are usually considered. In the first approach, usually denoted as intrinsic or ab initio, the only type of information used for gene prediction lies in the statistical properties of the various gene elements (exons, splice sites and other biological signals). On the contrary, so-called extrinsic approaches essentially rely on the existence of similarities between the sequence to annotate and other known sequences (either proteins, transcripts or other genomic sequences). Several existing gene finding tools are essentially intrinsic (or ab initio): this is the case for Genscan [4], HMMgene [5] or SLAM [6]. For such a gene finder, the predicted gene structure is defined as an optimal prediction, that is the most probable according to its underlying probabilistic model. In the presence of AS however, a unique prediction is not sufficient. One obvious possibility is to look for suboptimal predictions. This can be done for a classic HMM-based gene finder by a modification of the Viterbi algorithm, thus providing the set of the k best predictions. This approach has been applied eg. in HMMgene or in FGENES-M (unpub.). Another way to obtain suboptimal solutions from a HMM is to do HMM sampling [7]. This method, which consists in randomly generating parses according to the posterior probabilities, has been implemented in the gene finder SLAM. Usually, a very large amount of samples are needed to generate just a single prediction that differs from the optimal one. Genscan adopt a different approach and search for alternative exons not represented in the optimal prediction. This is done using a forward-backward algorithm to identify potential exons for which the a posteriori likelihood is larger than a given threshold.
In addition to the fact that all these exclusively intrinsic approaches cannot take into account transcript evidences, they suffer from two major problems of sensibility and specificity:
First of all, these methods assume that predictions representing AS variants should have a probability which is very close to the optimal probability according to the underlying gene model. This is however quite arguable, especially when the alternative structure significantly differs from the optimal one. Actually, when an AS variant eg. shifts from a strong to a weak or a non-consensus splice site or shows a complete coding exon skipping event, it is quite unlikely that the probability will remain in the neighborhood of the optimum since it will not be able to incorporate the corresponding splicing or coding score.
Moreover, a strong specificity problem has been observed for this approach. Since a very large number of alternative predictions can always be produced for any sequence, it is essential to be able to distinguish those reflecting real AS variants from in silico false positives. To perform this, and as long as AS sites dedicated prediction tools are unavailable, the probability of a prediction alone cannot be sufficient and additional evidence is required.
In opposition to the purely intrinsic approach, the analysis of experimental data can provide useful information. More specifically, sequences of mature transcripts resulting from AS provide reliable evidence of the existence of the AS event. Large scale studies have already been undertaken to detect AS evidences from transcript alignments and to collect them in databases such as eg. HASDB [8], ASDB [9], ASAP [10], ASD [11], EASED [12] or ProSplicer [13]. Some software tools have also been designed to perform and/or exploit transcript alignment with the aim of identifying alternative gene structures. Such extrinsic annotation tools include GeneSeqer [14], ASPic [15], TAP [16,17], and PASA [18]. Except for GeneSeqer which is more focused on performing spliced alignment, the three other software adopt the same strategy: using genomically aligned transcripts, the aim is to determine the exon-intron structure(s) compatible with the greatest number of transcripts. Another approach, Cluster Merge [19], has been recently used in the Ensembl annotation system [20] to identify minimal sets of transcript variants compatible with genomically aligned ESTs evidences.
Unlike intrinsic methods, extrinsic approaches take advantage of transcript information. However, they also suffer from some limitations : first they entirely depend on the availability of transcribed sequences which bounds their sensitivity. With little exceptions (like TAP that exploits genomic sequence properties to identify gene boundaries, including eg. a polyA site scanning step, or GeneSeqer, that contains an intrinsic splice sites scoring method), they cannot predict a splice site if it is not represented in a transcript-to-genome alignment and therefore require a total coverage of each gene with all exon-intron boundaries. This can be problematic considering the ESTs fragmented nature. Moreover, when such methods can take advantage of a total gene coverage, the CDS localization remains to be done and the pure transcript predictions may not respect elementary coding gene properties (such as the presence of an ORF w.r.t. a given frame). Furthermore, overlapping transcripts are sometimes assumed to come from the same mature mRNA and are therefore merged. This may lead to the fusion of two overlapping transcripts coming from exclusive inconsistent mRNA variants, thus forcing the prediction to respect a chimeric virtual assembly.
Finally, and because experimental transcripts cannot exist for every existing gene, both intrinsic and extrinsic information are needed inside an annotation pipeline [20]. The predictions provided by two different approaches can be different and even inconsistent, and merging them together requires a careful inspection of human curators, as performed in [18]. A fully integrative method alleviates all these problems. GrailEXP [21] seems to be the only gene finder that tried to go in this direction. However, it can only consider AS events leading to complete exon inclusion/retention, ignoring thus approximatively half of the AS cases [8,18]. The underlying approach remains unpublished.
To extend the domain of application of gene prediction to alternatively spliced gene structure prediction, we have designed an intrinsic/extrinsic integrative annotation method with the following aims:
• For a given genomic sequence, an optimal gene structure prediction is produced, as usual.
• In addition to this optimal prediction, for every transcript sequence providing evidence of AS, an optimal prediction consistent with this splicing form is also provided.
• Each additional or alternative gene structure prediction has to be supported by some biological evidence.
• Full-length transcript coverage is not required for a complete gene structure identification.
• Each prediction satisfies the usual constraints on gene structure. A correct proteic coding gene is defined by a succession of one or more exons separated by introns flanked by splice sites. It contains a CDS between a start and a stop codon, and no in-frame stop in coding exons.
Our aim is to combine the advantages of the intrinsic and extrinsic approaches in an integrative system allowing for AS detection based on the analysis of genomic aligned transcript sequences. The method has been implemented inside EuGÈNE-M, a new version of the Arabidopsis thaliana EuGÈNE gene finder [22,23], and applied to a reference genes set.
Results
To evaluate the interest of EuGÈNE-M compared to existing transcripts-based approaches, we applied it on the spl7 Arabidopsis thaliana gene. This gene codes for the squamosa promoter-binding protein-like 7, has 10 exons and two known alternative mature mRNA variants, both supported by a distinct full-length cDNA (accession AY063815 and AF367355, Figure 1). The genomic alignments of these cDNAs provide two correct and reliable gene structures used as reference annotations. The structures differ only by the 3' extremity of the 9th exon. However, beyond these 2 complete cDNAs, only the first and the two last exons are covered by ESTs. This partial EST coverage configuration is interesting because without the full-length cDNAs (unavailable in dbEST), finding a correct gene structure with pure extrinsic assembly tools would not be possible.
Given only the genomically ESTs alignments, we applied EuGÈNE-M on the genomic sequence containing the spl7 gene. Since two ESTs (T04465 and AI995153) show incompatible alignments (see Methods), EuGÈNE-M computes two additional predictions, each being consistent with one of them. The first alternative prediction is the same as the optimal one and corresponds to one variant; the second corresponds to the other variant.
For a more extensive test, we applied EuGÈNE-M on AraSet [24], a data set of Arabidopsis thaliana curated genes recently used in the assessment of GeneSeqer [25]. Since EuGÈNE has already been evaluated on this benchmark set, performing as one of the most accurate gene finder [22], the aim of this test is to provide an estimation of an alternatively spliced genes ratio on a reference set. Predictions are available in the additional files. On the 168 AraSet reference genes, 9 show at least two alternative predictions, that were carefully analyzed. This is summarized in Table 1. All these predictions but two correspond to potential alternative splicing events. Among the two remaining ones, a first predicted AS event corresponds to an incompatibility caused by an apparently incompletely spliced EST. The other is more interesting since it is caused by two ESTs from two different genes lying on opposite strands and overlapping on their 3' ends. In this case, EuGÈNE-M is forced to predict two overlapping genes, one on each strand, which effectively address the usual impossibility for existing gene models to predict overlapping genes. Of course, these predictions, as all in silico expertise, require experimental verifications to be confirmed. If we assume the 7 remaining genes are effectively subject to AS, this yield to an AS rate of ~4.2%, a ratio in the same order as previously estimated, from 1.5% [26] to 6.5% (computed from [18]).
Discussion
In the recent assessment of GeneSeqer on this AraSet data set, only three AS cases were reported [25]. However, the authors only reported AS cases that were detected in GeneSeqer high-quality alignments and producing introns differing from the AraSet annotated introns. We therefore verified that our alignments were consistent with the GeneSeqer assessment alignment data available in the Arabidopsis thaliana Genome Database AtGDB [27,28]. We noticed an alignment difference for only one of our alternative EST (CF652136), not present in the AtGDB because of its dbEST entry date (Oct. 2003). We also checked if the AS variants predicted by EuGÈNE-M were already reported in the AS sections of the AtGDB [26,29] and of the TIGRdb [18,30]. Only 3 of our detected AS predictions were already reported in both databases, and 3 were missing in all of them (Table 1), confirming that this methodology can help to automatically discover new potential AS cases, even on a well studied dataset.
The analysis of these AS cases confirms that AS seems to be much less frequent in A. thaliana than in Homo sapiens. Nevertheless, this AS ratio estimation is expected to increase in the future with the growth of transcript data availability. Another interesting point is the nature of the variants: on this gene set, the majority of AS cases involves a simple acceptor or donor alternative splice site. Notice however that since EuGÈNE-M's underlying model allows arbitrary alternative gene structure to be predicted, it is not limited to the prediction of such simple AS events and can perfectly cope with complex AS events, as found in mammals. This methodology can also be integrated in other existing gene finders where the score of a gene structure is defined as the sum of elementary scores of the signals and nucleotides involved in the gene structure (this includes HMM-based gene finders).
Conclusions
In this paper we have presented a new method to deal with alternative splicing in annotation and gene prediction. This integrative approach combines the advantages of an intrinsic and an extrinsic process to incorporate AS detection into ab initio gene finding. We showed that this method allows the discovery of new alternative spliced genes, with the reliability of extrinsic annotation and the potential exhaustiveness of ab initio gene prediction.
Methods
The process that goes from the original genomic sequence and associated aligned transcripts to the AS prediction is composed of three steps which we rapidly describe here :
• first, the set of genomically aligned transcripts is analysed to detect AS evidences on the basis of splicing inconsistency between transcripts variants.
• Then, the graph-model used in EuGÈNE to model potential gene structures is modified to take into account these aligned transcripts. For each transcript variant, the graph used in EuGÈNE for gene structure prediction is connected to an additional parallel graph subunit where local constraints are injected according to the exon-intron information provided by the corresponding transcript alignment.
• Finally, an extended version of the dynamic programming algorithm used for obtaining an optimal prediction allows to identify, for each graph subunit, the best prediction consistent with the corresponding transcript alignment.
Detection of AS evidences from transcripts analysis
Since EuGÈNE already exploits transcripts information to improve the gene prediction process [22], the AS prediction only requires to consider transcripts providing evidence of AS. With this purpose, we focus on inconsistencies between transcript alignments.
Transcript sequences are first aligned against the genomic sequence using a spliced alignment tool. The choice of the source transcript database and the alignment tool is not a priori imposed by the method. Transcript sequences in our analysis were extracted from the A. thaliana section of dbEST [31] (release Dec. 2003: 190, 708 entries), and aligned in two steps. For the first step we used sim4 [32], a fast software that can deal with huge EST datasets. In the second step, we used GeneSeqer [14], usually more accurate on splice junction identification, to realign all transcripts aligned by sim4 that passed the following filtering process.
A first filtering step is performed on the basis of the transcript sequence and alignment quality. To be considered, an alignment has to satisfy some constraints defined by filtering parameters. For Arabidopsis thaliana, default parameters values are set as following: transcript length between 30 and 10000 bp, minimum alignment length = 95% of the transcript length, minimum identity score of 97%, maximum gap length of 5000 bp, maximum match length of 4000 bp. By default, and to avoid genomic contamination, unspliced transcripts are removed from the analysis. Moreover, because of the frequent weak alignment quality at the terminal regions, alignments extremities are shortened (by 15 bp by default).
The second filtering step depends on the relation between transcript alignments. To detect AS evidences, every pair of overlapping transcript alignments is analyzed. We consider two special types of pairwise relation : a transcript alignment A is labeled as included in B if and only if for each genomic position of A the same genomic position in B shares the same alignment information (either gap or match). Every transcript included in another transcript is ignored by default. Transcript alignments A and B are labeled as incompatible if and only if there is a genomic position for which both ESTs are informative and give an inconsistent information, that is a gap (representative of the presence of an intron) is faced with a match (representative of the presence of an exon, coding or not). Examples of incompatible ESTs are displayed in Figure 1 and 5. Since we focus on AS evidences, we only keep transcripts labeled as incompatible after all pairwise comparisons. Considering orientation of ESTs, the information on the clone-sequencing orientation that can be found with ESTs is totally ignored in this filtering process because of its unreliability. In practice, spliced EST can be reliably oriented by looking for splice sites on the hit-match frontier of the EST alignments and by choosing the strand for which such splice sites exist. The parameters of these two automatic filtering steps can be modified by the user through a simple text file. We will denote the resulting transcript alignments kept as alternative transcripts.
The gene-finder EuGène
General description
EuGÈNE is a gene finding software based on a directed acyclic graph gene model [22]. For each nucleotide of the genomic sequence, every possible annotation of this nucleotide is represented in the graph. The graph is designed to model the whole prediction space: all consistent gene structures can be represented by a path through the graph, whose weight is defined as the sum of its edges weights. The minimum weight path defines the optimal prediction. Several sources of evidence are used to weight the edges of the graph and a shortest-path dynamic programming algorithm (linear in time and space) scans the graph to provide an optimal path which represents the best gene prediction according to available evidences.
Structure of the initial graph
Each path through the graph represents a potential gene structure prediction for the genomic sequence (Figure 2). The graph is composed of k tracks that represent the possible annotations that can be attributed to each nucleotide (coding, intronic, intergenic and UTR, with specific strand and frame). Let ℓ be the genomic sequence length. For a given nucleotide's position i with (1 <i < ℓ) and for each track j with (1 <j <k) the two flanking vertices and are defined. Two edges are also built : a contents edge linking to and a transition edge linking to (Figure 3). Additional transition edges are put from to all according to the occurrence of a potential biological signal allowing a switch from state j for the nucleotide at the position i to the state j' for the following nucleotide. For example, on the position i before the occurrence of an ATG, a transition edge linking to (where j corresponds to the UTR5' track and j' is the coding exon track on the appropriate frame) is present, as illustrated in Figure 3. Two special vertices and are added at the extremities of the graph. They are respectively connected to all and all . Initially, all edges are oriented from left (5') to right (3'). It is easy to see that all possible gene structure can be represented by a path from to .
Weighting the graph
The weight of a path is the sum of all the weights of the edges in the path. The edges are weighted according to the evidences used. EuGÈNE can combine several sources of evidence such as probabilistic coding models, output of splice site or start codon prediction software and sequence similarities with transcripts, proteins, or other genomic sequences [33]. Contents and transition edges c and t are penalized respectively by weights Wc and Wt according to a weighting function characterized by parameters specifically set for the corresponding source of evidence. The set of parameters is optimized on a learning dataset by maximizing the overall accuracy of the software. For more information about the weighting methods, please refer to [22].
Example of transcript alignment integration
A transcript-to-genome alignment can easily be taken into account by weighting the appropriate edges of the graph. To favor a gene prediction in the alignment region, the intergenic track edges included in this region can be penalized by increasing their weight. More finely, the exon and the intron tracks edges can also be penalized at all positions involved respectively in a gap and in a match in the alignment. Thus, all gene structure prediction inconsistent with the transcript alignment information tends to be penalized. More drastically, it is possible to force the prediction to be consistent with the alignment by applying infinite penalty weights. Note that there are several such predictions since the start codon used is unknown and the transcript may be incomplete.
Initial algorithm
To identify the optimal path defined by the lowest weight, EuGÈNE uses a dynamic programming algorithm inspired from Bellman's shortest-path algorithm [34], also used for HMM in its Viterbi's version. Improvements of this algorithm allow EuGÈNE to take into account constraints on gene element lengths. For simplicity, we will not describe these sophistications in this paper. The algorithm of EuGÈNE associates to each vertex a variable which contains the weight of the optimal path from to and a variable which contains the vertex that precedes in this optimal path. The weight of this path can be computed recursively from 5' to 3' as:
A short example is displayed in Figure 3. The vertex that minimizes this value provides the previous . At vertex , the best path is retrieved by a simple backtracing procedure through all π. This algorithm is linear in time and space in the length of the sequence (O(ℓ) complexity). It is important to note that the same algorithm can be used in a backward version (from to ), by computing at each vertex the weight of the best path from to as .
AS evidences integration
Given an alternative transcript genomic alignment, any prediction which is optimal among all the predictions that are consistent with the alignment evidence will be called an alternative prediction. Given the set of the previously detected alternative transcripts, we want EuGÈNE-M to produce a set of alternative predictions such that every alternative transcript has a corresponding prediction in this set. A simple way to produce such an alternative prediction would be to inject the exon-intron structure information given by the transcript alignment into the graph as described above (using infinite weights to force the prediction to strictly respect the alignment evidence), and then to execute EuGÈNE on the resulting graph. However, obtaining all alternative predictions would require one execution for each alternative transcript. n being the number of transcripts and l the genomic sequence length, this would result in a O(ln) time complexity, which is not appropriate for long genomic sequences and numerous transcripts.
Hopefully, this complexity can be drastically reduced. The general idea to achieve a realistic complexity is to duplicate the subsection of the graph region involved in an alignment to create a so called local "Parallel Graph Subunit" (PGS), connected to the main graph at its extremities. Each alignment information is taken into account as constraints in the corresponding PGS, in such a way that finding the optimal path going through the PGS provides a corresponding optimal alternative prediction.
Extending the graph model with PGS
For a transcript alignment that extends from position g to h on the genomic sequence, the entire subsection of the graph between g and h is duplicated to create a Parallel Graph Subunit (PGS) (Figure 4). This PGS is connected to the main graph at its extremities by special so-called deviation edges. For each track j, a deviation edge links the source vertex in the main graph to its copy at the PGS left extremity, and another connects the source vertex in the main graph to its copy at the PGS right extremity. The deviation edges are all oriented from the main graph to the PGS. The weights of the PGS edges, initially identical to the weight of the original edges, are modified according to the corresponding transcript alignment : gaps and matches forbid respectively the exonic and the intronic tracks, and the entire PGS intergenic track is forbidden.
Findingalternative predictions
The modified algorithm proceeds in two steps. A first scan starts from to and applies the recursive formula described above to compute all , branching into each PGS (Figure 5). Thus, at each nucleotide's position and for each track (including those in the PGS), the weight of the optimal path from the left extremity is identified. At , the optimal path is obtained by backtracing. Furthermore, for any given PGS, the cost of an optimal path going from , through the PGS and then to each of the righmost vertices is known.
Then all edges (except the deviation) are reversed, and the backward version of the same shortest-path algorithm is used from to to compute all . This step ignores the PGS.
For a given PGS A, if we now consider the vertices at the rightmost extremity of A, then the weight of an optimal path that goes from to through A can be computed as . From the given vertex, backtracing in both directions provides an optimal path that represents an optimal prediction in accordance with the transcript alignment evidence.
Output
Predictions are produced in the standard GFF format. The entire optimal annotation is first displayed, followed by the alternative ones. To enhance the readability and to avoid redundancy, for each alternative prediction the name of the corresponding transcript is mentioned and the region that differs from the optimal prediction is displayed. Besides, if several predictions are identical (regarding their predicted CDS only, UTR length differences being ignored), a single representative is displayed, along with the list of its associated transcripts.
Computation time
The initial filtering and incompatible transcripts identification requires O(n2) pairwise comparisons. Each comparison is itself linear in the maximum number of introns in the transcript compared, which is typically bounded by a small constant and the whole process is therefore in O(n2).
The step that corresponds to the two dynamic programming scans (application of the recursive formula) requires a time and space complexity which is linear in the size of the input data. Indeed, if L is the total nucleic sequences length (genomic + kept alternative transcript), the weights of all (alternative and optimal) predictions can be computed in O(L).
For the backtracing and output step, since each alternative prediction has to be displayed in the region where it differs from the optimal one, and because this can extend beyond the alignment region, it is not possible to obtain an algorithm which is linear in the size of the input. However, it is possible to reach a linear complexity in the size of the output. This can be done by a simple modification of the standard backtracing procedure to avoid a full backtrace for each prediction. This is yet not implemented in the current version of the software.
A typical run of EuGÈNE-M on an AMD Athlon 1.7 GHz takes 47 sec. for a 500 kb BAC (for which 945 transcript alignments were kept after the first quality filtering step).
Authors' contributions
SF designed and implemented the filtering algorithm as well as the double dynamic programming algorithm for AS prediction. He also ran the experiments on the Araset dataset. TS directed the research. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
EuGène's predictions on AraSet. The additional file (SupplementaryFiles.tar.gz) contains gene structure predictions in the standard GFF text format for every AraSet sequence. ESTs detected as Incompatible by the method are displayed at the top of each prediction (if any).
Click here for file
Figures and Tables
Figure 1 EST/cDNA alignments on the spl7 gene region. Thick lines represent matches an dotted lines, gaps. Above the genomic sequence, the 2 full-length cDNAs that provide the two correct reference gene structures are presented. Arrows indicate the start and stop codons. The ESTs T04465 and AI995153 present inconsistent splicing profiles and are labeled as incompatible.
Figure 2 EuGène's directed acyclic graph for a short example sequence. For simplicity purposes, only the forward strand is considered. The DNA sequence is shown above the graph. Horizontal tracks represent the different possible annotations: intergenic (bottom), UTR 5' and 3', exon in the 3 frames, intron in 3 phases (the phase of an intron is defined according to the splicing position in the last codon of the previous exon). On each track, 2 vertices are used to represent each nucleotide. These 2 vertices are linked horizontally by a contents and a transition edge (see the text and Figure 4 for details). Dotted arrows show occurrences of biological signals (like start/stop codons and donor/acceptor splice sites). They produce additional transition edges at the corresponding position. Since this version of EuGÈNE does not include any promoter or polyA site prediction tool, transitions from intergenic to UTR and vice-versa are allowed at every nucleotide position. All consistent gene structures can be represented by a path connecting the initial and terminal vertices and .
Figure 3 Detail of EuGène's directed acyclic graph and algorithm. The zoomed region contains the two first nucleotides of the example sequence of Figure 3 (C at position i - 1, and A at position i), and two annotation tracks (UTR5' for j and exon in frame 2 for j + 1). The contents edges c connect the l vertices to the following r vertices of the same track. Transition edges t are either horizontal and link the r vertices to the l vertices of the same track, or transversal and link the r vertices to all possible l vertices according to the occurrence of a biological signal in the sequence. In this example, between and a vertex allows the transition from the UTR5' track at position i - 1 to the exonic track at i because the A nucleotide at position i is the first nucleotide of a potential start codon ATG. The dynamic programming algorithm used in EuGÈNE determines, for each vertex r, which vertex precedes r in the optimal path. In this example, at position i for the track j the best path leading to from the left has a weight (only one origin is possible). For the track j + 1, the best path leading to will be attributed a weight of either , whatever the lower.
Figure 4 Extension of EuGène's graph by a PCS to incorporate a single alternative transcript alignment. From the main graph (bottom) described in Figure 3, a Parallel Graph Subunit (PGS) is built (above) by duplicating the whole graph section involved in the EST alignment (between the graphs). Gene structure evidences provided by the alignment are taken into account in the PGS by forbidding the intergenic track all along the alignment, intronic tracks at match positions (light grey), and exonic tracks in gap positions (dark grey). Dotted arrows represent the two algorithm scans, the forward version from left to right, and the backward version from right to left. At the junction point in the PGS, an optimal prediction is obtained. Figure not to scale.
Figure 5 Integration of several incompatible ESTs in EuGène-M's graph and algorithm. A: EST alignments (plain lines represent exons, dotted lines, intron) on a genomic sequence (thick line). Each displayed EST is incompatible with at least another one. B: Multiple extensions of EuGÈNE's graph model after having processed these alignments. Each PGS (Figure 3) contains the information provided by its source EST. The dotted arrows show the algorithm progression through the resulting graph during the first scan, from the left to the right.
Table 1 Analysis of the AS cases detected by EuGène-M in the AraSet genes data set. First, sequence IDs, genes and EST involved are reported. The TIGR and AtGDB columns indicate if each AS case is reported in these databases. The AS status is described as follows: ACC = alternative acceptor splice site, DON = alternative donor splice site, -EX = exon skipping (an entire exon lacks in the reported variant), +IN = intron inclusion (an internal part of an exon is spliced), FP = false positive AS. nt = nucleotide. Some ESTs of the At2g39780 gene in seq16 are not correctly aligned: the use of either GeneSeqer or sim4 with default options leads to a missed 4 nt exon (not involved in AS). In seq50, the 168 nt additional (+IN) intron from the EST CF652136 is flanked by GC-CT (instead of the canonical GT-AG). In seq62, the EST AV542276 from the gene At4g37040 overlaps with an intron of EST AV562725 from the neighboring gene At4g37050. In seq65, the EST BE521212 is not spliced between the exon 5 and 6 of the gene At2g44100 (intron retention case), and is thus suspected of incomplete maturation. Except for CF652136, all alignments can be browsed on the AtGDB site.
AraSet sequence Gene ID EST evidence TIGR AtGDB AS status Note
seq14 At2g47640 AI998209 Y N ACC skip 3 nt
seq16 At2g39780 AV832175 N N -EX incorrect alignments
seq50 At5g46290 CF652136 N N +IN non consensus splice sites
seq53 At3g51800 AV544387 Y Y ACC add 27 nt
seq62 At4g37070 AU236122 Y Y DON add 33 nt
seq62 At4g37050 AV542276 N N FP overlapping genes
seq65 At2g44100 BE521212 N N FP incomplete splicing
seq65 At2g44120 BE524396 Y N ACC skip 33 nt
seq69 At4gl4350 AV547538 Y Y +IN skip 105 nt
==== Refs
Modrek B Lee C A genomic view of alternative splicing Nat Genet 2002 30 13 9 11753382 10.1038/ng0102-13
International Human Genome Sequencing Consortium Initial sequencing and analysis of the human genome Nature 2001 409 860 921 11237011 10.1038/35057062
Johnson J Castle J Garrett-Engele P Kan Z Loerch P Armour C Santos R Schadt E Stoughton R Shoemaker D Genome-wide survey of human alternative pre-mRNA splicing with exon junction microarrays Science 2003 302 2141 4 14684825 10.1126/science.1090100
Burge C Karlin S Prediction of complete gene structures in human genomic DNA J Mol Biol 1997 268 78 94 9149143 10.1006/jmbi.1997.0951
Krogh A Using database matches with for HMMGene for automated gene detection in Drosophila Genome Res 2000 10 391 7 10779478 10.1101/gr.10.4.523
Alexandersson M Cawley S Pachter L SLAM: cross-species gene finding and alignment with a generalized pair hidden Markov model Genome Res 2003 13 496 502 12618381 10.1101/gr.424203
Cawley SL Pachter L HMM sampling and applications to gene finding and alternative splicing Bioinformatics 2003 19 II36 II41 14534169
Modrek B Resch A Grasso C Lee C Genome-wide detection of alternative splicing in expressed sequences of human genes Nucleic Acids Res 2001 29 2850 9 11433032 10.1093/nar/29.13.2850
Gelfand MS Dubchak I Dralyuk I Zorn M ASDB: database of alternatively spliced genes Nucleic Acids Res 1999 27 301 2 9847209 10.1093/nar/27.1.301
Lee C Atanelov L Modrek B Xing Y ASAP: the Alternative Splicing Annotation Project Nucleic Acids Res 2003 31 101 5 12519958 10.1093/nar/gkg029
Thanaraj TA Stamm S Clark F Riethoven JJ Le Texier V Muilu J ASD: the Alternative Splicing Database Nucleic Acids Res 2004 32 D64 9 14681360 10.1093/nar/gkh030
Pospisil H Herrmann A Bortfeldt RH Reich JG EASED: Extended Alternatively Spliced EST Database Nucleic Acids Res 2004 32 D70 4 14681361 10.1093/nar/gkh136
Huang HD Horng JT Lee CC Liu BJ ProSplicer: a database of putative alternative splicing information derived from protein, mRNA and expressed sequence tag sequence data Genome Biol 2003 4 R29 12702210 10.1186/gb-2003-4-4-r29
Usuka J Zhu W Brendel V Optimal spliced alignment of homologous cDNA to a genomic DNA template Bioinformatics 2000 16 203 211 10869013 10.1093/bioinformatics/16.3.203
Bonizzoni P Pesole G Rizzi R Benson G, Page R A Method to Detect Gene Structure and Alternative Splice Sites by Agreeing ESTs to a Genomic Sequence Algorithms in Bioinformatics, 3rd International Workshop (WABI), LNCS 2003 Springer Verlag 63 77
Kan Z Rouchka E Gish W States D Gene structure prediction and alternative splicing analysis using genomically aligned ESTs Genome Res 2001 11 889 900 11337482 10.1101/gr.155001
Kan Z States D Gish W Selecting for functional alternative splices in ESTs Genome Res 2002 12 1837 45 12466287 10.1101/gr.764102
Haas B Delcher A Mount S Wortman J Smith RJ Hannick L Maiti R Ronning C Rusch D Town C Salzberg S White O Improving the Arabidopsis genome annotation using maximal transcript alignment assemblies Nucleic Acids Res 2003 31 5654 66 14500829 10.1093/nar/gkg770
Eyras E Caccamo M Curwen V Clamp M ESTGenes: alternative splicing from ESTs in Ensembl Genome Res 2004 14 976 87 15123595 10.1101/gr.1862204
Curwen V Eyras E Andrews TD Clarke L Mongin E Searle SMJ Clamp M The Ensembl automatic gene annotation system Genome Res 2004 14 942 50 15123590 10.1101/gr.1858004
Xu Y Uberbacher E Automated gene identification in large-scale genomic sequences J Comput Biol 1997 4 325 38 9278063
Schiex T Moisan A Rouzé P EuGène, an eukaryotic gene finder that combines several type of evidence Computational Biology, selected papers from JOBIM' 2000, no 2066 in LNCS 2001 Springer Verlag 118 133
EuGène web site
Pavy N Rombauts S Déhais P Mathé C Ramana D Leroy P Rouzé P Evaluation of gene prediction software using a genomic data set: application to Arabidopsis thaliana sequences Bioinformatics 1999 15 887 99 10743555 10.1093/bioinformatics/15.11.887
Brendel V Xing L Zhu W Gene structure prediction from consensus spliced alignment of multiple ESTs matching the same genomic locus Bioinformatics 2004 20 1157 69 14764557 10.1093/bioinformatics/bth058
Zhu W Schlueter S Brendel V Refined annotation of the Arabidopsis genome by complete expressed sequence tag mapping Plant Physiol 2003 132 469 84 12805580 10.1104/pp.102.018101
Dong Q Schlueter SD Brendel V PlantGDB, plant genome database and analysis tools Nucleic Acids Res 2004 32 D354 9 14681433 10.1093/nar/gkh046
GeneSeqer evaluation on AtGDB
Alternative splicing on AtGDB
Arabidopsis splicing variations on TIGR db
Boguski MS Lowe TM Tolstoshev CM dbEST-database for expressed sequence tags Nat Genet 1993 4 332 3 8401577 10.1038/ng0893-332
Florea L Hartzell G Zhang Z Rubin G Miller W A computer program for aligning a cDNA sequence with a genomic DNA sequence Genome Res 1998 8 967 974 9750195
Foissac S Bardou P Moisan A Cros MJ Schiex T EUGENE'HOM: A generic similarity-based gene finder using multiple homologous sequences Nucleic Acids Res 2003 31 3742 5 12824408 10.1093/nar/gkg586
Bellman R Dynamic Programming 1957 Princeton, New Jersey: Princeton Univ Press
| 15705189 | PMC550657 | CC BY | 2021-01-04 16:35:47 | no | BMC Bioinformatics. 2005 Feb 10; 6:25 | utf-8 | BMC Bioinformatics | 2,005 | 10.1186/1471-2105-6-25 | oa_comm |
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-231572070710.1186/1471-2164-6-23Research ArticleNonrandom distribution and frequencies of genomic and EST-derived microsatellite markers in rice, wheat, and barley La Rota Mauricio [email protected] Ramesh V [email protected] Ju-Kyung [email protected] Mark E [email protected] Department of Plant Breeding and Genetics, 240 Emerson Hall, Cornell University, Ithaca, NY, 14853, USA2 Department of Plant & Soil Science, 138 ARC Building, Alabama A&M University, Normal, AL, 35762, USA2005 18 2 2005 6 23 23 27 7 2004 18 2 2005 Copyright © 2005 La Rota et al; licensee BioMed Central Ltd.2005La Rota 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
Earlier comparative maps between the genomes of rice (Oryza sativa L.), barley (Hordeum vulgare L.) and wheat (Triticum aestivum L.) were linkage maps based on cDNA-RFLP markers. The low number of polymorphic RFLP markers has limited the development of dense genetic maps in wheat and the number of available anchor points in comparative maps. Higher density comparative maps using PCR-based anchor markers are necessary to better estimate the conservation of colinearity among cereal genomes. The purposes of this study were to characterize the proportion of transcribed DNA sequences containing simple sequence repeats (SSR or microsatellites) by length and motif for wheat, barley and rice and to determine in-silico rice genome locations for primer sets developed for wheat and barley Expressed Sequence Tags.
Results
The proportions of SSR types (di-, tri-, tetra-, and penta-nucleotide repeats) and motifs varied with the length of the SSRs within and among the three species, with trinucleotide SSRs being the most frequent. Distributions of genomic microsatellites (gSSRs), EST-derived microsatellites (EST-SSRs), and transcribed regions in the contiguous sequence of rice chromosome 1 were highly correlated. More than 13,000 primer pairs were developed for use by the cereal research community as potential markers in wheat, barley and rice.
Conclusion
Trinucleotide SSRs were the most common type in each of the species; however, the relative proportions of SSR types and motifs differed among rice, wheat, and barley. Genomic microsatellites were found to be primarily located in gene-rich regions of the rice genome. Microsatellite markers derived from the use of non-redundant EST-SSRs are an economic and efficient alternative to RFLP for comparative mapping in cereals.
==== Body
Background
The genetic maps of grass species have been constructed using a variety of marker types. Most of the older species-specific molecular maps were constructed with RFLP markers, but in recent times there has been increased utilization of PCR-based markers because of accessibility and higher throughput. Conservation of gene content and order has been detected among grass genomes through the use of comparative maps [1,2]. The applications of comparative maps have been discussed many times in the past (See for example: [3]), however genetic maps are not always designed with a comparative study in mind, thus, current maps from different grass species (and in many cases, within the same species) seldom share an adequate number of common (anchor) markers to allow researchers to bridge across maps with an adequate resolution. This is especially true when comparing the genome maps of the Triticeae tribe with the maps of rice or maize, which on average share 3 to 4 markers per wheat homoeologous chromosome group. The lack of anchor markers for bridging across species is exacerbated as new maps are constructed using PCR-based markers such as AFLP, genomic microsatellites and single nucleotide polymorphisms (SNPs) rather than the transferable but laborious cDNA-based RFLP markers.
Genomic SSR (gSSR) markers are biased towards genome specificity [4,5] and generally do not transfer to other species, making them less useful for the generation of comparative maps. For comparative mapping, markers must identify orthologous loci and be polymorphic in two or more species [6].
Recently, several researchers [6-17] have addressed the lack of transferability of gSSRs to other genomes by limiting primer design to transcribed regions, that are expected to have higher levels of conservation across related organisms. Public EST sequence databases from the Poaceae family can be scanned for the presence of SSRs both in protein-coding regions and in untranslated regions of genes (5' or 3' UTRs).
When compared to gSSRs, EST derived SSRs (EST-SSRs) were less polymorphic in a study in hexaploid wheat [9] with only 25% polymorphism, but the successful markers were of high quality and were also polymorphic in durum wheat. Thiel [13] reported a higher level of polymorphism in barley (42%). Lower polymorphism requires more effort to design primers for testing a larger set of candidate markers, but the ease and speed of finding SSRs among freely available EST sequence data offsets this extra effort. This approach is only feasible in species for which there have been EST sequencing projects.
We used the rice genome sequence generated by the International Rice Genome Sequence Project (IRGSP) [18] to identify gSSRs that can potentially serve as sources of markers for mapping. In an equivalent experiment, non-redundant sets of transcript sequences from rice, wheat and barley were scanned (dataset was obtained from the TIGR gene-index databases [19]) and those transcripts containing SSRs were collected and mapped in-silico to the rice genome. SSR-containing transcripts derived from different species, sharing a pre-determined threshold of similarity and matching the same location in rice were considered putative orthologs that may be used as anchors in comparative mapping studies.
This paper describes a methodology for developing EST-SSR markers from wheat, barley and rice as markers for developing independent species maps as well as for homologous anchor markers for comparative maps. Over 13,000 untested PCR primer pairs for EST-SSRs were generated from the three gene indices and made available to the research community interested in grass genomes. Researchers are encouraged to evaluate a subset of primer pairs and send feedback regarding their utility to the GrainGenes [20] database for posting. The list (along with all other materials: scripts, programs source code and database schemas) is available from the additional files as well as from the Triticeae EST-SSR Coordination webpage in GrainGenes [21].
Results
Frequency of microsatellite types and motifs
Based on combinations of all four nucleotides, the canonical set of SSR motifs is represented by four different duplets (AC, AG, AT, CG), 10 different triplets, 33 different quadruplets and 102 different quintuplet motifs. In the source sequences, all these basic nucleotide motifs can be represented in variant forms of the same basic set or by their reverse complements but to keep a consistency in the database for estimating frequencies, they were transformed into the canonical motifs. Reverse complements and variants would include, for example, CT for AG and GAG for AGG. Sets of unigene sequences such as the TIGR gene indices have the advantage of built-in elimination of redundant SSR counts allowing for more precise estimates of EST-SSR frequency. The rice genomic SSR (gSSR) counts were processed a-posteriori to eliminate redundancy due to BAC/PAC clone overlaps (see methods). Mononucleotide repeats are common in genomic DNA and some are known to be polymorphic but these were deliberately avoided in the unigene database, because they are usually added by the RNA polymerase and are not present in the template DNA (e.g. poly A tails).
Table 1 summarizes the frequencies of SSRs in the TIGR gene indices and the rice genome, grouped by SSR type (di-, tri-, tetra-, and penta-nucleotides) and by several minimum acceptable microsatellite lengths starting at 12 bp or longer. Counts were cumulative, meaning that the totals for SSRs with 12 or more nucleotides included the number of longer SSRs displayed in columns to the right in this table (Table 1). If microsatellite expansion were motif-sequence and location independent, under the null hypothesis one would expect that all types of SSRs would expand at the same rate, and thus, the proportions of SSR types would remain equal from short to long SSRs. However, the proportions of SSR types changed with the length of the SSRs (as did the proportions of motifs, Table 2) and the different SSR types (and motifs) seemed to expand or contract at different rates in the genome of rice, and at different rates when compared to the unigenes of the other two cereals (Tables 1 and 2). SSRs of the trinucleotide type were the most frequent overall. The relatively higher proportion of short trinucleotide-based SSRs in rice unigenes was apparent when compared to the same length categories in wheat or barley unigenes and when compared to the rice gSSRs. The proportion of dinucleotide repeats was greater among genomic microsatellites than among EST-SSRs, and this proportion increased with longer SSRs, overtaking trinucleotide repeats in all datasets when the minimum length was set to 20 bp in gSSRs and 30 bp in EST-SSRs. For instance, in the rice genome 72% of the SSRs longer than 30 bp were of the dinucleotide type; thus, it appears that the rice genomic sequence is relatively richer in dinucleotide SSRs than the gene indices from rice (a subset of the genome) and the other two species.
Table 1 Frequency table of perfect and imperfect microsatellites grouped by type of repeat in the non-redundant rice genome (IRGSPnr), rice gene index (Osgi), barley gene index (Hvgi) and wheat's gene index (Tagi) under nine different constraints for minimum SSR length, starting at a minimum of 12 bp. 'n' is SSR count, 'f' is relative proportion (fraction). 'U.f' is unigene fraction: the percentage of unigenes from that species containing at least one SSR.
Dataset SSRLength > = 12 > = 14 > = 15 > = 16 > = 18 > = 20 > = 24 > = 30 > = 40
SSR type n f n f n f n f n f n f n f n f n f
Osgi dinucleotide 1751 0.07 1206 0.09 950 0.07 950 0.14 673 0.12 507 0.16 366 0.25 237 0.45 136 0.67
trinucleotide 17342 0.67 8921 0.63 8921 0.65 4272 0.62 4272 0.75 1993 0.62 912 0.61 242 0.46 56 0.28
tetranucleotide 4039 0.16 1242 0.09 1242 0.09 1242 0.18 275 0.05 275 0.09 84 0.06 18 0.03 9 0.04
pentanucleotide 2706 0.10 2706 0.19 2706 0.20 458 0.07 458 0.08 458 0.14 124 0.08 33 0.06 2 0.01
Total SSR 25838 14075 13819 6922 5678 3233 1486 530 203
U.f 50% 27% 27% 13% 11% 6% 3% 1% 0.4%
Hvgi dinucleotide 1304 0.07 916 0.10 706 0.08 706 0.15 508 0.15 387 0.17 237 0.25 162 0.48 96 0.67
trinucleotide 9176 0.52 4326 0.45 4326 0.46 1990 0.43 1990 0.57 971 0.42 464 0.48 115 0.34 34 0.24
tetranucleotide 4122 0.23 1359 0.14 1359 0.15 1359 0.29 379 0.11 379 0.16 103 0.11 10 0.03 4 0.03
pentanucleotide 2976 0.17 2976 0.31 2976 0.32 594 0.13 594 0.17 594 0.25 157 0.16 53 0.16 10 0.07
Total SSR 17578 9577 9367 4649 3471 2331 961 340 144
U.f 36% 20% 19% 10% 7% 5% 2% 1% 0.3%
Tagi dinucleotide 2606 0.08 1890 0.11 1526 0.09 1526 0.18 1122 0.18 890 0.22 668 0.32 478 0.54 317 0.70
trinucleotide 18650 0.55 8671 0.50 8671 0.51 3847 0.46 3847 0.61 1840 0.45 989 0.48 343 0.39 124 0.27
tetranucleotide 8189 0.24 2253 0.13 2253 0.13 2253 0.27 572 0.09 572 0.14 181 0.09 26 0.03 8 0.02
pentanucleotide 4570 0.13 4570 0.26 4570 0.27 773 0.09 773 0.12 773 0.19 243 0.12 33 0.04 5 0.01
Total SSR 34015 17384 17020 8399 6314 4075 2081 880 454
U.f 31% 16% 16% 8% 6% 4% 2% 1% 0.4%
IRGSPnr dinucleotide 32202 0.17 23056 0.21 18541 0.17 18541 0.33 14081 0.34 11286 0.39 8516 0.53 6307 0.72 4284 0.81
trinucleotide 78832 0.41 38759 0.35 38759 0.36 17615 0.31 17615 0.43 8392 0.29 4431 0.28 1400 0.16 582 0.11
tetranucleotide 48836 0.25 15030 0.14 15030 0.14 15030 0.26 3789 0.09 3789 0.13 1636 0.10 625 0.07 378 0.07
pentanucleotide 34102 0.18 34102 0.31 34102 0.32 5591 0.10 5591 0.14 5591 0.19 1401 0.09 385 0.04 54 0.01
Total SSR 193972 110947 106432 56777 41076 29058 15984 8717 5298
Table 2 Frequency table (counts and relative proportions) of the ten most common motifs from perfect and imperfect microsatellites under the same minimum length constraints used in Table 1 found in the non-redundant rice genome (IRGSPnr), rice gene index (Osgi), barley gene index (Hvgi) and wheat's gene index (Tagi).
IRGSPnr Osgi Hvgi Tagi
SSRLength motif prop. count motif prop. count motif prop. count motif prop. count
> = 12 CCG 0.18 34238 CCG 0.32 8309 CCG 0.19 3415 CCG 0.20 6643
AG 0.08 14603 AGG 0.10 2501 AGG 0.09 1536 AGG 0.08 2720
AT 0.06 11594 ACG 0.06 1677 AGC 0.07 1250 AGC 0.07 2506
AGG 0.05 10534 AGC 0.06 1583 AAG 0.04 730 AAC 0.05 1805
ACG 0.04 7151 AG 0.04 1074 AG 0.04 714 AG 0.04 1463
AGC 0.03 6576 ACC 0.04 1031 ACG 0.04 668 AAG 0.04 1305
AAG 0.03 4889 AAG 0.04 994 ACC 0.03 604 ACC 0.03 1184
AAAAG 0.02 4809 ATC 0.02 604 ATC 0.03 446 ACG 0.03 1162
AAAT 0.02 4756 ATCG 0.01 341 AGGGG 0.02 426 ATC 0.02 758
ACC 0.02 4517 AAAG 0.01 270 AGGG 0.02 391 AC 0.02 749
> = 14 CCG 0.16 17881 CCG 0.33 4622 CCG 0.18 1717 CCG 0.19 3270
AG 0.09 10293 AGG 0.09 1277 AGG 0.07 716 AGG 0.07 1256
AT 0.08 9385 AG 0.06 822 AGC 0.06 607 AGC 0.07 1190
AGG 0.05 5343 ACG 0.06 776 AG 0.06 543 AG 0.07 1140
AAAAG 0.04 4809 AGC 0.05 774 AGGGG 0.04 426 AAC 0.05 907
AAAAT 0.03 3221 AAG 0.03 482 AAG 0.04 359 AAG 0.03 559
ACG 0.03 3178 ACC 0.03 478 ACG 0.03 287 AC 0.03 530
AGC 0.03 2995 ATC 0.02 273 ACC 0.02 236 ACG 0.03 479
AGAGG 0.02 2627 CCGCG 0.02 249 AC 0.02 231 ACC 0.03 478
AAG 0.02 2448 AGAGG 0.02 248 CCCCG 0.02 224 ATC 0.02 322
> = 16 AT 0.15 8334 CCG 0.32 2238 CCG 0.17 784 CCG 0.17 1452
AG 0.14 7931 AG 0.10 674 AG 0.09 424 AG 0.11 951
CCG 0.14 7820 AGG 0.09 643 AGC 0.07 317 AGC 0.07 560
AGG 0.04 2438 AGC 0.05 354 AGG 0.07 316 AGG 0.06 542
AC 0.03 1714 ACG 0.05 339 AAG 0.04 186 AC 0.05 413
AAAG 0.03 1457 AAG 0.04 249 AC 0.04 175 AAC 0.04 344
AGC 0.03 1421 ACC 0.03 218 AGGG 0.03 154 AAG 0.03 291
AAT 0.02 1403 AT 0.02 127 AGGGG 0.03 150 ACG 0.03 221
AAAT 0.02 1390 ATC 0.02 114 ACC 0.02 115 AGGG 0.02 203
ACG 0.02 1345 ATCG 0.02 112 ACG 0.02 110 ACC 0.02 202
> = 18 CCG 0.19 7820 CCG 0.39 2238 CCG 0.23 784 CCG 0.23 1452
AT 0.17 7103 AGG 0.11 643 AG 0.09 318 AG 0.12 753
AG 0.14 5580 AG 0.09 503 AGC 0.09 317 AGC 0.09 560
AGG 0.06 2438 AGC 0.06 354 AGG 0.09 316 AGG 0.09 542
AGC 0.03 1421 ACG 0.06 339 AAG 0.05 186 AAC 0.05 344
AAT 0.03 1403 AAG 0.04 249 AGGGG 0.04 150 AAG 0.05 291
ACG 0.03 1345 ACC 0.04 218 AC 0.04 124 AC 0.04 267
AAG 0.03 1228 ATC 0.02 114 ACC 0.03 115 ACG 0.04 221
AC 0.03 1133 AT 0.02 100 ACG 0.03 110 ACC 0.03 202
AAAAG 0.03 1131 AGAGG 0.01 66 ATC 0.02 83 ATC 0.02 145
> = 20 AT 0.22 6360 CCG 0.32 1032 CCG 0.16 383 CCG 0.15 625
AG 0.14 4055 AG 0.12 385 AG 0.11 251 AG 0.15 618
CCG 0.12 3431 AGG 0.10 317 AGC 0.07 170 AGG 0.06 262
AGG 0.04 1190 AGC 0.05 156 AGGGG 0.06 150 AGC 0.06 250
AAAAG 0.04 1131 ACG 0.04 143 AGG 0.06 139 AAC 0.06 249
AAT 0.04 1121 AAG 0.04 134 AAG 0.04 101 AC 0.05 195
AC 0.03 806 ACC 0.03 97 AC 0.04 85 AAG 0.04 176
AGAT 0.02 693 AT 0.03 84 ACC 0.03 64 ACC 0.02 92
AAG 0.02 685 AGAGG 0.02 66 CCCCG 0.03 59 ACGAT 0.02 91
AGC 0.02 573 ATC 0.02 57 AGAGG 0.02 52 ACG 0.02 77
> = 24 AT 0.34 5501 CCG 0.29 436 CCG 0.19 180 AG 0.23 484
AG 0.16 2504 AG 0.19 278 AG 0.18 171 CCG 0.13 275
CCG 0.10 1574 AGG 0.10 152 AGC 0.10 93 AAC 0.10 203
AAT 0.05 866 AAG 0.06 83 AGGGG 0.06 56 AGG 0.06 131
AGG 0.04 615 AGC 0.05 69 AAG 0.05 48 AC 0.06 127
AC 0.03 487 ACG 0.04 65 AGG 0.05 47 AAG 0.06 122
AGAT 0.03 486 AT 0.04 65 AC 0.04 41 AGC 0.06 118
AAG 0.03 447 ACC 0.03 48 ACC 0.04 37 ACGAT 0.04 86
AAAAG 0.02 345 ATC 0.02 26 AT 0.03 25 AT 0.03 56
ACAT 0.02 302 AC 0.01 21 ATC 0.02 21 ATC 0.02 39
> = 30 AT 0.54 4680 AG 0.35 184 AG 0.39 133 AG 0.42 368
AG 0.15 1343 CCG 0.17 90 AGGGG 0.09 29 AAC 0.15 132
AAT 0.05 420 AT 0.09 46 CCG 0.07 24 AC 0.08 68
CCG 0.04 310 AGG 0.08 40 AAG 0.07 23 AAG 0.07 59
AGAT 0.03 295 AAG 0.06 32 AGC 0.07 23 AT 0.05 42
AC 0.03 280 AGC 0.05 24 AC 0.05 16 CCG 0.05 42
AAG 0.03 236 ACG 0.03 16 AT 0.04 13 AGC 0.04 37
ACAT 0.02 198 ACC 0.02 13 AAC 0.03 10 AGG 0.03 27
AGG 0.02 176 AAC 0.02 10 ACC 0.03 9 AAT 0.01 13
AAAAG 0.01 119 AGAGG 0.02 9 CCCCG 0.02 8 ATC 0.01 12
> = 40 AT 0.69 3631 AG 0.51 104 AG 0.60 86 AG 0.57 260
AG 0.09 494 AT 0.14 29 AGC 0.07 10 AAC 0.14 62
AAT 0.07 353 AAG 0.08 17 AAG 0.06 9 AC 0.07 31
AGAT 0.04 192 AGC 0.04 8 AT 0.06 8 AAG 0.07 31
AC 0.03 157 CCG 0.03 7 AGGGG 0.05 7 AT 0.06 26
ACAT 0.03 153 ACG 0.03 6 AAC 0.03 4 AGC 0.03 14
AAG 0.02 109 AGG 0.03 6 ACC 0.01 2 ACAT 0.01 5
CCG 0.01 27 AAC 0.02 5 ACAT 0.01 2 AGG 0.01 3
AAC 0.00 24 ATC 0.02 4 ATC 0.01 2 CCG 0.01 3
AGG 0.00 19 ACAT 0.02 4 ACT 0.01 2 AAT 0.01 3
ESTs are a rich source of SSRs
The abundance of SSRs (perfect and imperfect) in the unigenes can range from one in every 100 to one in every two unigenes depending on the minimum length (Table 1). When all SSRs with a minimum length of 12 bp are tabulated, 50%, 36% and 31% of rice, barley and wheat unigenes have at least one SSR. When the minimum length was raised to 16 bp the proportion was reduced to 13, 10 and 8%, respectively. Rice unigenes had a higher frequency of SSRs than did barley and wheat for most minimum lengths, but not for SSRs longer than 20 bp, where the relative abundance was similar in all three gene indices. The nearly two-fold difference at a minimum length of 18 bp was mostly due to the high abundance of trinucleotide SSRs in rice unigenes relative to wheat and barley. The abundance of trinucleotide repeats decreased by about one half for each repeat unit added to the series. The decline in abundance was steeper for tetranucleotide and pentanucleotide repeats but was less than one half for dinucleotide repeats, which at lengths greater than or equal to 30 bp, became the predominant type in all datasets. At 30 bp or longer, the AT motif was most common among gSSRs while AG was more numerous among EST-SSRs motifs (Table 2).
Wheat unigenes contained a larger number of SSRs for all repeat length categories, followed by rice and barley unigenes. This is probably because wheat has more than twice the number of unigenes than rice or barley with 109,782 for wheat, 51,569 for rice and 48,159 for barley. The larger number of unigenes in hexaploid wheat may result from divergence of the genes in the three genomes, but also from a relatively larger EST dataset, i.e., more ESTs have been sequenced for wheat, with a sequence redundancy of 3.8×, versus 6× and 2.7× for barley and rice, respectively (see methods).
The number of the ten most frequent motifs was tabulated for different minimal SSR lengths (Table 2). The relative proportions of motifs fluctuated with different length constraints as well as source species. At a minimum SSR length of 12 bp, CCG was predominant in all datasets, but AT and AG were more frequent in the higher range of minimal SSR lengths. Among dinucleotides in the rice EST-SSRs, AG and AT were the most common, but AG and AC were more common in wheat and barley EST-SSRs. Besides CCG, other frequent trinucleotide motifs were AGG and AGC. The trinucleotide (CCG)n microsatellite was present in both coding regions and UTRs. In coding regions, this triplet has the potential to code for the amino acids proline (CCG), arginine (CGG), alanine (GCC), glycine (GGC), but among these, expansion of the motif leading to additions of the amino acid proline could have the strongest effects on protein structure while alanine and glycine would have relatively small effects.
The longest SSRs were genomic microsatellites (as long as 726 bp). Unigenes had few SSRs longer than 40 bp, up to 333 bp in wheat ESTs. Often these were not useful for developing SSR markers because no flanking sequence was available to design primers. The overall mean length for rice gSSRs equal to or longer than 12 bp was 16.5 (s.d = 12.7), with no significant differences in mean gSSR lengths among the chromosomes in rice, while the mean length for EST-SSRs was 15.3 (s.d = 6) with no significant differences among the three species gene indices (t-test 2-sample with unequal variances, p > 0.01).
Density of gSSRs and comparison to EST-SSRs mapped in silico in rice chromosome 1
The contiguity of the pseudomolecule sequence of rice chromosome 1 (R1) (a virtual sequence composed of the assembly of tiling path clones), with only eight gaps for the whole chromosome, provided a convenient framework of coordinates for calculating density estimates of gSSRs features (by in-silico scanning with the Sputnik program), and for anchoring rice, wheat and barley unigenes associated with microsatellites (by sequence similarity). Best similarity matches between the rice genome and 8,259 barley, 16,917 rice and 13,565 wheat EST-SSR unigenes were identified using BLASTN. From these, a total of 6,373 EST-SSRs mapped to R1 pseudomolecules including 1,104 from barley, 3,568 from rice and 1,701 from wheat with 88.8%, 98.4% and 89.1% average sequence similarity respectively.
Density of gSSRs equal to or longer than 12 bp in R1 ranged from 1 gSSR in 2.8 kbp near the centromere to 1 gSSR per 1.1 kbp in the distal regions (Figure 1A). For a more stringent subset of gSSRs (≥ 16 bp tetranucleotides, ≥ 18 bp dinucleotides and trinucleotides, and ≥ 20 bp pentanucleotides), the density ranged from 1 gSSR in 10 kbp around the centromere to 1 gSSR in 3.8 kbp in the densest region of the short arm (Figure 1A).
Figure 1 A) Features in the contiguous 42.5 Mb of rice chromosome 1: Comparison between the density (counts per 500 kbp) of the stringent subset of gSSRs (gSSR Stringent: ≥ 16 bp tetranucleotides, ≥ 18 bp dinucleotides and trinucleotides, and ≥ 20 bp pentanucleotides; in orange), density of the best matches to rice unigenes associated with microsatellites (OsgiSSR_vs_chr1; in green) and density of genomic microsatellites (gSSR; in blue) with length ≥ 12 bp. The centromere location is indicated by "CEN". B) Features in the contiguous 42.5 Mb of rice chromosome 1: Comparison between the density of gSSRs ≥ 12 bp (in blue), the density of best sequence similarity matches to rice unigenes associated with microsatellites (OsgiSSR_vs_chr1; in green), and the density of best sequence similarity matches between rice unigenes (Osgi_vs_chr1) and rice chromosome 1 (gray line and red points). C) Rice chromosome 1 density plots of the stringent subset of gSSRs and its component types (di-, tri-, tetra- and pentanucleotides)
The comparison of the density of gSSRs to, a) the density of all rice unigenes mapped to R1, and b) the density of EST-SSRs (a subset from the unigenes) mapped to R1 provided an estimate of the relationship between gSSRs and gene regions in the rice genome. There was a striking resemblance in the patterns of the plots for the density of gSSRs and the density of unigene-derived EST-SSRs in R1 pseudomolecules (Figure 1A). The similarity in density patterns was less apparent but still present between gSSRs and R1 matches to all rice unigenes (Figure 1B), and these densities were significantly correlated (r = 0.45, p ≤ 1E-5; Figure 2).
Figure 2 Linear regression of the density of genes in rice chromosome 1 (roughly estimated by the matches of OsGI sequences to this chromosome: Osgi_vs_chr1) on the density of gSSRs (≥ 12 bp) in rice chr1. Pearson correlation is 0.45. Regression coefficients are highly significant (P-value = 1.2E-05). Density is expressed in counts per 500 kbp.
A decomposition of the set of stringent gSSRs (see the methods section for criteria defining the "stringent gSSRs") by types in R1 (Figure 1C) showed that the relative proportions of the pentanucleotide gSSRs ≥ 20 bp were consistently lower over the majority of the chromosome, while the proportion of the other three types of microsatellites was higher, indicating that pentanucleotides are only a small component in the non-homogeneous distribution of the stringent subset.
Development of primers for cereal EST-SSRs
We designed primer pairs for 5,425 wheat, 3,036 barley and 4,726 rice EST-SSRs conforming to the stringent restrictions described in the methods. The average product size expected from the set of designed primers was 217 bp for rice EST-SSRs, 213 for wheat and 218.9 for barley.
Of those EST-SSRs, 42% of the wheat and 56% of the barley were mapped in-silico to the rice genome. The additional file 1 contains the list of primer pairs that can be downloaded for testing.
Discussion
Are microsatellites preferentially associated with gene-rich DNA in rice?
Morgante and colleagues [22] reported that in plants, gSSRs were preferentially associated with non-repetitive DNA such as the gene-rich regions. They found a highly significant, positive, linear relationship (r2 = 0.94, p < 0.006) between genomic microsatellite frequency and the percentage of single copy DNA in several plant species with a wide range of genome sizes. Estimates of repetitive and non-repetitive single-copy DNA fractions were based on reviews of the literature describing renaturation kinetics experiments for each of the species. Plant species that have gone through genome expansion due to retrotransposon amplification, such as maize and wheat, had a lower genomic microsatellite frequency indicating that SSR frequency is not a function of overall genome size but rather the relative proportion of single-copy DNA.
In this study, the best similarity matches between rice unigene sequences and the genomic sequence of rice chromosome 1 were used to estimate the density of transcribed regions along R1. This estimate was compared to both the density of gSSRs and the density of EST-SSRs (the latter group being the intersection of the set of gSSRs and the set of transcribed regions, or unigenes). The density pattern of transcribed regions (unigenes in R1) and of SSRs within transcribed regions (unigenes in R1 with SSRs or EST-SSRs) followed closely the density pattern of gSSRs in rice chromosome 1 (r = 0.45, p < 1 × 10-5 and r = 0.62, p < 1 × 10-10, respectively) (Figures 1A, 1B and 2).
The density (counts per 500 Kbp) of gSSRs along R1 was higher than both the density of transcribed regions and the density of EST-SSRs. A large number of gSSRs that are not already included in the set of EST-SSRs could still be associated with genes because, as Figure 1B suggests, they are preferentially found in genic regions near promoter regions or inside introns and away from the highly repetitive and gene-poor DNA in heterochromatin. In rice, (AT)n SSRs are rare among ESTs, but are the most common gSSRs among the long group (= 20 bp). The (AT)n SSR motif is frequently found along with sequences of the Micropon family of MITEs [23,24] which are associated with gene-rich regions.
Other reports have documented a role for SSRs that are associated with genes in the control of gene expression. For example, several human diseases have been linked with events of triplet expansions in the past [25]. Chromatin remodeling and gene silencing via histone-deacetylation/cytosine-methylation are among the putative functions of SSRs in the vicinity of genes, especially if GC rich. Coffee [26] showed that histone deacetylation (and methylation of CpG bases) leading to lower expression at the FMR1 locus in fragile X was a consequence of CCG repeat expansion. In another example, the expansion of a (AGC)n SSR in the 3' UTR of the myotonic dystrophy (DM) protein kinase gene could potentially affect the expression due to changes in local chromatin structure [27]. It has been found that DM patients have a reduced or complete loss of a nuclease-hypersensitive site in the region of the gene. Further analysis showed that the majority of DM protein kinase transcripts from cells carrying the repeat expansion also lacked the last two exons of a normal transcript, showing that the repeat expansions affected the splicing at the 3' end.
In rice, although the presence of a single-base mutation breaking an intron splice site is more directly responsible for the difference in phenotypes of the waxy gene, polymorphism due to SSR expansion has been associated with variation of expression levels in different japonica and indica varieties [28,29]. The effect that microsatellites might have in gene expression in plants may be observed as natural phenotypic variation.
A strategy to exploit the EST database for microsatellite markers
One strategy to better exploit a database of EST-SSRs in order to find polymorphic markers is to first sample the longest SSRs (≥ 30 bp), favoring dinucleotide repeats, then follow with trinucleotide, tetranucleotide and pentanucleotide repeats [23]. After exhausting the longest SSRs, one would then proceed with another cycle to select shorter SSRs. Short trinucleotide-based microsatellites such as (CCG)n, the most abundant group overall (Table 2), are more likely to derive from coding regions, thus reducing the chances for finding polymorphism [13]. This strategy is based on the following observations from our results and the literature: 1) Dinucleotides are a better source of polymorphic markers than the other types [30]. 2) Longer SSRs generally have a higher tendency to be polymorphic [23]. 3) SSRs deriving from UTRs have the potential for a higher polymorphism than those derived from coding regions, which are constrained by purifying selection [30].
One percent of unigenes from the three species examined in this study have SSRs starting with a minimum of 30 bp (Table 1). Overall, 880 wheat unigenes, 530 rice unigenes and 340 barley unigenes contain at least one of these long microsatellites and primer pairs were successfully designed for 451, 276 and 148 of these long EST-SSRs in wheat, rice and barley, respectively. At this minimum length, trinucleotide repeats were not the most frequent. Nearly 50% of the EST-SSRs longer than 30 bp were based on dinucleotide repeats (Table 1), with (AG)n being the most common motif (Table 2). Yet, the frequency of SSR types among those for which primers could be designed did not follow this pattern. Trinucleotide repeats were still the most common type in this group, followed by dinucleotides. This was due to the fact that dinucleotides are found preferentially in the UTRs of transcripts, and their sequences had fewer surrounding bases to anchor acceptable primers.
After relaxing the microsatellite length constraint to a minimum of 20 bp, the overall number of SSRs in the unigenes increased to around 5%. An additional 3,195, 2,703 and 1,991 EST-SSRs in wheat, rice and barley become available for primer design. Acceptable primer pairs were designed for 2,622 wheat, 2,183 rice and 1,476 barley EST-SSRs in this category (which included the set mentioned previously).
The set of EST-SSRs with acceptable primer pairs (Additional file 1) were selected from among all dinucleotide and trinucleotide EST-SSRs with a minimum of 18 bp, tetranucleotide EST-SSRs longer than 16 bp and pentanucleotide EST-SSRs 20 bp or longer. However, from 5,424 wheat unigenes associated with microsatellites and having a set of PCR primers in our database, only 2,323 had a best match in the rice BAC/PACs with our stringency settings. The rest (57%) are not anchored to the rice genome but still have potential to provide polymorphic wheat microsatellite markers. The same applies to 44% of the barley EST-SSRs with acceptable primers. The reasons for a large number of wheat/barley unigenes without matches to rice genomic sequence include not having the complete sequence of the rice genome available (the majority of clones were still in sequencing phase 2, with gaps) and having a relatively high stringency setting for filtering wheat and barley sequence comparisons to rice. In previous comparisons between wheat EST unigenes and the same version of the rice genome sequence draft [31,32], we found that 40% of the unigenes did not significantly match a sequence in the rice genome.
Conclusion
The relative proportions of di-, tri-, tetra-, and penta nucleotide repeats and motifs varied widely depending on length and were not consistent among the species examined. We have shown that ESTs are a good source of SSRs that can be exploited to develop microsatellite markers for wheat, barley and rice. The advantage to this approach is that the sequences are already available resulting in a lower cost than designing and testing microsatellites from anonymous genomic libraries, even if the polymorphism rate for EST-derived markers is lower.
EST-SSRs are useful for enhancing individual species maps, but can be used as anchor probes for creating links between maps in comparative studies when designed from sets of orthologous genes, as demonstrated by Yu et al [33]. The annotation and/or the sequence similarity between putative orthologous genes from two related species can provide the basis for their use in comparative maps. More than 13,000 primer pairs were designed to amplify fragments from a stringent subset of EST-SSRs in wheat, rice and barley and are available to the public for testing.
Using a different methodology, our results substantiated the report by Morgante et al [22] suggesting that microsatellites are predominantly found in the vicinity of genes. In some instances, their presence in the vicinity of genes may implicate a regulating function by mechanisms involving chromatin remodelling and DNA methylation.
Methods
Source of unigene sequences
TIGR's non-redundant gene indices [34] from wheat, barley and rice were downloaded in January of 2003. The databases were: OsGi rel.11 (January 2003) for rice transcripts, with 51,569 non-redundant sequences after processing 139,918 ESTs; HvGi rel. 5 (December 2002) for barley transcripts, with 48,159 non-redundant sequences from 304,061 initial sequences and TaGi rel. 6 (January 2003) for wheat transcripts with 109,782 non-redundant sequences (415,125 initial sequences).
Source of rice genomic sequences
All analyses of the rice genome used the version released in December 2002 by the International Rice Genome Sequencing Consortium [18] and consisted of a minimum tiling path of 3,280 BAC or PAC clones for the 12 rice chromosomes. There were nine pseudomolecules of assembled, contiguous sequence available for rice chromosome 1 that replaced the overlapping clones in that chromosome [35]. For the rest of the genome, accession numbers for the individual BAC/PAC clones in the tiling path were used to download the corresponding sequence from NCBI GenBank [36]. The tiling path for chromosomes 2 to 12 was used to facilitate the posterior ordering of clones.
Scanning of the rice genome and the non-redundant EST-datasets for SSRs
The TIGR gene indices and the genome of rice were scanned with a modified version of Sputnik [22] available from the University of Delaware [37] to find all perfect and imperfect SSRs having 2 to 5 nucleotides in the basic repeat unit and at least 12 bp in total length. For imperfect SSRs, up to 10% sequence deviation from a perfect SSR was included. We modified the way the program handles input sequences in the NCBI FASTA format and the format of the program's output, making it easier to export to relational databases. No changes were made to the underlying algorithms written by C. Abajian and modified by Morgante's group. The version of Sputnik used to generate the microsatellite data for this report can be obtained from the GrainGenes EST-SSR coordination webpage [21], or by downloading the additional file 2.
In order to eliminate the problem of counting the same microsatellites several times in the rice genome due to the redundancy created by overlapping regions between contiguous BAC/PACs in chromosomes 2 to 12, the gSSRs were annotated as redundant or not, according to their location in the tiling path. When located to a region in the BAC/PAC that overlapped with a neighbor clone (based on the tiling path information as well as MegaBLAST [38] pairwise alignments) only the SSRs belonging to the overlapped region of the top (northern) clone were counted while those present in the overlapped region of the bottom (south) clone were ignored. Of course, all SSRs found in unique, non-overlapping regions of rice clones were counted. A perl script that performed queries and updates to the SQL database (via the DBI perl module) scanned the tables of genomic microsatellites and flagged them according to the procedure explained above. Thus 24% of the genomic microsatellites (of length ≥ 12 bp) found in the rice BAC/PAC clones were ignored, as they were duplicates due to clone overlaps.
Table 1 shows the counts and relative proportions of SSRs found in the four datasets (rice, wheat and barley gene indices as well as in the non-redundant rice genomic) for dinucleotides, trinucleotides, tetranucleotides and pentanucleotides when having different minimum microsatellite lengths (greater than or equal to 12 bp) as the starting point. Table 2, on the other hand, shows the relative proportions of the ten most common motifs for each dataset when subject to different constraints for minimum microsatellite lengths.
In-silico mapping of grass Non-redundant EST-SSRs
The set of EST unigenes associated with SSRs from wheat and barley was matched against the sequence of the rice genome to provide a putative map location in rice. Only the best hits were recorded for any given EST unigene. The similarity threshold was set at an E-value < 1 × 10-10 and at least 80% similarity over 100 bp of minimum alignment. Rice EST unigenes were matched with the same criteria except for a higher similarity threshold of 95%. The inferred location in the rice genome for rice EST unigenes was used to estimate the proportion of rice gSSRs that were associated with regions containing genes.
PCR primer design
We developed a perl script (see Additional file 3) that automatically queries the database of EST-SSRs to design primers in batch based on what was learned in previous experiments and on recommendations found in the literature to maximize the chance of selecting polymorphic microsatellite markers. The script used the BioPerl module [39] to control the Primer3 core program [40,41], feeding each of the SSR source sequences and specifying the target regions to be amplified via PCR.
EST-SSRs were selected for primer design when conforming to the following more stringent restrictions (referred to as the set of stringent SSRs or gSSR stringent):
a) The SSRs are dinucleotides or trinucleotides of length equal or larger than 18 bp, tetranucleotides equal or larger than 16 bp or pentanucleotides equal or larger than 20 bp.
b) The imperfect SSRs have less than 10% mismatches or gaps relative to a perfect SSR of the same length and motif.
c) There is a minimum of 50 bp surrounding the SSR edges in the source sequence to allow for possible primer design.
The parameters used for the Primer3 program specified an optimal Tm of 60°C with a minimum and maximum of 57°C and 65°C, respectively, and a 30% to 70% GC content with a low chance of dimer or hair-loop formation. The range for PCR product length was set to be between 100 and 300 bp.
Abbreviations
EST: Expressed Sequence Tag. SSR: Simple Sequence Repeat. gSSR: genomic SSR. EST-SSR: EST-derived SSR. UTR: Untranslated region flanking a coding region in DNA and messenger RNA. IRGSP: International Rice Genome Sequence Project. R1: rice chromosome 1. BAC/PAC: Bacterial artificial chromosomes or bacteriophage P1 artificial chromosomes (for cloning of large DNA fragments). MITE: Miniature Inverted-repeat Transposable Element.
Authors' contributions
ML did all programming and design of computational experiments and databases. RVK contributed in the first database design. JY did wet-lab testing of a subset of primer pairs.
ML and MES drafted the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
EST-SSR designed primers Table listing the stringent subset of SSRs (≥ 16 bp tetranucleotides, ≥ 18 bp dinucleotides and trinucleotides, and ≥ 20 bp pentanucleotides) found in rice, barley and wheat gene indices for which primer sequences were designed, the source sequences, the primers, their in-silico mapping in the rice genome (to BAC/PAC clones or pseudo-molecule) and relevant metadata. File is a spreadsheet table, compressed with the zip program. The file (14 Mb uncompressed) is also available at the GrainGenes Triticeae EST-SSR Coordination page
Click here for file
Additional File 2
Modified Sputnik source code and executable This is the source code with modifications, to the microsatellite searching program "Sputnik", originally written by Chris Abajian from the University of Washington at Seattle. The set of files are compressed using the zip program. File is also available at
Click here for file
Additional File 3
Perl script to design primers in batch The script uses the Bioperl perl modules to control the Primer3 program in order to design primers from a microsatellite database stored in a MySQL database. The script can be modified to accommodate similar schemas on any database engine supported by the perl DBI module. File is also available at
Click here for file
Acknowledgements
ML acknowledges and is grateful for the financial support from Colciencias during the first years of his MS/PhD studies and from AstraZeneca(Syngenta) for support during subsequence years.
==== Refs
Ahn S Anderson JA Sorrells ME Tanksley SD Homoeologous relationships of rice, wheat and maize chromosomes Mol Gen Genet 1993 241 483 90 7903411 10.1007/BF00279889
Gale MD Devos KM Comparative genetics in the grasses Proc Natl Acad Sci U S A 1998 95 1971 4 9482816 10.1073/pnas.95.5.1971
Devos KM Gale MD Comparative genetics in the grasses Plant Mol Biol 1997 35 3 15 9291955 10.1023/A:1005820229043
Pestsova EG Ganal MW Roder MS Isolation and mapping of microsatellite markers specific for the D genome of bread wheat Genome 2000 43 689 97 10984182 10.1139/gen-43-4-689
Chen X Cho YG McCouch SR Sequence divergence of rice microsatellites in Oryza and other plant species Mol Genet Genomics 2002 268 331 43 12436255 10.1007/s00438-002-0739-5
Kantety RV La Rota M Matthews DE Sorrells ME Data mining for simple sequence repeats in expressed sequence tags from barley, maize, rice, sorghum and wheat Plant Mol Biol 2002 48 501 10 11999831 10.1023/A:1014875206165
Scott KD Eggler P Seaton G Rossetto M Ablett EM Lee LS Henry RJ Analysis of SSRs derived from grape ESTs Theor Appl Genet 2000 100 723 726 10.1007/s001220051344
Cordeiro GM Casu R McIntyre CL Manners JM Henry RJ Microsatellite markers from sugarcane (Saccharum spp.) ESTs cross transferable to erianthus and sorghum Plant Sci 2001 160 1115 1123 11337068 10.1016/S0168-9452(01)00365-X
Eujayl I Sorrells ME Baum M Wolters P Powell W Isolation of EST-derived microsatellite markers for genotyping the A and B genomes of wheat Theor Appl Genet 2002 104 399 407 12582712 10.1007/s001220100738
Rossetto M McNally J Henry RJ Evaluating the potential of SSR flanking regions for examining taxonomic relationships in the Vitaceae Theor Appl Genet 2002 104 61 6 12579429 10.1007/s001220200007
Rohrer GA Fahrenkrug SC Nonneman D Tao N Warren WC Mapping microsatellite markers identified in porcine EST sequences Anim Genet 2002 33 372 6 12354146 10.1046/j.1365-2052.2002.00880.x
Varshney RK Thiel T Stein N Langridge P Graner A In Silico analysis on frequency and distribution of microsatellites in ESTs of some cereal species Cell Mol Biol Lett 2002 7 537 46 12378259
Thiel T Michalek W Varshney RK Graner A Exploiting EST databases for the development and characterization of gene-derived SSR-markers in barley (Hordeum vulgare L.) Theor Appl Genet 2003 106 411 22 12589540
Gupta PK Rustgi S Sharma S Singh R Kumar N Balyan HS Transferable EST-SSR markers for the study of polymorphism and genetic diversity in bread wheat Mol Genet Genomics 2003 270 315 23 14508680 10.1007/s00438-003-0921-4
Eujayl I Sledge MK Wang L May GD Chekhovskiy K Zwonitzer JC Mian MA Medicago truncatula EST-SSRs reveal cross-species genetic markers for Medicago spp Theor Appl Genet 2004 108 414 22 13679975 10.1007/s00122-003-1450-6
Saha MC Mian MA Eujayl I Zwonitzer JC Wang L May GD Tall fescue EST-SSR markers with transferability across several grass species Theor Appl Genet 2004 109 783 91 15205737 10.1007/s00122-004-1681-1
Han ZG Guo WZ Song XL Zhang TZ Genetic mapping of EST-derived microsatellites from the diploid Gossypium arboreum in allotetraploid cotton Mol Genet Genomics 2004 272 308 27 15368122 10.1007/s00438-004-1059-8
International Rice Genome Sequence Project
TIGR gene-index databases
Matthews DE Carollo VL Lazo GR Anderson OD GrainGenes, the genome database for small-grain crops Nucl Acids Res 2003 31 183 186 12519977 10.1093/nar/gkg058
Triticeae EST-SSR coordination webpage
Morgante M Hanafey M Powell W Microsatellites are preferentially associated with nonrepetitive DNA in plant genomes Nat Genet 2002 30 194 200 11799393 10.1038/ng822
Temnykh S DeClerck G Lukashova A Lipovich L Cartinhour SW McCouch SR Computational and experimental analysis of microsatellites in rice (Oryza sativa L.): frequency, length variation, transposon associations, and genetic marker potential Genome Res 2001 11 1441 52 11483586 10.1101/gr.184001
Akagi H Yokozeki Y Inagaki A Mori K Fujimura T Micron, a microsatellite-targeting transposable element in the rice genome Mol Genet Genomics 2001 266 471 80 11713677 10.1007/s004380100563
Cummings CJ Zoghbi HY Fourteen and counting: unravelling trinucleotide repeat diseases Hum Mol Genet 2000 9 909 16 10767314 10.1093/hmg/9.6.909
Coffee B Zhang F Ceman S Warren ST Reines D Histone modifications depict an aberrantly heterochromatinized FMR1 gene in fragile x syndrome Am J Hum Genet 2002 71 923 32 12232854 10.1086/342931
Frisch R Singleton KR Moses PA Gonzalez IL Carango P Marks HG Funanage VL Effect of triplet repeat expansion on chromatin structure and expression of DMPK and neighboring genes, SIX5 and DMWD, in myotonic dystrophy Mol Genet Metab 2001 74 281 91 11592825 10.1006/mgme.2001.3229
Bligh HF-J Larkin P-D Roach P-S Jones C-A Fu H Park W-D Use of alternate splice sites in granule-bound starch synthase mRNA from low-amylose rice varieties Plant Molecular Biology 1998 38 407 415 9747848 10.1023/A:1006021807799
Bao S Corke H Sun M Microsatellites in starch-synthesizing genes in relation to starch physicochemical properties in waxy rice (Oryza sativa L.) Theor Appl Genet 2002 105 898 905 12582915 10.1007/s00122-002-1049-3
Cho YG Ishii T Temnykh S Chen X Lipovich L McCouch SR Park WD Ayres NM Cartinhour SW Diversity of microsatellites derived from genomic libraries and GenBank sequences in rice (Oryza sativa L.) Theor Appl Genet 2000 100 713 722 10.1007/s001220051343
Sorrells ME La Rota M Bermudez-Kandianis CE Greene RA Kantety RV Munkvold JD Miftahudin Mahmoud A Ma X Gustafson PJ Qi LL Echalier B Gill BS Matthews DE Lazo GR Chao S Anderson OD Edwards H Linkiewicz AM Dubcovsky J Akhunov ED Dvorak J Zhang D Nguyen HT Peng J Lapitan NLV Gonzalez-Hernandez JL Anderson JA Hossain K Kalavacharla V Kianian SF Choi DW Close TJ Dilbirligi M Gill KS Steber C Walker-Simmons MK McGuire PE Qualset CO Comparative DNA sequence analysis of wheat and rice genomes Genome Res 2003 13 1818 27 12902377
La Rota M Sorrells ME Comparative DNA sequence analysis of mapped wheat ESTs reveals the complexity of genome relationships between rice and wheat Funct Integr Genomics 2004 4 34 46 14740255 10.1007/s10142-003-0098-2
Yu J-K La Rota M Kantety RV Sorrells ME EST derived SSR markers for comparative mapping in wheat and rice Molecular Genetics and Genomics 2004 271 742 751 15197579 10.1007/s00438-004-1027-3
Quackenbush J Cho J Lee D Liang F Holt I Karamycheva S Parvizi B Pertea G Sultana R White J The TIGR Gene Indices: analysis of gene transcript sequences in highly sampled eukaryotic species Nucleic Acids Res 2001 29 159 64 11125077 10.1093/nar/29.1.159
Sasaki T Matsumoto T Yamamoto K Sakata K Baba T Katayose Y Wu J Niimura Y Cheng Z Nagamura Y Antonio BA Kanamori H Hosokawa S Masukawa M Arikawa K Chiden Y Hayashi M Okamoto M Ando T Aoki H Arita K Hamada M Harada C Hijishita S Honda M Ichikawa Y Idonuma A Iijima M Ikeda M Ikeno M Ito S Ito T Ito Y Iwabuchi A Kamiya K Karasawa W Katagiri S Kikuta A Kobayashi N Kono I Machita K Maehara T Mizuno H Mizubayashi T Mukai Y Nagasaki H Nakashima M Nakama Y Nakamichi Y Nakamura M Namiki N Negishi M Ohta I Ono N Saji S Sakai K Shibata M Shimokawa T Shomura A Song J Takazaki Y Terasawa K Tsuji K Waki K Yamagata H Yamane H Yoshiki S Yoshihara R Yukawa K Zhong H Iwama H Endo T Ito H Hahn JH Kim HI Eun MY Yano M Jiang J Gojobori T The genome sequence and structure of rice chromosome 1 Nature 2002 420 312 6 12447438 10.1038/nature01184
Benson DA Karsch-Mizrachi I Lipman DJ Ostell J Wheeler DL GenBank Nucleic Acids Res 2003 31 23 7 12519940 10.1093/nar/gkg057
Bioinformatics Group, Delaware Biotechnology Institute. University of Delaware
Zhang Z Schwartz S Wagner L Miller W A greedy algorithm for aligning DNA sequences J Comput Biol 2000 7 203 14 10890397 10.1089/10665270050081478
BioPerl.org
Rozen S Skaletsky HJ Krawetz S, Misener S Primer3 on the WWW for general users and for biologist programmers Bioinformatics Methods and Protocols: Methods in Molecular Biology 2000 Totowa, NJ: Humana Press 365 386
Primer3 core program
| 15720707 | PMC550658 | CC BY | 2021-01-04 16:39:33 | no | BMC Genomics. 2005 Feb 18; 6:23 | utf-8 | BMC Genomics | 2,005 | 10.1186/1471-2164-6-23 | oa_comm |
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-261570519210.1186/1471-2105-6-26Research ArticleComparison of seven methods for producing Affymetrix expression scores based on False Discovery Rates in disease profiling data Shedden Kerby [email protected] Wei [email protected] Rork [email protected] Debashis [email protected] James [email protected] Kathleen R [email protected] Thomas J [email protected] Stephen B [email protected] Eric R [email protected] Jeremy MG [email protected] Samir [email protected] Department of Statistics, University of Michigan, Ann Arbor, Michigan, USA2 Department of Biostatistics, University of Michigan, Ann Arbor, Michigan, USA3 Department of Pediatrics, University of Michigan, Ann Arbor, Michigan, USA4 Comprehensive Cancer Center, University of Michigan, Ann Arbor, Michigan, USA5 Department of Pathology, University of Michigan, Ann Arbor, Michigan, USA6 Department of Internal Medicine, University of Michigan, Ann Arbor, Michigan, USA2005 10 2 2005 6 26 26 26 10 2004 10 2 2005 Copyright © 2005 Shedden 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 critical step in processing oligonucleotide microarray data is combining the information in multiple probes to produce a single number that best captures the expression level of a RNA transcript. Several systematic studies comparing multiple methods for array processing have used tightly controlled calibration data sets as the basis for comparison. Here we compare performances for seven processing methods using two data sets originally collected for disease profiling studies. An emphasis is placed on understanding sensitivity for detecting differentially expressed genes in terms of two key statistical determinants: test statistic variability for non-differentially expressed genes, and test statistic size for truly differentially expressed genes.
Results
In the two data sets considered here, up to seven-fold variation across the processing methods was found in the number of genes detected at a given false discovery rate (FDR). The best performing methods called up to 90% of the same genes differentially expressed, had less variable test statistics under randomization, and had a greater number of large test statistics in the experimental data. Poor performance of one method was directly tied to a tendency to produce highly variable test statistic values under randomization. Based on an overall measure of performance, two of the seven methods (Dchip and a trimmed mean approach) are superior in the two data sets considered here. Two other methods (MAS5 and GCRMA-EB) are inferior, while results for the other three methods are mixed.
Conclusions
Choice of processing method has a major impact on differential expression analysis of microarray data. Previously reported performance analyses using tightly controlled calibration data sets are not highly consistent with results reported here using data from human tissue samples. Performance of array processing methods in disease profiling and other realistic biological studies should be given greater consideration when comparing Affymetrix processing methods.
==== Body
Background
Affymetrix microarrays are high throughput assays for measuring the expression levels of thousands of gene transcripts simultaneously. This type of microarray measures the expression of each transcript multiple times through a set of "probe pairs". Since the advent of the Affymetrix microarray, numerous methods have been proposed for producing numerical expression summaries for each transcript based on the probe pair data. Several systematic studies have appeared comparing a number of methods on a common basis (e.g. [1-5]). These studies rely heavily on calibration data sets derived from spike-in, dilution series, and mixture experiments for comparing methods. Our goal here was to carry out a comparative study of Affymetrix array processing methods using data sets from typical biological experiments seeking differentially expressed genes in human tissue samples.
The following seven methods are considered here: Dchip [10], GCRMA-EB and GCRMA-MLE [11], MAS5 [12], PDNN [13], RMA [2,3], and TM [[6,7], and ]. While not every popular method is included in our study, several highly distinctive and original approaches are studied. For example, Dchip was one of the first approaches to attempt to learn probe weights directly from the probe data, and RMA pioneered the approach of disregarding the control mismatch probes. PDNN uses physical modeling to determine probe weights, while the two GCRMA methods use GC content of the probe sequences to reduce variance in the mismatch (control) probe levels. The MAS5 method is the current default method provided by Affymetrix.
In addition to the six methods cited previously, we also include a method designated TM (trimmed mean). This is a simple method that has been used in a number of published investigations (e.g. [6,7]), but has not been considered in any previous systematic comparison of Affymetrix processing methods. To produce the probe-set summary score, the PM-MM differences are rank ordered, and the brightest 20% and dimmest 20% of values are deleted. The mean of the remaining values is used as the summary score. The scores for all probe-sets are then quantile normalized to a reference array using a piecewise linear spline with 100 knots.
An important feature of this study is the use of False Discovery Rate (FDR) to quantify the sensitivity of a processing method in terms of its ability to distinguish differentially expressed genes from genes having invariant expression. This is a highly relevant property, as differential expression analysis is the most common application of microarray data. A key advantage of using FDR to compare processing methods is that FDR values can be calculated accurately using real disease profiling data where the identities of differentially expressed genes are uncertain. In contrast, most previous systematic comparisons of array processing methods have focused on calibration data sets in which concentrations of certain genes were experimentally manipulated.
When it is highly likely that at least one gene is differentially expressed, false discovery rate may be defined as the expected ratio of the number of false positive calls to the total number of positive calls in a differential expression analysis between two groups of samples [8]. If the groups are biologically distinct, a sensitive processing method should result in many genes with low FDR. Thus to compare the performances of different array processing methods, we looked at two datasets in which a verified biological characteristic divided the samples into two classes, and compared the methods based on the number of genes having FDR smaller than various thresholds. For this to be a valid basis for comparison, the FDR values must be estimated with reasonable accuracy. Following other recent work (e.g. [9]), we used a permutation approach for this estimation, arguing that there is no reason that this approach favors or disfavors any particular array processing method.
A small FDR is due either to a small numerator, a large denominator, or both. The denominator of the FDR depends on the actual data distribution, so variation in this value may be due to factors such as accuracy in modeling the physical and chemical nature of probe binding. Variation in the FDR numerator, however, depends only on the distribution of values produced for randomized data, a purely statistical quantity reflecting the tendency of the method to incorrectly produce test statistic outliers. Our results suggest that both factors are important in determining sensitivity. The best methods produce many large test statistic values in the actual data, and also produce consistently small test statistic values for randomized data. Poor performance of one method can be directly explained by the tendency of the method to produce outlier expression values, leading to greater numbers of incorrectly large test statistics.
For overall comparison, we evaluated every pair of methods on the basis of whether the first method is expected to call at least one truly differentially expressed gene that is not also called by the second method. If this is not expected to occur, the second method is said to strongly outperform the first. Based on this comparison, two of the methods considered are clearly favored, two are inferior, and results for the other three methods are mixed.
Results
Sensitivity differences
Our primary basis for comparison is sensitivity – the number of genes detected at a given FDR0 level, where FDR0 is a rescaled FDR (see methods). Figure 1 shows the key sensitivity results, using both the t-test statistic and the rank-sum statistic to assess differential expression. Setting aside at first differences between the seven processing methods, we note two findings. First, in the colon data, analysis using the rank-sum statistic is substantially more sensitive than analysis using the t-test statistic. For the ovary data, where the sample sizes would not naturally suggest a robust analysis, there is no harm to sensitivity in using the rank-sum statistic. Second, the ovary curves are substantially higher overall than the colon curves. This may be due to a greater number of true positives in the ovary data, or it may be that the small sample size for the MSI group makes it difficult to attain high evidence levels for differential expression in the colon data. In any case, both data sets have many genes with small FDR values, supporting the biological relevance of the tumor groupings for both colon and ovary samples.
The more challenging colon set distinguishes the seven processing methods to a greater extent than the ovary set. Using FDR0 = 0.1 as a reference point, there is roughly 7-fold variation across the seven methods in the number of detected genes in colon data using t-test statistics, while for rank-sum statistics the range is roughly 2-fold. In the ovary data, the range is around 1.25-fold for both statistics. Also notable is that variation in sensitivity due to the choice of test statistic (t-test or rank-sum) is smaller than variation in sensitivity due to the seven processing methods.
No single method stands out as having the best or worst performance in every case. However some methods generally perform better than others. The Dchip and TM methods perform consistently well, while the GCRMA-EB and MAS5 methods consistently perform poorly. PDNN performs well on the ovary data, but poorly on the colon data, and results for the other methods are mixed.
Level of agreement between methods
Identities of probe sets falling below a given FDR0 threshold vary across the methods. Figure 2 summarizes this variation. The ratio of the number of probe sets falling below various FDR0 thresholds in k or more of the seven methods to the number of probe sets falling below the threshold for at least one method is plotted against the FDR0 threshold, for k = 3, 4, 5, 6, 7. In the ovary data there is a very high level of agreement in this measure. For the rank-sum analysis, almost 90% of called genes are called by at least four methods, and more than 70% of called genes are called by all seven methods. For the t-test analysis, the agreement is slightly higher yet. For the colon data, the methods are much more inconsistent. For the rank-sum analysis, three of the methods agree on up to 90% of genes, but all seven methods only agree on around 30% of genes. The t-test analysis is even worse, with only around 10% of genes common to all seven methods.
Turning to pairwise agreement, Table 1 shows the percentage of genes called by both members of a pair of methods out of the genes called by at least one of the two methods. In the ovary data, MAS5 shares the fewest calls with the other methods for both t-test and rank-sum analysis, while GCRMA-EB has relatively weak agreement for the t-test analysis. In the colon data, the GCRMA-EB method is highly inconsistent, with less than a quarter of calls in common with four of the six other methods for t-tests. A notable similarity is that the DChip and TM methods have at least 90% agreement in all analyses.
Complementing comparison of the statistical tests, we also compared the expression levels produced by the seven processing methods. For each pair of methods, and for each pair of samples within one of the two data sets, we calculated Pearson correlation coefficients of expression levels over all genes. These values were summarized by taking the median over all pairs of samples within a data set, shown in Table 2. Interestingly, methods calling similar genes as differentially expressed do not exhibit particularly strong correlation in expression levels. For example, TM and DChip perform very similarly in terms of which genes are identified as significant, but the pairwise correlation between expression levels for these two methods is less than the average. On the other hand, the TM and MAS5 methods are generally at the extreme high and low ends of the sensitivity scale respectively, but their expression levels are the most strongly correlated of any pair of methods.
Calibration
Variation in FDR across the seven methods is due to two factors – variation in the number of transcripts with large test statistics, and variation in the expected number of transcripts with large test statistics when there is no real differential expression. Here we investigate the second factor, which is driven by the tendency of each method to produce outlier expression values.
The numerator of the FDR aims to correct for variation in the number of false positives, so that a method claiming large numbers of differentially expressed genes is not considered superior unless it also produces relatively small numbers of false positives. This can be viewed as a calibration, in which for each method, the test statistic must reach a certain threshold in order that the proportion of false positives is no greater than a specified value.
Calibration results are summarized in Figure 3. For each method, the threshold test statistic value required to obtain FDR0 less than f was calculated, and plotted against f. For example, to achieve any FDR0 value between 0.05 and 0.1 in the colon rank-sum data, GCRMA-MLE requires the lowest test statistics, RMA requires a rank-sum statistic 0.15 units larger than that of GCRMA-MLE, and MAS5 requires a rank-sum statistic 0.3 units larger than that of GCRMA-MLE.
Figure 3 indicates that the methods differ substantially in terms of calibration. Notably, the ordering of the seven methods in Figures 1 and 3 are quite similar, suggesting that calibration plays a major role in determining sensitivity. Variation in thresholds among the seven processing methods is greater in the colon than the ovary data, particularly for the t-test analysis.
Since calibration depends only on randomized data, it should be possible to trace variation in thresholds across the processing methods to statistical properties of the expression levels. For example, if one method produces expression levels with heavier tails, it is easier to get a large t-test statistic value by chance, particularly for the colon data with small sample sizes. This would necessitate a higher threshold. To quantify this, let denote the log2 expression level of transcript i in sample j for method k, where k = 1, ..., 7 denotes the seven processing methods, and let
where is the pth quantile of , and med is the median value. This is an affine-invariant measure of the size of the right tail of the expression values. Values of Bk for the seven methods and two data sets are shown in Table 3. For reference, a Gaussian distribution has a B value of 3.74 when the sample size is as in the ovary data, and 3.56 when the sample size is as in the colon data. The GCRMA-EB method is seen to have a much greater propensity for producing extreme expression values, explaining its low sensitivity, poor agreement with other methods, and conservative calibration.
Variation in observed test statistics
In addition to calibration differences, FDR variation is also influenced by the observed test statistic values. This is summarized in Figure 4. For each method, and for a range of test statistic values t, the number of probe sets for which the observed test statistic value exceeds t was calculated and plotted against t. For example, in the colon rank-sum data, PDNN had the smallest test statistics, with MAS5 having around 500 more probe sets meeting a log test statistic threshold of 5 compared to PDNN. The Dchip and TM methods have over a thousand more probe sets meeting this threshold.
Variation in test statistic values across the methods is greater in colon than in ovary data, and generally tracks with sensitivity. However note that in the colon rank-sum data, Dchip has substantially larger test statistics than GCRMA-MLE, even while GCRMA-MLE has better sensitivity (Figure 1), due to its less stringent calibration (Figure 3).
Identification of genes with large fold changes
An interesting possibility that can not always be excluded is that the intergroup differences are so vast that nearly every gene is affected to a small degree. If this were the case, the FDR values for the t-test and Wilcoxon statistics would converge to zero for every gene as the number of samples grows, making FDR values difficult to interpret. To further investigate this issue, we repeated the analysis using t-statistics truncated to zero when the fold change is less than 1.5 as test statistics for FDR analysis. The corresponding FDR values remain bounded away from zero for genes having true fold change smaller than 1.5, while genes with true fold change exceeding 1.5 have FDR values converging to 0. Thus the statistic identifies a meaningful subset of genes even when all genes are differentially expressed to some degree.
Results for this analysis are shown in Figure 5. In the ovary data, the GCRMA-EB method performs best, with GCRMA-MLE, MAS5, and TM slightly inferior. Several of the methods, specifically PDNN, DChip, and RMA exhibit flat curves indicating that only a limited number of genes meet the 50% change criterion. In the colon data, GCRMA-MLE and TM are nearly tied as the best performers. Overall, variation in sensitivity across the methods exists at a similar level to that found in the t-test and Wilcoxon analyses. Only the GCRMA-MLE and TM methods give consistently good performances in the two data sets for this analysis.
Strong outperformance
Thus far we have focused on sensitivity as a criterion for comparing methods. However even if one method is less sensitive than another, if the overlap in the called gene sets is not too great then the less sensitive method may still contribute to our understanding of which genes are differentially expressed. Suppose two methods denoted 1 and 2 give N1 and N2 genes respectively at a given FDR level. Then nk = (1 - p0FDR0)·Nk estimates the expected number of truly differentially expressed genes called by method k. Now suppose that I is the number of genes called by both methods. Then nk - I is an estimated lower bound for the expected number of genes correctly called by method k but not by the other method. We will say that method 1 strongly outperforms method 2 if n1 - I ≥ 0 but n2 - I < 0. This means that in terms of differential expression, method 2 is not expected to contribute any true positives that were not called by method 1.
Table 4 summarizes the results of this analysis using p0 = 1 and FDR0 = 0.05, showing the number of times that each method was strongly outperformed by other methods in our study. This analysis clearly favors the TM and Dchip methods, while the MAS5 and GCRMA-EB methods are nearly always found to be strongly outperformed by the other 5 methods. These results are not sensitive to choices of p0 between 0.5 and 1 (more than half of values are constant within this range and non-constant values do not vary by more than 1).
Discussion
Impact of processing method choice
The choice of processing method for Affymetrix array data evidently has a major impact on the ability to confidently report the results of differential expression analysis. The effect is greater, for example, than the choice of using a robust or a non-robust analysis, even in the colon data where robust analysis results in substantial improvements. Differences among processing methods are much greater in the more challenging colon data set compared to the ovary data, yet it should be noted that the sample sizes in the colon data are not atypical in real investigations.
While results from two data sets can never conclusively determine the optimal method, it is notable that across both data sets, using both t-statistic and rank-sum analyses, there is a high degree of similarity in the rank ordering of the methods from the best to the worst performer. The trimmed mean (TM) and Dchip methods consistently perform as well or better than any of the other methods. A possible explanation for this is that the weights used by the Dchip may tend to downweight the least and greatest PM-MM differences, just as the TM method excludes these differences.
Interpretation of FDR comparisons
When comparing array processing methods using experimental data in which the identities of differentially expressed genes are unknown, great care must be taken to ensure that apparent differences in sensitivity are not due to other factors. One critical point is that the null distribution providing the expected number of false positives at a given test statistic threshold (the numerator of the FDR) must fairly reflect the statistical behavior of null genes. Permutation approaches have been extensively used to produce empirical p-values (e.g. [14]) and were used by Efron et al. [9] to estimate FDR values. Although permutation approaches are known to be slightly biased for estimating the FDR, the size of the bias (e.g. as shown in figure 5 of Efron et al. [9]) can not explain the magnitude of differences found here. In addition, for a comparative analysis, as carried out here, it is more crucial that the biases be relatively constant across the methods. However, since permutation approaches may not be highly accurate when the sample size is small, it is important to check performance on multiple data sets before conclusions about performance are drawn.
While we have focused on FDR as the basis of comparison, the pursuit of small FDR values is not the only desirable operating characteristic of an array processing method, and other reports have also emphasized the accuracy of estimating the precise size of concentration differences. However to the extent that most actual studies seek to find differential expression between groups, the use of small FDR values seems more instrumental as the basis for judging methods.
Variation due to choice of test statistic
Although our primary aim was to investigate variation in sensitivity due to the seven processing methods, all analysis was carried out independently for two test statistics. The t-statistic is widely used in practice, but is well-known to be sensitive to outliers, particularly when the sample size is small. We found that certain processing methods, particularly EB-GCRMA, had a tendency to produce outlier expression values in the colon data set. Thus the combination of using the EB-GCRMA method with t-statistics in the colon data led to particularly poor performance.
Variation due to log transform and array normalization
In practice, the approach used for array normalization and for forming log-transformed expression values may be equally or more influential than the method used for producing probe set summaries [15]. In this study, we used implementations of the seven processing methods as prepared by their developers, and thus array normalization and and log-transforms were applied in a method-specific fashion. This provides a comparative analysis of the various methods as they are used in practice, which is most directly relevant since few investigators will override the default normalization and log-transform methods provided by the developers of each method.
Nevertheless it remains of interest whether these routine processing steps are the determining factor of performance. In a future study it will be important to investigate this question further by modifying the implementations of the processing methods so that uniform log transforms and array normalizations are applied.
Comparison of methods using data from disease profiling data sets
A key point that we advocate in this work is that false discovery rates in actual disease profiling data constitute a valuable complement to benchmarking results obtained from spike-in, dilution series, and mixture experiments (e.g. [4,5]). The primary obstacle that must be overcome is that proper null sampling distributions are essential to ensure that the methods are compared on a common basis. Since numerous data sets covering a wide range of Affymetrix platforms are available, to the extent that multiple data sets are in agreement about relative performances it is unlikely that the randomization procedure used to calculate FDR values is systematically biased against a particular method.
In spite of the statistical challenges in using disease profiling data for benchmarking, we argue that these data sets also offer some unique advantages. Calibration data sets are relatively few in number and are not available for all platforms. Newer platforms in particular are under-represented. Therefore overtraining to the available calibration data through manipulation of the many tuning parameters in the more complicated processing methods is an unavoidable concern. In addition, the calibration data sets likely do not represent the same degree of challenge as disease profiling data in that reproducibility of fold changes for affected and unaffected genes is quite high compared to data from, say, human tissues where a large number of uncontrolled sources of variability are present.
Conclusions
Performances of multiple array processing methods on disease profiling data sets vary widely across the seven methods studied here, but results are generally consistent between the two data sets studied. Results of our analysis generally do not parallel results obtained using calibration data sets [4,5], suggesting that such comparisons may not completely capture the most relevant aspects of performance.
A major determinant of sensitivity is test statistic variability for randomized data. Such variability will affect false discovery rates as well as empirical p-values, which are an often-used alternative approach for identifying differentially expressed genes (e.g. [14]). Therefore it will be important in future work to seek a better understanding of statistical sampling properties of array processing methods. A particular focus should be the way that sampling variance in probe masking and probe weighting is controlled. Methods seeking to incorporate mechanistic information about the dynamics of probe binding, such as the two GCRMA methods and PDNN, should in principal outperform more generic approaches such as the TM method. Our results, particularly in the colon data, suggest that in medium-sized data sets this potential is not yet reached.
In this comparative analysis we did not seek to draw definitive conclusions about the "best" or "worst" methods. Such conclusions may be made after investigating a greater number of data sets, including disease profiling data, data from controlled experiments, and calibration data. Moreover, it may be that the correct choice of method may depend on the scientific question being asked. The key message of this work is that the wide range of data sets collected in actual scientific investigations may be used for comparison of processing methods, and that in at least the two data sets considered here, similar results were obtained in the rank ordering of the methods.
Methods
Data sets
We used two data sets – one consisting of 79 ovary tumors and the other consisting of 47 colon tumors. Both sets were generated at the University of Michigan using Affymetrix HG_U133A arrays, which consist of 22283 probe-sets, each of which is designed to assay a RNA transcript. Each probe-set consists of a set of (typically 11) probe-pairs, with each probe-pair comprising a "perfect match" (PM) probe which is a 25-base oligonucleotide complementary to the transcript, and a "mismatch" (MM) probe that is identical to the PM sequence except for alteration of the central base. The MM probe is intended as a control for nonspecific hybridization, so that the difference PM-MM measures only specific binding. However not all processing methods use the MM data in this way.
For differential expression analysis, the 79 ovary samples were partitioned according to histological class into 38 endometrioid and 41 serous samples. The 47 colon samples were partitioned into 40 microsatellite stable (MSS) samples and 7 microsatellite instable samples (MSI). In both data sets, the partition is based on an independently measured biological characteristic, so there almost certainly are differentially expressed genes to be found. However in neither case are the two classes highly distinct, and numerous other sources of biological variation are undoubtedly present in the data.
Normalization across arrays
Array normalization refers to an adjustment of data distributions within each array in order to make the arrays more comparable. Each array processing method has been coupled with a normalization procedure by its developers (see references). We followed these method-specific normalization practices in our analysis. All methods other than MAS5 use some form of quantile normalization.
Log transform and truncation
All analysis was based on log-transformed data. Log-transformed values, including truncations where needed, were calculated in the manner recommended by the developers of each method (see references).
Methodology of comparison
We compared the seven methods based on their sensitivity in detecting differential expression at a fixed false discovery rate (FDR). For each method, two different two-sample test statistics were calculated for each gene – the standard two-sample t-statistic, and the Wilcoxon rank-sum statistic (equivalent to the Mann-Whitney statistic). The t-test statistic T is always analyzed as |T|, and the rank-sum statistic R is standardized as , where m0, m1 are the numbers of samples in the two classes, and m = m0 + m1 is the total number of samples.
Our FDR approach closely follows the "global estimate" of Efron et al. ([9] equation 5.9). For a given test statistic threshold t, the FDR was estimated as follows. Randomized data sets were constructed by randomly reassigning the class identifiers to the samples. The average number of transcripts with test statistic value exceeding t was calculated over 1000 randomized data sets. This number was divided by the number of actual transcripts with test statistic value exceeding t to produce a value that we denote FDR0. In practice the value of FDR0 should be scaled by the proportion p0 of non-differentially expressed genes, giving FDR = p0FDR0. Although various estimates of p0 exist, we elected to ignore this factor since it is constant across the methods for a given data set, and any estimate of p0 would add an additional source of uncertainty to our results. Thus it should be noted that the reported FDR0 values, while comparable across methods, are somewhat larger than the usual estimates. Since p0 would generally be greater than 1/2, the bias is likely less than a factor of 2.
Authors' contributions
KS, JMGT, RK, and DG participated in all phases of design and analysis. WC and JM performed the data analysis. KRC, TJG, SBG, ERF, and SH assisted in study design and supervised data collection. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Sensitivity results for colon and ovary data. Top row: number of significant probe sets at a range of FDR0 values using the t-test statistic. Bottom row: number of significant probe sets at a range of FDR0 values using the rank-sum statistic. The left column shows the results for colon data and the right column shows the results for ovary data.
Figure 2 FDR agreement between methods. The ratio of the number of probe sets with FDR0 value below a given threshold in k or more of the seven methods to the number of probe sets with FDR0 value below the threshold in at least one method was calculated for k = 3, 4, 5, 6, 7, and plotted against the FDR0 threshold. Results are shown for the colon data (left column), the ovary data (right column), and for the t-test statistic (top row), and the rank-sum statistic (bottom row).
Figure 3 Calibration results for ovary and colon data. The threshold test statistic required to obtain a given FDR0 for each method is plotted against the FDR0 value. Results are shown for the colon data (left column), the ovary data (right column), and for the t-test statistic (top row), and the rank-sum statistic (bottom row).
Figure 4 Test statistics for ovary and colon data. For each of the seven processing methods, the number of probe sets exceeding a test statistic threshold t was calculated and plotted against log2 t. Results are shown for the colon data (left column), the ovary data (right column), and for the t-test statistic (top row), and the rank-sum statistic (bottom row).
Figure 5 Sensitivity for detecting genes with at least 50% change in expression magnitude. The number of significant probe sets at a range of FDR0 values is shown for analysis in which the test statistic is the t-statistic truncated to zero when the fold change is less than 50%.
Table 1 Pairwise agreement between methods. For each pair among the seven processing methods, the ratio of the number of probe sets with FDR0 < 0.05 in both methods to the number of probe sets with FDR0 < 0.05 in either method was calculated. Results are displayed as percentages.
Colon Ovary
t-test EB 9 77
MAS5 19 45 68 88
MLE 15 60 60 94 81 72
PDNN 68 14 28 23 93 83 73 98
RMA 58 16 32 26 86 92 83 74 98 100
DCHIP 90 8 17 14 60 52 95 81 71 99 97 97
Rank-Sum EB 39 90
MAS5 38 88 73 82
MLE 84 32 32 99 89 72
PDNN 45 86 84 38 98 91 74 97
RMA 71 54 53 60 63 88 98 83 87 89
DCHIP 94 41 40 79 48 75 99 91 74 98 99 89
TM EB MAS5 MLE PDNN RMA TM EB MAS5 MLE PDNN RMA
Table 2 Median pairwise correlations over all sample pairs. For each pair of processing methods, expression levels were computed for each sample in the colon and ovary data sets. Results shown are the median Pearson correlation coefficients over all sample pairs between log-scale expression levels for all genes.
Colon Ovary
GCRMA-EB 0.86 0.86
MAS5 0.94 0.84 0.95 0.84
GCRMA-MLE 0.88 0.80 0.91 0.89 0.80 0.92
PDNN 0.82 0.77 0.74 0.71 0.82 0.78 0.76 0.73
RMA 0.88 0.81 0.85 0.82 0.85 0.89 0.80 0.86 0.85 0.85
DChip 0.81 0.73 0.76 0.75 0.81 0.89 0.83 0.74 0.79 0.77 0.83 0.90
TM GCRMA-EB MAS5 GCRMA-MLE PDNN RMA TM GCRMA-EB MAS5 GCRMA-MLE PDNN RMA
Table 3 Tendencies of the processing methods to produce outlier expression values. Values of the B statistic (see text) are shown for the seven processing methods and two data sets.
Colon Ovary
TM 3.71 4.03
GCRMA-EB 6.75 6.21
MAS5 3.44 3.74
GCRMA-MLE 5.09 5.43
PDNN 3.49 4.84
RMA 4.34 4.90
Dchip 3.82 4.39
Table 4 Strong outperformance of each method. For each of the seven processing methods, and for each of the four analyses, the number out of the other 6 processing methods that strongly outperform the given method at FDR = 0.05 was determined.
Colon Ovary
t RS t RS
TM 1 1 0 0
GCRMA-EB 5 5 5 4
MAS5 4 5 6 6
GCRMA-MLE 4 0 1 0
PDNN 2 4 1 0
RMA 3 3 1 4
DChip 0 2 0 0
==== Refs
Bolstad B Irizarry R Astrand M Speed T A comparison of normalization methods for high density oligonucleotide array data based on variance and bias Bioinformatics 2003 19 185 193 12538238 10.1093/bioinformatics/19.2.185
Irizarry R Bolstad B Collin F Cope L Hobbs B Speed T Summaries of Affymetrix GeneChip probe level data Nucleic Acids Research 2003 31 e15 12582260 10.1093/nar/gng015
Irizarry R Hobbs B Collin F Beazer-Barclay Y Antonellis K Scherf U Speed T Exploration, Normalization, and Summaries of High-Density Oligonucleotide Array Probe Level Data Biostatistics 2003 4 249 264 12925520 10.1093/biostatistics/4.2.249
Rajagapolan D A comparison of statistical methods for analysis of high density oligonucleotide array data Bioinformatics 2003 19 1469 76 12912826 10.1093/bioinformatics/btg202
Cope L Irizarry R Jaffee H Wu Z Speed T A benchmark for Affymetrix GeneChip expression measures Bioinformatics 2004 20 323 331 14960458 10.1093/bioinformatics/btg410
Giordano T Shedden K Schwartz D Kuick R Taylor J Lee N Misek D Greenson J Kardia S Beer D Rennert G Cho K Gruber S Fearon E Hanash S Organ-specific molecular classification of primary lung, colon, and ovarian adenocarcinomas using gene expression profiles Am J Pathol 2001 159 1231 8 11583950
Rickman D Bobek M Misek D Kuick R Blaivas M Kurnit D Taylor J Hanash S Distinctive molecular profiles of high-grade and low-grade gliomas based on oligonucleotide microarray analysis Cancer Research 2001 61 6885 91 11559565
Storey J A direct approach to false discovery rates J R Statist Soc B 2002 64
Efron B Tibshirani R Storey J Tusher V Empirical Bayes analysis of a microarray experiment Journal of the American Statistical Association 2001 96
Li C Wong W Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection Proc Natl Acad Sci USA 2001 98 31 6 11134512 10.1073/pnas.011404098
Wu Z Irizarry R Gentleman R Murillo F Spencer F A Model Based Background Adjustment for Oligonucleotide Expression Arrays Technical Report, John Hopkins University, Department of Biostatistics Working Papers, Baltimore, MD 2003
Hubbell E Liu W Mei R Robust estimators for expression analysis Bioinformatics 2002 18 1585 92 12490442 10.1093/bioinformatics/18.12.1585
Zhang L Miles M Aldape K A model of molecular interactions on short oligonucleotide microarrays Nat Biotechnol 2003 21 818 21 12794640 10.1038/nbt836
Tusher V Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci USA 2001 98 5116 21 11309499 10.1073/pnas.091062498
Hoffmann R Seidl T Dugas M Profound effect of normalization on detection of differentially expressed genes in oligonucleotide microarray data analysis Genome Biol 2002 3 RESEARCH0033 12184807 12184807
| 15705192 | PMC550659 | CC BY | 2021-01-04 16:02:51 | no | BMC Bioinformatics. 2005 Feb 10; 6:26 | utf-8 | BMC Bioinformatics | 2,005 | 10.1186/1471-2105-6-26 | oa_comm |
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-331572071910.1186/1471-2105-6-33Research ArticlePSSM-based prediction of DNA binding sites in proteins Ahmad Shandar [email protected] Akinori [email protected] Department of Bioinformatics and Bioscience, Kyushu Institute of Technology, Iizuka 820 8502, Fukuoka, Japan2 Department of Biosciences, Jamia Millia Islamia University, New Delhi-110025, India2005 19 2 2005 6 33 33 18 11 2004 19 2 2005 Copyright © 2005 Ahmad and Sarai; 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
Detection of DNA-binding sites in proteins is of enormous interest for technologies targeting gene regulation and manipulation. We have previously shown that a residue and its sequence neighbor information can be used to predict DNA-binding candidates in a protein sequence. This sequence-based prediction method is applicable even if no sequence homology with a previously known DNA-binding protein is observed. Here we implement a neural network based algorithm to utilize evolutionary information of amino acid sequences in terms of their position specific scoring matrices (PSSMs) for a better prediction of DNA-binding sites.
Results
An average of sensitivity and specificity using PSSMs is up to 8.7% better than the prediction with sequence information only. Much smaller data sets could be used to generate PSSM with minimal loss of prediction accuracy.
Conclusion
One problem in using PSSM-derived prediction is obtaining lengthy and time-consuming alignments against large sequence databases. In order to speed up the process of generating PSSMs, we tried to use different reference data sets (sequence space) against which a target protein is scanned for PSI-BLAST iterations. We find that a very small set of proteins can actually be used as such a reference data without losing much of the prediction value. This makes the process of generating PSSMs very rapid and even amenable to be used at a genome level. A web server has been developed to provide these predictions of DNA-binding sites for any new protein from its amino acid sequence.
Availability
Online predictions based on this method are available at
==== Body
Background
There has been a growing interest in the prediction of DNA-binding sites in proteins which play crucial roles in gene regulation [1-4]. We have previously developed a method of predicting DNA-binding sites of proteins from the sequence information [5]. We reported development of a neural network and corresponding web server to predict amino acid residues which are likely to bind DNA. The only input to the neural network in this algorithm was the identity of the amino acid residue and its two sequence neighbors on C- and N- terminals. We also developed a method to identify DNA-binding proteins using electrical moments from structural information of proteins [6]. On the other hand, several investigators have reported that the use of evolutionary information in sequence-based predictions of secondary structure and solvent accessibility can improve the prediction capacity of a neural network [7-10].
Here we report the use of such evolutionary information in improving the prediction of DNA-binding sites of proteins. We note that one of the major problems in applying evolutionary information by way of position specific scoring matrices (PSSMs) for sequence based prediction is that such matrices are generated over large data sets and take a long time to complete. Thus large scale predictions remain inaccessible to moderately capable computers. This is a serious limitation in the portability of neural network based predictions using PSSMs [8]. In this work, we report that evolutionary profiles or PSSMs against much smaller representative reference data sets may be utilized to achieve almost the same levels of prediction as would be obtained from alignments with large sequence data sets representing entire available sequence space. We have used four different reference data sets of PSSMs for 62 representative protein sequences. These are (1) PDNA-RDN: a data set of protein sequences from all Protein-DNA complexes from the PDB, (2) PDNA-NR90: a non-redundant data set compiled from PDNA-RDN, (3) PDB-ALL: a data set of all amino acid sequences from PDB and (4) NCBI-NR: a non-redundant data set of all protein sequences taken from sequence and structure databases and compiled by NCBI (see Methods). We find that the net prediction (an average of sensitivity and specificity) of the best of these systems (using PIR sequence data as reference) improves to 67.1% from the value of 58.4% reported earlier for a sequence-only prediction. We also report that a small reference data set of 375 sequences (PDNA-NR90) can give a 64.6% net prediction – just 2.5% poorer than the best- while reducing the PSSM calculation time from more than two hours (against NCBI-NR) to just about one minute. A better compromise could be the use of PDNA-RDN data for which 65.2% net prediction #150; 1.9% less than the best- was obtained, while about 2 and a half minutes are taken to generate their PSSMs. It is also reported that the presence of redundancy is helpful in improving the prediction whereas presence of data not relevant for DNA-binding may in some cases reduce predictive performance.
Results and discussion
Position Specific Iterative BLAST (PSI BLAST) is a strong measure of residue conservation in a given location. In the absence of any alignments, PSI BLAST simply returns a 20-dimensional vector representing probabilities of conservation against mutations to 20 different amino acids including itself. A matrix consisting of such vector representations for all residues in a given sequence is called Position Specific Scoring Matrix or PSSM. When a residue is conserved through cycles of PSI BLAST, it is likely to be due to a purpose i.e. biological function. It has been established by several authors cited in the introduction that the prediction of structural properties is significantly enhanced by the use of PSSMs compared to predictions based on unique representations of amino acid sequence and its environment. Protein structure universe is vast and a prediction of structural properties should span the entire range of this diversity. However, the question of predicting DNA-binding sites is much narrower and hence the significance of conservation of residues at specific locations may be limited to a subset of this protein space. Such reduction in the protein search space or the reference data sets against which PSSM-based predictions should be attempted is desired for a rapid prediction of binding sites as well as portability of prediction methods. Compact reference data size can not only answer these questions of speed and portability but also try to minimize noise in information contents and improve prediction quality.
Table 1 shows the results of DNA-binding site prediction using different sets of PSSMs as the neural network inputs. The best net prediction results were 67.1% which is 8.7% better than the predictions with sequence information only. These results were obtained for PSSMs against PIR sequence data. An even larger NCBI-NR data set showed a slightly smaller (66.7%) net prediction. The fact that NCBI-NR reference data sets produce somewhat worse results than PIR sequence suggests that the redundancy present in the PIR sequence data could be the factor responsible for giving better PSSMs than those of a non-redundant NCBI sequence data. Thus an overall redundancy in the data turns out to be helpful in improving the prediction of binding sites. The question is how rapidly the prediction ability will fall if we reduce the redundancy even further, replacing the larger data sets with smaller ones until a small representative data set is left. This question is partly answered by first using a sequence database of the entire protein data bank (PDB-ALL), which gives an accuracy of 64.7% (about 2.4% poorer than the best). Further reducing the data set to protein sequences from only the Protein-DNA complexes surprisingly increases the net prediction to 65.2%. We suggest that the increase in net prediction on the PDNA-RDN over the entire PDB is caused by the fact that PDNA-RDN contains all the data from PDB which is relevant for the DNA-binding. However, an additional data in the PDB-ALL represents conservation scores in regions not involved in DNA-binding and hence lead to a somewhat lower net prediction. Going further down from a redundant (PDNA-RDN) to a non-redundant (PDNA-NR90) sequence data of Protein-DNA complexes, we observe a 0.6% fall in net prediction- just about the same we observed from PIR to NCBI-NR. We attribute this fall in net prediction to the reduction in the redundancy in the sequence data sets, which is concluded to be useful in better prediction of DNA-binding sites.
In terms of CPU time, it may be noted that the time taken by 62 protein sequences used here is about one hour for the best (PIR) data sets. These times are prohibitively large for making predictions at a genomic scale or for providing rapid web services. A compromise could be obtained by using PDNA-RDN instead, which reduces the CPU time by a factor more than 8. The loss of net prediction for this compromise is about 1.9%, which is still 6.8% better than the predictions obtained from sequence information only. PSSMs against this data set for a typical protein of 500 residues can be generated in about 1 s, making it possible to run large scale predictions. A smaller size of reference data and high speed of PSSMs also make this method portable and light weight with a strong predictive ability.
Binary decision function of the neural network (see Methods) assigns a value of zero (not binding) or 1 (binding) based on a threshold on the real value output received at the output node. Most of the accuracy scores presented here have been obtained by using 0.5 as the cutoff (mid point of the transfer function range). By changing this threshold from 0.5 to higher and lower values, the balance between sensitivity and specificity can be adjusted. In our online prediction we also present the scores obtained for a ROC analysis of such adjustments (Figure 2). ROC for only one reference data set has been shown here as most other graphs show a similar behavior.
Online predictions
We have provided online predictions based on the above method at our web site [16]. The raw probability scores, their annotations at different sensitivity thresholds, and a reference scale for expected sensitivity and specificity have been provided. In addition, results of sequence alignments obtained after PSI BLAST iterations against a reference data (PDNA-RDN) are also provided. This allows us to have a complete picture of similarity of a given sequence with known DNA-binding proteins and predictions based on neural network using alignment profiles in the form of PSSM. The only input to this neural network is the amino acid sequence of the protein. The web server will automatically generate PSSMs of the given sequence against a reference data and use them as the input to a neural network, trained for predictions of 62 DNA-binding proteins.
Conclusion
A PSSM-based neural network method for predicting DNA-binding sites in proteins has been developed. PSSMs were developed against different data sets and it was observed that significant computer time can be saved by replacing the reference data sets with much smaller reference data sets without loss of much prediction ability. Redundant reference data sets show a better prediction than the non-redundant data sets. A web server was developed to provide prediction of DNA-binding sites based on this method. In addition, the web server provides BLAST alignments against a reference data set of known DNA-binding proteins.
Methods
Data sets
PDNA-62
This is the (non redundant) target data set of 62 DNA-binding proteins from Protein Data Bank (PDB) [11]. The same data set has been used in our related studies [5,12].
PDNA-RDN
This is a new data set, developed for this work. We have selected all Protein-DNA complexes from PDB and separated their chains. 1386 protein chains were obtained in this way. FASTA formatted sequences were subsequently formatted using formatdb program of the BLAST package [13].
PDNA-NR90
The data set (PDNA-RDN), obtained from the procedure mentioned above was filtered to remove redundancy at 90% sequence identity level by using sequence clustering program BLASTCLUST [13]. Resulting data set now contains 375 sequences which are formatted for use as a reference data set using formatdb. This data set is called PDNA-NR90.
Other data sets
PDB-ALL (47,189 sequences) is a data set of all protein sequences obtained from NCBI. PIR is the sequence data set (283,177 sequences) of Protein Information Resource at Georgetown University [14]. NCBI-NR is a non-redundant data set of all protein sequences compiled from GeneBank, PIR, SwissProt, PDB and other resources by NCBI [17].
Generation of PSSMs
Target sequences are scanned against the reference data sets to compile a set of alignment profiles or position specific scoring matrices (PSSMs) using Position Specific Iterative BLAST (PSI BLAST) program [15]. Three cycles of PSI-BLAST were run for each protein and the scores were saved as profile matrices (PSSMs).
Neural network
Neural network inputs
Conservation scores in 20 amino acid positions for every residues form 20 columns (column 3 onwards) of corresponding row in a PSI-BLAST PSSM. For every residue, we make a binary or real-value (interpreted as probability) prediction of that residue being a binding site or not. Input for every prediction is the PSSM score on the row corresponding to this target residue and two more rows on either side, totaling 20 × 5 = 100 inputs (Figure 1).
Network architecture and transfer function
We use a neural network with one hidden layer (two nodes) in addition to the input layer described above and a single node output layer. Large number of units in the hidden layer and additional layers were not tried because the data size does not justify an unreasonably large neural network. Network signal is transferred to subsequent layers by an algebraic summation of inputs from the previous layer. Total signal in the last unit is transformed to a real output by a binary decision function much in the same way as in our previous work except that the input to the network is now replaced by PSSM scores rather than 20 bit binary coding [5].
Training and validation
A six-fold cross-validation has been used in this work. Out of 62 proteins, 10 were removed at one time and the remaining 52 were trained until the accuracy on the left-out 10 also improved. Six random sets are created in this way and the figures in Table 1 report the averages on all six runs of each set of 10 proteins.
Training error function and measure of prediction quality
Data imbalance in the two binary categories for this neural network makes the choice of error function particularly important. We have used an accuracy score called Net Prediction, which is the average of sensitivity and specificity values defined below. Neural network learns to maximize this accuracy score rather than minimizing an error function.
Sensitivity is defined as the number of correct prediction in the binding category relative to total number of such items in the original data and specificity is the number of correctly rejected residues in this category relative to the total number of non-binding residues in the original data.
Sensitivity (S1) = 100 * TP/(TP+FN) (1)
Specificity (S2) = 100 * TN/(TN+FP) (2)
where TP: True Positive; TN: True Negative; FP: False Positive; FN: False Negative
Relative number of true positive (TP) values in the prediction was termed as accuracy in our previous work. We have avoided using that term here, as we prefer a more operational definition of accuracy measures here. The imbalance of sensitivity (S1) and specificity (S2) is taken care of by comparing the Net Prediction of the models which gives a better comparison when S1 and S2 vary from one sample to the other. Thus,
Net Prediction (NP) = (S1+S2)/2 (3)
List of abbreviations
PSSM: Position Specific Substitution Matrix
PSI BLAST: Position Specific Iterative Basic Local Alignment Search Tool
Authors' contributions
SA conceived and implemented this project. AS contributed in manuscript preparation, results analysis and discussions.
Acknowledgements
Corresponding author would like to thank Advanced Technology Institute Inc. for partially supporting this research. This work was supported in part by Grants-in-Aid for Scientific Research 16014219 and 16041235 to A.S.
Figures and Tables
Figure 1 Rows of Position Specific Scoring Matrices selected for neural network input: Network inputs consist of the PSSM of the target residue and its two neighboring residues on C- and N-terminals. Each residue is thereby represented by a 20 dimensional vector with integer values. These values represent (logarithmic) effective frequencies of occurrence at respective positions in a multiple alignment. Neural network input layer is therefore made of 20 × 5 = 100 units. Two units in the only hidden layer and one unit in the output layer add up to a total of 202 neural units to be trained in the fully connected neural network.
Figure 2 ROC analysis of binding site prediction using PSSMs against PDNA-RDN reference data set, compared with results obtained from sequence based predictions. The sensitivity of the prediction could be adjusted by changing the threshold on predicted probabilities, to annotate that residue to be DNA-binding or otherwise. As may be noted the area under the PSSM based prediction curve is significantly greater than that obtained from sequence based predictions. In addition, sensitivity versus specificity values also seems to be difficult to manipulate in case of sequence based predictions as points on the curve are very closely spaced. PDNA-RDN curve also shows the levels of prediction scores expected on our web-based predictions.
Table 1 Prediction results for binding sites in 62 Proteins with different data sets used for generating PSSM.
Reference Data Overall Correct predictions (%) Sensitivity (S1) % Specificity (S2) % Net Prediction (S1+S2)/2 %
Sequence only (No PSSM) 73.6 40.6 76.2 58.4(2.5)
PDNA-NR90 375 sequences 63.8 65.9 63.4 64.6(2.1)
PDNA-RDN 1386 sequences 64.0 67.1 63.3 65.2(2.1)
NCBI-NR 1,547,365 sequences 66.7 69.5 63.9 66.7(1.4)
PDB-ALL 47,179 sequences 62.6 65.6 61.8 64.7(1.8)
PIR 283,177 sequences 66.4 68.2 66.0 67.1(2.7)
PDNA refers to sequences from Protein-DNA complexes in the Protein Data Bank; NR90 means non-redundant at 90% sequence identity; RDN means data is redundant because similar proteins have not been removed. Values in the brackets show the standard deviation in values obtained from six cross-validation sets. Note that the sensitivity and specificity values shown in this table only refer to those values which sum up to give the best net prediction. These two scores can be mutually adjusted by changing cutoff threshold as described in the text and hence comparison between the data sets should only be made for the net prediction value (the last column) which is the score optimized during training.
==== Refs
Gutfreund MY Margalit H Quantitative parameters for amino acid-base interaction: implications for prediction of protein-DNA binding sites Nucleic Acids Res 1998 26 2306 2312 9580679 10.1093/nar/26.10.2306
Pabo CO Nekludova L Geometric analysis and comparison of protein-DNA interfaces: why is there no simple code for recognition? J Mol Biol 2000 301 597 624 10966773 10.1006/jmbi.2000.3918
Luscombe NM Thornton JM Protein-DNA Interactions: Amino Acid Conservation and the Effects of Mutations on Binding Specificity J Mol Biol 2002 320 991 1009 12126620 10.1016/S0022-2836(02)00571-5
Stawiski EW Gregoret LM Mandel-Gutfreund Y Annotating Nucleic Acid binding function based on protein structure J Mol Biol 2003 326 1065 1079 12589754 10.1016/S0022-2836(03)00031-7
Ahmad S Gromiha MM Sarai A Analysis and Prediction of DNA-binding proteins and their binding residues based on Composition, Sequence and Structural Information Bioinformatics 2004 20 477 486 14990443 10.1093/bioinformatics/btg432
Ahmad S Sarai A Moments based prediction of DNA-binding proteins J Mol Biol 2004 341 65 71 15312763 10.1016/j.jmb.2004.05.058
Rost B Sander C Improved prediction of protein secondary structure by using sequence profiles and neural networks Proc Natl Acad Sci USA 1993 90 7558 7562 8356056
Jones DT Protein secondary structure prediction based on position specific scoring matrices J Mol Biol 1999 292 195 202 10493868 10.1006/jmbi.1999.3091
Cuff JA Barton GJ Application of multiple sequence alignment profiles to improve protein secondary structure prediction Proteins 2000 40 502 11 10861942 10.1002/1097-0134(20000815)40:3<502::AID-PROT170>3.0.CO;2-Q
Adamczak R Porollo A Meller J Accurate prediction of solvent accessibility using neural networks based regression Proteins 2004 56 753 767 15281128 10.1002/prot.20176
Berman HM Westbrook J Feng Z Gilliland G Bhat TN Weissig H Shindyalov IN Bourne PE The Protein Data Bank Nucleic Acids Res 2000 28 235 242 10592235 10.1093/nar/28.1.235
Selvaraj S Kono H Sarai A Specificity of Protein-DNA RecognitionRevealed by Structure-based Potentials: Symmetric/Asymmetric and Cognate/Non-cognate Binding J Mol Biol 2002 322 907 915 12367517 10.1016/S0022-2836(02)00846-X
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
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
Altschul SF Madden TL Schäffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
DBS-PSSM: Prediction of DNA-binding sites by PSSM and sequence homology
NCBI BLAST databases download web site:
| 15720719 | PMC550660 | CC BY | 2021-01-04 16:02:52 | no | BMC Bioinformatics. 2005 Feb 19; 6:33 | utf-8 | BMC Bioinformatics | 2,005 | 10.1186/1471-2105-6-33 | oa_comm |
==== Front
BMC Dev BiolBMC Developmental Biology1471-213XBioMed Central London 1471-213X-5-31571004310.1186/1471-213X-5-3Research ArticleSea urchin vault structure, composition, and differential localization during development Stewart Phoebe L [email protected] Miriam [email protected] Jennifer [email protected] Carrie [email protected] Anthony J [email protected] James A [email protected] Kathy A [email protected] Department of Molecular Physiology and Biophysics, Vanderbilt University Medical Center, Nashville, TN USA2 Department of Molecular and Medical Pharmacology, Crump Institute for Molecular Imaging, David Geffen School of Medicine at UCLA, Los Angeles, CA USA3 Department of Molecular Biosciences, University of Kansas, Lawrence, KS USA4 Stowers Institute for Medical Research, Kansas City, MO USA2005 14 2 2005 5 3 3 2 11 2004 14 2 2005 Copyright © 2005 Stewart et al; licensee BioMed Central Ltd.2005Stewart 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
Vaults are intriguing ribonucleoprotein assemblies with an unknown function that are conserved among higher eukaryotes. The Pacific coast sea urchin, Strongylocentrotus purpuratus, is an invertebrate model organism that is evolutionarily closer to humans than Drosophila and C. elegans, neither of which possesses vaults. Here we compare the structures of sea urchin and mammalian vaults and analyze the subcellular distribution of vaults during sea urchin embryogenesis.
Results
The sequence of the sea urchin major vault protein (MVP) was assembled from expressed sequence tags and genome traces, and the predicted protein was found to have 64% identity and 81% similarity to rat MVP. Sea urchin MVP includes seven ~50 residue repeats in the N-terminal half of the protein and a predicted coiled coil domain in the C-terminus, as does rat MVP. A cryoelectron microscopy (cryoEM) reconstruction of isolated sea urchin vaults reveals the assembly to have a barrel-shaped external structure that is nearly identical to the rat vault structure. Analysis of the molecular composition of the sea urchin vault indicates that it contains components that may be homologs of the mammalian vault RNA component (vRNA) and protein components (VPARP and TEP1). The sea urchin vault appears to have additional protein components in the molecular weight range of 14–55 kDa that might correspond to molecular contents. Confocal experiments indicate a dramatic relocalization of MVP from the cytoplasm to the nucleus during sea urchin embryogenesis.
Conclusions
These results are suggestive of a role for the vault in delivering macromolecules to the nucleus during development.
==== Body
Background
The sea urchin Strongylocentrotus purpuratus is an important model system in developmental biology and its genome is currently being sequenced by the Human Genome Sequencing Center at the Baylor College of Medicine under the auspices of the National Human Genome Research Institute (NHGRI). Sea urchin embryos are well suited for biochemical approaches to studying the cell biology of development, as large quantities of eggs can easily be obtained, and their fertilization initiates the synchronous development of optically transparent embryos [1]. Sea urchins occupy an important phylogenetic position as basal deuterostomes, and are thus more closely related to humans than are other invertebrate model organisms such as Drosophila and C. elegans. In addition, the echinoderm lineage leading to sea urchins diverged from chordates prior to the large scale gene duplication events that occurred early in the evolution of the vertebrates, and because of this many of the genes that are found as multiple paralogues in vertebrates have only a single homolog in sea urchins. Therefore sea urchins provide a system that avoids the problem of functional redundancies between multiple paralogues that often occurs in vertebrates, and serves as a useful comparison for assessing the significance of conserved genes and regulatory linkages within the genome.
Sea urchins cells, like mammalian cells, contain abundant quantities of vaults, which are ~13MDa ribonucleoprotein particles of as yet unknown function [2]. Vaults are barrel-shaped assemblies composed of multiple copies of three proteins and small untranslated RNA molecules, called vRNA. Intriguingly vaults are up-regulated in certain human multidrug-resistant cancer cell lines, although their role in multidrug resistance remains unclear [3-6]. The high level of conservation among higher eukaryotes of both the major vault protein (MVP) sequence and the barrel-shaped vault structure suggests an important cellular role for the vault. Nucleocytoplasmic transport, sequestration of macromolecules, and protection from xenobiotics have all been proposed as possible functions for the vault [4,7-9]. Several recent publications have supported a role for vaults as either a transporter [10,11] or a scaffold protein [12]. Knock-out mice have been produced lacking one each of the three mammalian vault proteins: MVP; the vault poly(ADP-ribose) polymerase (VPARP); and the telomerase associated protein one (TEP1) [13-15]. These knock-out mice appear to be healthy, indicating that if the vault does perform a critical cellular function there must be a redundant or compensatory pathway in the mouse.
Immunofluorescence studies in sea urchins indicate that MVP is present throughout the cytoplasm in cleavage-stage zygotes and in the nucleus in adult somatic cells [16]. Within the nucleus of coelomocytes, the MVP is present in particularly high concentrations in the nucleolus. This is in contrast to localization studies in most other eukaryotic cells, which show MVP as primarily cytoplasmic [17-21]. During isolation from sea urchins, vaults are found to co-purify with both microtubules and ribosomes [22-24]. The sea urchin MVP cell localization and co-purification results led to a hypothesis that vaults may play a role in nucleocytoplasmic transport of ribosomes and/or mRNA [16].
The present study provides a comparison of vaults from sea urchins and rats in terms of their molecular composition and MVP protein sequence, as well as by cryoEM imaging and three-dimensional reconstruction. In addition, images obtained by confocal microscopy are presented that indicate differential localization of MVP during sea urchin embryogenesis, suggesting that the highly conserved ribonucleoprotein vault might play a role during development.
Results
Molecular composition of sea urchin vaults and comparison with mammalian vaults
Vaults isolated from mammals, including rats, mice, and humans, are composed of three proteins (MVP, VPARP, and TEP1) and one or more small untranslated RNA molecules, called vRNA. The protein stoichiometry is thought to be 96 copies of MVP with 2–16 copies of VPARP and TEP1 per particle [25]. The human (99 kDa) and rat (96 kDa) homologs of MVP display 91% similarity. In humans the two high molecular weight vault proteins, VPARP and TEP1, have been sequenced and have masses of 193 and 290 kDa, respectively [18,26-28]. vRNA accounts for ~5% of the total mass of the mammalian vault particle. Four human vRNA genes have been identified, encoding vRNAs that range from 86 to 99 bases in length [8,28]. Rat vaults in contrast have only one species of vRNA, 141 bases in length [29], which appears as a ~37 kDa band by SDS-PAGE [25]. All of the mammalian vRNA sequences that have been sequenced thus far show ~80% sequence identity and have similar predicted secondary structures [18]. In rat vaults vRNA can be degraded by a harsh treatment with two RNases leaving the rest of the vault particle structurally intact [25].
When purified under identical conditions, the molecular composition of the sea urchin vault is more complex than that of the rat vault (Fig. 1A). For both rat and sea urchin vaults, the strongest protein band is that of MVP at ~100 kDa. We have previously demonstrated that polyclonal antiserum generated against the sea urchin ~100 kDa protein has cross-reactivity with the rat vault MVP, as well as with the Dictyostelium MVPα and MVPβ vault proteins [16]. SDS-PAGE indicates that the sea urchin vault has two or more high molecular weight components at ~200 kDa, similar to the VPARP and TEP1 bands for rat and mouse vaults [9,30,31]. Searches of the S. purpuratus genome identified clear sequence homologs of both VPARP and TEP1 (data not shown). Purified sea urchin vaults also appear to contain vRNA, as demonstrated by the loss of a 26.5 kDa band after RNase treatment (Fig. 1B). We note that the apparent molecular weight of the sea urchin RNA molecule is smaller than that of the rat vault vRNA, 26.5 vs. 37 kDa.
Figure 1 The molecular composition of the sea urchin vault.(A) SDS-PAGE analysis (4–16% acrylamide gradient gel) of isolated sea urchin (S) and rat (R) vaults. The band at ~100 kDa in both lanes represents the major vault protein (MVP). (B) SDS-PAGE analysis of sea urchin vaults both without (-) and with (+) RNase treatment. The 26.5 kDa band in the (-) RNase lane, which is missing in the (+) RNase lane, is thought to correspond to the sea urchin vRNA. Mr represents the molecular weight marker lane.
What is noticeably different between the sea urchin and rat vaults is the number of protein bands within the molecular weight range of 14 to 55 kDa in the sea urchin vault that are not observed in the rat vault. Since sea urchin vaults were purified in the same manner as rat vaults, we conclude that either the molecular composition of the sea urchin vault is more complex than that of the mammalian vaults, or that the sea urchin vaults have a more varied composition of molecular cargo.
Amino acid sequence of the sea urchin major vault protein
To more fully characterize and identify the sea urchin MVP, we used expressed sequence tags (ESTs) and trace sequences from the S. purpuratus genome to assemble the coding sequence of the SpMVP gene (see Methods), which is apparently present as a single homolog per haploid genome. The fact that this sequence was found in several blastula-stage ESTs shows that it is expressed in the embryo. The deduced sequence of sea urchin MVP encodes a 95 kDa protein which is slightly smaller than the 96 kDa rat MVP, with 847 vs. 861 aa. These two MVP homologs exhibit 64% identity and 81% strong similarity (Fig. 2A). The sea urchin MVP sequence is composed of seven repeats (each 41 to 62 residues) in the N-terminal half of the protein, and a predicted coiled coil region (aa 663–762) in the C-terminal half of the protein (Fig. 2B). The rat MVP sequence has seven similar repeats of unknown function in the N-terminal half and also has a predicted coiled coil region in the C-terminal half [32,33].
Figure 2 The sequence of the sea urchin major vault protein (SpMVP). (A) Amino acid sequence of SpMVP aligned with that of rat MVP. Identical residues are highlighted in yellow, and similar residues are highlighted in green. (B) Bar diagram of SpMVP. The sequence has seven repeats in the N-terminal half of the protein (R1-R7), each repeat consisting of 41 to 62 aa, spanning residues 32–400; and a predicted coiled coil region in the C-terminal half of the sequence (C.C.), residues 663–762. The positions of two possible nuclear localization sequences (NLS1 470-KKAR, and NLS2 498-KPKR), a putative nuclear export sequence (NES, 792–816), and two probable sumoylation sites that are conserved across six species (S1 K308-VKGE, and S2 K707-AKAE) are indicated.
In addition to comparing rat and sea urchin MVP sequences, we used pair-wise alignments to compare the amino acid sequences of several other known MVPs, including the two from Dictyostelium (Table 1). These comparisons indicate that the MVP sequence is highly conserved between phylogenetically distant species, and that the similarity between sea urchin and rat MVP sequences is comparable to that among distantly related vertebrates, and also to the intraspecific similarity of Dictyostelium MVPs.
Table 1 Differences between MVP amino acid sequences of various species
Rat Xenopus Zebrafish Sea urchin Dictyostelium A Dictyostelium B
Rat 0 81 (9.38 %) 120 (13.67 %) 112 (12.86 %) 151 (17.34 %) 161 (18.51 %)
Xenopus 0 91 (10.50 %) 90 (10.48 %) 124 (14.44 %) 154 (17.91 %)
Zebrafish 0 113 (12.96 %) 148 (16.91 %) 165 (18.97 %)
Sea Urchin 0 121 (14.07 %) 157 (18.15 %)
Dictyostelium A 0 137 (15.91 %)
Dictyostelium B 0
Pair-wise sequence alignments of MVPs from each species were used to obtain number of amino acids that are not identical, strongly similar, or weakly similar between each pair of sequences. The numbers in parentheses represent corresponding percent difference. The two Dictyostelium MVPs are designated A and B.
The sea urchin MVP sequence was analyzed for potential nuclear localization signals (NLS's) and nuclear export signals (NES's), as well as possible sumoylation sites. Although the PredictNLS server [34] did not find any putative NLS's, visual examination of the sequence resulted identification of two basic regions with three lysine (K) or arginine (R) residues within a four amino acid stretch that are similar to the smallest consensus sequence of the monopartite type NLS [35]. Comparison of the sea urchin MVP sequence with the NES logo defined by la Cour et al. [36], indicates that a region of 25 residues near the C-terminus of protein (aa 792 – 816) has a close resemblance to the NES logo. Within this window we find the three most highly conserved leucine (L) residues (L804, L807 and L809); 3 additional residues matching the most favored residue type in the logo; 6 additional residues matching the second or third most favored residue type; and 3 additional residues matching the fourth, fifth of sixth most favored residue type. The Abgents Sumoylation Calculator predicts five highly probable sumoylation sites in order of probability: K308 (VKGE), K731 (LKAE), K707 (AKAE), K736 (AKIE), and K427 (AKDP) within the sea urchin MVP. With the exception of K427, the amino acid sequence surrounding the lysine acceptor residue follow the consensus sequence, ΨKXE, where Ψ represents a hydrophobic residue, K represents the target lysine, X represents any amino acid and E represents glutamic acid [37]. Sequence alignment of MVP homologs from six species showed that two of these five motifs (K308 and K707) are conserved.
Epitopes are also conserved between the rat MVP and the sea urchin MVP. Previously, we generated and purified a rabbit polyclonal antibody against the MVP that copurified with sea urchin microtubules [16]. These antibodies were shown to recognize both Dictyostelium and rat MVP [16]. In this report, we show that these affinity-purified antibodies also recognize the MVP in purified sea urchin vault preparations (Fig. 3). Two minor bands of ~80 kDa and 50 kDa cross-react with the anti-MVP antibodies indicating that these may be breakdown products of the 100 kDa MVP. In addition to the antibodies generated against the sea urchin MVP, anti-peptide antibodies were generated against a peptide located at the amino terminus of the rat MVP sequence [38]. This sequence is approximately 70% identical in the sea urchin and rat MVP sequence and the affinity-purified anti-peptide antibodies bind to purified sea urchin MVP (Fig. 3). This observation provides experimental support to relate the assembled coding sequence of the SpMVP gene to the major protein component of the purified sea urchin vault.
Figure 3 Western blots of isolated sea urchin vaults. (A) Peptide amino acid sequence of rat MVP (aa 19–35) aligned with Sp MVP (25–41). Anti-peptide antibodies were generated against the rat MVP sequence shown and affinity-purified as previously described [38]. Twelve of the rat MVP amino acids are conserved in the Sp MVP peptide sequence. (B) 150 μg of sea urchin egg extract proteins (lane E) and 10 μg of purified sea urchin vault proteins (lane V) were separated on this Coomassie-blue stained SDS-8% polyacrylamide mini-gel. The arrow shows the position of the 100 kDa MVP in (B, C and D). (C) Alkaline-phosphatase stained western blot showing the migration of the pre-stained protein ladder (lane M: 176.5, 113.7, 80.9, 63.8 (pink), 49.5, and 8.4 kDa polypeptides). Affinity-purified anti-sea urchin MVP antibodies [16] recognize a 100 kDa polypeptide in egg extracts (lane E) and in purified vault preparations (lane V). Asterisks indicate two bands of approximately 80 kDa and 50 kDa that may be breakdown products of the MVP. (D) Affinity-purified anti-peptide antibodies (anti-LDQN, [38]) bind to the 100 kDa MVP polypeptide in purified sea urchin vaults. This peptide antibody appears to bind non-specifically to a large number of polypeptides in the sea urchin egg extracts.
Electron microscopy and reconstruction of the sea urchin vault
Negative-stain EM and cryoEM images of isolated sea urchin vaults reveal that they have the same overall morphology as mammalian vaults (Fig. 4). As was noted in cryomicrographs of the rat vault [9], sea urchin vaults are occasionally observed to be 'open' at the midsection. One advantage of cryoEM over negative-stain TEM is that a cryomicrograph is essentially a projection image of all of the density, external and internal, within a macromolecular assembly. In cryomicrographs, both sea urchin and rat vaults appear to have molecular contents of varying mass per particle (Fig. 4B, white arrow and [9]).
Figure 4 Negative-stain and cryoEM images of isolated sea urchin vaults. (A) Negative-stain electron micrograph. (B) Cryoelectron micrograph. The black arrow indicates a vault that is opening at the midsection, and the white arrow indicates the dark molecular contents within another particle. The scale bar represents 1,000 Å. Note that the magnification of the cryoEM image (B) is slightly higher than that of the negative-stain EM image (A).
In order to characterize the three dimensional structure of the sea urchin vault, a three-dimensional structure was calculated using cryoEM images and single particle reconstruction methods. Only particle images of well-formed, fully closed sea urchin vaults were selected for image processing. Refinement of the cryoEM data set produced a three-dimensional reconstruction at 33 Å resolution (Fig. 5). The exterior shape of the sea urchin vault reconstruction is nearly identical to the shape of the rat vault reconstruction with a central barrel section and two protruding caps [9]. When the sea urchin vault reconstruction is cropped in half the large internal cavity is revealed (Fig. 5B and 5C). As has been noted for the rat vault, the volume of the internal cavity is large enough to enclose an intact ribosome [9].
Figure 5 The sea urchin vault reconstruction at 33 Å resolution. (A) The full reconstruction, which reveals that the sea urchin vault has essentially the same exterior structure as rat and mouse vaults. (B and C) The reconstruction is shown cropped along two perpendicular axes to reveal the hollow interior. The crop planes are displayed with the strongest density in red and the weakest density in green. Note that the strongest density is in the "shoulder" region at the top and bottom of the central barrel section. The flat portion of one cap is indicated by an arrow in (B). The scale bar represents 100 Å.
CryoEM reconstructions of recombinant and tissue derived vaults often show small holes at the cap/barrel junction [9,39]. The sea urchin vault reconstruction does not show these holes, but they are probably obscured by the low resolution of the sea urchin vault structure. Another subtle difference between the sea urchin and rat vault reconstructions is that less density is observed in the flat portion of the sea urchin vault cap (Fig. 5B, arrow). This is the same region of the vault that was identified as the vRNA binding site within the rat vault [25]. It is also a region that tends to be variable between vault reconstructions [39].
Cellular localization of MVP during sea urchin embryonic development
Earlier immunofluorescence studies indicated that maternal and zygotic MVP is located in the cytoplasm, whereas adult MVP is predominantly nuclear [16]. The subcellular location of MVP during embryogenesis was not determined by immunofluorescence. However, cell fractionation and immunoblot analyses of embryos showed that while total amount of MVP remained constant during embryogenesis, MVP became progressively concentrated in the nuclear fractions as development proceeded from the mesenchyme blastula stage to the larval stage [16]. Here we confirm and extend this result using antibody staining and confocal microscopy to localize MVP subcellularly during embryonic development. The confocal images in Figure 6 demonstrate that sea urchin MVP moves from a largely cytoplasmic distribution in the cleavage stage embryo, to a predominantly nuclear and/or perinuclear location at blastula and gastrula stages.
Figure 6 Confocal images of sea urchin embryos at various developmental stages. The cellular distribution of MVP is revealed by staining with an affinity-purified polyclonal antiserum against the sea urchin MVP and secondary staining with Oregon Green-conjugated anti-rabbit antibody (A, D, G, J; left column). The nuclei are stained with DAPI (B, E, H, K; middle column). By merging the MVP and DNA images coincident staining is observed as cyan (C, F, I, L; right column). Note that as development proceeds the cellular localization of MVP progressively shifts from cytoplasmic to nuclear.
Discussion
In this study we used the sea urchin genome to deduce the amino acid sequence of the sea urchin MVP and characterized the structure of the sea urchin vault particle using cryoEM single particle reconstruction methods. It is known that multiple copies of MVP form the recognizable vault structure, as expression of rat MVP in insect cells leads to the assembly of vaults [40]. Given the relatively high (81%) sequence similarity observed between sea urchin and rat MVP, it is not surprising that the exterior sea urchin vault structure is quite similar to that of rat vaults [9,39]. In addition to MVP, the rat vault is composed of the high molecular weight proteins VPARP and TEP1, as well as the small vRNA. The electrophoretic analysis of purified sea urchin vault presented here shows protein bands and one RNA band that might correspond to homologs of the mammalian vault components VPARP, TEP1, and vRNA. Moreover, sequence homologs of VPARP and TEP1 are found in the sea urchin genome.
SDS-PAGE analysis has indicated that the molecular composition of sea urchin vaults is more complex than that of rat vaults. Protein bands in the range of 14 to 55 kDa are observed that might either be additional vault components or macromolecular contents that are present in high copy numbers within the sea urchin vault. One protein in this mass range (~50 kDa) cross-reacts with MVP antisera and is likely to be a breakdown product of MVP. The identities of the additional proteins will be the subject of future investigations. It is tempting to speculate that the difference in protein composition between sea urchin and rat vaults might indicate that sea urchin vaults are transporting or sequestering a larger assortment of macromolecules.
Vaults were originally named because of their similarity to arched cathedral ceilings [41]. This name may be apt in another sense in that the vault might serve to provide a safe enclosure for its molecular contents [2]. It has been postulated that mammalian vaults might open under certain physiological conditions to allow encapsulation or release of their molecular cargo. Sea urchin vaults are occasionally observed to be partially open at the midsection in cryomicrographs. Thus sea urchin vaults, as well as mammalian vaults, might be able to regulate what is contained within the large internal cavity. Both sea urchin and rat vaults [9] appear to have molecular contents in cryomicrographs. However as the features of the contents vary from vault to vault they are inappropriately averaged or "smeared" during the image reconstruction process.
A difference map analysis between cryoEM reconstructions of intact and RNase-treated rat vaults led to the localization of the vRNA to the flat portions of the rat vault caps [25]. The reconstruction of the sea urchin vault reveals a difference in this region of the vault and appears more similar to the RNase-treated rat vault [25]. Although RNase-treatment does not harm the structure of the rat vault, it does appear to affect the sea urchin vault. Several attempts were made to collect negative-stain and cryoEM images of the RNase-treated sea urchin vaults, but no intact vault particles were observed. The difference observed in the cap density and the differing response to RNase-treatment could imply that the vRNA has a different location or a more important structural role in the sea urchin vault than in the rat vault.
The confocal images reveal a progressive enrichment of MVP in the nucleus and perinuclear cytoplasm during sea urchin embryogenesis. Our previous biochemical fractionation studies confirm that the MVP accumulates in the nucleus beginning at the mesenchyme blastula stage [16]. This important stage of development marks the transition between proliferation and differentiation. As a consequence of differentiation the cells are preparing to exit from the rapid cell cycles that followed fertilization. For example, mRNAs which were abundant in the early cleavage stages, such as cyclin mRNA, decrease dramatically at the blastula stage [42]. New post-cleavage stage mRNAs begin to accumulate at blastula stage. We suggest that vaults participate in nucleocytoplasmic transport that accompanies remodeling of nuclear architecture in preparation for terminal cell differentiation [43].
While MVP localization studies in mammalian cells reveal a largely cytoplasmic location, the published staining pattern is consistent with there also being a small amount of nuclear MVP [17,18]. While no classical NLS sequence is found in either the sea urchin or rat MVP, there are basic sequences that might potentially fulfill the requirements of an NLS. The C-terminal region of sea urchin MVP has a region that strongly resembles the NES logo assembled from known NES's [36] and this region is fairly well conserved in MVP from six species. In addition to possible NLS and NES signals, there are several sumoylation signals within the sea urchin MVP.
Sumoylation is a dynamic and reversible process that involves the covalent attachment of the small ubiquitin-like modifier (SUMO) protein to protein substrates. Similar to ubiquination, sumoylation requires four enzymatic steps to attach the carboxyl-terminal glycine of SUMO to the ε-amino group of lysine (reviewed in [44]). However, unlike traditional ubiquination, sumoylation does not mark the protein for degradation. The wide variety of SUMO targets identified to date does not reveal a single function for this protein modification (reviewed in [45]). A variety of transcription factors and cofactors, viral proteins, proteins associated with nuclear architecture and genome surveillance, as well as signal transduction molecules appear to by sumoylated [44]. Importantly for this study, two proteins of the nuclear pore complex, RanGAP1 and Ran-binding protein 2 are both SUMO substrates. Sumoylation of RanGAP1 is important for the nucleocytoplasmic transport of this cytoplasmic nuclear import factor [46-48]. In addition, RanBP2, a docking protein at the cytoplasmic surface of the nuclear protein complex is a SUMO E3 ligase, also known as Nup358 [49-51]. These observations suggest that SUMO plays an important role in nucleocytoplasmic trafficking. In this regard, sumoylation of the major vault protein may regulate the transport of the vault particle through the nuclear pore complex.
Taken together these results suggest that the vault may play a role in delivering macromolecules to the nucleus during embryonic development in sea urchins. A trafficking role has also been proposed for vaults in human cells based on the observations that MVP interacts with the estrogen receptor in MCF-7 breast cancer cells [10] and with the tumor suppressor PTEN in HeLa cells [11]. Estrogen receptor is involved in regulating eukaryotic gene expression and can affect cellular proliferation and differentiation. Estrogen treatment is observed to increase the association of the estrogen receptor with MVP in the nuclei of MCF-7 cells, thus leading to the hypothesis that vaults may be involved in nucleocytoplasmic shuttling of the estrogen receptor and in modulating the effect of steroid hormones. PTEN negatively regulates the phosphoinositide 3-kinase pathway and cell growth. MVP has been shown by a yeast two-hybrid screen to be a dominant PTEN-binding protein. It has been postulated that perhaps the vault serves to mediate the cellular localization PTEN, thus potentially affecting cell growth.
The results presented here suggest that sea urchin vaults contain various macromolecules in addition to the two main vault-associated proteins, VPARP and TEP1. Human MVP is known to interact specifically with the tyrosine phosphatase SHP-2, an enzyme that plays an important role in intracellular signaling [12]. It has been demonstrated using MVP-deficient mouse fibroblasts that MVP helps to support cell survival. MVP is proposed to act as a scaffold protein for SHP-2 and other extracellular-regulated kinases (Erks) and thus facilitate, or somehow modulate, growth factor signaling. Perhaps the varied macromolecular contents of the sea urchin vault include proteins involved in sea urchin growth factor signaling.
Conclusions
It would seem likely that the sea urchin MVP is transported to the nucleus for a reason. Presumably cells must expend a considerable amount of energy to achieve this differential localization, as MVP is a relatively abundant protein. The MVP concentration in the unfertilized sea urchin egg is on the order of 10 μM, as determined by immunoblotting [16]. This maternal store of MVP is sufficient to assemble 107 intact vault particles. By comparison, the mature sea urchin egg contains approximately 4 × 108 ribosomes [52]. Although the function of the vault is not yet clearly delineated, our results on the highly conserved structure, molecular composition, and differential cellular localization of MVP suggest that vaults may be important for nucleocytoplasmic trafficking during embryonic development.
Methods
Isolation of sea urchin and rat vaults
Vaults were purified to homogeneity from unfertilized eggs of the Pacific coast sea urchin Strongylocentrotus purpuratus. Animals were spawned and the eggs were collected and washed as described [53]. Washed eggs (150 ml) were homogenized with a motor driven Potter-Elvehjem-type tissue grinder in an equal volume of buffer (100 mM PIPES-KOH, pH 7.3, 1 mM MgSO4, 4 mM EGTA, 2 mM DTT, 1 mM GTP, 0.2 mM PMSF, 10 μg/ml leupeptin and 1 μg/ml pepstatin A). The homogenate was centrifuged at 18 K rpm (~39,000 g) in a Beckman J20 rotor at 4°C for 45 min. The resulting supernatant fluids were decanted into a 500-ml Erlenmeyer flask at room temperature. A fresh aliquot of GTP was added such that the GTP concentration was increased by 1 mM. The DMSO was added in three aliquots with gentle mixing until a final concentration of 15% (v/v) was reached. Microtubule assembly was promoted by incubating the supernatant fluids in a 24°C water bath for 30–45 min. The microtubules were pelleted by centrifugation at 18K rpm (~39,000 g) in a Beckman J20 rotor at 24°C for 30 min. The supernatant fluids (H1S) were carefully removed and drop-frozen in liquid nitrogen and stored in a -80°C freezer. Microtubules in the pellet were purified by two subsequent cycles of assembly and disassembly as previously described [54] and used for purposes other than the current study.
Sea urchin vaults were purified from the microtubule-depleted supernatant fluids that had been previously frozen in liquid nitrogen. Approximately 150 ml of frozen supernatant (H1S) was thawed and mixed with an equal volume of 50 mM Tris-HCL, pH 7.4, 1.5 mM MgCl2, 75 mM NaCl, 1 mM DTT and 1 mM PMSF. This pooled supernatant was used as the starting point for the vault purification. This extract was treated as if it was a rat liver homogenate and all subsequent purification steps were as described for rat liver vaults, including purification on sucrose density gradients and two cesium chloride gradients [9].
Rat vaults were purified from liver by sucrose-density and cesium chloride gradient centrifugation as described by the Rome laboratory [9].
Molecular composition of sea urchin vaults
Vault samples were analyzed by SDS-PAGE with the discontinuous buffer formulation of Laemmli et al., [55] and stained with Coomassie Blue R-250 (Fig. 1A) or silver (Fig. 1B).
Assembly and analysis of the sea urchin major vault protein sequence
Trace sequences from the S. purpuratus genome project [56]were used to assemble the coding sequence of the SpMVP gene. Toward this end the TBLASTN algorithm [57,58] was performed using the rat MVP as a query sequence to search the NCBI trace archive of the sea urchin genome that is now at near 10x coverage. Due to the high level of sequence conservation it was possible to assemble the urchin sequence from the TBLASTN results using the rat MVP as a scaffold. The urchin MVP sequence was assembled in such a way that maximum similarity to the rat scaffold sequence was maintained. The fully assembled SpMVP sequence was then verified by using it to query the NCBI sea urchin EST collection (representing mRNAs expressed in the embryo and larva) by TBLASTN. Except for a small region (A728 to G785) near the C-terminal sequence, the entire SpMVP was covered by S. purpuratus ESTs, with the following NCBI accession numbers: BG784484, BG784394, CD305511, CD295832, CD295616, CD304000, CD310423, CD307400, CD305382, CD292179, CD305646, CD333690, CD309547, CD292477, and BG783194. The gap from A728 to G785 was covered by a sequence in the NCBI trace archives (gnl|ti|287010960). The assembled SpMVP sequence has been deposited in the NCBI database under accession number BK005641.
We note that the sea urchin MVP has a molecular weight of 95 kDa, calculated from the protein sequence. Previously the sea urchin MVP was referred to as a 107-kDa polypeptide based on its apparent molecular weight by SDS-PAGE [16]. The program COILS [59] was used to identify a likely coiled coil region in the C-terminal half of the protein (aa 663–762) [60]. The sea urchin MVP sequence was submitted to the PredictNLS server [34,61]. No potential NLS sequences were found by this server. The two possible NLS sequences shown in Figure 2B are based on sequence observation. The sea urchin MVP sequence was compared by hand to the NES logo described by la Cour et al. and based on the alignment of 58 high-quality NES's [36]. The sea urchin MVP sequence was submitted to the Abgent Sumoylation Calculator [62]. Five motifs with high probability of being sumoylation sites were identified (scores 0.6889 to 0.9278). Two of these five motifs are conserved among MVP sequences of six species (Strongy. purp., Homo sapiens, Mus musculus, Rattus norv., Danio rerio, and Xenopus laevis) and are indicated in Figure 2B. The CLUSTALW alignment between sea urchin and rat MVP in Figure 2A was produced by submitting the sequences to the CLUSTALW server [63].
Negative-stain electron microscopy
Isolated sea urchin vaults were applied to freshly glow-discharged formvar-coated copper grids (Electron Microscopy Sciences). Excess liquid was wicked away and the samples were briefly air dried. Samples were stained with uranyl acetate and observed using a JEOL 1200EXII transmission electron microscope.
CryoEM imaging and three-dimensional reconstruction
CryoEM grids were prepared using cryogenic plunge freezing methods described previously [9,64]. Digital cryoelectron micrographs were collected on an FEI/Philips CM120 transmission electron microscope with a LaB6 filament, a Gatan 626 cryotransfer holder, and a Gatan slow scan CCD camera (1024 × 1024 pixels, YAG scintillator). One hundred and forty-one cryomicrographs were collected with a nominal magnification of 45,000×, and with two different defocus values (-1.0 and -0.6 μm). The pixel size in the cryomicrographs is 4.1 Å on the molecular scale as determined by calibration with a catalase crystal.
The QVIEW software package was used to extract 481 individual vault particle images with a selection box size of 200 × 200 pixels [65]. The initial translation step was performed by cross-correlating each particle image with an 180° rotated version of itself. The IMAGIC software package was used for all 3D image processing steps [66]. The published RNase-treated rat vault reconstruction was used as a search model in the first round of refinement [9]. C8 symmetry was assumed for the sea urchin vault for the first several rounds of refinement, as was used for the intact rat vault reconstruction [9]. After a few rounds of refinement the two ends of the vault appeared nearly identical, and from then on D8 symmetry was imposed, as was done for the RNase-treated rat vault reconstruction [25]. The particles images were corrected for the contrast transfer function (CTF) of the microscope using the CTF equation published by Baker et al. [67]. The following CTF parameters were used: Cs = 2 mm, fraction of amplitude contrast = 0.1, and kV = 120. A subset of 409 particle images that agreed best with the rest of the set were selected for the final reconstruction. The resolution was calculated using the Fourier shell correlation method with an elliptical mask applied to remove noise and disordered contents. The resolution was found to be 33 Å by the 0.5 correlation threshold criterion. The sea urchin vault reconstruction is shown filtered to 33 Å resolution and contoured so that the surface appears continuous.
All of the image processing was performed on HP/Digital alpha unix workstations. The graphics representations were produced with the AVS (Advanced Visual Systems, Inc.) software package.
Antibody staining and confocal microscopy of whole mounted embryos
Staged embryos were fixed in -20°C Methanol for 20 minutes on ice, then washed three times in PBS plus 0.2% Tween-20 (PBST). Embryos were blocked for 30 minutes on ice in PBST plus 5% Bovine Serum Albumin, then incubated with an appropriate dilution of an affinity purified antibody against MVP [16] overnight at 4°C. Following three washes for 5 minutes each in PBST, the embryos were incubated for 1 hour at room temperature in 1 μg/ml Oregon Green 488 goat anti-rabbit antibody. Embryos were washed three times in PBST and stained in PBST with 300 nM DAPI for 10 minutes at room temperature during the second wash. Digital confocal images were collected using a Leica TCS confocal microscope.
Authors' contributions
PLS directed the cryoEM and three-dimensional image processing as well as helped to draft the manuscript. MM performed the cryomicroscopy and image processing. JL provided technical assistance. CDS performed the antibody staining and confocal imaging of whole-mounted embryos. AJR assembled the SpMVP sequence and performed the sequence alignments used to calculate sequence similarities and differences. JAC directed the research pertaining to vault localization in the embryo, and drafted Figure 6 and the relevant parts of the manuscript. KAS conceived of the study, participated in its design and coordination, purified the sea urchin vaults, and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by grants from the National Institutes of Health (COBRE award 1P20RR15563) with matching support from the State of Kansas and the University of Kansas (K.A.S.); the National Science Foundation (MCB-9722353 to P.L.S. and MCB-9982377 to K.A.S.); and institutional funding by the Stowers Institute for Medical Research (J.A.C.). The authors thank Dr. Kristi Neufeld for her advice on predicting NES and NLS.
==== Refs
Suprenant KA Rebhun LI Assembly of unfertilized sea urchin egg tubulin at physiological temperatures J Biol Chem 1983 258 4518 4525 6833265
Suprenant KA Vault ribonucleoprotein particles: sarcophagi, gondolas, or safety deposit boxes? Biochemistry 2002 41 14447 14454 12463742 10.1021/bi026747e
Scheffer GL Wijngaard PL Flens MJ Izquierdo MA Slovak ML Pinedo HM Meijer CJ Clevers HC Scheper RJ The drug resistance-related protein LRP is the human major vault protein Nat Med 1995 1 578 582 7585126 10.1038/nm0695-578
Izquierdo MA Scheffer GL Flens MJ Giaccone G Broxterman HJ Meijer CJ van der Valk P Scheper RJ Broad distribution of the multidrug resistance-related vault lung resistance protein in normal human tissues and tumors Am J Pathol 1996 148 877 887 8774142
Ikeda K Oka M Yamada Y Soda H Fukuda M Kinoshita A Tsukamoto K Noguchi Y Isomoto H Takeshima F Murase K Kamihira S Tomonaga M Kohno S Adult T-cell leukemia cells over-express the multidrug-resistance-protein (MRP) and lung-resistance-protein (LRP) genes Int J Cancer 1999 82 599 604 10404077 10.1002/(SICI)1097-0215(19990812)82:4<599::AID-IJC21>3.0.CO;2-R
Siva AC Raval-Fernandes S Stephen AG LaFemina MJ Scheper RJ Kickhoefer VA Rome LH Up-regulation of vaults may be necessary but not sufficient for multidrug resistance Int J Cancer 2001 92 195 202 11291045 10.1002/1097-0215(200102)9999:9999<::AID-IJC1168>3.0.CO;2-7
Chugani DC Rome LH Kedersha NL Evidence that vault ribonucleoprotein particles localize to the nuclear pore complex J Cell Sci 1993 106 23 29 8270627
Kickhoefer VA Rajavel KS Scheffer GL Dalton WS Scheper RJ Rome LH Vaults are up-regulated in multidrug-resistant cancer cell lines J Biol Chem 1998 273 8971 8974 9535882 10.1074/jbc.273.15.8971
Kong LB Siva AC Rome LH Stewart PL Structure of the vault, a ubiquitous cellular component Structure Fold Des 1999 7 371 379 10196123 10.1016/S0969-2126(99)80050-1
Abbondanza C Rossi V Roscigno A Gallo L Belsito A Piluso G Medici N Nigro V Molinari AM Moncharmont B Puca GA Interaction of vault particles with estrogen receptor in the MCF-7 breast cancer cell J Cell Biol 1998 141 1301 1310 9628887 10.1083/jcb.141.6.1301
Yu Z Fotouhi-Ardakani N Wu L Maoui M Wang S Banville D Shen SH PTEN associates with the vault particles in HeLa cells J Biol Chem 2002 277 40247 40252 12177006 10.1074/jbc.M207608200
Kolli S Zito CI Mossink MH Wiemer EA Bennett AM The major vault protein is a novel substrate for the tyrosine phosphatase SHP-2 and scaffold protein in epidermal growth factor signaling J Biol Chem 2004 279 29374 29385 15133037 10.1074/jbc.M313955200
Liu Y Snow BE Hande MP Baerlocher G Kickhoefer VA Yeung D Wakeham A Itie A Siderovski DP Lansdorp PM Robinson MO Harrington L Telomerase-associated protein TEP1 is not essential for telomerase activity or telomere length maintenance in vivo Mol Cell Biol 2000 20 8178 8184 11027287 10.1128/MCB.20.21.8178-8184.2000
Mossink MH van Zon A Franzel-Luiten E Schoester M Kickhoefer VA Scheffer GL Scheper RJ Sonneveld P Wiemer EA Disruption of the murine major vault protein (MVP/LRP) gene does not induce hypersensitivity to cytostatics Cancer Res 2002 62 7298 7304 12499273
Liu Y Snow BE Kickhoefer VA Erdmann N Zhou W Wakeham A Gomez M Rome LH Harrington L Vault poly(ADP-ribose) polymerase is associated with mammalian telomerase and is dispensable for telomerase function and vault structure in vivo Mol Cell Biol 2004 24 5314 5323 15169895 10.1128/MCB.24.12.5314-5323.2004
Hamill DR Suprenant KA Characterization of the sea urchin major vault protein: a possible role for vault ribonucleoprotein particles in nucleocytoplasmic transport Dev Biol 1997 190 117 128 9331335 10.1006/dbio.1997.8676
Kedersha NL Rome LH Vaults: large cytoplasmic RNP's that associate with cytoskeletal elements Mol Biol Rep 1990 14 121 122 1694556 10.1007/BF00360441
Kickhoefer VA Siva AC Kedersha NL Inman EM Ruland C Streuli M Rome LH The 193-kD vault protein, VPARP, is a novel poly(ADP-ribose) polymerase J Cell Biol 1999 146 917 928 10477748 10.1083/jcb.146.5.917
Herrmann C Golkaramnay E Inman E Rome L Volknandt W Recombinant major vault protein is targeted to neuritic tips of PC12 cells J Cell Biol 1999 144 1163 1172 10087261 10.1083/jcb.144.6.1163
Berger W Spiegl-Kreinecker S Buchroithner J Elbling L Pirker C Fischer J Micksche M Overexpression of the human major vault protein in astrocytic brain tumor cells Int J Cancer 2001 94 377 382 11745417 10.1002/ijc.1486
Schroeijers AB Reurs AW Scheffer GL Stam AG de Jong MC Rustemeyer T Wiemer EA de Gruijl TD Scheper RJ Up-regulation of drug resistance-related vaults during dendritic cell development J Immunol 2002 168 1572 1578 11823484
Suprenant KA Tempero LB Hammer LE Association of ribosomes with in vitro assembled microtubules Cell Motil Cytoskeleton 1989 14 401 415 2479489
Suprenant KA Microtubules, ribosomes, and RNA: evidence for cytoplasmic localization and translational regulation Cell Motil Cytoskeleton 1993 25 1 9 8519063
Hamill D Davis J Drawbridge J Suprenant KA Polyribosome targeting to microtubules: enrichment of specific mRNAs in a reconstituted microtubule preparation from sea urchin embryos J Cell Biol 1994 127 973 984 7962079 10.1083/jcb.127.4.973
Kong LB Siva AC Kickhoefer VA Rome LH Stewart PL RNA location and modeling of a WD40 repeat domain within the vault RNA 2000 6 890 900 10864046 10.1017/S1355838200000157
Nakayama J Saito M Nakamura H Matsuura A Ishikawa F TLP1: a gene encoding a protein component of mammalian telomerase is a novel member of WD repeats family Cell 1997 88 875 884 9118230 10.1016/S0092-8674(00)81933-9
Harrington L McPhail T Mar V Zhou W Oulton R Bass MB Arruda I Robinson MO A mammalian telomerase-associated protein Science 1997 275 973 977 9020079 10.1126/science.275.5302.973
Kickhoefer VA Stephen AG Harrington L Robinson MO Rome LH Vaults and telomerase share a common subunit, TEP1 J Biol Chem 1999 274 32712 32717 10551828 10.1074/jbc.274.46.32712
Kickhoefer VA Searles RP Kedersha NL Garber ME Johnson DL Rome LH Vault ribonucleoprotein particles from rat and bullfrog contain a related small RNA that is transcribed by RNA polymerase III J Biol Chem 1993 268 7868 7873 7681830
Kedersha NL Heuser JE Chugani DC Rome LH Vaults. III. Vault ribonucleoprotein particles open into flower-like structures with octagonal symmetry J Cell Biol 1991 112 225 235 1988458 10.1083/jcb.112.2.225
Kickhoefer VA Liu Y Kong LB Snow BE Stewart PL Harrington L Rome LH The Telomerase/vault-associated protein TEP1 is required for vault RNA stability and its association with the vault particle J Cell Biol 2001 152 157 164 11149928 10.1083/jcb.152.1.157
Herrmann C Kellner R Volknandt W Major vault protein of electric ray is a phosphoprotein Neurochem Res 1998 23 39 46 9482265 10.1023/A:1022445302710
van Zon A Mossink MH Schoester M Scheffer GL Scheper RJ Sonneveld P Wiemer EA Structural domains of vault proteins: a role for the coiled coil domain in vault assembly Biochem Biophys Res Commun 2002 291 535 541 11855821 10.1006/bbrc.2002.6472
Cokol M Nair R Rost B Finding nuclear localization signals EMBO Rep 2000 1 411 415 11258480 10.1093/embo-reports/kvd092
Tinland B Koukolikova-Nicola Z Hall MN Hohn B The T-DNA-linked VirD2 protein contains two distinct functional nuclear localization signals Proc Natl Acad Sci USA 1992 89 7442 7446 1502156
la Cour T Gupta R Rapacki K Skriver K Poulsen FM Brunak S NESbase version 1.0: a database of nuclear export signals Nucleic Acids Res 2003 31 393 396 12520031 10.1093/nar/gkg101
Rodriguez MS Dargemont C Hay RT SUMO-1 conjugation in vivo requires both a consensus modification motif and nuclear targeting J Biol Chem 2001 276 12654 12659 11124955 10.1074/jbc.M009476200
Eichenmüller B Kedersha N Solovyeva E Everley P Lang J Himes RH Suprenant KA Vaults bind directly to microtubules via their caps and not their barrels Cell Motil Cytoskeleton 2003 56 225 236 14584025 10.1002/cm.10147
Mikyas Y Makabi M Raval-Fernandes S Harrington L Kickhoefer VA Rome LH Stewart PL Cryoelectron microscopy imaging of recombinant and tissue derived vaults: localization of the MVP N termini and VPARP J Mol Biol 2004 344 91 105 15504404 10.1016/j.jmb.2004.09.021
Stephen AG Raval-Fernandes S Huynh T Torres M Kickhoefer VA Rome LH Assembly of vault-like particles in insect cells expressing only the major vault protein J Biol Chem 2001 276 23217 23220 11349122 10.1074/jbc.C100226200
Kedersha NL Rome LH Isolation and characterization of a novel ribonucleoprotein particle: large structures contain a single species of small RNA J Cell Biol 1986 103 699 709 2943744 10.1083/jcb.103.3.699
Kelso-Winemiller L Yoon J Peeler MT Winkler MM Sea urchin maternal mRNA classes with distinct development regulation Dev Genet 1993 14 397 406 8293581
Angerer LM Angerer RC Patterning the sea urchin embryo: gene regulatory networks, signaling pathways, and cellular interactions Curr Top Dev Biol 2003 53 159 198 12509127
Seeler JS Dejean A Nuclear and unclear functions of SUMO Nat Rev Mol Cell Biol 2003 4 690 699 14506472 10.1038/nrm1200
Muller S Ledl A Schmidt D SUMO: a regulator of gene expression and genome integrity Oncogene 2004 23 1998 2008 15021887 10.1038/sj.onc.1207415
Matunis MJ Coutavas E Blobel G A novel ubiquitin-like modification modulates the partitioning of the Ran-GTPase-activating protein RanGAP1 between the cytosol and the nuclear pore complex J Cell Biol 1996 135 1457 1470 8978815 10.1083/jcb.135.6.1457
Mahajan R Delphin C Guan T Gerace L Melchior F A small ubiquitin-related polypeptide involved in targeting RanGAP1 to nuclear pore complex protein RanBP2 Cell 1997 88 97 107 9019411 10.1016/S0092-8674(00)81862-0
Mahajan R Gerace L Melchior F Molecular characterization of the SUMO-1 modification of RanGAP1 and its role in nuclear envelope association J Cell Biol 1998 140 259 270 9442102 10.1083/jcb.140.2.259
Pichler A Gast A Seeler JS Dejean A Melchior F The nucleoporin RanBP2 has SUMO1 E3 ligase activity Cell 2002 108 109 120 11792325 10.1016/S0092-8674(01)00633-X
Kirsh O Seeler JS Pichler A Gast A Muller S Miska E Mathieu M Harel-Bellan A Kouzarides T Melchior F Dejean A The SUMO E3 ligase RanBP2 promotes modification of the HDAC4 deacetylase EMBO J 2002 21 2682 2691 12032081 10.1093/emboj/21.11.2682
Miyauchi Y Yogosawa S Honda R Nishida T Yasuda H Sumoylation of Mdm2 by protein inhibitor of activated STAT (PIAS) and RanBP2 enzymes J Biol Chem 2002 277 50131 50136 12393906 10.1074/jbc.M208319200
Griffith JK Griffith BB Humphreys T Regulation of ribosomal RNA synthesis in sea urchin embryos and oocytes Dev Biol 1981 87 220 228 7286427 10.1016/0012-1606(81)90145-7
Suprenant KA Foltz Daggett MA Sea urchin microtubules Curr Top Dev Biol 1995 31 65 99 8746662
Suprenant KA Marsh JC Temperature and pH govern the self-assembly of microtubules from unfertilized sea-urchin egg extracts J Cell Sci 1987 87 71 84 3667717
Laemmli UK Cleavage of structural proteins during the assembly of the head of bacteriophage T4 Nature 1970 227 680 685 5432063
Sea Urchin Genome Project [http://www.hgsc.bcm.tmc.edu/projects/seaurchin/]
Karlin S Altschul SF Methods for assessing the statistical significance of molecular sequence features by using general scoring schemes Proc Natl Acad Sci USA 1990 87 2264 2268 2315319
Karlin S Altschul SF Applications and statistics for multiple high-scoring segments in molecular sequences Proc Natl Acad Sci USA 1993 90 5873 5877 8390686
COILS - Prediction of Coiled Coil Regions in Proteins [http://www.ch.embnet.org/software/COILS_form.html]
Lupas A Van Dyke M Stock J Predicting coiled coils from protein sequences Science 1991 252 1162 1164 2031185
PredictNLS Server [http://cubic.bioc.columbia.edu/services/predictNLS/]
Abgent SUMOplot Sumoylation Calculator [http://www.abgent.com/default.php?page=sumoplot]
CLUSTALW Server
[http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_clustalw.html]
Adrian M Dubochet J Lepault J McDowall AW Cryo-electron microscopy of viruses Nature 1984 308 32 36 6322001 10.1038/308032a0
Shah AK Stewart PL QVIEW: software for rapid selection of particles from digital electron micrographs J Struct Biol 1998 123 17 21 9774540 10.1006/jsbi.1998.4011
van Heel M Harauz G Orlova EV Schmidt R Schatz M A new generation of the IMAGIC image processing system J Struct Biol 1996 116 17 24 8742718 10.1006/jsbi.1996.0004
Baker TS Olson NH Fuller SD Adding the third dimension to virus life cycles: three-dimensional reconstruction of icosahedral viruses from cryo-electron micrographs Microbiol Mol Biol Rev 1999 63 862 922 10585969
| 15710043 | PMC550661 | CC BY | 2021-01-04 16:40:17 | no | BMC Dev Biol. 2005 Feb 14; 5:3 | utf-8 | BMC Dev Biol | 2,005 | 10.1186/1471-213X-5-3 | oa_comm |
==== Front
BMC MicrobiolBMC Microbiology1471-2180BioMed Central London 1471-2180-5-71569137710.1186/1471-2180-5-7Research ArticleProteolytic cleavage of pertussis toxin S1 subunit is not essential for its activity in mammalian cells Carbonetti Nicholas H [email protected] R Michael [email protected] Galina V [email protected] Roger D [email protected] Zoë EV [email protected] Department of Microbiology and Immunology, University of Maryland School of Medicine, Baltimore, MD 21201, USA2005 3 2 2005 5 7 7 23 9 2004 3 2 2005 Copyright © 2005 Carbonetti 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
Pertussis toxin (PT) is an exotoxin virulence factor produced by Bordetella pertussis, the causative agent of whooping cough. PT consists of an active subunit (S1) that ADP-ribosylates the alpha subunit of several mammalian G proteins, and a B oligomer (S2–S5) that binds glycoconjugate receptors on cells. PT appears to enter cells by endocytosis, and retrograde transport through the Golgi apparatus may be important for its cytotoxicity. A previous study demonstrated that proteolytic processing of S1 occurs after PT enters mammalian cells. We sought to determine whether this proteolytic processing of S1 is necessary for PT cytotoxicity.
Results
Protease inhibitor studies suggested that S1 processing may involve a metalloprotease, and processing does not involve furin, a mammalian cell protease that cleaves several other bacterial toxins. However, inhibitor studies showed a general lack of correlation of S1 processing with PT cellular activity. A combination of replacement, insertion and deletion mutations in the C-terminal region of S1, as well as mass spectrometry data, suggested that the cleavage site is located around residue 203–204, but that cleavage is not strongly sequence-dependent. Processing of S1 was abolished by each of 3 overlapping 8 residue deletions just downstream of the putative cleavage site, but not by smaller deletions in the same region. Processing of the various mutant forms of PT did not correlate with cellular activity of the toxin, nor with the ability of the bacteria producing them to infect the mouse respiratory tract. In addition, S1 processing was not detected in transfected cells expressing S1, even though S1 was fully active in these cells.
Conclusions
S1 processing is not essential for the cellular activity of PT. This distinguishes it from the processing of various other bacterial toxins, which has been shown to be important for their cytotoxicity. S1 processing may be mediated primarily by a metalloprotease, but the cleavage site on S1 is not sequence-dependent and processing appears to depend on the general topology of the protein in that region, indicating that multiple proteases may contribute to this cleavage.
==== Body
Background
Pertussis toxin (PT) is a complex exotoxin and an important virulence factor produced by Bordetella pertussis, a bacterial pathogen of the human respiratory tract that causes the disease whooping cough. PT holotoxin is a multi-subunit complex with an AB5 structure [1,2]: the enzymatically active A subunit (S1) is an ADP-ribosyltransferase that modifies the alpha subunit of several heterotrimeric G proteins (primarily Gi proteins) in mammalian cells [3,4], and the B oligomer (S2, S3, 2 copies of S4, and S5) binds unidentified glycoconjugate receptors on cells [5,6]. The events in the intracellular trafficking of PT between surface binding and ADP-ribosylation of target G proteins on the cytoplasmic side of cellular membranes are relatively obscure. Electron microscopy studies and experiments with inhibitors suggest that the holotoxin is internalized by endocytosis [7-9]. Subcellular fractionation experiments and inhibition of cytotoxicity by Brefeldin A (BFA), which disrupts the Golgi apparatus [10], provide evidence for subsequent retrograde transport of PT to the Golgi apparatus [7-9]. Trafficking of PT beyond the Golgi apparatus is relatively uncharacterized, though it has been hypothesized that further retrograde transport of PT through the secretory pathway to the endoplasmic reticulum (ER) occurs [11-13]. After dissociation of S1 from the holotoxin, the liberated S1 subunit is then proposed to traverse the ER membrane to gain access to its target G proteins in the cytosol [13]. Evidence supporting this ER-to-cytosol translocation was obtained from transfection studies with constructs expressing S1 with a signal peptide for ER localization [12].
Another observation that may bear on the cell biology and cytotoxicity of PT is that the S1 subunit appears to be proteolytically processed to a lower molecular weight form upon interaction of PT with mammalian cells [14]. This processing was shown to be dependent upon entry of PT into cells and seemed to involve an early endosome function. The size of the processed form of S1 (approximately 22 kDa versus 26 kDa for the full-length S1) suggested that processing may be targeted at a protease-sensitive loop near the C-terminus of S1 that contains primary sites for trypsin and chymotrypsin cleavage [15]. However, evidence for the location of the cellular cleavage site on S1 was not presented. In addition, a link between processing of S1 and activity of PT in cells was not established. Proteolytic processing is a common theme in the activation of bacterial toxins upon interaction with mammalian cells. For example, anthrax toxin, diphtheria toxin, Pseudomonas exotoxin A and shiga toxin are all activated after cleavage by the endogenous eukaryotic protease furin [16], a subtilisin-like protease residing in the secretory pathway of eukaryotic cells [17], or by closely-related proteases [18]. Cholera toxin (CT) and Escherichia coli heat-labile toxin (LT) A subunits are cleaved at a protease-sensitive loop to promote maximal activity [19,20], and CT A subunit was found to be cleaved upon interaction of CT with T84 epithelial cells, by an unidentified protease [21].
In this study we extend the analysis of proteolytic processing of cell-associated S1 and conclude that S1 processing is not essential for the cellular activity of PT.
Results and discussion
Processing and fractionation of S1 in PT-treated CHO cells
In a previous study, 125I-labelled PT was used for analysis of S1 processing in mammalian cells [14]. As an alternative to radiolabeled toxin, we analyzed detergent lysates of cells treated with unlabeled PT to determine whether we could detect S1 processing. Near-confluent Chinese hamster ovary (CHO) cells were treated with PT (20 nM) for 4 h at 37°C, and then cells were washed, recovered by trypsinization and lysed on ice with either Triton X-100 lysis buffer or RIPA lysis buffer. The detergent-soluble and -insoluble fractions were then analyzed by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blotting. As shown in Fig. 1A, we were able to detect processing of S1 by this method, with the majority of cell-associated S1 present as the lower molecular weight form (S1p, approximately 22 kDa, versus 26 kDa for full length S1). This processing event was independent of PT enzymatic activity, since the enzymatically inactive PT-9K/129G (PT*) was similarly processed by CHO cells (Fig. 1B). The kinetics of S1 processing (data not shown) were very similar to those previously reported [14], which, along with the similar size of the processed form, strongly suggested that we were observing the same event as previously studied. Surprisingly, however, the great majority (>80%) of S1 fractionated with the detergent-insoluble pellet material in this assay. Triton lysis at 25°C rather than on ice increased the proportion of S1 that was solubilized, but at least half of the processed form remained in the insoluble pellet material (Fig. 1A). To determine whether S1 fractionation with the detergent-insoluble material represented a potentially interesting feature of its intracellular transport, or merely an artifact of the lysis procedure, we added 500 ng PT to a Triton lysate of untreated CHO cells, incubated this on ice 30 min, centrifuged to separate the mix into soluble and insoluble fractions and analyzed these by SDS-PAGE and western blotting. Almost all (>95%) of the S1 fractionated to the insoluble pellet (Fig. 1C), demonstrating that S1 association with detergent-insoluble material occurs in the lysate and is independent of PT transport within cells. The Triton-insoluble fraction typically contains nuclei and cytoskeletal components [22], so it is possible that PT, or at least S1, has an affinity for one or more of these components.
Effects of protease and cell trafficking inhibitors on S1 processing in CHO cells
We first ruled out the possibility that furin, a protease that cleaves several other bacterial toxins in mammalian cells, is responsible for S1 cleavage, by finding that S1 processing occurs normally in furin-deficient FD11 cells [23] (data not shown). This observation was not surprising since there are no consensus furin cleavage sites in the S1 sequence. In order to determine the catalytic type of protease responsible for S1 processing in CHO cells, we preincubated CHO cells with serine, cysteine, aspartic and metalloprotease inhibitors (following the suggestions of Barrett [24]) before addition of PT, and studied S1 processing by these cells as before (Fig. 2). The broad specificity serine protease inhibitor 3,4-DCI did not significantly inhibit processing (though the inhibitor was somewhat toxic to the CHO cells and reduced the amount of S1 recovered), and neither did the serine protease inhibitor aprotinin (Fig. 2A). The serine protease inhibitor pefabloc SC (BMB/Roche) did have a consistent inhibitory effect on S1 processing, with approximately 45% of the cell-associated S1 in the unprocessed form (versus approximately 10% on average in the absence of inhibitor). Since the other serine protease inhibitors had no effect on S1 processing, the reason for the inhibitory effect of pefabloc SC is unclear, but may be related to its ability to bind covalently to proteins (BMB/Roche). Neither the aspartic protease inhibitor pepstatin nor the cysteine protease inhibitor E-64 had any significant inhibitory effect on S1 processing (Fig. 2A) (nor did the cysteine/serine protease inhibitor leupeptin – data not shown). However, EDTA had a strong inhibitory effect on S1 processing by CHO cells (Fig. 2A,B), with 56% and 74% of cell-associated S1 in the unprocessed form in the presence of 0.5 mM and 1 mM EDTA, respectively. Since the inhibitory effect of a non-specific metal chelator such as EDTA could be due to the cation-dependence of activity of other proteases [24], we also used the metalloprotease inhibitor 1,10-phenanthroline, which has a high affinity for zinc and is considered the most useful inhibitor for metalloproteases [24]. This inhibitor also had a strong inhibitory effect on S1 processing by CHO cells (Fig. 2B), with 61% and 63% of cell-associated S1 in the unprocessed form in the presence of 0.5 mM and 1 mM phenanthroline, respectively. Therefore, we conclude that the cellular protease responsible for S1 processing in CHO cells is most likely a metalloprotease, which presumably resides in the secretory (endocytic) pathway and the identity of which remains to be determined. This is a novel observation in the sense that other bacterial toxins are cleaved by cellular proteases of the subtilisin family (such as furin) [16,23] or by other serine proteases [21], although one report demonstrated that CT activity on several different cell types was blocked by a competitive substrate for metalloproteases [25], suggesting that metalloproteases may also be involved in the cellular activity of other bacterial toxins. However, the possibility remains that multiple proteases, possibly of different classes, are involved in this S1 processing event.
We also determined the inhibitory activity on S1 processing by CHO cells of two inhibitors of cellular trafficking and secretion, bafilomycin A1, which inhibits vacuolar proton ATPase and therefore prevents endosome acidification [26], and BFA, which disrupts the Golgi apparatus [10]. Bafilomycin A1 had a significant inhibitory effect on S1 processing, with 51% of cell-associated S1 in the unprocessed form (Fig. 2B), consistent with the hypothesis that S1 processing occurs in the endosomal compartment of CHO cells and demonstrating a role for endosome acidification in this processing event. However, BFA had no inhibitory effect on S1 processing (Fig. 2B), consistent with previously reported results [14] and with the hypothesis that S1 processing occurs prior to the Golgi apparatus in the putative retrograde trafficking pathway.
Effects of S1 processing inhibitors on cellular activity of PT
Proteolytic processing of bacterial toxins is a common theme in their activation within mammalian cells [16], but whether cellular processing of S1 plays a role in the activity of PT had not been previously addressed. As a preliminary investigation of this question, we sought to determine whether the inhibitors of S1 processing by CHO cells had any effect on the ability of PT to ADP-ribosylate target G proteins in CHO cells. CHO cells were preincubated with inhibitors before addition of PT (1 nM) as before. Controls were cells to which either PT or PT* (which has no ADP-ribosylation activity) was added in the absence of inhibitor. Cells were recovered after 3 h, and lysates were prepared and tested in the ADP-ribosylation assay. In this assay, active PT within cells will ADP-ribosylate available G proteins, so that when a lysate is prepared from these cells and used in an in vitro ADP-ribosylation assay with PT and 32P-labelled NAD, there is no labeling of G proteins in the lysate, since they were already modified by the PT added to the cells. If PT activity within cells is inhibited, then a proportion of the G proteins in the lysate from these cells will be unmodified and therefore labeled in the in vitro reaction with PT. The assay was repeated twice and the result of one experiment is shown in Fig. 2C. The mean percent inhibition for the various inhibitors (in decreasing order of their inhibitory effect) was as follows: pefabloc SC – 95%, BFA – 79%, bafilomycin A1 – 49%, EDTA – 44%, phenanthroline – 28%, 3,4-DCI – 9%, pepstatin – 8%. Therefore there was no strong correlation between the extent of inhibition of S1 processing by the inhibitors and their inhibitory activity on ADP-ribosylation of G proteins by PT in CHO cells. The cell trafficking/secretion inhibitors had a significant inhibitory effect (BFA has previously been shown to inhibit cellular activity of PT [7-9], presumably due to its disruption of PT trafficking in cells, despite its lack of inhibition of S1 processing), but the metalloprotease inhibitors had only a mild inhibitory effect. Of the other protease inhibitors, pefabloc SC had the greatest inhibitory effect, as it did on S1 processing, whereas pepstatin had no significant inhibitory effect on ADP-ribosylation (or S1 processing). To rule out an effect of the inhibitors on the enzymatic activity of PT (independent of its cellular activity), we also performed in vitro ADP-ribosylation assays with PT in the presence of the various inhibitors at the concentrations used on the CHO cells. No significant inhibitory effect was seen with any of the inhibitors (data not shown) with the exception of 3,4-DCI, which was also somewhat toxic to the CHO cells but did not significantly inhibit processing of S1. Altogether, these data are inconclusive with regard to the hypothesis that cellular processing of S1 plays a role in the ADP-ribosylation of G proteins in CHO cells by PT.
Location of processing site on S1
We reasoned that if we could identify the precise processing cleavage site on S1, then we would be able to mutagenize this site, obtain a mutant form of S1 that was resistant to cellular processing, and then determine whether processing was required for cellular activity of PT. The size of S1p compared to unprocessed S1 (approximately 22 kDa versus 26 kDa) suggested a cleavage site close to either the N-terminus or the C-terminus of S1, although processing at both termini was also a possibility. Processing at the N-terminus of S1 seems unlikely in view of the fact that the arginine at amino acid 9 (R9) is crucial to enzymatic activity [27]. A protease-sensitive loop that has primary sites for trypsin and chymotrypsin cleavage [15] is located towards the C-terminus of S1 (amino acids 211–220; Fig. 3) [1]. To help define the location on S1 of processing by CHO cells, we compared the cellular processing of a modified form of PT containing an extension of 9 amino acids at the N-terminus of S1 (PT*-CSP/N) to that of native PT. As shown in Fig. 4A, the processed form of S1 in cells incubated with PT*-CSP/N was slightly larger (approximately 1 kDa) than that of cells incubated with PT, leading to the conclusion that processing occurs towards the C-terminus of S1. The presence of 2 processed forms of S1 in cells incubated with PT*-CSP/N presumably reflects some additional processing of the N-terminal extension, but since each of these forms is larger than that of S1p from native PT, the conclusion is the same. Fig. 4B shows that the major trypsin-digested fragment of S1 is slightly larger than S1p from cellular processing of PT, suggesting that the site on S1 of cellular processing is close to the protease-sensitive loop, but is N-terminal to the trypsin cleavage site (at R218). However, we cannot rule out the possibility of intracellular modification of S1 that would alter its migration on SDS-PAGE gels and complicate the interpretation of this experiment.
We performed extensive site-specific mutagenesis in the region between M202 and R218 of S1 in an attempt to obtain a mutant form of PT that was no longer processed in mammalian cells. However, we were unable to obtain such a mutant by this approach – almost all PT mutants were processed normally, including those with changes at the primary cleavage sites for trypsin (R218A) and chymotrypsin (W215A) (Fig. 5A). Certain changes at residue A203 (A203D, A203R) resulted in very low levels (<1%) of PT secretion by B. pertussis (data not shown) so that we were unable to determine whether processing was affected. However, other changes at this residue (A203G, A203S, A203V) had no effect on PT secretion, processing or activity (data not shown). In an attempt to locate the processing site, we constructed mutant forms of PT with deletions of various stretches of residues in this region. Each of 2 adjacent deletions of 8 residues (Δ203–210, Δ211–218; Fig. 3) resulted in the complete loss of S1 processing in CHO cells (as far as could be determined by western blotting – Fig. 5A), and the same was true of the overlapping 8 residue deletion Δ207–214 (data not shown), strongly suggesting that the processing site is located in this region. However, smaller overlapping deletions of 4 or 5 residues in this region did not disrupt S1 processing (Fig. 5 and data not shown), indicating that no particular sequence in this region served as a specific cleavage site for the processing event. In further support of this idea, 2 PT constructs in which residues 210–218 of S1 were replaced by random amino acid sequence (210–218/R1 and 210–218/R2) were both processed normally (Fig. 5B). Equivalent replacement of residues 203–210 abrogated assembly and secretion of PT by B. pertussis, so we could not make the same assessment for this region. However, two lines of evidence suggested that the processing site may be located in the region of residue A203. First, a PT construct (I-205–206) with an insertion of 8 amino acids between residues Q205 and A206 in S1 was processed to the same size S1p as wild type PT (Fig. 5A), indicating that the processing site was upstream of the insertion (although it may have been within the insertion). Second, mass spectrometry analysis of the processed and unprocessed forms of S1 in lysates from PT-treated CHO cells identified a difference in mass between S1 and S1p (3577.4) most closely corresponding to the theoretical mass of a peptide from R204 to the C-terminus of S1 (3603.9) (data not shown). However, as mentioned above, substitutions at A203 either did not affect processing of S1 or greatly reduced secretion of PT, and substitutions at M202 and R204 also did not affect S1 processing (data not shown), so we were unable to verify this putative cleavage site by mutagenesis. It is also possible that the major processing site is further downstream (within the 211–220 loop region for example) with subsequent additional cleavage occurring during cell lysis, resulting in the apparent size of S1p.
Effect of processing site mutations on PT activity
We determined the effect of several of these deletion and replacement mutations on PT activity using the ADP-ribosylation assay. Cells were incubated with the purified PT construct (2 nM) for 3 h to determine cellular activity. As shown in Fig. 6A, all mutant constructs retained significant activity, relative to the inactive mutant PT*. The two deletion mutants that were not processed in cells were analyzed further and were found to possess 88.5% (Δ203–210) and 78.1% (Δ211–218) of wild type PT cellular activity (Fig. 6B). The in vitro activity of these constructs was also assayed, using a fusion of GST with the C-terminal 20 amino acids of human Giα3 (GST-αC20) as the substrate, with the finding that the Δ203–210 mutant possessed 100% of wild type PT activity while the Δ211–218 mutant had a 26.5% reduction in activity (Fig. 6B). We also found that the 210–218 replacement mutants retained full cellular activity (Fig. 6C). Together these data demonstrate that S1 processing is not essential for PT cellular activity, since the mutants in which processing was apparently abolished retained significant activity. However, this conclusion must remain tentative in the absence of a single substitution mutant that is unprocessed, which would be the best reagent to answer this question. A large deletion of eight amino acid residues may affect other PT-associated properties or may mask the effects of the loss of processing. The stability of these mutant forms of PT may have been altered, though apparently not significantly reduced from the processing assay data (Fig. 5). It is also possible that the number of PT molecules necessary for full activity is low enough to be undetectable in the processing western blot assay. In previous studies, limited trypsin cleavage of PT increased its activation in vitro [15], indicating that S1 processing may play a role in activity, and domains on S1 for interaction with target proteins and enzymatic activity were found to lie within the N-terminal 204 amino acids [28], so the processed form of S1 should retain these functions.
We also tested several strains expressing mutant PT constructs in our mouse infection model of B. pertussis virulence [29], and none of these strains was significantly defective in colonization compared to the parental wild type strain (data not shown). Although these data suggest that S1 processing is not essential for the virulence of B. pertussis, it is possible that S1 could be processed in vivo by alternative proteases absent from CHO cells.
Processing in stable CHO cell transfectants expressing S1
Previously we constructed stable CHO cell transfectants expressing S1 either with a signal peptide (for ER localization) or without one (for cytoplasmic localization), and showed that S1 fully ADP-ribosylated target G proteins in each transfectant [12]. In that study we did not observe significant processing of S1 in whole cell lysates of these transfectants, but we re-examined this issue using Triton X-100 lysis of transfectants and western blotting of the insoluble pellet material. As seen in Fig. 7A, there was no detectable processing of S1 to S1p in either transfectant. Exogenous addition of PT to these cells resulted in EDTA-inhibitable processing of S1 to S1p (Fig. 7B), demonstrating that S1 expression and activity in the transfectants did not prevent S1 processing of exogenously added PT, though the level of processing in the S1+SP transfectant was apparently quite low.
The data from these transfectants do not rule out the possibility that S1 processing may contribute to the trafficking of PT to the ER. However, the data are consistent with the idea that S1 processing is not essential for its activity in mammalian cells, and therefore that translocation of the full-length S1 across the ER membrane occurs. This would distinguish PT from several other bacterial toxins of similar subunit structure, such as cholera toxin, heat-labile toxin and shiga toxin, for which processing of the A subunit is apparently important for cellular activity [16]. The difference may be in the association of the A subunit with the B oligomer. The other toxin A subunits have a relatively long helix that protrudes through a central pore in the B oligomer [30], and cleavage of the A subunit is required to release the enzymatic domain from this complex. S1 has a relatively short helix associated with the B oligomer [1], and therefore cleavage may be unnecessary for its release from the holotoxin complex in the ER, or whichever compartment translocation occurs from.
Conclusions
In this study we have further characterized the cellular processing of the S1 subunit after PT interacts with mammalian cells. Our major conclusion is that this processing event is not essential for PT activity in mammalian cells, based on several lines of evidence: (1) protease inhibitor studies showed a general lack of correlation between S1 processing and PT cellular activity; (2) mutant forms of PT in which S1 processing was apparently abolished retained significant (>75% of wild type) cellular activity; and (3) no S1 processing was apparent in transfected cells expressing active S1. Although we have not definitively ruled out a contribution of S1 processing to the cellular activity of PT due to the imperfect nature of our unprocessed mutants, it is possible that the processing event is completely unrelated to PT cytotoxicity and instead is an irrelevant activity occurring, possibly in lysosomes, on the large majority of the intracellular pool of PT molecules that do not enter the putative retrograde transport pathway to the ER and then on to the cytosolic target proteins.
Methods
Bacterial strains and growth conditions
The B. pertussis strain used in this study was a streptomycin- and nalidixic acid-resistant derivative of W28 (Wellcome). The PT9K/129G (PT*) derivative of this strain was constructed as previously described [27]. The other PT mutant derivatives used in this study were constructed as described below. B. pertussis was grown on Bordet-Gengou agar (Difco) plates containing 15% defibrinated sheep blood and the following antibiotics at the indicated concentrations where necessary: streptomycin 400 μg ml-1, nalidixic acid 20 μg ml-1, gentamicin 10 μg ml-1; or in Stainer-Scholte liquid medium [31] containing heptakis-dimethylcyclodextrin (Sigma). Escherichia coli strains used were DH10B [32] for standard cloning experiments and S17.1 [33] for conjugation with B. pertussis, and these were grown on LB agar plates containing 10 μg ml-1gentamicin where necessary or in LB broth containing 100 μg ml-1ampicillin.
Plasmid and strain construction
Plasmid pJ-PT was obtained by subcloning a 2.3 kb EcoRI-BsrGI fragment (containing the PT S1 and S2 genes) into EcoRI/Acc65I-digested allelic exchange vector plasmid pJHC1 [34]. Mutations were engineered into the S1 sequence of this plasmid, which was confirmed by DNA sequencing, transformed into E. coli S17.1 and introduced into the B. pertussis chromosome by conjugation and allelic exchange as described previously [35]. Deletion, insertion and substitution mutations were constructed by overlap extension PCR [36], and the 210–218/R1 and R2 replacement mutations were constructed by using a degenerate oligonucleotide (CARBON 1097; 5'-GATAAGAGCTCCVNNVNNVNNVNNVNNVNNVNNVNNVNNGCCGGCGAGGCCTCGCC-3' where V = G, A, or C and N = any nucleotide) which was allowed to anneal, extended with DNA polymerase Klenow fragment, digested with SacI and BglI and inserted into SacI- and BglI-digested pJ-PT. The GST-αC20 construct was obtained by inserting annealed complementary oligonucleotides (encoding the C-terminal 20 amino acids of human Giα3) into the plasmid pGEX-2T (Pharmacia) to derive pGEX-αC20. The PT*-CSP/N construct was obtained as previously described [11].
Western blotting
Samples were run on 12% SDS-PAGE gels and transferred to nitrocellulose filters. To detect S1, blocked filters were incubated with S1-specific monoclonal antibody X2X5 or 1C7, followed by peroxidase conjugated anti-mouse IgG secondary antibody (Amersham). Blots were developed using ECL Plus (Amersham) and exposed to X-ray film.
Protein purification
PT and mutant derivatives were prepared from B. pertussis culture supernatants by the fetuin affinity method of Kimura et al. [37], resuspended in PBS and stored at -80°C until use. The protein concentration was determined by BCA assay (Pierce). GST-αC20 protein was purified from a culture of E. coli DH10B containing the plasmid pGEX-αC20. For induction of the fusion protein, the strain was grown in LB to A600 1.0 and then IPTG was added to a concentration of 0.5 mM. 2 h after IPTG addition, cells were centrifuged, lysed in BPER lysis reagent (Pierce), cleared by centrifugation, and passed through a GSTrap column (Amersham-Pharmacia). Fusion protein was eluted in reduced glutathione buffer, dialyzed against PBS and analyzed by SDS-PAGE and BCA assay (Pierce) to determine protein concentration.
Cell lysis and fractionation
CHO cells were grown in 6-well plates to near confluency and then PT was added and incubated for the indicated times at 37°C. Cells were then collected by trypsinization, washed in PBS, resuspended in 50–100 μl of either NET/Triton lysis buffer (150 mM NaCl, 5 mM EDTA, 50 mM Tris, pH 7.4, 0.01% NaN3, and 0.5% Triton X-100) or RIPA buffer (10 mM Tris, pH 7.4, 0.1% SDS, 1% sodium deoxycholate, 1% NP-40, 150 mM NaCl) and incubated 30 min on ice. The lysate was then centrifuged 15 min at 13,000 rpm at 4°C in a microfuge, the supernatant was removed to a fresh tube and the pellet was resuspended in sample buffer. Samples were boiled 5 min and loaded onto an SDS-PAGE gel.
Inhibitors
Inhibitors were added to CHO cells 30 min prior to the addition of PT. For protease inhibitors we followed the suggestions of Barrett [24] to determine the catalytic type of protease involved. Protease inhibitors (BMB/Roche) and concentrations used were: aprotinin (0.15 μM), 3,4-dichloroisocoumarin (3,4-DCI, 1 mM) and 4-(2-aminoethyl)-benzenesulfonyl-fluoride, hydrochloride (pefabloc SC, 1 mM) for inhibition of serine proteases; trans-epoxysuccinyl-L-leucylamido(4-guanidino)butane (E-64, 1 mM) and leupeptin (1 μM) for inhibition of cysteine proteases; pepstatin (1 μM) for inhibition of aspartic proteases; and EDTA (0.5–1 mM) and 1,10-phenanthroline (0.5–1 mM) for inhibition of metalloproteases. Other inhibitors used were BFA (Sigma, 5 μg/ml) and bafilomycin A1 (ICN, 0.5 μM). After SDS-PAGE and western blotting of samples, band intensities were measured by densitometry and used to calculate the extent of inhibition of S1 processing.
Trypsin digestion of PT
PT (100 or 200 ng) was digested with trypsin (Sigma) in a volume of 20 μl at room temperature for 1 h in 50 mM Tris, pH 8, and 2 mM CaCl2. Trypsin was present at 70 μg/ml. Sample buffer was added and the samples were then boiled 5 min before loading onto an SDS-PAGE gel.
Western blotting
Samples were run on 12% SDS-PAGE gels and transferred to nitrocellulose filters. To detect S1, blocked filters were incubated with S1-specific monoclonal antibody X2X5 (3) (a generous gift from Drusilla Burns) or 3CX4, followed by peroxidase conjugated anti-mouse IgG secondary antibody (Amersham). Blots were developed using ECL Plus (Amersham) and exposed to X-ray film.
ADP-ribosyltransferase assays
To determine the cellular activity of PT samples, PT (0.5–2 nM) was added to near confluent CHO cells in 12-well culture plates, and after 3 h at 37°C cells were recovered from plates, washed in PBS and lysed in NET/Triton lysis buffer 30 min on ice. The lysate was then centrifuged 15 min at 13,000 rpm at 4°C and the supernatant was stored at -20°C until the assay. The ADP-ribosylation assay contained, in 25 μl, 0.1 M Tris, pH 7.5, 20 mM dithiothreitol (DTT), 0.5 mM ATP, 1 μM 32P-NAD (specific activity 30 Ci/mmol; NEN), 10 ng PT, and an aliquot of the lysate sample. For assessment of in vitro enzymatic activity of PT samples by ADP-ribosylation assay, reactions contained, in 25 μl, 0.1 M Tris, pH 7.5, 20 mM DTT, 0.5 mM ATP, 1 μM 32P-NAD (specific activity 30 Ci/mmol; NEN), 10–50 ng PT, and 0.5 μg GST-αC20 protein as substrate. The mixture was incubated 90 min at room temperature, sample buffer was added, and the sample was boiled 5 min and loaded onto 15% SDS-PAGE gels. After electrophoresis, gels were fixed, dried and exposed to X-ray film. Band intensities were measured by densitometry and used to calculate the extent of ADP-ribosylation of target proteins.
Mass spectrometry analysis
CHO cells were incubated overnight with PT and then lysates were made using Triton X-100 buffer at room temperature (to maximize the proportion of S1 in the soluble fraction). Purified 1C7 monoclonal antibody was allowed to bind to a PS10 chip (Ciphergen Biosystems) for 2 h at room temperature, which was then washed with PBS. The CHO cell lysates (soluble fraction) were diluted 1:1 with PBS and 100 μl of each lysate was added to a spot on the chip and incubated at 5°C for 4 days. The chip was washed 3 times with PBS containing 0.1% Triton X-100, 3 times with PBS, and twice with 5 mM HEPES (pH 7.0). After the chip was air-dried, matrix (Sinapinic acid) was added to each spot and allowed to dry. Matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry analysis of these samples was performed in a model PBS-II machine (Ciphergen).
Mouse infection
Six-week-old female BALB/c mice (Harlan) were used for infection experiments. Bacterial inocula were prepared and intranasal inoculation of mice was performed as previously described [29]. Seven days after inoculation, mice were sacrificed by carbon dioxide inhalation and the respiratory tract (trachea + lungs) was removed, homogenized in 2 ml PBS, and dilutions were plated on BG-blood agar plates containing streptomycin to determine the number of colony forming units (CFU) per respiratory tract.
Authors' contributions
NHC conceived of the study, performed some experiments, supervised personnel and drafted the manuscript. RMM performed most of the plasmid and strain constructions, protein purification, processing assays and cell culture work. GVA performed the mouse infection experiments, some plasmid and strain constructions, protein purification, processing assays and ADP-ribosylation assays. RDP performed some of the plasmid and strain constructions and processing assays. ZEVW performed the GST fusion constructions and purification, some ADP-ribosylation assays, cell culture work and contributed to supervision of the other personnel.
Acknowledgements
We thank Drusilla Burns for antibodies, Steve Leppla for the FD11 cells, Joe Barbieri for advice on ADP-ribosylation assays, Pat Campbell for performing the Ciphergen analysis, and Ulrike McNamara, Susan Kinnear and Monica Castro for performing some of the initial experiments. This project was supported by NIH grant AI50022.
Figures and Tables
Figure 1 S1 processing and detergent fractionation in CHO cells. (A) Cells were incubated with PT (20 nM) for 4 h followed by lysis with RIPA or Triton X-100 (TX100) buffer on ice or at room temperature (RT). A western blot of the detergent-insoluble (I) and -soluble (S) fractions is shown, with bands corresponding to full length S1 or processed S1 (S1p) indicated. Lane marked C is PT (200 ng) loading control. (B) Western blot of Triton X-100 lysate fractions of CHO cells treated with PT or PT* (catalytically inactive mutant of PT) for 4 h. (C) Western blot of Triton X-100 lysate fractions of untreated CHO cells after addition of 500 ng PT to the lysis mixture.
Figure 2 Effect of inhibitors on S1 processing and PT activity in CHO cells. (A, B) CHO cells were preincubated with the indicated inhibitor before addition of PT (20 nM) and then analyzed for S1 processing as before. Western blots are shown with unprocessed and processed (S1p) forms of S1 indicated. Lanes marked C are PT (100–200 ng) loading controls. DMSO = dimethylsulfoxide (solvent control). Metalloprotease inhibitor concentrations are indicated (mM). 0 = no inhibitor. BafA1 = bafilomycin A1. (C) Effect of inhibitors on ADP-ribosylation of CHO cell G proteins by PT. Autoradiogram of cellular ADP-ribosylation assay after treatment of CHO cells with the indicated inhibitor and either PT or PT* (last lane). Radiolabeled band (approximately 41 kDa) corresponds presumably with Giα2 and Giα3 [9].
Figure 3 Diagram of the S1 protein and the putative cleavage site. The region of investigation between A200 and E221 is highlighted and details of some of the mutant constructs are shown. The line with a Δ symbol indicates the deleted amino acids.
Figure 4 Localization of the processing site on S1. (A) CHO cells were incubated with PT or PT*-CSP/N (PT*/N) for 4 h and analyzed for S1 processing as before. Western blots are shown of the processing samples (proc) or purified protein loading controls (C). (B) Western blot comparing S1p after processing of PT in CHO cells (proc) or trypsin cleavage (tryp) of PT (200 or 100 ng).
Figure 5 Processing of mutant PT proteins in CHO cells. (A) Western blot showing purified protein (lanes marked [C]) or processing of S1 (lanes marked [proc]) after addition of various mutant PT proteins (20 nM) to CHO cells. (B) Western blot showing processing of S1 (lanes marked [proc]) after addition of the 210–218/R replacement mutant PT proteins or PT and PT* controls to CHO cells.
Figure 6 ADP-ribosylation assay of mutant PT constructs. (A) Autoradiogram of cellular ADP-ribosylation assay after treatment of CHO cells with the indicated PT protein. The upper radiolabeled band (approximately 41 kDa) corresponds presumably with Giα2 and Giα3, while the lower band is PT-independent endogenously labeled band often seen in this assay with CHO cell lysates [9]. (B) Graphed data showing the mean (n = 3) cellular activity of the 2 deletion constructs that are not processed (activity of wild type PT was set at 100%) and the in vitro ADP-ribosylation activity of the same proteins. (C) Autoradiogram of cellular ADP-ribosylation assay after treatment of CHO cells with the indicated PT protein, demonstrating the retention of activity of the replacement mutant PT constructs.
Figure 7 S1 processing in CHO cell transfectants expressing S1. (A) Western blot showing lack of processing of endogenous S1 in transfectants expressing S1 in the cytosol (S1-SP) or in the ER (S1+SP). (B) Western blot showing EDTA-inhibitable processing of S1 after exogenous addition of purified PT (20 nM) to these transfectants.
==== Refs
Stein PE Boodhoo A Armstrong GD Cockle SA Klein MH Read RJ The crystal structure of pertussis toxin Structure 1994 2 45 57 8075982 10.1016/S0969-2126(00)00007-1
Tamura M Nogimori L Murai S Yajima M Itio K Katada T Ui M Ishii S Subunit structure of the islet-activating protein, pertussis toxin, in conformity with the A-B model Biochemistry 1982 21 5516 5522 6293544
Katada T Tamura M Ui M The A protomer of islet-activating protein, pertussis toxin, as an active peptide catalyzing ADP-ribosylation of a membrane protein Arch Biochem Biophys 1983 224 290 298 6683482 10.1016/0003-9861(83)90212-6
Moss J Stanley SJ Burns DL Hsia JA Yost DA Myers GA Hewlett EL Activation by thiol of the latent NAD glycohydrolase and ADP-ribosyltransferase activities of Bordetella pertussis toxin (islet-activating protein) J Biol Chem 1983 258 11879 11882 6311827
Brennan MJ David JL Kenimer JG Manclark CR Lectin-like binding of pertussis toxin to a 165 kilodalton Chinese hamster ovary cell glycoprotein J Biol Chem 1988 263 4895 4899 3350815
Witvliet MH Burns DL Brennan MJ Poolman JT Manclark CR Binding of pertussis toxin to eukaryotic cells and glycoproteins Infect Immun 1989 57 3324 3330 2478471
el Baya A Linnermann R von Olleschik-Elbheim L Robenek H Schmidt MA Endocytosis and retrograde transport of pertussis toxin to the Golgi complex as a prerequisite for cellular intoxication Europ J Cell Biol 1997 73 40 48 9174670
Xu Y Barbieri JT Pertussis toxin-mediated ADP ribosylation of target proteins in chinese hamster ovary cells involves a vesicle trafficking mechanism Infect Immun 1995 63 825 832 7868253
Xu Y Barbieri JT Pertussis toxin-catalyzed ADP ribosylation of Gi-2 and Gi-3 in CHO cells is modulated by inhibitors of intracellular trafficking Infect Immun 1996 64 593 599 8550212
Fujiwara T Oda K Yokota S Takatsuki A Ikehara Y Brefeldin A causes disassembly of the Golgi complex and accumulation of secretory proteins in the endoplasmic reticulum J Biol Chem 1988 263 18545 18552 3192548
Carbonetti NH Irish TJ Chen CH O'Connell CB Hadley GA McNamara U Tuskan RG Lewis GK Intracellular delivery of a cytolytic T-lymphocyte epitope peptide by pertussis toxin to major histocompatibility complex class I without involvement of the cytosolic class I antigen processing pathway Infect Immun 1999 67 602 607 9916065
Castro MG McNamara U Carbonetti NH Expression, activity and cytotoxicity of pertussis toxin S1 subunit in transfected mammalian cells Cell Microbiol 2001 3 45 54 11207619 10.1046/j.1462-5822.2001.00092.x
Hazes B Read RJ Accumulating evidence suggests that several AB-toxins subvert the endoplasmic reticulum-associated degradation pathway to enter target cells Biochemistry 1997 36 11051 11054 9333321 10.1021/bi971383p
Finck-Barbançon V Barbieri JT Preferential processing of the S1 subunit of pertussis toxin that is bound to eukaryotic cells Mol Microbiol 1996 22 87 95 8899711
Krueger KM Mende-Mueller LM Barbieri JT Protease treatment of pertussis toxin identifies the preferential cleavage of the S1 subunit J Biol Chem 1991 266 8122 8128 1850738
Gordon VM Leppla SH Proteolytic activation of bacterial toxins: role of bacterial and host cell proteases Infect Immun 1994 62 333 340 8300195
Barr PJ Mammalian subtilisins: the long-sought dibasic processing endoproteases Cell 1991 66 1 3 2070411 10.1016/0092-8674(91)90129-M
Gordon VM Rehemtulla A Leppla SH A role for PACE4 in the proteolytic activation of anthrax toxin protective antigen Infect Immun 1997 65 3370 3375 9234799
Clements JD Finkelstein RA Isolation and characterization of homogeneous heat-labile enterotoxins with high specific activity from Escherichia coli Infect Immun 1979 24 760 769 89088
Mekalanos JJ Collier RJ Romig WR Enzymic activity of cholera toxin. II. Relationships to proteolytic processing, disulfide bond reduction and subunit composition J Biol Chem 1979 254 5855 5861 221485
Lencer WI Constable C Moe S Rufo PA Wolf A Jobling MG Ruston SP Madara JL Holmes RK Hirst TR Proteolytic activation of cholera toxin and Escherichia coli labile toxin by entry into host epithelial cells J Biol Chem 1997 272 15562 15568 9182593 10.1074/jbc.272.24.15562
Solomon F Direct identification of microtubule-associated proteins by selective extraction of cultured cells Meth Enzymol 1986 134 139 147 3821557
Gordon VM Klimpel KR Arora N Henderson MA Leppla SH Proteolytic activation of bacterial toxins by eukaryotic cells is performed by furin and by additional cellular proteases Infect Immun 1997 63 82 87
Barrett AJ Classification of peptidases Meth Enzymol 1994 244 1 15 7845199
de Wolf MJ A dipeptide metalloendoprotease substrate completely blocks the response of cells in culture to cholera toxin J Biol Chem 2000 275 30240 30247 10831601 10.1074/jbc.M004434200
Bowman EJ Siebers A Altendorf K Bafilomycins: a class of inhibitors of membrane ATPases from microorganisms, animal cells, and plant cells Proc Natl Acad Sci USA 1988 85 7972 7976 2973058
Pizza M Covacci A Bartoloni A Perugini M Nencioni L DeMagistris MT Villa L Nucci D Manetti R Bugnoli M Giovannoni F Olivieri R Barbieri JT Sato H Rappuoli R Mutants of pertussis toxin suitable for vaccine development Science 1989 246 497 500 2683073
Krueger KM Barbieri JT Assignment of functional domains involved in ADP-ribosylation and B-oligomer binding within the carboxyl terminus of the S1 subunit of pertussis toxin Infect Immun 1994 62 2071 2078 8168972
Carbonetti NH Artamonova GV Mays RM Worthington ZEV Pertussis toxin plays an early role in respiratory tract colonization by Bordetella pertussis Infect Immun 2003 71 6358 6366 14573656 10.1128/IAI.71.11.6358-6366.2003
Sixma TK Pronk SE Kalk KH Wartna ES van Zanten BA Witholt B Hol WG Crystal structure of a cholera toxin-related heat-labile enterotoxin from E. coli Nature 1991 351 371 377 2034287 10.1038/351371a0
Stainer DW Scholte MJ A simple chemically defined medium for the production of phase I Bordetella pertussis J Gen Microbiol 1970 63 211 220 4324651
Raleigh EA Murray NE Revel H Blumenthal RM Westaway D Reith AD Rigby PW Elhai J Hanahan D McrA and McrB restriction phenotypes of some E. coli strains and implications for gene cloning Nucleic Acids Res 1988 16 1563 1575 2831502
Simon R Priefer U Pühler A A broad host range mobilization system for in vivo genetic engineering: transposon mutagenesis in gram negative bacteria Bio/Technology 1983 1 784 791 10.1038/nbt1183-784
Kinnear SM Boucher PE Stibitz S Carbonetti NH Analysis of BvgA activation of the pertactin gene promoter in Bordetella pertussis J Bacteriol 1999 181 5234 5241 10464192
Stibitz S Black W Falkow S The construction of a cloning vector designed for gene replacement in Bordetella pertussis Gene 1986 50 133 140 2884169 10.1016/0378-1119(86)90318-5
Ho SN Hunt HD Horton RM Pullen JK Pease LR Site-directed mutagenesis by overlap extension using the polymerase chain reaction Gene 1989 77 51 59 2744487 10.1016/0378-1119(89)90358-2
Kimura A Mountzouros KT Schad PA Cieplak W Cowell JL Pertussis toxin analog with reduced enzymatic and biological activities is a protective immunogen Infect Immun 1990 58 3337 3347 2119344
| 15691377 | PMC550662 | CC BY | 2021-01-04 16:03:40 | no | BMC Microbiol. 2005 Feb 3; 5:7 | utf-8 | BMC Microbiol | 2,005 | 10.1186/1471-2180-5-7 | oa_comm |
==== Front
BMC Pulm MedBMC Pulmonary Medicine1471-2466BioMed Central London 1471-2466-5-41572070610.1186/1471-2466-5-4Research ArticleMold sensitization is common amongst patients with severe asthma requiring multiple hospital admissions O'Driscoll B Ronan [email protected] Linda C [email protected] David W [email protected] Department of Respiratory Medicine, Salford Royal Hospitals NHS Trust, Hope Hospital, Salford M6 8HD, UK2 Department of Medicine University of Manchester Clinical Sciences Building Hope Hospital Salford M6 8HD UK2005 18 2 2005 5 4 4 25 2 2004 18 2 2005 Copyright © 2005 O'Driscoll et al; licensee BioMed Central Ltd.2005O'Driscoll 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
Multiple studies have linked fungal exposure to asthma, but the link to severe asthma is controversial. We studied the relationship between asthma severity and immediate type hypersensitivity to mold (fungal) and non-mold allergens in 181 asthmatic subjects.
Methods
We recruited asthma patients aged 16 to 60 years at a University hospital and a nearby General Practice. Patients were categorized according to the lifetime number of hospital admissions for asthma (82 never admitted, 53 one admission, 46 multiple admissions). All subjects had allergy skin prick tests performed for 5 mold allergens (Aspergillus, Alternaria, Cladosporium, Penicillium and Candida) and 4 other common inhalant allergens (D. pteronyssinus, Grass Pollen, Cat and Dog).
Results
Skin reactivity to all allergens was commonest in the group with multiple admissions. This trend was strongest for mold allergens and dog allergen and weakest for D. pteronyssinus. 76% of patients with multiple admissions had at least one positive mold skin test compared with 16%-19% of other asthma patients; (Chi squared p < 0.0001). Multiple mold reactions were also much commoner in the group with multiple admissions (50% V 5% and 6%; p < 0.0001). The number of asthma admissions was related to the number and size of positive mold skin allergy tests (Spearman Correlation Coefficient r = 0.60, p < 0.0001) and less strongly correlated to the number and size of non-mold allergy tests (r = 0.34, p = 0.0005). Hospital admissions for asthma patients aged 16–40 were commonest during the mold spore season (July to October) whereas admissions of patients aged above 40 peaked in November-February (Chi Squared, p < 0.02).
Conclusion
These findings support previous suggestions that mold sensitization may be associated with severe asthma attacks requiring hospital admission.
==== Body
Background
Most asthma patients have mild symptoms which are well controlled with anti-inflammatory and bronchodilator therapy but a minority of asthma patients have severe airway inflammation and airflow obstruction requiring multiple hospital admissions. The reasons for these differences in asthma severity are complex and not fully understood [1]. At least two thirds of asthmatic patients are atopic with skin reactivity to common allergens [2-4]. It has also been reported that individuals with severe ("brittle") asthma may have a greater degree of atopy than other asthmatic patients [5]. Skin reactivity to fungal allergens such as Alternaria species has been reported to be especially common in patients with life-threatening asthma [6]. Asthma deaths, hospital admissions, respiratory symptoms, and Peak Expiratory Flow rates can be adversely affected by high fungal spore concentrations in outdoor air [7-11]. Mold sensitivity has been associated with increased asthma severity and intensive care admissions in adults and with increased bronchial reactivity in children [12-17].
Indoor mold exposure might also contribute to asthma severity. Many patients believe that their asthma is aggravated by damp housing, especially if there is visible mold growth [18]. It has been reported that asthma patients are more likely than control patients to live in a damp dwelling and their asthma severity was correlated to the degree of dampness and mold growth in their home (measured by a building surveyor) [19]. It is also known that asthma deaths in young adults in England and Wales are commonest in the months of July, August and September which coincides with peak levels of mold spores in the outdoor air in the UK [20-22].
These studies prompted us to undertake a study of asthmatic patients in Salford in Greater Manchester (North-West England) to assess whether atopy in general and mold sensitization in particular was associated with increased asthma severity. There are few agreed definitions of severe asthma so we used hospital admissions and treatment category (according to British Thoracic Society Asthma Guidelines) as surrogate markers for asthma severity [23].
Methods
This study was undertaken in Hope Hospital, a 900 bedded University Hospital in Salford, Greater Manchester, UK. The study was designed to record the atopic status for mold and non-mold allergens of asthma patients at every level of severity from very mild to very severe (defined as multiple hospital admissions despite intensive asthma medication). We studied patients with severe asthma during hospital admissions or at subsequent visits to the hospital chest clinic. Patients were recruited opportunistically during hospital admissions or during routine consultations with the Respiratory Nurse Specialist (RNS) in the hospital Chest Clinic. Recruitment was evenly spread over 30 months from January 1996 to June 1998. During this period, the Respiratory Nurse service was under development with only one part-time RNS available to serve a large population of hospital in-patients and ambulatory care patients at chest clinics. Ward-based doctors and nurses referred patients with acute asthma to the RNS on an opportunistic basis based on availability of the part-time RNS and whether the ward teams were aware of the developing RNS service. About 20% of all adult asthma patients admitted during the study period were referred to the RNS and most of these were recruited in the study (provided the RNS had time available to do so). Although recruitment was not systematic (based mainly on availability of RNS service), we were not aware of any potential bias which might have affected the recruitment process and we believe that the recruited patients were typical of adult patients with acute asthma admitted to our hospital. We recruited ambulatory patients (some of whom had previous hospital admissions) at the same hospital chest clinic and in a single Primary Care practice where 25 patients were recruited during routine consultations with the practice nurse. This recruitment was also opportunistic, based mainly on the availability of time to complete the study protocol during busy clinics. We believe that these patients were typical of patients seen at hospital chest clinics and General Practice asthma clinics in this area. Patients were grouped according to their lifetime history of hospital admissions for asthma (multiple admissions, single admission, or no admissions). Patients with no hospital admissions were further categorized according to the treatment steps described in the British Thoracic Society guidelines for asthma management; Step 1 requiring only occasional bronchodilator treatment, Step 2 requiring low-doses (<800 mcg per day) of inhaled steroid, Step 3 requiring high doses of inhaled steroids or long acting beta agonists, Step 4 requiring additional therapy such as domiciliary nebulized therapy [23]. We attempted to recruit approximately equal numbers of patients at BTS steps 1, 2 and 3–4 to allow analysis of mold sensitization according to severity in non-admitted patients as well as in admitted patients.
All patients were Caucasians who were lifelong residents of the United Kingdom. Inclusion criteria were a diagnosis of asthma by the patient's doctor, age 16 to 60 and ability to give informed consent. All subjects gave written informed consent prior to partaking in the study which was approved by the Salford Research Ethics Committee. Exclusion criteria included a diagnosis of COPD, non-European ethnic group (98% of Salford residents are Caucasian) and consumption of any antihistamine in the previous 48 hours. All subjects completed a questionnaire concerning respiratory symptoms, smoking status and allergies.
All tests were conducted by one author (LCH) or by one other respiratory nurse specialist using a standardized technique in an open manner. A small drop of allergen was placed on the volar surface of the forearm. Allergens were purchased from Allergopharma (Reinbek, West Germany), a single batch of each allergen was used throughout the study. We used skin test lancets with a 1 mm tip (Bayer Prick Lancetter supplied by Miles Pharmaceutical Division, Spokane, Washington, USA). The lancet was introduced vertically into the skin through the allergen solution. Allergens studied were: negative control, histamine 0.1%, Dermatophagoides pteronyssinus, cat, dog, mixed grass pollen, Aspergillus fumigatus, Alternaria alternata, Cladosporium herbarum, Penicillium notatum and Candida albicans.
Weal diameter (if any) was recorded at 15 minutes. If a weal was asymmetrical, the mean of two perpendicular measurements was calculated. Weals less than 3 mm greater than the negative control reaction were regarded as negative in accordance with guidance issued by the European Academy of Allergology [24]. We devised an arbitrary numerical "sensitization score" for mold sensitization and non-mold sensitization to compare the number and size of positive allergy tests between groups of patients. This "sensitization score" was the sum of all positive weal diameters for each individual patient after subtraction of the negative control. For example, a patient with a 1 mm reaction to negative control, a 6 mm reaction to Aspergillus and a 5 mm reaction to Cladosporium would have a "mold sensitization score" of (6-1)+(5-1) = 9 mm.
As a subsidiary study (not part of the initial trial protocol), we also studied the seasonality of asthma admissions in Salford by reviewing the electronic records of 520 asthma admissions under the care of two pulmonary physicians who kept a computerized database at this hospital between 1995 and June 2000 (approximately 30% of all adult asthma admissions to the hospital). This covered a wide time-span before, during and after the period of the mold sensitization study; some of the allergy study patients were admitted and recruited during this time but they represented only a small (and random) proportion of the admissions studied for seasonality. Asthma admissions were analyzed by month of admission and divided into 3 four-month "seasons"; March to June (Spring and early summer season with maximum airborne levels of shrub, tree and grass pollens; July to October (late summer and fall season with maximum airborne levels of mold spores); November to February (winter peak of general respiratory infections involving COPD and older asthma patients); [20-22]. Patients were analyzed in two age groups as it is known that asthma admissions and asthma deaths in Britain are commonest in July to September for patients aged under 35 but older patients are more likely to suffer asthma death in winter [20,21]. We used two age bands (16–40 and >40) because our patients aged 35–40 had a seasonal profile for asthma admissions which was identical to the 16–35 group and different to the group aged above 40.
Statistical analysis was performed using Prism II software (GraphPad Prism, San Diego, California, USA). Chi squared test was used to compare the number of positive allergy skin tests between groups of asthma patients. Spearman Correlation Coefficient was used to compare each patients lifetime number of hospital admissions with their "Mold sensitization score" (as described above). Mann Whitney tests were used to compare mean sensitization scores between groups of patients.
Results
One hundred eighty-one asthmatic patients were recruited. Their characteristics are given in table 1. No subject was on antihistamine medication and none had a negative reaction to histamine. No subjects had dermatographism (>3 mm reaction to negative control skin test). No eligible patient declined to partake in the study. There was a predominance of female subjects in all groups. Patients with hospital admissions were more likely to report current smoking than patients with no admissions (p = 0.03). The habit was commoner amongst patients with one admission than those with multiple admissions but this difference was not significant (p = 0.18). Patients with multiple admissions were more likely to have developed their asthma in childhood and they had a stronger family history of asthma.
Table 1 Patient characteristics.
Asthma No Hospital Admissions Asthma One Hospital Admission Asthma >1 Hospital Admission
Number 82 53 46
32 BTS step 1 (Ref 23)
25 BTS step 2
25 BTS steps 3–4
Percent male 35% 25% 46%
Mean Age (Range) 37 36 36
16–59 17–58 16–60
%Smokers 15% 34% 22%
%Ex-Smokers 29% 23% 17%
%Non-Smokers 56% 43% 61%
Asthma onset before age 16 (Percent) 44% 23% 70%
Family history of Asthma in parents, siblings or children (%) 56% 55% 76%
Positive skin tests to all allergens were commoner in the group with severe asthma (multiple hospital admissions) than patients with milder asthma. (Table 2 and Figure 1). Atopic sensitization was common in all groups, especially the severe asthma group This tendency was most marked for dog sensitization (Table 2). Dog ownership was 31% amongst patients with mild asthma (no admissions) and 30% amongst those with multiple admissions (who were more dog-allergic).
Table 2 Prevalence of mold and non-mold sensitization in asthma patients. "Sensitization score" refers to number and size of positive skin tests as defined in methods section.
Asthma No admissions Asthma One Admission Asthma >1 admissions Chi squared p value
Mold allergens
Aspergillus 7% 6% 37% <0.0001
Alternaria 5% 6% 26% <0.0001
Cladosporium 1% 0% 41% <0.0001
Penicillium 2% 4% 30% <0.0001
Candida 10% 9% 33% 0.001
Any mold sensitization 16% 19% 76% <0.0001
>1 mold sensitisation 5% 6% 50% <0.0001
Mean mold sensitization score (95% CI) 0.9 mm 0.4–1.4 0.9 mm 0.3–1.5 6.7 mm 4.8–8.5 Mann Whitney See below
Other allergens
D pteronyssinus 56% 47% 67% 0.13
Grass pollen 46% 38% 63% 0.025
Cat 37% 36% 59% 0.029
Dog 18% 19% 48% 0.005
Any non-mold Sensitisation 70% 47% 74% 0.008
>1 non-mold sensitisation 43% 38% 70% 0.002
Mean non-mold sensitization score (95% CI) 8.6 mm 6.6–10.6 7.0 mm 4.7–9.3 14.5 mm 11.1–17.9 Mann Whitney See below
Mann Whitney Analysis of "Sensitization Scores" between different categories of asthmatic patients.
Mold sensitization score:
No admission V single admission P = 0.87
No admission V multiple admissions P = < 0.0001
Single admission V multiple admissions P = < 0.0001
Non- Mold sensitization score:
No admission V single admission P = 0.29
No admission V multiple admissions P = < 0.009
Single admission V multiple admissions P = 0.002
Figure 1 Mean mold sensitization scores and mean non-mold sensitization scores for asthma patients and controls (mean and 95% CI) Clear bar: 82 asthma patients with no hospital admissions. Grey bar : 53 patients with one hospital admission Striped bar: 46 patients with more than one hospital admission.
Mold sensitization was uncommon in mild asthma but very common in asthma patients with multiple admissions (Figure 1 and Table 2). There was a striking difference in the prevalence of mold sensitization amongst the three asthma groups. Three quarters of patients with multiple hospital admissions were sensitized to molds and half of them reacted to multiple mold allergens. The patients with a single hospital admission were more similar to those with no admissions than to the multiple admission group. This trend was seen for all five mold allergens studied (table 2). The frequency of sensitization to any individual mold ranged from 26% (Alternaria) to 41% (Cladosporium) in the severe asthma group compared with 0–10% in the milder asthma groups.
Aspergillus and Candida precipitins and specific IgE were not measured in this study. None of these patients had any clinical features suggestive of allergic bronchopulmonary aspergillosis (ABPA) such as pulmonary infiltrates, bronchiectasis or marked eosinophilia. The number of admissions correlated with the number and size of positive skin tests using the scoring system described previously. For mold sensitization, the Spearman Correlation Coefficient was 0.60 (two-tailed p < 0.0001) and for non-mold allergens was 0.34 (two-tailed p = 0.0005). The cumulative "mold sensitization score" and the "non-mold sensitization score" for each group of patients is shown in table 2. Only two of the 99 patients with asthma admissions had ever required admission to an Intensive Care Unit. Both were sensitized to a single mold (one Aspergillus, one Penicillium).
Of the patients not admitted to hospital, 32 had very mild asthma (BTS Step 1), 25 had mild-moderate asthma (BTS Step 2) and 25 had moderate to severe asthma (BTS Steps 3–4). There was no significant difference in mold or non-mold sensitization between these groups of non-admitted patients with different grades of asthma severity.
Our review of asthma admissions to this hospital between 1995 and 2000 identified 520 patients admitted under the care of the two chest physicians who kept a computerized database (approximately 30% of all asthma admissions to the hospital). There were 173 asthma admissions in the 16–40 age group, these admissions peaked in late summer and fall (figure 2). Of these admissions 24.3% occurred between March and June, 43.4% between July and October and 32.4% between November and February. By contrast, 347 asthma patients aged over 40 had a winter peak of admissions (30.3% March-June, 30.5% July-Oct, 39.2% Nov-Feb) These patterns of admissions were significantly different (Chi squared p < 0.02). The summer-fall peak in the 16–40 age group amounted to 33 additional admissions above the spring baseline. This represented 6.4% of all asthma admissions or 19.1% of admissions in the 16–40 age group.
Figure 2 Asthma admissions aggregated by "season", comparing 16–40 age group (black bars) with age >40 (white bars).
Discussion
Although the present study was larger than most previous studies of mold sensitization in severe asthma, it must be regarded as a "pilot study" due to the non-systematic recruitment of asthma patients and the cross-sectional nature of the study. Our data indicate that mold (and dog) sensitization is common in patients with severe asthma requiring multiple hospital admissions in Manchester. The results are consistent with previous evidence that atopy (especially to mold allergens) is related to asthma severity or bronchial hyper-reactivity [4-6,12-17]. A recent cross-sectional study of 1132 adults with asthma found that sensitization to Alternaria or Cladosporium is a powerful risk factor for severe asthma[16] in several European countries and also in Australia, New Zealand and in Portland, Oregon. The link between dog sensitization and asthma severity is in agreement with previous studies [5,14]. We had also expected to find an excess of house dust mite (D. pteronyssinus) sensitization in our patients with more severe asthma [15,25]. However, reactivity to this allergen was common in all asthma groups and only slightly commoner in patients with multiple admissions.
There has been some debate about the best cut-off point for weal size to define a positive skin-prick test. We accepted the European Academy of Allergology figure of 3 mm greater than the negative control [24]. However, re-analysis of our data using a 2 mm or 4 mm difference from the negative control would make no difference to the results. Furthermore, the number and size of positive skin tests to mold allergens was greatest in patients with a high number of admissions suggesting that the relationship is a genuine one. It was also notable that, although sensitization to non-mold allergens was common in the informal control population and in patients with mild asthma, mold sensitization was uncommon in these groups. This indicates that the positive skin tests to mold allergens are unlikely to be due to irritant reactions of a non-allergic nature. As skin test reagents from different manufactures are not standardized, different results might be obtained with different manufacturers' reagents. Until such antigens are standardized, this remains unsatisfactory. However, the consistency between the present study and the recent European Community respiratory health survey (using different antigens) supports the validity of the association between mold sensitization and severe asthma.
A key question is whether severe asthma is actually caused by sensitivity to molds or is simply associated with it. In any case, mold sensitivity will certainly not be the only cause of severe attacks of asthma; upper respiratory tract virus infections and some drugs being two of other well documented causes. The greater degree of mold sensitization in the severe asthma group could simply reflect an extreme example of the generalized increase in atopy amongst this group. However, we believe that mold allergy may be responsible for severe asthma attacks for several reasons.
First, the temporal relationship between high environmental spore counts and asthmatic attacks is strong. Airborne spore levels may be up to 1000 times higher than pollen levels [26]. The data of Targonski and colleagues provide strong evidence that asthma deaths in Chicago are more likely to occur on days when local mold spore counts are high [7]. High mold spore counts have been associated with asthma admissions in New Orleans (adults) and in Derby, UK (adults and children) [10,11]. Asthma symptoms are increased in California and Pennsylvania on days when mold spore counts are high [8,9]. The young patients of O'Hollaren and colleagues who were Alternaria-sensitive had their near-fatal asthma episodes in summer and early fall when mold spore levels would be expected to be high [6].
Second, the seasonal (summer -fall) peak of asthma admissions occurs when ambient air counts of molds are high. We have documented a late summer-fall peak of asthma admissions involving young adults in Manchester which coincides with the summer-fall peak of asthma deaths in UK patients aged under 35 years [20,21]. These asthma admissions also coincide with the peak months for outdoor levels of fungal spores [22,27]. Although there is no aero-biology service in Manchester, data for surrounding towns have shown a consistent summer-fall peak in mold spore counts. In Cardiff, for example, a city 150 miles south-west of Manchester with a similar climate, the highest spore counts were measured in late summer and fall [22]. The Cardiff authors reported maximal levels of Cladosporium in July, Alternaria and hyaline basidiospores in August, uredospores in September and coloured basidiospores in October. The data from Derby (52 miles south-east of Manchester) are similar[11]. In addition, similar findings have been reported from Copenhagen (600 miles north-east of Manchester) where 87% of the microfungal flora in outdoor air is accounted for by Cladosporium, Alternaria, Penicillium and Aspergillus with maximal levels between June and October [27]. A pan-European study with centres in Oregon, USA, Australia and New Zealand involving questionnaires and skin prick tests in 17,000 patients identified 1132 patients with asthma the severity of which could be determined [16]. Sensitisation to A. alternata and C. herbarium was common and associated with asthma severity – OR of 2.03 for the former, of 3.2 for the latter fungus and 2.34 for both. No such association was found for pollens or cats, although sensitization to house dust mite was slightly more frequent in those with severe asthma (OR 1.61) [16]. The present study extends these findings to a wider range of fungal allergens and a greater degree of asthma severity.
Third, there is evidence that indoor mold exposure may contribute to asthma severity. Many patients report respiratory symptoms in damp and moldy houses and a review of nine population-based studies found that seven reported one or more positive associations between fungal levels and health outcomes [28]. The study of Williamson and colleagues in Scotland reported that asthmatic patients were more than twice as likely than control patients to live in a house that was considered damp or moldy by a building surveyor [19]. Furthermore, in that study, there was a positive association between a patient's asthma severity and the degree of dampness which the surveyor measured in the patient's home (r = 0.3, p = 0.006) and independently with an index of visible mold growth in their dwelling (r = 0.23, p = 0.035). In a study of German children, it was found that bronchial hyper-reactivity was associated with damp housing [29] that was only partly explained by exposure to house dust mite antigen, suggesting that other factors such as mold growth may also be important. Taskinen et al found that the prevalence of asthma was similar (4.8%) amongst children attending a school with moisture and mold problems compared with a control school but asthma symptoms such as wheeze and cough were commoner in the damp moldy school as were emergency visits to hospital (OR 2.0, p < 0.01) [30].
Fourth, we know that Aspergillus in particular is a major respiratory allergen causing the vast majority of cases of allergic bronchopulmonary mycosis [31-33]. It is likely that these cases represent the extreme of a spectrum of mold allergy, the slightly less severe manifestation of which is severe asthma as described in this study and without all the serological and radiological markers characterisitic of ABPA. The reactivity of asthmatic patients to multiple mold allergens could be due to genuine sensitization to a variety of molds or it could be due to cross-reactivity between mold allergens. The paper of Hemmann and colleagues suggests that Aspergillus and Candida allergens may share IgE-binding epitopes [34]. However, it is believed that multiple mold sensitization skin test reactions are usually due to sensitivity to multiple antigens rather than cross reactivity [25]. Few fungi of the >1 million species of fungi thought to exist worldwide have been subjected to the antigenic scrutiny that Aspergillus and a few other common airborne fungi have and it is likely that sensitization to other fungi will be discovered in the future.
It is not known why mold allergens should produce more severe airway disease than other common allergens such as house dust mites, cat dander, or grass pollen. Fungi are very common in the environment and Candida is present in the gut of most, if not all, humans. The difference may relate to the nature or intensity of exposure to mold allergens or to their ability to become airborne and to gain entry to human airways due to their small size. Also many potent allergenic proteins have been described in Aspergillus and some in other fungi [35]. There is the probability that some fungal antigens, such as Asp f6 being a manganese dependant superoxide dismutase which is closely related to the human enzyme, might set up a self perpetuating allergic response, which is aggravated every time Aspergillus is inhaled, which is almost hourly. Some fungal antigens are proteases (Asp f5, Asp f10, Asp f13, Asp f15, Asp f18) and as DP1 is also a protease – a similar pathogenic role can be postulated [35,36].
It therefore seems likely that there is a causal relationship between mold allergy and asthma severity for some younger asthma patients. Our seasonal admissions data suggest that up to 6% of adult asthma admissions in Manchester and 19% of asthma admissions in the 16–40 age group may be attributable to mold allergy. These unlucky individuals seem to have more severe asthma than patients with sensitization to other common allergens and an increased risk of fatal or near-fatal asthma or hospital admissions during the mold spore season, especially on days when the local mold spore count is high. The risk may be further increased if the patient lives in a damp, moldy house or attends a damp moldy school [19,29,30,37]. It is not yet known which mold species are most important in causing such reactions or whether indoor or outdoor mold exposure is more important {outdoor levels are usually higher}[38]. It is very difficult to make accurate measurements of indoor mold exposure and most studies have used surrogate markers such as dampness or visible mold growth. This phenomenon needs to be studied further in a variety of geographic locations with different climates. It is not yet known if environmental modification or the use of airway protection would be of any value in the management of mold-allergic asthma patients.
One of the fungi which we investigated and for which antigenic extracts are available, Candida albicans, is a yeast, does not become airborne but does produce hyphae in tissue. Thus it is possible that human asthma due to fungal allergens may have three sources of allergen exposure – outdoor mold spores, indoor mold spores and endogenous fungal growth on body surfaces including the skin and gut [33].
Fungal sensitization is relatively uncommon in our British asthma patients and in Finnish schoolchildren [30] compared with Arizona [12] and Australia where up to 31% of asthmatic children and up to 23% of non-asthmatic controls react to at least one fungal allergen [15,17,37]. The prevalence of Alternaria sensitization in Italian patients with respiratory symptoms ranges from 2% in Northern Italy to 29% in Southern Italy [39]. Some of these differences are probably due to difficulties in the standardization of mold allergen extracts or skin testing techniques [26]. Also fungal sensitization is commoner in children and declines with age. However, it is likely that the significance of a positive skin test to fungal allergens varies in different climatic zones. Our data and those of O'Hollaren, Sureik and Black [6,16,17] suggests that fungal skin sensitization tests may identify adults who are at risk of especially severe asthma. The difference in the prevalence of mold sensitization between patients with multiple asthma admissions and the other groups in our study was so striking that it is extremely unlikely to have occurred by chance. Mold allergy tests may be useful to screen for children and adults who are at greatly increased risk of developing severe or fatal asthma. A large prospective study will be required to confirm these preliminary findings.
Conclusion
The findings of this study support previous suggestions that mold sensitization may be associated with severe asthma attacks requiring hospital admission.
Abbreviations
BTS Guidelines = British Thoracic Society Guidelines for Asthma Management, RNS = Respiratory Nurse Specialist
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
ROD originated and co-ordinated the study and contributed to the analysis of the data and preparation of the paper.
LCH contributed to the design of the study and was the main clinical investigator. She also contributed to the analysis of the data and preparation of the paper.
DWD contributed to the design of the study and contributed to the analysis of the data and preparation of the paper.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We are grateful to Mrs Melanie Bainbridge RGN for assistance with allergy skin tests and to Rob Taylor PhD and Andy Vail PhD for assistance with data handling and statistical analysis
==== Refs
Barnes PJ Woolcock AJ Difficult asthma Eur Respir J 1998 12 1209 18 9864023 10.1183/09031936.98.12051209
Kalliel JN Goldstein BM Braman SS Settipane GA High frequency ofatopic asthma in a pulmonary clinic population Chest 1989 96 1336 40 2582840
Herbert FA Weimer N Salkine ML RAST and skin tests in the investigation of asthma Ann Allergy 1982 49 311 14 7149346
Burrows B Martinez FD Halonen M Barbee RA Cline MG Association of asthma with serum IgE levels and skin test reactivity to common allergens N Engl J Med 1989 320 271 277 2911321
Miles J Cayton R Ayres J Atopic status in patients with brittle and non-brittle asthma: a case- control study Clint Exp Allergy 1995 25 1074 82
O'Hollaren MT Yunginger JW Offord KP Somers MJ O'Connell EJ Ballard DJ Sachs MI Exposure to an aeroallergen as a possible precipitating factor in respiratory arrest in young patients with asthma N Engl J Med 1991 324 359 63 1987459
Targonski PV Persky VW Ramekrishnan V Effect of environmental molds on risk of death from asthma during the pollen season J Allergy Clin Immunol 1995 95 955 61 7751516
Delfino RJ Zeiger RS Seltzer JM Street DH Matteucci RM Anderson PR Koutrakis P The effect of outdoor fungal spore concentrations on dailyasthma severity Environ Health Perspect 1997 105 622 35 9288497
Neas LM Dockery DW Burge H Koutrakis P Speizer FE Fungus spores, Air Pollutants and Other Determinants of Peak Expiratory Flow Rate in Children Am J Epidemiol 1996 143 797 807 8610690
Salvaggio J Seabury J Schoenhardt E New Orleans Asthma V. Relationship between Charity Hospital admission rates, semiquantitative pollen and fungal spore counts and total particulate aerometric sampling data J Allergy Clin Immunol 1971 48 96 114 5283607
Newson R Strachan D Corden J Millington W Fungal and other spore counts as predictors of admissions for asthma in the Trent region Occup Environ Med 2000 57 786 92 11024204 10.1136/oem.57.11.786
Halonen M Stern DA Wright AL Taussig LM Martinez FD Alternaria as a major allergen for asthma in children raised in a desert environment Am J Respir Crit Care Med 1997 155 1356 61 9105079
Martinez FD Progression of asthma from childhood to adolescence Eur Respir Rev 1997 40 8 10
Nelson HS Szefler SJ Jacobs J Huss K Shapiro G Sternberg AL The relationships among environmental allergen sensitization, allergen exposure, pulmonary function and bronchial hyperresponsivenes in the Chidhood Asthma Management Program J Allergy Clin Immunol 1999 104 775 85 10518821
Peat JK Tovey E Mellis CM Leeder SR Woolcock AJ Importance of house dust mite and Alternaria allergens in childhood asthma: an epidemiological study in two climatic regions of Australia Clin Exp Allergy 1993 23 812 20 10780887
Zureik M Neukirch C Leynaert B Liard R Bousquet J Neukirch F Sensitisation to airborne moulds and severity of asthma: cross sectional study from European Community respiratory health survey BMJ 2002 325 411 5 12193354 10.1136/bmj.325.7361.411
Black PN Udy AA Brodie SM Sensitivity to fungal allergens is a risk factor for life-threatening asthma Allergy 2000 55 501 504 10843433 10.1034/j.1398-9995.2000.00293.x
Platt SD Martin CJ Hunt SM Lewis CW Damp housing, mould growth, and symptomatic health state B M J 1989 298 1673 78
Williamson IJ Martin CJ McGill G Monie RD Fennerty AG Damp housing and asthma; a case-control study Thorax 1997 52 229 34 9093337
Khot A Burn R Seasonal variation and time trends of deaths from asthma in England and Wales 1960–82 B M J 1984 289 233 4
Jarvis D Burney P The epidemiology of allergic disease Brit Med J 1998 316 607 10 9518918
Jenkins PF Mullins JK Davies BH Williams DA The possible role of aero-allergens in the epidemic of asthma deaths Clin Allergy 1981 11 611 20 7333005
British Thoracic Society The British Guidelines on Asthma Management: 1995 Review and Position Statement Thorax 1997 52 S1 S21
Sub-Committee on Skin Tests of the European Academy of Allergology and Clinical Immunology Skin tests used in type 1 allergy testing; Position paper Allergy 1989 44 1 59 2683837
Sporik R Chapman MD Platts-Mills TAE House dust mite exposure as a cause of asthma Clin Exp Allergy 1992 897 906 1464045
Salvaggio J Aukrust L Mold-induced asthma J All Clin Immunol 1981 68 327 46 10.1016/0091-6749(81)90131-7
Larsen LS A three-year survey of microfungi in the air of Copenhagen 1977–79 Allergy 1981 36 15 22 7224107
Verhoeff AP Burge HA Health risk of fungi in home environments Ann Allergy Asthma Immunol 1997 78 544 54 9207717
Nicolai T Illi S von Mutius E Effect of dampness at home in childhood on bronchial hyperreactivity in adolescence Thorax 1998 53 1035 40 10195075
Taskinen T Hyvarinen A Meklin T Husman T Nevalainen A Korppi M Asthma and respiratory infections in school children with special reference to moisture and mold problems in the school Acta Paediatr 1999 88 1373 79 10626525 10.1080/080352599750030112
Greenberger PA Allergic bronchopulmonary aspergillosis J Allergy Clin Immunol 1984 74 645 53 6438210
Akiyama K Mathison DA Riker JB Greenberger PA Patterson R Allergic bronchopulmonary candidiasis Chest 1984 85 699 701 6370621
Ward GW Karlsson G Rose G Platts-Mills TAE Trichophyton asthma: sensitization of bronchi and upper airways to dermatophyte antigen Lancet 1989 1 589 62 2564114
Hemmann S Blaser K Crameri R Allergens of Aspergillus fumigatus and Candida boidinii share IgE-binding epitopes Am J Respir Crit Care Med 1997 156 1956 62 9412580
Kurup VP Cremeri R Aspergillus antigens Posted January 13th 2001
Kauffman HF Tomee JF van de Riet MA Timmerman AJ Borger P Protease-dependent activation of epithelial cells by fungal allergens leads to morphologic changes and cytokine production J Allergy Clin Immunol 2000 105 1185 93 10856154 10.1067/mai.2000.106210
Garrett MH Rayment PR Hooper MA Abramson MJ Hooper BM Indoor airborne fungal spores, house dampness and associations with environmental factors and respiratory health in children Clin Exp Allergy 1998 28 459 67 9641573 10.1046/j.1365-2222.1998.00255.x
Verhoeff AP van Wijnen JH Brunekreef B Fischer P van Reenen-Hoekstra ES Samson RA Presence of viable mould propagules in outdoor air in relation to house damp and outdoor air Allergy 1992 47 83 91 1632482
Corsico R Cinti B Feliziani V Gallesio MT Liccardi G Loreti A Lugo G Marcucci F Marcer G Meriggi A Minelli M Gherson G Nardi G Negrini AC Piu G Passaleva A Pozzan M D'Ambrosio FP Venuti A Zanon P Zerboni R Prevalence of sensitization to Alternaria in allergic patients in Italy Ann Allergy Asthma Immunol 1998 80 71 76 9475571
| 15720706 | PMC550663 | CC BY | 2021-01-04 16:30:12 | no | BMC Pulm Med. 2005 Feb 18; 5:4 | utf-8 | BMC Pulm Med | 2,005 | 10.1186/1471-2466-5-4 | oa_comm |
==== Front
BMC BiotechnolBMC Biotechnology1472-6750BioMed Central London 1472-6750-5-51568924110.1186/1472-6750-5-5Methodology ArticleAssembly of a gene sequence tag microarray by reversible biotin-streptavidin capture for transcript analysis of Arabidopsis thaliana Wirta Valtteri [email protected] Anders [email protected] Morten [email protected] Peter [email protected] Pierre [email protected]én Mathias [email protected] Rishikesh P [email protected] Joakim [email protected] Department of Molecular Biotechnology, KTH-Royal Institute of Technology, AlbaNova University Center, SE-106 91, Stockholm, Sweden2 Department of Plant Systems Biology, VIB – Ghent University, B-9052 Ghent, Belgium3 Department of Forest Genetics and Plant Physiology, The Swedish University of Agricultural Sciences, SE-901 83, Umeå, Sweden2005 3 2 2005 5 5 5 20 8 2004 3 2 2005 Copyright © 2005 Wirta 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
Transcriptional profiling using microarrays has developed into a key molecular tool for the elucidation of gene function and gene regulation. Microarray platforms based on either oligonucleotides or purified amplification products have been utilised in parallel to produce large amounts of data. Irrespective of platform examined, the availability of genome sequence or a large number of representative expressed sequence tags (ESTs) is, however, a pre-requisite for the design and selection of specific and high-quality microarray probes. This is of great importance for organisms, such as Arabidopsis thaliana, with a high number of duplicated genes, as cross-hybridisation signals between evolutionary related genes cannot be distinguished from true signals unless the probes are carefully designed to be specific.
Results
We present an alternative solid-phase purification strategy suitable for efficient preparation of short, biotinylated and highly specific probes suitable for large-scale expression profiling. Twenty-one thousand Arabidopsis thaliana gene sequence tags were amplified and subsequently purified using the described technology. The use of the arrays is exemplified by analysis of gene expression changes caused by a four-hour indole-3-acetic (auxin) treatment. A total of 270 genes were identified as differentially expressed (120 up-regulated and 150 down-regulated), including several previously known auxin-affected genes, but also several previously uncharacterised genes.
Conclusions
The described solid-phase procedure can be used to prepare gene sequence tag microarrays based on short and specific amplified probes, facilitating the analysis of more than 21 000 Arabidopsis transcripts.
==== Body
Background
Extensive transcriptional profiling of the plant model system Arabidopsis thaliana has been limited when compared to other model organisms, such as human and mouse, mainly due to the lack of high-quality cDNA microarrays offering genome-wide coverage. However, during the recent years both academic and commercial alternatives to these cDNA arrays have emerged. The public initiative by the CATMA consortium has aimed at the production of high-quality probes for each of the 29 787 genes predicted in the Arabidopsis genome [1,2]. The design of the CATMA gene sequence tag (GST) probes is based on de novo gene prediction from the genome sequence [1,3,4], since only a relatively now number of ESTs are available for Arabidopsis (dbEST at NCBI contains only about 320 000 Arabidopsis ESTs compared with 6 million for the human species). Commercial alternatives for genome-wide monitoring of the Arabidopsis transcriptome have been developed by Affymetrix, Agilent Technologies, MWG Biotech, Operon and others. In a recent study the CATMA, Affymetrix and Agilent arrays were found to perform equally, but with a minor advantage for the CATMA arrays in terms of dynamic range [5].
In the first phase of the CATMA program 21 120 GSTs covering more than 70% of the predicted genes were designed. The length of the probes is kept low to ensure specificity and ranges from 150 bp to 500 bp, with the size distribution heavily shifted towards the shorter fragments. To further increase the specificity of the GSTs their distribution is shifted towards the 3'-end of the genes, with 60%, 16% and 24% representing 3'-, centre and 5'-regions, respectively [1,3]. Both 3' and 5' untranslated regions of the genes were included in the design. As a consequence of the GST fragment length, an efficient and robust high-throughput method for purification of short fragments is needed. Here we demonstrate that the purification can be accomplished by taking advantage of the recent finding that the streptavidin-biotin bond can be broken, in a fully reversible fashion, without denaturation of the protein [6]. The approach is based on incorporation of a biotin molecule during PCR amplification of the GSTs, binding of the products to streptavidin-coated paramagnetic beads using high ionic-strength conditions and elution through disruption of the streptavidin-biotin bond in a non-denaturing and fully reversible fashion with deionised water. We exemplify that this feature can be applied for generation of high-quality gene sequence tag microarrays in a cost-effective and high-throughput manner. We also demonstrate the use of these arrays by presenting results on the alteration in gene expression levels at different time points in Arabidopsis plants treated with physiological concentrations of the well-known plant hormone indole-3-acetic acid (auxin). Finally, we compare our results with those obtained in two previous studies [7,8] carried out on the Affymetrix 8 k Gene Chip platform to identify auxin regulated gene expression.
Results and discussion
In this study we present a method suitable for purification of gene sequence tags, which have recently been designed and successfully used for transcriptional profiling of the plant model system Arabidopsis thaliana [1,3]. The purification method is based on reversible biotin-streptavidin binding, utilises streptavidin-coated paramagnetic beads and can be automated on a robotic workstation dedicated for magnetic separation and equipped with a temperature control [6]. We exemplify the performance of the method by studying the purification of three representative biotinylated amplification products and subsequently show that arrays prepared using this method can successfully be used for large-scale transcriptional profiling.
The amplification products we use to study the purification process are 500 bp, 1 kb and 1.3 kb in length, covering the size range typically used for probes on cDNA arrays. For successful purification of the probe both efficient capture by the beads and release is important. An example of the capture of a 50-μl PCR product and subsequent release is shown in Figure 1A. As shown, the initial capture and release is highly efficient (upper left panel), and with no product detectable in the eluate corresponding to the second release (lower left panel). Next we analyse the efficiency of the capture reaction using an increasing amount of beads while keeping the amount of PCR product and length of incubation time constant. The results indicate that for a standard 50-μl PCR product highly efficient capture is achieved already at approximately 100 μg of beads for all products up to 1.3 kb (Figure 1B), but for a highly optimised amplification reaction a higher amount of beads may be necessary (data not shown). As expected, the binding of the biotinylated product to the streptavidin moiety is a rapid and efficient process with the majority of the binding taking place during the first minutes of incubation (Figure 1C). Shorter products appear to have faster binding kinetics reaching saturation at earlier time points. Also important to note is that the molar amount of captured and eluted product is not equal for the different-sized products, indicating that other factors such as steric hindrance also contribute to capacity of the beads and should be considered for purification of longer products > 1000 bp.
Repeated use of the magnetic beads after a single round of capture and elution is the key feature of the described strategy. To investigate the cross-contamination between iterative cycles of purification as well as the total number of bead purifications that can be used without significant loss in performance, we used agarose gel electrophoresis, DNA Lab-on-chip technology as well as a more sensitive approach based on printing of eluates onto glass slides and hybridisation using a fluorescently labelled oligonucleotide complementary to the purified and printed probes. A carry-over free purification requires that all captured product is released at the first elution so that no product is transferred to the next sample to be purified using the same set of beads. We analysed the presence of cross-contamination by analysing the eluates of two consecutive release reactions from a single immobilised product (Figure A, left panels). Furthermore, the cross-contamination issue was analysed by using the reversible beads in six sequential capture reactions containing either a PCR product or a water-only control in an alternating order (Figure A, right panel). A released product is detected in the first eluate, as expected, but not in the second, by all three methods including the sensitive fluorescence assay. As shown in Figure 1A, right panel, hybridisation with a labelled oligonucleotide complementary to the purified probes shows a signal in features originating from a PCR product of the three sizes, but not in features originating from the negative water-only control.
We studied the capacity of the beads after multiple capture and release cycles by using a constant amount of the three PCR products as input for each iterative purification cycle. An extra washing step was carried out between the purification cycles. Data for nine consecutive binding, washing, elution and regeneration rounds of the PCR products is presented in Figure 1D. The yields of purified products are similar for six rounds of reuse with a minor decline during the subsequent cycles, which more likely correlates with loss of beads during the washing steps than with reduced capacity. We continued to analyse the efficiency of the bead regeneration and reuse by using amplification products of twelve additional clones (range 0.3 – 2 kb) and a hybridisation-based quantification approach. The clones were amplified, purified using beads reused up to nine times, printed onto glass slides and finally hybridised with a DNA-binding dye to determine the amount of purified product. A clone-wise scaling of the hybridisation signal of each subsequent reuse versus the signal corresponding to the first use was carried out, followed by a calculation of the overall average, which is shown as the solid black line in Figure 1D. The results from the quantification through hybridisation are in close agreement with the pattern observed using the probes discussed in more detail above.
Assembly of the Arabidopsis gene sequence tag microarray
We applied the described method for purification of 21 120 Arabidopsis biotinylated gene sequence tags (GSTs), with sizes ranging from 150 to 500 bp. The use of GSTs in transcript profiling offers improved specificity when compared to the more common EST or cDNA libraries since each GST has been designed to have minimal cross-hybridisation to other genes, including members of the same gene family. The investigated set of GSTs covers approximately 70% of the genes in the genome, as described in more detail by the CATMA consortium [1]. The consortia amplification strategy is based on a two-step PCR system. This facilitates, as shown in this study, the incorporation of a biotin label in the second PCR by generic handle sequences introduced at the initial amplification step. This circumvents the need to design individual gene-specific biotinylated primers. The products were purified in an automated fashion onto 200 micrograms of magnetic, streptavidin-coated beads that were reused up to six times. To compensate for the higher molar amount of the GST amplification products, an initially higher amount of beads was used for the purification of the GSTs than was used for the optimisation of the method. After elution with 12 μl water an equal amount of DMSO is added to eluted products, which are then printed onto the glass slides.
Changes in gene expression caused by auxin treatment
The arrays generated by large-scale purification of GSTs are used in a pilot time-point study where the plant hormone indole-3-acetic acid (IAA), also known as auxin, is used to cause transcriptional changes in Arabidopsis thaliana seedlings. Total RNA is collected at three post-treatment time points and compared, using a reference design, to RNA extracted from untreated plants. A general overview of the data is shown in Figure 2. Using the filtered and normalised data (for details see Materials) genes which are differentially expressed upon auxin treatment are identified using a Bayesian approach [13,14]. Using a false discovery rate adjusted p-value of less than 0.001 as threshold level for differential expression, a total of 120 and 150 genes are found to be up- and down-regulated, respectively, at one or more of the three time points (see Additional data files 1 and 2). As expected, none of these genes are differentially regulated in the control self-to-self hybridisation experiment (Figure 2A). It is previously known that auxin influences several key processes during plant growth and development and several lines of evidence indicate that auxin regulation of gene expression plays a key role in its mode of action. A particularly well-studied pathway of auxin-regulated gene expression is the auxin induction of the Aux/IAA genes. The Aux/IAA genes encode small short-lived nuclear proteins that interact with the ARF (auxin response factor) family of transcription factors and are thought to modulate the transcriptional activity of the ARFs in an auxin-dependent manner. These ARFs have been shown to bind to auxin-responsive elements (AuxREs) that are found in promotors of several auxin-regulated genes [16]. The ARFs function as both transcriptional activators and repressors [17], and the combination of ARF and Aux/IAA proteins is thought to mediate the tissue-specific effects of auxin [18,19]. Thirteen of the up-regulated genes identified in this study are previously shown to be auxin-regulated and include several members of the Aux/IAA family. These exhibit different induction patterns, with for example IAA5 and IAA19 being strongly up-regulated (>30- and 10-fold, respectively) already at 30 minutes, while a two-hour treatment is required for the IAA7 transcripts to reach a two-fold up-regulation. In a recent independent study where Arabidopsis seedlings were treated for only 15 min with 1 μM IAA, all of the Aux/IAA genes listed in Additional data file 1, with the exception of IAA7, were found to be up-regulated [7]. In addition to the Aux/IAA genes four members of the GH3 family that have also been shown to be induced by auxin exhibited a rapid and sustained 2- to 8-fold up-regulation in our study, again confirming the findings reported in a previous study on auxin regulation of gene expression using Affymetrix 8 k oligonucleotide arrays [8]. A key feature of auxin regulation of development is the polar transport of auxin that is mediated by auxin transporters. Our data indicate that the polar auxin transporters PIN1 [20] and PIN7 are up-regulated by auxin whereas in contrast the expression of several members of the Aquaporin gene family [21] are down-regulated. Expression of the PIN transporters is up-regulated already at 30 minutes and remains high throughout the studied time frame. These observations, of control of auxin transporters by auxin, are interesting since it is known that auxin regulates its own transport but to date there has been little data on this type of feedback. Other genes that are influenced by auxin in our study include transcription factors (8 up-regulated and 15 down-regulated), genes involved in signal transduction (7 and 6, respectively), metabolic enzymes (15 and 30, respectively), as well as several genes classified to other categories and also currently unknown genes. The most down-regulated gene at all time points (CATMA5a08790), for example, shows no sequence similarity to any known sequence and has no recorded expression in any of the sequenced tissue libraries deposited into the public domain. These 270 genes are interesting candidates for further research, but it is important that additional validations are carried out to identify and separate the immediate auxin target genes from the indirect.
Conclusions
We have described an efficient procedure for large-scale purification of gene sequence tags that can be used for several purposes including microarray fabrication. We demonstrate the utility of the technology by applying it to generate more than 21 000 short (150 – 500 bp) and highly specific Arabidopsis gene sequence tags for use as microarray probes in transcriptional profiling. Biotinylated amplification products are rapidly captured and eluted using a reusable streptavidin-coated solid-phase support in an automated high-throughput manner directly compatible with subsequent microarray printing. Our results demonstrate that the assembly and purification of gene-specific tags is an alternative to currently used purification methods, especially suitable for short amplification products such as gene sequence tags. In addition, the possibility to generate single-strand probes in the range of 150–500 nucleotides by a sodium hydroxide treatment of immobilised probes with subsequent elution of the remaining biotinylated strand, opens up for new microarray applications that would extend probe length beyond current oligonucleotide synthesis limits.
Methods
Optimisation of the purification procedure
The performance of the described purification method was investigated by varying either the amount of beads, the length of the incubation time for binding of the biotinylated product to the streptavidin-coated beads and the number of times the beads were reused. We also investigated if multiple reuses of the beads did introduce a well-to-well cross-contamination. The section below describes the general aspects of the purification and is followed by a more detailed description of the experiments carried out to investigate the different above-mentioned aspects of the purification method.
For all experiments three randomly chosen EST clones (0.5 kb, 1 kb and 1.3 kb) were amplified in 50-μl reactions containing 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 1.5 mM MgCl2, 200 nM dNTPs (Amersham Biosciences Europe GmbH, Sweden), 5 pmole universal sequencing primer (USP, 5'-TAAGCTAGGCACTGGCCGTCGTTTTACAACG-3', MWG Biotech AG, Germany), 5 pmole biotinylated reverse sequencing primer (RSP, 5'-AGGCCTAATGGTCATAGCTGTTTCCTGTGTG-3', MWG Biotech AG) and 1.5 units Platinum Taq DNA Polymerase (Invitrogen AB, Sweden). The temperature cycling (5 min at 95°C, 30 × (30 s at 95°C, 30 s at 64°C, 2 min at 72°C), 10 min at 72°C) was carried out in a Hybaid thermal cycler (Thermo Electron Molecular Biology, MA, USA). Pooling and splitting into aliquots of 50 μl in a clone-wise manner was used to remove variances introduced by the amplification step.
All purification steps, including bead dispensing, binding of the biotinylated product to the streptavidin moieties, washing, elution and regeneration of beads, were carried out in the Magnatrix 1200 automated workstation (Magnetic Biosolutions AB, Sweden). The biotinylated amplification products were bound to Dynabeads M-270 Streptavidin beads (Dynal Biotech ASA, Norway) during an incubation at room temperature using a high-salt binding buffer [1 M NaCl, 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 5% PEG-6000 and 0.1% Tween-20] and when bound, washed with 1 × TE-buffer [10 mM Tris-HCl, pH 7.5, 1 mM EDTA]. During incubation the beads were kept in suspension by mixing through pipetting every third minute. Elution was achieved by breaking the streptavidin-biotin bond in a 20-μl volume using deionised H2O. By use of a peltier thermal element, the immobilised products kept in suspension were heated in deionised water to 80°C (1°C / 2 s) for 1 second and cooled to room temperature (1°C / 2 s). Efficient elution is achieved through a combination of elevated temperature, appropriate temperature ramping and incubation at the elevated temperature, as described in more detail elsewhere [6]. The beads were separated from released products by magnetic separation, reconditioned through a repeated wash procedure with 1 × TE-buffer and, finally, prepared for the next round of purification. Quantification of DNA was carried out using the Nanodrop ND-1000 spectrophotometer (NanoDrop Technologies Inc, DE, USA).
For all the purifications described below an aliquot of the pooled PCR product corresponding to a 50-μl reaction was used. The binding capacity of the beads was studied using an increasing amount (10 μg, 50 μg, 100 μg and 150 μg) of fresh beads (first use) while keeping the incubation time constant at 20 minutes. To estimate the variability of the method eight independent purifications were carried out (n = 8). The effect of the length of the incubation time was studied using 150 μg of beads and eight different incubation times (1, 5, 10, 15, 20 or 30 min) with four replicates of each (n = 4). To study the effects of multiple reuses of the beads, the same beads were reconditioned and used up to nine times. The amount of beads in the first capture was 150 μg and the capture time 20 minutes. The variability was estimated using four independent replications. The multiple reuse of the beads was also analysed using a hybridisation based approach. Twelve clones ranging from 0.3 to 2 kb were amplified and purified multiple times using reconditioned beads. The purified products were subsequently printed in eight replicates onto glass slides and quantified using Syto61 (Molecular Probes Inc, OR, USA). The well-to-well carry-over of product was analysed by first purifying one of the PCR products, followed by a purification reaction with no PCR product added (water-only control). This pattern was repeated six times for all three products, while the same set of beads was used for all purifications. Eluates from all these purifications were printed on slides and hybridised with a Cy5-labelled oligonucleotide complementary to the common vector sequence present in all products. Hybridisation was carried out using 10 pmole of the labelled oligonucleotide for 1 hour at 35°C in a hybridisation solution containing 50% formamide, 5 × SSC and 0.1% SDS. Slides were washed with 2 × SSC containing 0.1% SDS (5 min at room temperature) and three times with 1 × SSC (1 min at room temperature). Scanning using the G2565BA DNA microarray scanner (Agilent Technologies) was carried out at the highest possible photo multiplier tube setting in order to reveal low-level signals.
Preparation and arraying of gene sequence tags
Initial amplification from BAC-clones or genomic DNA was carried out by the CATMA consortium at different nodes throughout Europe [1,2]. One percentage of the first amplification product, obtained using gene-specific primers with 5' handle sequences, was used as template for the second amplification. A total of 51 cycles [11 × (15 s at 94°C, 15 s at 55°C (-1°C / cycle), 30 s at 72°C), 40 × (15 s at 94°C, 30 s at 55°C, 30 s at 72°C)] were carried out in the presence of 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 1.5 mM MgCl2, 200 nM dNTPs, 20 pmole of each primer (forward primer biotinylated, Thermo Hybaid GmbH, Germany) and 1.5 units Platinum Taq DNA Polymerase in a total volume of 50 μl. Biotinylated products from each amplification reaction were bound to 200 μg of Dynabeads M-270 Streptavidin beads (reused up to six times) during a 15-minute incubation at room temperature using a high ionic-strength buffer, washed with 1 × TE-buffer and, finally, eluted with 12 μl deionised water. After each capture round the beads were reconstituted, pooled plate-wise and randomly assigned to a new plate. An equal volume of 99.9% dimethyl sulfoxide (DMSO) (Sigma-Aldrich Sweden AB, Sweden) was added and the purified amplification products were arrayed into 22 by 22 patterns in 48 individual blocks with the QArray arrayer (Genetix Limited, UK) and SMP2.5 pins (TeleChem International Inc, CA, USA). When dried, printed products were immobilised to the reactive surface of the Ultra-GAPS slides (Corning B.V. Life Sciences, The Netherlands) using 250 mJ/cm2 UV-light (Stratalinker, Stratagene Europe, The Netherlands).
Auxin treatment and sample preparation
10-day-old Arabidopsis Col-0 seedlings were grown at 22°C in MS medium (Duchefa AB, The Netherlands) supplemented with 0.5% sucrose and using a 24-h photoperiod with 16 h of light at 75 mE m-2 sec-1 PAR. The samples were treated with 1 μM indole-acetic acid for a period of 0, 30, 120 and 240 minutes, washed once with an excess of MS medium with 0.5% sucrose for 5 minutes, frozen in liquid nitrogen and stored at -70°C. For each time point, frozen seedlings from three independent vials were pooled, grinded and total RNA extracted using the RNeasy kit (Qiagen GmbH, Germany). The quality of the RNA was determined using the RNA 6000 Nano kit and the Bioanalyzer instrument (Agilent Technologies, CA, USA).
Target labelling, hybridisation, washing and scanning
Ten μg anchored oligo dT primer (dT20VN, MWG Biotech AG) was annealed to 20 μg total RNA after a denaturating step (10 min at 70°C). The cDNA synthesis reaction was carried out at 42°C for 1 h 45 min in a 30-μl reaction containing 2 mM dNTPs (dTTP:aminoallyl-dUTP in 1:4, unmodified Amersham Biosciences, modified Sigma-Aldrich), first-strand buffer (50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl2), 0.01 mM DTT and 400 units Superscript II (Invitrogen AB). The synthesis reaction was terminated by addition of EDTA, the RNA strand hydrolysed with NaOH (15 minutes at 70°C) and the reaction neutralised with HCl (final concentrations 20 mM, 150 mM and 150 mM, respectively). The cDNA strands were purified using the MinElute spin columns (Qiagen GmbH) with the provided wash and elution buffers replaced by 80% ethanol and 100 mM NaHCO3, pH 9.0, respectively. Monofunctional NHS-ester Cy3 or Cy5 fluorophores (Amersham Biosciences) were coupled to the amino-allyl groups during a 90-minute incubation at room temperature after which unincorporated ester groups were inactivated through a hydroxylamine treatment (final concentration of 730 mM). The pooled labelling reactions were purified using MinElute spin columns and hybridised to the arrays using a two-step protocol in the GeneTac hybridisation station (Genomic solutions Ltd, UK). The pre-hybridisation at 42°C for 45 min (5 × SSC, 1% BSA (Sigma-Aldrich), 0.1% SDS, 40 μg poly(dA) (Sigma-Aldrich) and 20 μg tRNA (Sigma-Aldrich)) was followed by a 16–18 h hybridisation at 42°C with the labelled material and a hybridisation buffer containing 5 × SSC, 25% formamide, 0.1% SDS, 40 μg poly(dA) and 20 μg tRNA. The slides were washed with 2 × SSC and 0.1% SDS at 42°C, followed by 0.1 × SSC + 0.1% SDS at room temperature and finally by three repeated washes with 0.1 × SSC at room temperature. Slides were scanned at 10-μm resolution using the G2565BA DNA microarray scanner for which the photo multiplier tube (pmt) setting was adjusted so that the images for the Cy3 and Cy5 channels were in balance as determined by visual observation. Each time point-to-reference sample comparison was carried out on two arrays, with the dye labels exchanged between the replicated hybridisations in order to avoid sequence-dependent labelling and hybridisation effects. A control self-to-self hybridisation was also carried out for the untreated sample in order to assess the level of noise in the experimental system.
Image processing and data analysis
The acquired tiff-images were processed using the GenePix 4.1 software (Axon instruments Inc, CA, USA) and the data with the R environment for statistical computing [9], Bioconductor [10] and the aroma package for microarray data analysis [11]. Expression values for each feature and dye channel were obtained by subtracting the median of the local background value from the median of the foreground value. Features for which the background subtracted value were zero or below in one of the channels, but not in the other, were given the expression value of 1. A feature was considered uncertain and removed from subsequent data analysis by setting its value to NA (not available) if a) it was flagged as Not Found by GenePix, b) it was manually flagged as bad (dust particles etc), c) the signals for both channels were saturated, d) the percentage of foreground pixels above the median background + 2 SD were below 60 for both channels or e) the feature diameter was <70 μm or >120 μm. Filtered data was normalised separately for each individual block on the slide using the intensity-dependent lowess method [12] and no between-slides scaling of the ratio values was deemed necessary. Differentially expressed genes were identified using a moderated t-test based on gene-wise standard errors estimated by an empirical Bayes method [13,14]. Genes with a false discovery rate adjusted p-value of less than 0.001 for any of the three time points were considered as potentially differentially expressed and are included in the Additional data files 1 and 2. The MIAME compatible data set, including processed and unprocessed data, is made available to the research community through the ArrayExpress expression data repository at the EMBL using the accession number E-MEXP-140 [15].
List of abbreviations
CATMA a complete Arabidopsis thaliana transcriptome microarray
EST expressed sequence tag
GST gene sequence tag
IAA indole-3-acetic acid
MIAME minimum information about a microarray experiment
Authors' contributions
VW carried out the laboratory work and the data analysis and participated in the design of the study and drafting of the manuscript. AH and ML participated in the solid-phase purification procedure. PN participated in the array production. PH designed and provided the CATMA probes. MU participated in the automation of the solid-phase procedure. RB carried out the auxin treatment and participated in the interpretation of the expression data. JL conceived of the study, participated in the drafting of the manuscript and coordinated the study. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
List of genes identified as up-regulated by the indole-3-acetic acid treatment. Group labels refer to (A) previously identified auxin regulated, (B) transcription related, (C) signal transduction related, (D) transport, (E) cell wall establishment related, (F) metabolic enyzmes, (G) light related, (H) disease repsonse related, (I) other and (J) unknown genes.
Click here for file
Additional File 2
List of genes identified as down-regulated by the indole-3-acetic acid treatment. Group labels refer to (A) previously identified auxin regulated, (B) transcription related, (C) signal transduction related, (D) transport, (E) cell wall establishment related, (F) metabolic enyzmes, (G) light related, (H) disease response related, (I) other and (J) unknown genes.
Click here for file
Acknowledgements
The authors wish to thank Annelie Waldén for helping with the printing of the arrays and the CATMA consortium for providing amplicons to carry out amplification with biotinylated handle primers. This work was supported by grants from the Wallenberg Consortium North, the Knut and Alice Wallenberg foundation and the Swedish Scientific Research Council.
Figures and Tables
Figure 1 Solid-phase purification parameters. In A the agarose gel image shows the first (upper left image) and second (bottom left image) elution of the captured 1.0 kb product. The first and fourth lanes contain a size marker, while the second and third lanes contain an equal amount, assuming 100% yield at purification step, of unpurified PCR product (PP) and purified PCR product (eluate), respectively. The images on the right shows the results from the carry-over test, where PCR product and water were used as input samples for multiple consecutive purification reactions in an alternating order. The upper part shows the hybridisation results (PCR product, water, PCR product and water), while the box-and-whiskers plot below shows the quantifications of the signals (n = 18, six replicates of the three different products). Purification of three amplification products (red line 0.5 kb, green line 1.0 kb and blue line 1.3 kb) is investigated using an increasing amount of streptavidin-coated magnetic beads (number of independent replications, n = 8, panel B), varying binding time (n = 4, panel C) and repeatedly used beads (n = 4, panel D). The black line in (D) is based on fluorescence data and is plotted using the y-axis on the right, while the three other lines are based on absorbance measurements and use the y-axis on the left side. The presented data originates from repeated independent experiments and the error bars denote the calculated standard error.
Figure 2 General overview of the filtered and normalised data for each of the time point comparisons. (A-D) The average of the replicated hybridisations for each time point is presented using MA-plots (samples are labelled with a Red dye and the reference with a Green dye). An increased noise at lower signal intensities is observed, as expected, as no absolute cut-off level for signal intensity is used in the data filtering process. However, genes at the low-intensity region are not identified as differentially expressed genes due to variation between the replicated hybridisations. (B) At 30 minutes the overall pattern is drastically changed, with several genes showing a >2-fold change in expression levels. (C-D) The same general pattern is observed for the data corresponding to 120- and 240-min treatments. An increasing number of differentially expressed genes are, however, observed. For (A-D) genes differentially expressed at one time point ● (30 min blue, 120 min red, 240 min green), two time points + (30 and 120 min red, 30 and 240 min blue, 120 and 240 min green) or three time points ▲ (purple) time points are listed in Additional data files 1 and 2 available online. Dashed lines indicate two-fold differential expression.
==== Refs
Hilson P Allemeersch J Altmann T Aubourg S Avon A Beynon J Bhalerao RP Bitton F Caboche M Cannoot B Chardakov V Cognet-Holliger C Colot V Crowe M Darimont C Durinck S Eickhoff H de Longevialle AF Farmer EE Grant M Kuiper MT Lehrach H Leon C Leyva A Lundeberg J Lurin C Moreau Y Nietfeld W Paz-Ares J Reymond P Rouze P Sandberg G Segura MD Serizet C Tabrett A Taconnat L Thareau V Van Hummelen P Vercruysse S Vuylsteke M Weingartner M Weisbeek PJ Wirta V Wittink FR Zabeau M Small I Versatile gene-specific sequence tags for Arabidopsis functional genomics: transcript profiling and reverse genetics applications Genome Res 2004 14 2176 89 15489341 10.1101/gr.2544504
Crowe ML Serizet C Thareau V Aubourg S Rouze P Hilson P Beynon J Weisbeek P van Hummelen P Reymond P Paz-Ares J Nietfeld W Trick M CATMA: a complete Arabidopsis GST database Nucleic Acids Res 2003 31 156 8 12519971 10.1093/nar/gkg071
Thareau V Dehais P Serizet C Hilson P Rouze P Aubourg S Automatic design of gene-specific sequence tags for genome-wide functional studies Bioinformatics 2003 19 2191 8 14630647 10.1093/bioinformatics/btg286
Schiex T Moisan A Rouzé P EUGÉNE: an eukaryotic gene finder that combines several sources of evidence Lect Notes Comput Sci 2001 2066 111 125
Allemeersch J Durinck S Vanderhaeghen R Alard P Maes R Seeuws K Bogaert T Coddens K Deshouwer K van Hummelen P Vuylsteke M Moreau Y Kwekkeboom J Wijfjes AHM May S Beynon J Hilson P Kuiper MTR Benchmarking the CATMA microarray: a novel tool for Arabidopsis transcriptome analysis Plant Physiol
Holmberg A Blomstergren A Nord O Lukacs M Lundeberg J Uhlen M The biotin-streptavidin interaction can be reversibly broken using water at elevated temperatures Electrophoresis
Sawa S Ohgishi M Goda H Higuchi K Shimada Y Yoshida S Koshiba T The HAT2 gene, a member of the HD-Zip gene family, isolated as an auxin inducible gene by DNA microarray screening, affects auxin response in Arabidopsis Plant J 2002 32 1011 22 12492842 10.1046/j.1365-313X.2002.01488.x
Tian Q Uhlir NJ Reed JW Arabidopsis SHY2/IAA3 inhibits auxin-regulated gene expression Plant Cell 2002 14 301 19 11884676 10.1105/tpc.010283
R Development Core Team R: A language and environment for statistical computing
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 Leisch F Li C Maechler M Rossini AJ Sawitzki G Smith C Smyth G Tierney L Yang JY Zhang J Bioconductor: open software development for computational biology and bioinformatics Genome Biol 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Bengtsson H aroma – An R Object-oriented Microarray Analysis environment
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 10.1093/nar/30.4.e15
Lonnstedt I Speed TP Replicated microarray data Stat Sinica 2002 12 31 46
Smyth G Linear models and empirical Bayes methods for assessing differential expression in microarray experiments Statistical Applications in Genetics and Molecular Biology 2004 3 Article 3
ArrayExpress – a public repository for microarray gene expression data at the EBI
Ulmasov T Hagen G Guilfoyle TJ ARF1, a transcription factor that binds to auxin response elements Science 1997 276 1865 8 9188533 10.1126/science.276.5320.1865
Ulmasov T Hagen G Guilfoyle TJ Activation and repression of transcription by auxin-response factors Proc Natl Acad Sci U S A 1999 96 5844 9 10318972 10.1073/pnas.96.10.5844
Tiwari SB Hagen G Guilfoyle T The roles of auxin response factor domains in auxin-responsive transcription Plant Cell 2003 15 533 43 12566590 10.1105/tpc.008417
Liscum E Reed JW Genetics of Aux/IAA and ARF action in plant growth and development Plant Mol Biol 2002 49 387 400 12036262 10.1023/A:1015255030047
Vernoux T Kronenberger J Grandjean O Laufs P Traas J PIN-FORMED 1 regulates cell fate at the periphery of the shoot apical meristem Development 2000 127 5157 65 11060241
Quigley F Rosenberg JM Shachar-Hill Y Bohnert HJ From genome to function: the Arabidopsis aquaporins Genome Biol 2002 3 RESEARCH0001 11806824
| 15689241 | PMC550664 | CC BY | 2021-01-04 16:02:58 | no | BMC Biotechnol. 2005 Feb 3; 5:5 | utf-8 | BMC Biotechnol | 2,005 | 10.1186/1472-6750-5-5 | oa_comm |
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-71569138010.1186/1477-7819-3-7Technical InnovationsA novel surgical procedure for bridging of massive bone defects Knothe Ulf R [email protected] Dempsey S [email protected] Department of Orthopedic Surgery and Orthopedic Research Center, The Cleveland Clinic Foundation, Cleveland, OH, USA2 Department of Orthopedic Surgery, The Mount Sinai School of Medicine, New York, NY, USA2005 3 2 2005 3 7 7 4 11 2004 3 2 2005 Copyright © 2005 Knothe and Springfield; licensee BioMed Central Ltd.2005Knothe and Springfield; 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
Bony defects arising from tumor resection or debridement after infection, non-union or trauma present a challenging problem to orthopedic surgeons, as well as patients due to compliance issues. Current treatment options are time intensive, require more than one operation and are associated with high rate of complications. For this reason, we developed a new surgical procedure to bridge a massive long bone defect.
Methods
To bridge the gap, an in situ periosteal sleeve is elevated circumferentially off of healthy diaphyseal bone adjacent to the bone defect. Then, the adjacent bone is osteotomized and the transport segment is moved along an intramedullary nail, out of the periosteal sleeve and into the original diaphyseal defect, where it is docked. Vascularity is maintained through retention of the soft tissue attachments to the in situ periosteal sleeve. In addition, periosteal osteogenesis can be augmented through utilization of cancellous bone graft or in situ cortical bone adherent to the periosteal sleeve.
Results
The proposed procedure is novel in that it exploits the osteogenic potential of the periosteum by replacing the defect arising from resection of tissue out of a pathological area with a defect in a healthy area of tissue, through transport of the adjacent bone segment. Furthermore, the proposed procedure has several advantages over the current standard of care including ease of implementation, rapid patient mobilization, and no need for specialized implants (intramedullary nails are standard inventory for surgical oncology and trauma departments) or costly orthobiologics.
Conclusions
The proposed procedure offers a viable and potentially preferable alternative to the current standard treatment modalities, particularly in areas of the world where few surgeons are trained for procedures such as distraction osteogenesis (e.g. the Ilizarov procedure) as well as areas of the world where surgeons have little access to expensive, complex devices and orthobiologics.
==== Body
Background
Replacement of bone where there is none is one of the most challenging problems facing orthopedic surgeons today. In the case of tumor resection or trauma, massive bone defects must be filled with regenerate bone as quickly as possible in order to restore function. Current standards for bridging of massive bone defects in long bones generally follow a theme of i) filling the defect with bone autograft or allograft (including cancellous bone graft or bone transplantation via vascularized or non-vascularized fibula transfer) and ii) accelerating functional remodeling and integration through addition of physical and/or chemical stimuli such as tension (e.g. Ilizarov technique, which is a standard surgical treatment modality for bone transport whereby an osteotomy is performed far from the defect site and the transport segment thus created is moved, approximately one millimeter per day, under the constant tension by wires attached to a cumbersome external fixator, until the defect is bridged and the segment can be docked onto the other side of the defect), and orthobiologics (e.g. bone graft or bone graft replacement, BMP's). Surgical treatment modalities involving auto/allografting and bone regeneration via distraction osteogenesis are complex, time intensive procedures of inherently high risk due to vagaries of organ donation, in the case of allografts, and the complexity of soft and hard tissue salvage during the process of distraction osteogenesis. In addition, orthobiologics are costly and their dosage regimes as well as efficacy are currently the subject of much research [1-3].
Both bone grafting and bone transport procedures are complex for the surgeon as well as for the patient. Furthermore, they are susceptible to complications such as delayed union, extensive treatment time periods, infections, and insufficient mechanical function outcomes that can result in fractures. The high complication rates of these procedures exacerbate the previously mentioned difficulties associated with these treatment modalities, from the perspective of the surgeon as well as that of the patient. In short, the inherent risk of complications increases the need for patient compliance and clinical follow-up. Despite the effort associated with these procedures, their results are often less than satisfactory. Hence, the complexity and shortcomings of current state-of-the-art surgical procedures have provided impetus to develop a new treatment modality that provides a relatively straightforward, single step procedure with a high probability of success for the bridging of massive bone defects in long bones. The procedure is straightforward and can be implemented in operating rooms across the world without the need for high-tech equipment or expensive orthobiologics. The purpose of this manuscript is to describe the novel procedure.
Technical innovation – methodology and proof of feasibility
The proposed procedure is applicable for clinical scenarios including tumor resection as well as debridement after an infection or non-union. In the case of reconstruction after tumor resection (Fig. 1A), a transport-segment of the diaphysis adjacent to the defect is pealed out of the surrounding periosteum (Fig 1B,C), an osteotomy is performed and the transport-segment is moved out of the periosteal sleeve and docked to the other side of the defect (Fig. 1D). The periosteal sleeve is then closed like a tube surrounding the newly created defect. Either an internal fixation device such as an intramedullary nail, a plate or an internal fixator or an external fixator can be used to provide stabilization through the healing and regeneration phase.
Figure 1 Schematic diagram showing the concept for the new surgical procedure.
The proposed procedure depends to a large degree on bone's inherent healing strategies. Bone is a remarkably resilient tissue capable of adaptation to the most extreme biological and mechanical environments; this capacity for self-regeneration without scarring is based on bone's endogenous healing strategies. First, bone remodels itself through osteoclastic resorption and osteoblastic matrix apposition; by constantly reweaving itself, the structure is dynamic and optimal for prevailing mechanical function. Furthermore, the natural healing cascade of bone after trauma recapitulates embryonic endochondral ossification. Hence, modeling, growth and remodeling confer a means to regenerate functional tissue at any time in the life cycle of a bone. The "raw materials" necessary to replace bone are located in the environment or produced by the cells that do the work of regeneration, i.e. osteoclasts and osteoblasts. In the case of regeneration of bone in defects, further potentially key constituents to the formation of a functional regenerate in situ include a patent blood supply, chemical gradients of morphogens and/or cytokines, a template onto which the cells can anchor themselves during the rebuilding process (e.g. graft or a scaffold), and biophysical stimuli such as fluid flow and/or cell level strains. The proposed procedure essentially replaces the defect site in a pathological zone with a defect site in a healthy bed of tissue and provides for progenitor cells through the surrounding, healthy periosteum as well as many of the other key constituents for successful tissue regeneration, as defined above.
A clinical case described below demonstrates the osteoinductive potential of the periosteum and serves is a proof of feasibility for the proposed procedure. An 11-year old male presented with a low grade surface osteosarcoma of the tibia. After resection of the tumor, the fibula was resected for transfer and the surrounding periosteum was left behind to serve as an osteo-inductive and -conductive sleeve (Figure 2A). Already 3 weeks after the procedure, bone regenerate is visible within this sleeve (Figure 2A). Impressive remodeling of the fibula is also evident in follow up radiographs and includes extensive remodeling of the intramedullary canal by three months post procedure (Figure 2B and 2C). Based on this clinical case as well as one author's previous experience with an in vivo segmental defect in an ovine model [4,5], the osteogenic potential of the periosteum as a source of progenitor cells and as a "membrane" or boundary template for guided bone generation is demonstrated. Taking this one step further, the corresponding author conceived of the idea to exploit the potential of the healthy periosteum by moving the defect site from a pathological zone to a healthy one and then providing sufficient mechanical stability to let bone's endogenous healing capacity regenerate functional tissue within the new defect zone.
Figure 2 11 year male with malignant tumor. Fibula-pro-tibia following local resection. Cortical regeneration from periosteum. Performed at Mt. Sinai Medical Center, NYC, 2000. A: 3 weeks post-operative radiograph, B: 6 weeks post-operative, C: 3 months post-operative, D: approximately 6 months post-operative.
Discussion
The proposed procedure is novel in that it introduces for the first time the possibility to bridge a massive defect in a long bone using a single stage procedure. Furthermore, the proposed procedure has several advantages over the current standard of care including ease of implementation, lack of requirement for specialized implants (intramedullary nails are standard inventory for surgical oncology and trauma departments) or costly orthobiologics, and rapid patient mobilization. This makes the proposed procedure a viable and potentially preferable alternative to the current standard treatment modalities, particularly areas of the world where few surgeons are trained for procedures such as distraction osteogenesis (e.g. the Ilizarov procedure) as well as where surgeons have less access to expensive, complex devices and orthobiologics.
Conclusion
In summary, the authors propose a new procedure which obviates the need for several surgical procedures, reduces the risk for complications, reduces the time frame for the treatment and is much more comfortable for and requires less compliance of the patient. This novel, one stage procedure exploits the osteogenetic potential of the periosteum for bone formation to bridge the defect with concomitant bone transport and does not require the use of expensive hardware or orthobiologics.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
UK conceived of the technical innovation, organized its design and coordination, and drafted the manuscript. DSS was the senior surgeon in the clinical case showing feasibility. Both authors read and approved the final manuscript.
Acknowledgements
Permission was obtained from the patient's guardian regarding publication of his case report and x-ray photographs.
==== Refs
Salkeld SL Patron LP Cook SD The effect of osteogenic protein-1 on the healing of segmental bone defects treated with allograft bone J Bone Joint Surg Am 2001 83-A 803 816 11407788
Torricelli P Fini M Giavaresi G Rimondini L Giardino R Characterization of bone defect repair in young and aged rat femur induced by xenogenic demineralized bone matrix J Periodontol 2002 73 1003 1009 12296584 10.1902/jop.2002.73.9.1003
Sheller MR Crowther RS Kinney JH Yang J Di Jorio S Breunig T Carney DH Ryaby JT Repair of rabbit segmental defects with the thrombin peptide, TP508 J Orthop Res 2004 22 1094 1099 15304284 10.1016/j.orthres.2004.03.009
Klaue K Knothe U Anton C Masquelet AC Petten SM Biological implementation of autologous foreign body membranes in corticalization of massive cancellous bone grafts Trans Orthopedic Trauma Association 1996
Hertel R Knothe U Gerber A Cordey J Rahn R The osteogenic potential of vascularized periosteum and cancellous bone graft in sheep Trans Orthopedic Research Society 1997
| 15691380 | PMC550665 | CC BY | 2021-01-04 16:39:05 | no | World J Surg Oncol. 2005 Feb 3; 3:7 | utf-8 | World J Surg Oncol | 2,005 | 10.1186/1477-7819-3-7 | oa_comm |
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-101570519510.1186/1477-7525-3-10ResearchRole of religion and spirituality in medical patients: Confirmatory results with the SpREUK questionnaire Büssing Arndt [email protected] Thomas [email protected] Peter F [email protected] Krebsforschung Herdecke, Department of Applied Immunology and Cancer Service, Herdecke Community Hospital, Gerhard-Kienle-Weg 4, 58313 Herdecke, Germany2 Department of Medical Theory and Complementary Medicine, University Witten/Herdecke, Gerhard-Kienle-Weg 4, 58313 Herdecke, Germany2005 10 2 2005 3 10 10 15 12 2004 10 2 2005 Copyright © 2005 Büssing et al; licensee BioMed Central Ltd.2005Büssing 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
Spirituality has become a subject of interest in health care as it is was recognized to have the potential to prevent, heal or cope with illness. There is less doubt that values and goals are important contributors to life satisfaction, physical and psychological health, and that goals are what gives meaning and purpose to people's lives. However, there is as yet but limited understanding of how patients themselves view the impact of spirituality on their health and well-being, and whether they are convinced that their illness may have "meaning" to them. To raise these questions and to more precisely survey the basic attitudes of patients with severe diseases towards spirituality/religiosity (SpR) and their adjustment to their illness, we developed the SpREUK questionnaire.
Methods
In order to re-validate our previously described SpREUK instrument, reliability and factor analysis of the new inventory (Version 1.1) were performed according to the standard procedures. The test sample contained 257 German subjects (53.3 ± 13.4 years) with cancer (51%), multiple sclerosis (24%), other chronic diseases (16%) and patients with acute diseases (7%).
Results
As some items of the SpREUK construct require a positive attitude towards SpR, these items (item pool 2) were separated from the others (item pool 1). The reliability of the 15-item the construct derived from the item pool 1 respectively the 14-item construct which refers to the item pool 2 both had a good quality (Cronbach's alpha = 0.9065 resp. 0.9525). Factor analysis of item pool 1 resulted in a 3-factor solution (i.e. the 6-item sub-scale 1: "Search for meaningful support"; the 6-item sub-scale 2: "Positive interpretation of disease"; and the 3-item sub-scale 3: "Trust in external guidance") which explains 53.8% of variance. Factor analysis of item pool 2 pointed to a 2-factor solution (i.e. the 10-item sub-scale 4: "Support in relations with the External life through SpR" and the 4-item sub-scale 5: "Support of the Internality through SpR") which explains 58.8% of variance. Generally, women had significantly higher SpREUK scores than male patients. Univariate variance analyses revealed significant associations between the sub-scales and SpR attitude and the educational level.
Conclusions
The current re-evaluation of the SpREUK 1.1 questionnaire indicates that it is a reliable, valid measure of distinct topics of SpR that may be especially useful of assessing the role of SpR in health related research. The instrument appears to be a good choice for assessing a patients interest in spiritual concerns which is not biased for or against a particular religious commitment. Moreover it addresses the topic of "positive reinterpretation of disease" which seems to be of outstanding importance for patients with life-changing diseases.
QuestionnairesReligion and MedicineSpirituality and Religioncopingchronic disease, cancer
==== Body
Background
Spirituality has become a subject of interest in health care, and an increasing number of studies, commentaries and reviews examine the connection between religiosity/spirituality and health, its potential to prevent, heal or cope with diseases [1-10]. Moreover, research has confirmed that spiritual well-being is positively associated with quality of life, fighting-spirit, but also fatalism, yet negatively correlated with helplessness/hopelessness, anxious preoccupation, and cognitive avoidance [11]. Indeed, there is evidence that spirituality is important in coping with illness, as spiritual well-being offers some protection against hopelessness and despair in terminally ill patients [12-16].
However, although religiosity and spirituality were interchangeable words, these constructs may not be identical. It is well established to divide Religiosity into three sub-constructs: Intrinsic, Extrinsic, and Quest Religiosity [17-20], while the construct Spirituality was divided into the following sub-constructs: Cognitive Orientation Towards Spirituality, Experiential/Phenomenological Dimension of Spirituality, Existential Well-Being, Paranormal Beliefs, and Religiousness [21].
The measurability and operability of spirituality and religiosity remains a problem and thus several questionnaires address this topic. Most of them measure beliefs of specific religious groups, and ask about the relationship with God (i.e. the Spiritual Well-Being Scale [22], the Daily Spiritual Experience Scale [23], or the Santa Clara Strength of Religious Faith Questionnaire [24], while only a few took into account that several patients are offended by institutional religion, but may have an interest in distinct forms of spirituality, respectively in a more personal search for spiritual fulfilment [25,26]. The Functional Assessment of Chronic Illness Therapy – Spiritual Well-Being (FACIT-Sp) scale has a much more open design [27], but, however, the 12 items of this instrument which made up 2 main factors (labelled "Meaning/Peace" and "Faith") may not really meet the situation of patients with severe and life-threatening diseases. In the post-treatment orientation phase of cancer patients, more existentialistic issues in the patients' attempt to manage the implications of their disease in daily life are of outstanding importance [28,29] The same is true for hospitalised cancer patients [30].
There is less doubt that values and goals are important contributors to life satisfaction, physical and psychological health, and that goals are what gives meaning and purpose to people's lives [31-33]. Moreover, health can be conceptualized as a competence to gain control for the design of the biography [34]. But in face of a life-threatening diseases, do patients find meaning and purpose in their life? Many of them rely on religious beliefs to relieve stress, retain a sense of control, maintain hope and their sense of meaning and purpose in life [35], while others may lose faith in their religious beliefs, and seek for alternatives [28,29]. There is as yet but limited understanding of how patients themselves view the impact of spirituality on their health and well-being, and whether they are convinced that spirituality may offer some beneficial effects.
To raise these questions and to more precisely survey the basic attitudes of those patients towards spirituality/religiosity (SpR) and their adjustment to their illness, we developed the SpREUK questionnaire [28,29,36-39]. We defined the multi-dimensional construct "Spirituality" as an "individual and open approach in the search for meaning and purpose in life, as a search for transcendental truth which may include a sense of connectedness with others, nature, and/or the divine" [28]. The main sub-scales of our instrument may thus correspond to MacDonald's spirituality constructs of an "Existential Well-Being" [21] which describes a meaning and purpose for existence, and the perception of self as being competent and able to cope with the difficulties of life and limitations of human existence, and to the construct of an "Cognitive Orientation Towards Spirituality" which is identified by beliefs, attitudes, and perceptions regarding the nature and significance of spirituality, as well as having relevance and importance for personal functioning.
In this article we report the re-validation of the SpREUK 1.1 questionnaire (SpREUK is an acronym of the German translation of "Spiritual and Religious Attitudes in Dealing with Illness"), an instrument designed to examine attitudes of patients with life-threatening and chronic diseases towards spirituality/religiosity.
Methods
Procedure and subjects
All individuals were informed of the purpose of the study, were assured of confidentiality, and gave informed consent to participate. The patients were recruited consecutively in the cancer service, the multiple sclerosis service, and two internal medical units of the Communal Hospital in Herdecke (Germany). All subjects completed the questionnaire by themselves. Demographic information is provided in Table 1.
Table 1 Demographic data and SpREUK scores of 257 subjects
% Search Meaning (50.6 ± 25.9) Message Disease (70.4 ± 20.9) Trust Guidance (70.0 ± 26.5) Support External (59.0 ± 23.9) Support Internal (62.3 ± 24.3)
sex ** ** * ** *
female 70 54.7 ± 26.0 73.4 ± 20.3 72.9 ± 24.4 61.8 ± 23.1 64.9 ± 22.6
male 30 40.8 ± 22.9 63.4 ± 20.8 63.0 ± 30.0 52.4 ± 24.5 55.6 ± 27.1
age (*) ** * *
< 30 years 3 31.8 ± 19.8 60.9 ± 24.6 40.6 ± 20.5 38.0 ± 18.1 43.0 ± 20.7
30–49 years 38 49.7 ± 26.2 70.4 ± 20.2 63.4 ± 26.9 55.3 ± 23.4 58.8 ± 23.9
50–69 years 45 54.2 ± 26.2 72.2 ± 19.6 74.7 ± 24.9 62.5 ± 22.4 65.8 ± 23.2
> 70 years 12 46.2 ± 23.6 67.7 ± 25.7 79.8 ± 23.5 63.4 ± 28.2 64.5 ± 28.0
marital status * * *
married 65 47.2 ± 26.0 69.3 ± 21.3 70.1 ± 27.0 56.1 ± 24.6 60.5 ± 26.0
living with partner 11 52.5 ± 20.1 73.3 ± 21.3 57.1 ± 24.7 58.3 ± 16.8 63.2 ± 13.9
divorced 9 65.0 ± 26.9 75.4 ± 19.3 74.1 ± 25.8 69.9 ± 19.7 68.5 ± 24.7
alone 10 55.9 ± 29.6 71.3 ± 20.4 72.3 ± 25.8 61.6 ± 27.2 65.9 ± 23.7
widowed 5 51.4 ± 17.8 68.2 ± 19.6 84.3 ± 17.8 73.5 ± 18.2 62.5 ± 22.9
education1 ** ** ** **
level 1 25 36.6 ± 24.9 58.6 ± 20.4 69.0 ± 26.1 55.0 ± 24.1 58.9 ± 24.6
level 2 29 44.4 ± 27.8 67.8 ± 19.5 68.9 ± 29.6 50.6 ± 25.7 53.5 ± 29.2
level 3 37 65.5 ± 24.7 76.9 ± 16.7 75.8 ± 19.8 68.2 ± 19.2 69.1 ± 15.3
other 9 65.5 ± 24.94 74.6 ± 21.6 74.2 ± 21.2 67.9 ± 19.8 75.0 ± 19.0
disease ** ** ** ** **
Cancer 51 55.4 ± 24.6 73.8 ± 19.7 74.2 ± 23.4 62.3 ± 22.5 64.3 ± 22.3
Multiple Sclerosis 24 35.8 ± 22.5 59.0 ± 19.1 56.8 ± 28.0 48,0 ± 23.3 52.9 ± 25.8
Chronic diseases 16 56.5 ± 25.7 75.0 ± 19.3 69.8 ± 27.5 63.1 ± 26.1 68.6 ± 26.1
Acute diseases 7 44.7 ± 26.4 74.8 ± 27.1 76.4 ± 30.6 60.6 ± 23.9 67.6 ± 23.9
duration of disease
< 0.5 years 19 48.2 ± 26.1 73.3 ± 18.6 68.8 ± 27.7 58.4 ± 21.7 63.3 ± 20.2
0.5–1 years 12 53.1 ± 24.9 79.0 ± 19.7 67.9 ± 28.6 58.1 ± 20.6 58.0 ± 26.0
1–3 years 26 54.1 ± 24.4 72.5 ± 20.2 71.1 ± 23.1 60.6 ± 22.9 63.7 ± 22.9
3–5 years 12 46.8 ± 27.6 61.9 ± 19.4 66.4 ± 28.2 56.7 ± 26.1 56.7 ± 28.3
> 5 years 31 47.3 ± 26.4 70.8 ± 22.1 68.0 ± 28.1 56.7 ± 27.0 61.9 ± 26.0
confession ** ** ** **
Christian 80 52.9 ± 25.4 71.2 ± 20.8 76.0 ± 21.3 62.2 ± 22.8 63.8 ± 24.3
Others 3 58.3 ± 18.0 70.8 ± 17.1 85.4 ± 15.7 65.7 ± 23.5 83.0 ± 16.8
None 17 38.7 ± 26.9 67.0 ± 22.2 38.2 ± 28.3 41.7 ± 22.8 50.9 ± 22.4
Spiritual attitude ** ** ** ** **
R+S+ 32 71.1 ± 20.2 77.9 ± 18.5 85.4 ± 15.0 75.7 ± 14.0 74.6 ± 19.9
R+S- 36 42.1 ± 21.0 68.2 ± 19.8 81.3 ± 16.6 59.1 ± 20.3 61.8 ± 22.2
R-S+ 9 66.8 ± 14.7 76.7 ± 22.0 50.2 ± 15.6 63.8 ± 20.6 67.9 ± 16.7
R-S- 23 29.0 ± 17.6 61.1 ± 21.3 37.8 ± 22.6 33.3 ± 19.0 42.9 ± 23.6
1Increasing educational level (based on German school system): 1 = secondary education (Hauptschule), 2 = secondary education (junior high; Realschule), 3 = high school education (Gymnasium). Scores are significantly different (** p < 0.01; * p < 0.05; (*) 0.05 < p < 0.10; Kruskal-Wallis-Test for asymptomatic significance).
Deviations of >15% from the mean were highlighted.
The sample contained 257 subjects of whom 70% were women. The mean age was 53.3 ± 13.4 years. The majority had a Christian nomination (80%), 17% had no religious orientation, and 3% other nominations. Cancer was diagnosed in 51%, multiple sclerosis in 24%, and other chronic diseases in 16% (i.e. Hepatitis C, liver cirrhosis, inflammatory bowel disease, severe hypertension etc.); 7% of the individuals were patients with acute diseases (i.e. prolapsed intervertebral disc, stomach ulcer, heart arrhythmia etc). Patients in final stages of their disease were not enrolled.
Measures
The items of the SpREUK 1.0 were developed with the patients' input (cancer service of the Herdecke Community Hospital) and experts' statements (physicians, priest and chaplains working with patients) [28,36], rather than from theoretical concepts. Nevertheless, the original SpREUK 1.0 questionnaire heeded the concept of "internal resp. external locus of control" by Rotter [40] and Levenson [41], "passive, active or collaborative religious coping" by Pargament [41], and the search for "meaning in life" described by Emmons [32,33]. In the final step of the questionnaire design, the items were improved with respect to already existing questionnaires dealing with the topics of religion and spirituality in patients care [36].
According to a previously conducted reliability and factor analysis [36,37] the SpREUK 1.0 version had the following scales: (1) Search for meaningful support, (2) Guidance, control and message of disease, (A) Support in relations with the external through spirituality/religiosity, and (B) Stabilizing the inner condition through spirituality/religiosity.
In order to more precisely differentiate the three topics guidance, control and message of disease in scale 2 of the version 1.0, for the current version of the questionnaire, six new items were added (i.e. F3.5: "My illness is a chance for my own development."; F3.7: "Because of my illness, I reflect on what is essential in my life"). All items were scored on a 5-point scale from disagreement to agreement (0 – does not apply at all; 1 – does not truly apply; 2 – don't know; 3 – applies quite a bit; 4 – applies very much). The SpREUK scores are referred to a 100% level (4 "applied very much" = 100%).
Statistical analysis
Reliability and factor analysis of the new inventory were performed according to the standard procedures. Next, to combine several items with similar content, we relied on the technique of factor analysis which examines the correlations among a set of variables, and to achieve a set of more general "factors." Factor analyses were repeated rotating different numbers of items in order to arrive at the solution which demonstrated both the best simple structure and the most coherence.
Differences in the SpREUK scores were tested using the Kruskal-Wallis-Test for asymptomatic significance. We judged p < 0.05 significant, and 0,05 < p < 0.10 as a trend.
To tested the impact of several variables on the SpREUK sub-scales, we performed analysis of univariate variance (ANOVA). As in several cases Levene's test for equality of variances was significant, and we judged p < 0.01 as significant.
All statistical analyses were performed with SPSS for Windows 10.0.
Results
Reliability
In order to eliminate items from the item pool that were not contributing to the questionnaire reliability, the reliability of the scale and distinct sub-scales was evaluated with internal consistency coefficients, which reflect the degree to which all items on a particular scale measure a single (unidimensional) concept.
Our item pool consisted of the previously established set of items [28,36,37] and 6 new items which were added to differentiate the topic "Guidance, control and message of disease" of the version 1.0. As some of the questions require a positive attitude towards SpR, these items (item pool 2) were separated from the others (item pool 1).
Reliability analysis revealed that 6 items from the new item pool 1 had a poor corrected item-total correlation and thus were eliminated (however, several from the previous "Locus of Control" topic): F1.2 ("I do not need spiritual advice, I know by myself what should be done"; 0.033), F1.3 ("Spiritual/religious ideas are out-of-date"; 0.091), F2.1 ("I have no influence on my life, it is fixed by fate"; 0,025), F2.2 ("I accept my illness and bear it calmly"; 0.073). F2.3 ("My doctor or therapist helps me to keep my illness at bay"; 0.037), and F3.1 ("Whatever happens, I have trust in my inner strength"; 0.173). One item (F3.6 "The "true being" ("inner core") can not be affected by illness") was omitted because of a weak reliability (0.2997) and – even more important – it points to a distinct "field of meaning" that would need more items in the questionnaire, and thus will be used as marker item until the construct will be revised for this topic.
As shown in Table 2, the 15-item construct derived from the item pool 1 had a good quality (Cronbach's alpha = 0.9065). The 14-item construct which refers to the item pool 2 (which is identical to the old item pool 2 as described in [36]) had a very good quality (Cronbach's alpha = 0.9525).
Table 2 Mean values of the items from SpREUK 1.1 and reliability analysis
Factors and Items Mean value (Score 0–4) Standard deviation loading corrected Item-Total correlation Alpha if Item deleted (α = 0.9065)
1: Search for meaningful support
1.5 finding access to a spiritual source can have a positive influence on illness 2.21 1.30 .776 .7597 .8940
1.1 Spiritual attitude 1.96 1.37 .733 .6231 .8994
1.6 searching for an access to SpR 1.81 1.38 .730 .7641 .8935
1.9 urged to spiritual/religious insight 1.99 1.32 .721 .7593 .8940
1.7 others might teach and help to develop spirituality 2.13 1.31 .704 .6881 .8970
1.4 illness has brought renewed interest in SpR questions 1.97 1.40 .636 .6054 .9002
2: Positive interpretation of disease
3.5 illness as a chance for development 2.55 1.30 .813 .7135 .8959
3.4 illness has meaning 2.40 1.33 .703 .6901 .8968
3.2 illness as a hint to change life 2.86 1.05 .604 .6061 .9006
3.7 reflect on what is essential in life because of the illness 3.35 0.77 .570 .2975 .9085
2.4 able to affect the course of illness by themselves 2.58 1.18 .526 .4067 .9066
3.3 illness encourages me to get to know myself better 2.98 1.05 .457 .5026 .9037
3: Trust in external guidance
2.6 Religious attitude 2.71 1.24 .846 .5087 .9033
2.5 trust in a higher power. 2.71 1.24 .810 .5327 .9028
1.8 looking for purpose and meaning in life 3.02 1.21 .295 .4005 .9072
Factors and Items Mean value (Score 0–4) Standard deviation loading corrected Item-Total correlation Alpha if Item deleted (α = 0.9525)
4: Support in relations with the External life through SpR
4.1 plays a major role in life 2.23 1.38 .819 .7934 .9479
4.3 helps to manage life more consciously 2.64 1.18 .814 .9187 .9452
4.2 provides deeper connection with the world around 2.52 1.20 .807 .82112 .9475
4.4 helps to cope better with illness 2.49 1.23 .806 .8718 .9463
4.7 helps to restore mental and physical health 2.30 1.17 .720 .8530 .9465
4.8 practicing with others deepens SpR 1.80 1.34 .663 .6189 .9523
4.6 helps to view disease as a beneficial challenge for own development 2.06 1.24 .657 .8130 .9474
4.9 practicing alone and in silence deepens SpR 2.56 1.21 .594 .6139 .9523
4.10 distinct places stimulate SpR 2.64 1.32 .584 .5709 .9537
4.5 People who share SpR attitudes are important 2.47 1.17 .385 .7956 .9479
5: Support of the Internal life through SpR
5.4 refers to an inner power 2.13 1.27 .765 .4374 .9572
5.1 provides feeling of contentment and inner peace 2.64 1.19 .732 .8671 .9465
5.2 promotes inner strength. 2.46 1.21 .713 .8757 .9462
5.3 refers to a higher (external) power 2.74 1.30 .574 .7470 .9491
Thus, the internal consistency of the 29-item SpREUK 1.1 construct was sufficiently high. The level of difficulty (LoD = 2.482 [mean value] / 4) is 0.6205 for item pool 1 resp. (LoD = 2.406 [mean value] / 4) 0.6014 for item pool 2. With the exception of item F3.7 ("I reflect on what is essential in my life because of the illness"; LoD = 0.838), all values are in the acceptable range from 0.2 to 0.8.
Factor analysis
To combine several items with similar content, we relied on the technique of factor analysis which examines the correlations among a set of variables, and to achieve a set of more general "factors." Factor analyses were repeated rotating different numbers of items in order to arrive at the solution which demonstrated both the best simple structure and the most coherence.
With a Kaiser-Mayer-Olkin value of 0.850 (item pool 1) resp. 0.939 (item pool 2), which measures the degree of common variance, the 15 resp. 14-item-pool seems to be suitable. Barlett's test for non-sphericity was highly significant (p < 0,001).
Primary factor analysis of item pool 1 pointed to a 5-factor solution. However, due to a low item number in the tentative subscales 2–5 (with 2 or 3 items each), we favoured the more appropriate 3-factor solution which explains 53.8% of variance (Table 2). Sub-scale 1 ("Search for meaningful support") with its 6 items had a Cronbach's alpha of 0.8549, sub-scale 2 ("positive interpretation of disease ") with its 6 items had an alpha of 0.8000, while the 3-item sub-scale 3 ("trust in external guidance") had an alpha of 0.6625. An spiritual attitude loads to sub-scale 1, while a religious orientation loads to sub-scale 3.
Factor analysis of item pool 2 (Table 2) pointed to a 2-factor solution which explains 58.8% of variance. The 10-item subs-scale 4 ("Support in relations with the External life through SpR") had a Cronbach's alpha of 0.9400, while sub-scale 5 ("Support of the Internality through SpR") with its 4 items had an alpha of 0.7828.
Thus, this the internal consistency of the item pool 1 was sufficiently high. However, there are several inter-correlations between the sub-scales (Table 3). "Search for meaningful support" correlated negatively with "trust in external guidance" and slightly with the "positive interpretation of disease". Moreover, "trust in external guidance" negatively correlated with "positive interpretation of disease".
Table 3 Component Transformation Matrix
Scale Search Meaning Message Disease Trust Guidance Support External Support Internal
1 Search Meaning .745 .566 .352
2 Message Disease -.304 .758 -.577
3 Trust Guidance -.594 .323 .737
4 Support External .861 -.509
5 Support Internal -.509 .861
Components 1, 2 and 3 explain 53.8% of variance, while components 4 and 5 explain 58.8% of variance.
Analysis of the "side-loadings" of item pool 1 (only values > 0.35 were take into account) reveal that items F1.8 ("looking for purpose and meaning in life") load good on sub-scale 2 (0.414). Analysis of the side-loadings of item pool 2 revealed that several items load also on the other sub-scale. Moreover, sub-scale 4 showed a strong but negative inter-correlation with sub-scale 5 (Table 3).
Relation between SpREUK scores and demographic variables
The highest scores were found for the sub-scales 2 and 3 ("positive interpretation of disease" resp. "trust in external guidance"), the lowest for sub-scale 1 ("Search for meaningful support"). Means and standard deviations for study variables are provided in Table 1.
Women had significantly higher SpREUK scores than male patients. With respect to age, the lowest SpREUK scores were found in the group of < 30 years of age. With increasing age, the trust in a higher supporting presence (sub-scale 3) and the beneficial effects of resp. support through SpR increased.
With respect to the marriage status, widowed patients obviously has to rely on external guidance (sub-scale 3) but not the patients living with a partner not married with. Widowed and divorced patients find support in external relations through their SpR engagement (sub-scale 4), while – in contrast to married patients which may find hold in their partnership – especially divorced patients are in search for meaningful support (sub-scale 1).
Search for meaningful support and positive interpretation of diseases were depending on the educational level, as patients with lower educational level had significantly lower scores than those with a higher level. A higher educational level was associated with higher scores in the sub-scales 4 and 5 which deals with the beneficial effects of SpR.
Illness itself (but not the duration of disease) has a significant impact on the SpREUK scores, as MS patients had the lowest scores in all 5 sub-scales. The SpREUK scores of cancer patients revealed slight differences when compared to patients with other chronic diseases.
With the exception of sub-scale 2, patients without confessional affiliations had the lowest scores for all sub-scales, indicating that the "message of disease" was not depending on a denomination. Surprisingly, the few patients with other than a Christian orientation had the highest scores for sub-scales 1, 3, 4, 5.
Since nominational affiliation is not necessarily identical with religiosity or spirituality, we asked whether the patients would describe themselves as religious or spiritual [28,34,35]. Thirty-two % reported themselves as both religious and spiritual (R+S+); 35% as religious, but not spiritual (R+S-); 23% as neither religious nor spiritual (R-S-); 10% claimed that they were spiritual, but not religious (R-S+). Thus, the numbers of patients with denominational affiliation and self-reported spiritual/religious attitudes is somewhat similar. A spiritual attitude (R+S+ and R-S+) was associated with "search for meaningful support" and "positive interpretation of disease", while a religious attitude (R+S+ and R+S-) was associated with the highest scores for the "trust in external guidance" sub-scale 3.
The living area and the duration of diseases had no significant impact on the SpREUK scores.
Correlation with SpR practice
As shown in Table 4, there were moderate to strong correlations between the SpREUK sub-scales and the engagement in a SpR practice as measured by the SpREUK-P manual [28,29]. The new version of SpREUK-P [29] measures (1) conventional religious practice (praying, church attendance etc.), (2) nature-oriented practice (healing effect on environment etc.), (3) existentialistic practice (self-realization, spiritual development, higher level of consciousness etc.), (4) unconventional spiritual practice (meditation, rituals, body-mind discipline etc.), and (5) humanistic practice (make an effort for other people etc.). The "nature-oriented practice", "humanistic practice" and the "unconventional spiritual practice" were only weakly associated with "Trust in external guidance". In contrast, a "conventíonal religious practice" was strongly correlated with "Search for meaningful support", "Trust in external guidance" and both "Support through SpR" scales 4 and 5. An "unconventional spiritual practice" was associated more with "Search for meaningful support" and " Support in relations with the External life through SpR", while an "esistentialistic practice" was associated stronger with "positive interpretation of disease" and "Support in relations with the External life through SpR.
Table 4 Pearson correlation between SpREUK sub-scales and SpR practice1
Search Meaning Message Disease Trust Guidance Support External Support Internal
SpREUK-P engagement scores
conventional religious practice .577 ** .424 ** .642 ** .691 ** .624**
nature-oriented practice .247 ** .246 ** .266 * .358 ** .334 **
existentialistic practice .437 ** .530 ** .411 ** .479 ** .439 **
unconventional spiritual practice .498 ** .459 ** .223 * .506 ** .431 **
humanistic practice .362 ** .306 ** .241 ** .381 ** .327 **
Selected SpREUK-P items
praying .432 ** .348 ** .670 ** .528 ** .516 **
church attendance .290 ** .145 .473 ** .425 ** .324 **
meditation .452 ** .378 ** .137 .421 ** .374 **
make an effort for others .160 .167 -.077 .065 -.006
1engagement in SpR practice was measured with an additional manual of the SpREUK questionnaire, the SpREUK-P manual (Büssing et al., 2005).
Bivariate correlations are statistically significant with ** p < 0.01; * p < 0.05 (2-tailed significance)
In detail (Table 4), praying and church attendance were strongly correlated with "Trust in external guidance" and both "Support through SpR" scales 4 and 5, while church attendance did not correlate at all with "Message of disease" and only weakly with "Search for meaningful support". In agreement with the results of our study, meditation did not correlate with "Trust in external guidance". However, an attitude of "making and effort for others" did not correlate at all with our SpREUK sub-scales.
Analyses of variance
Next we tested the impact of several variables on the SpREUK sub-scales, such as sex and marital status, educational level and confession, age and SpR attitude, and disease and duration of disease. Using the method of univariate analyses of variance we identified several sources of variability (Table 5):
Table 5 Univariate variance analyses
Variables F-value significance
(1) Search for meaningful support SpR attitude
age 46.429
0.784 0.000
n.s.
Confession
educational level 0.205
4.205 n.s.
0.007
disease
duration of disease 3.784
0.991 0.011
n.s.
(2) Positive interpretation of disease SpR attitude
age 7.710
1.128 0.000
n.s
(3) Trust in external guidance SpR attitude
age 82.148
1.717 0.000
0.007
Confession
educational level
Confession * education 9.117
3.630
3.822 0.000
0.015
0.006
(4) Support in relations with the External life through SpR SpR attitude
age 51.319
1.120 0.000
n.s
Confession
educational level 0.511
2.697 n.s
0.049
(5) Support of the Internal life through SpR SpR attitude
age 51.319
1.120 0.000
n.s
In this table, only significant results were given. Levene's test for equality of variances was significant and thus the level of significance should be p < 0.01.
• The SpR attitude is an important covariate for the "Search for meaningful support", "Positive interpretation of disease", "Trust in external guidance", and both "Support through SpR" sub-scales.
• The educational level is an important covariate for "Search for meaningful support", "Trust in external guidance" and to a minor content for "Support in relations with the External life through SpR" – but not for the "Positive interpretation of disease".
• Age is an important covariate only for "Trust in external guidance".
• Confession is an important covariate for "Trust in external guidance".
• Disease itself has an impact on the "Search for meaningful support
Discussion
Data from the current analysis demonstrate the reliability and validity of the SpREUK construct. Moreover, the sub-scales 1 and A (= 4) and B (= 5) of the preliminary version 1.0 were confirmed in the new version 1.1. In order to more precisely differentiate the three topics guidance, control and message of disease from the SpREUK version 1.0, six new items were added. Due to this fact, some items from the original item pool decreased the reliability of the construct and thus, two items from the sub-scale 1 had to be deleted ("I do not need spiritual advice" and "Spiritual/religious ideas are out-of-date", and four items which deal with the "internal/external locus of control" topic as described by Rotter [40] and Levenson [41] ("I know by myself what should be done"; "Whatever happens, I have trust in my inner strength"; "I have no influence on my life, it is fixed by fate"; "I accept my illness and bear it calmly"; "My doctor or therapist helps me to keep my illness at bay"). However, the current item pool 1 made up the new sub-scale 2 which highlights the positive interpretation of disease ("message of disease") and the new sub-scale 3 which asks for the trust in an external guidance ("God"). To improve the quality of this 3-item-scale, we have added two additional items.
The search for "meaning in life" as described by Emmons [32,33] respectively the concept of "meaning-based coping" are important topics of our questionnaire. However, the item "looking for purpose and meaning in life" loads to the sub-scale 3 which is obviously not identical with the "Search for meaningful support" through spirituality as measured in sub-scale 1. The items of sub-scale 3 fit well to the concept of "external locus of control" and share several topic with Belschner's scale "Transpersonal Trust" [43,44], while the items which made up the new sub-scale 2 (which addresses the "message of disease" and how the patients actively respond to their illness) may fit to the concept of "internal locus of control". This topic of "meaning of disease" is of outstanding importance for cancer patients [45-49], in as much as health can be conceptualized as a competence to gain control for the design of the biography [34]. In consequence, loss of control due to a life-threatening illness might be interpreted by patients as "punishment", "weakness" or "irreparable loss" – illness has no positive meaning, no "signal" to change aspects of life. As reported by Degner et al. [46], women who ascribed a negative meaning of illness had significantly higher levels of depression and anxiety and poorer quality of life than women who indicated a more positive meaning. As Spiritual Well-Being can be described as a 2-factor construct, i.e. Religious Well-Being and Existential Well-Being [48], addressing existentialistic concerns and the possibility to find some kind of sense and meaning even in illness are thus functions of spiritual well-being. In breast cancer patients, Levine and Tarq [8] found significant correlations of spirituality and spiritual well-being with functional well-being, while items pertaining to meaning and peace tended to correlate significantly with physical well-being. Moreover, the spirituality scales accounted for 40% of the variance in functional well-being, thus confirming the importance of spirituality and spiritual well-being in both physical and functional well-being of cancer patients.
Conclusions
The SpREUK questionnaire may have important strengths. First, it appears to be a good choice for assessing a patients interest in spiritual concerns which is not biased for or against a particular religious commitment. Moreover, as several patients may be offended by institutional religion, even terms such as God, Jesus, praying, church etc. were avoided. Moreover, the subscale "Search for meaningful support" thus had a good correlation with both, an engagement in conventional religious practice and unconventional spiritual practice.
A second strength is that the subscale "positive reinterpretation ("message") of disease" has a good correlation with an existentialistic practice, which seems to be of outstanding importance for patients with life-changing diseases. It may be desirable to use such a measure that allows to assess attitudes which are independent of any religion or specific belief.
A third strength is that the validation was performed in a sample with at least two different types of life-changing diseases (cancer and MS, and other chronic diseases) and a healthy control group.
Beyond conceptual boundaries, our instrument differentiates the self-addressed "religious" and "spiritual" attitudes of the patients with life-threatening diseases and heeds their search for support and meaning, and integrates the topic of "meaning in illness". We cannot exclude the possibility that these topics are not relevant for healthy individuals.
In future studies we have to correlate our scales with other relevant instruments which measure aspects of SpR. Nevertheless, evaluation of the SpREUK questionnaire indicates that it is a reliable, valid measure of distinct topics of SpR that may be especially useful of assessing the role of non-religious spirituality in health related research. The focus of a larger study is to enrol patients from the highly secular Eastern Europe, and to run longitudinal studies with cancer, multiple sclerosis patients, but also cardiac failure and spinal cord damage.
The SpREUK with its additional SpREUK-P manual to measure a patient's engagement in distinct forms of SpR practice is currently available in English and German language.
Authors' contributions
AB conceived the study, designed and developed the questionnaire, performed statistical analysis and drafted the manuscript. TO participated to conceive and design the study, performed additional statistical analysis and helped to draft the manuscript. PFM conceived the study and participated in the design and development of the questionnaire. All authors read and approved the final manuscript.
Acknowledgements
We are grateful to our patients, and to Dr. Cristina Stumpf and Dr. Mette Kaeder for their help to recruit them.
==== Refs
Ellison CG George LK Religious involvement, social ties, and social support in a southeastern community J Scient Study Religion 1994 33 46 61
Sloan RP Bagiella E Powell T Religion, spirituality, and medicine The Lancet 1999 353 664 667 10030348 10.1016/S0140-6736(98)07376-0
Thoresen CE Spirituality and Health: Is There a Relationship? J Health Psychol 1999 4 291 300
Lukoff D Provenzano R Lu F Turner R Religious and spiritual case reports on Medline: A Systematic analysis of records from 1980–1996 Altern Therap Health Med 1999 5 64 70 9893317
McCullough ME Hoyt WT Larson DB Koenig HG Thoresen C Religious involvement and mortality: A meta-analytic review Health Psychol 2000 19 211 222 10868765 10.1037//0278-6133.19.3.211
Luskin FM A review of the effect of religious and spiritual factors on mortality and morbidity with a focus on cardiovascular and pulmonary disease J Cardiopulm Rehabil 2000 2 8 15 10680093 10.1097/00008483-200001000-00002
Sloan RP Bagiella E Claims about religious involvement and health outcomes Annals Behav Med 2002 24 14 21 10.1207/S15324796ABM2401_03
Levine EG Targ E Spiritual correlates of functional well-being in women with breast cancer Integr Cancer Ther 2002 1 166 174 14664742 10.1177/1534735402001002008
Powell LH Shahabi L Thoresen CE Religion and spirituality. Linkages to physical health Am Psychol 2003 58 36 52 12674817 10.1037/0003-066X.58.1.36
Seemann T Dubin LF Seemann M Religiosity/Spirituality and Health. A critical review of the evidence for biological pathways Am Psychol 2003 58 53 63 12674818 10.1037/0003-066X.58.1.53
Cotton SP Levine EG Fitzpatrick CM Dold KH Targ E Exploring the relationships among spiritual well-being, quality of life, and psychological adjustment in women with breast cancer Psychooncol 1999 8 429 438 10.1002/(SICI)1099-1611(199909/10)8:5<429::AID-PON420>3.0.CO;2-P
Post-White J Ceronsky C Kreitzer MJ Nickelson K Drew D Mackey KW Koopmeiners L Gutknecht S Hope, spirituality, sense of coherence, and quality of life in patients with cancer Oncol Nurs Forum 1996 23 1571 1579
McGrath P Spirituality and discourse: a postmodern approach to hospice research Aus Health Rev 1997 20 116 128
Fehring RJ Miller JF Shaw C Spiritual well-being, religiosity, hope, depression, and other mood states in elderly people coping with cancer Oncol Nurs Forum 1997 24 663 671 9159782
Nelson CJ Rosenfeld B Breitbart W Galietta M Spirituality, religion, and depression in the terminally ill Psychosomatics 2002 43 213 220 12075036 10.1176/appi.psy.43.3.213
McClain CS Rosenfeld B Breitbart W Effect of spiritual well-being on end-of-life despair in terminally-ill cancer patients Lancet 2003 361 1603 1607 12747880 10.1016/S0140-6736(03)13310-7
Allport GW Ross JM Personal religious orientation and prejudice J Personal Social Psychol 1967 5 432 443 10.1037/0022-3514.5.4.432
Batson CD Schoenrade PA Measuring religion as Quest: Validity concerns J Scien Study Religion 1991 30 416 429
Maltby J Lewis CA Measuring Intrinsic and Extrinsic Orientation Toward Religion: Amendments for its use among religious and non-religious samples Personal Indiv Differences 1996 21 937 946 10.1016/S0191-8869(96)00154-7
Maltby J Day L Amending a measure of the Quest Religious Orientation: Applicability of the scale's use among religious and non-religious persons Personal Indiv Differences 1998 25 517 522 10.1016/S0191-8869(98)00078-6
MacDonald DA Spirituality and the Five Factor Model J Personal 2000 68 153 197 10.1111/1467-6494.00094
Bufford RK Paloutzian RF Ellison CW Norms for the spiritual well-being scale J Psychol Theol 1991 19 56 70
Underwood LG Teresi JA The Daily Spiritual Experience Scale: Development, Theoretical Description, Reliability, Exploratory Factor Analysis, and Preliminary Construct Validity Using Health-Related Data Ann Behav Med 2002 24 22 33 12008791 10.1207/S15324796ABM2401_04
Plante TG Boccaccini MT The Santa Clara Strength of Religious faith Questionnaire Pastoral Psychol 1997 45 375 387
Walach H Bausteine für ein spirituelles Welt- und Menschenbild Transpers Psychol Psychother 2001 2 63 77
Walach H Spiritualität und Wissenschaft. Zum Verständnis (und zur Überwindung) eines Tabus Erfahrungsheilkunde 2003 10 650 659
Peterman AH Fitchett G Brady MJ Hernandez L Cella D Measuring spiritual well-being in people with cancer: the functional assessment of chronic illness therapy – Spiritual Well-being Scale (FACIT-Sp) Ann Behav Med 2002 24 49 58 12008794 10.1207/S15324796ABM2401_06
Büssing A Ostermann T Patzek M Caritas und ihre neuen Dimensionen: Spiritualität und Krankheit Caritas plus Qualität hat einen Namen 2004 Kevelaer: Butzon & Bercker 110 133
Büssing A Ostermann Th Matthiessen PF Spirituelle Bedürfnisse krebskranker Menschen – Einstellung und Praxis German J Oncol 2005 37 12 21
Tamburini M Gangeri L Brunelli C Boeri P Borreani C Bosisio M Karmann CF Greco M Miccinesi G Murru L Trimigno P Cancer patients' needs during hospitalisation: a quantitative and qualitative study BMC Cancer 2003 3 12 12710890 10.1186/1471-2407-3-12
Atkinson MJ Wishart PM Wasil BI Robinson JW The Self-Perception and Relationships Tool (S-PRT): A novel approach to the measurement of subjective health-related quality of life Health and Quality of Life Outcomes 2004 2 36 15257754 10.1186/1477-7525-2-36
Emmons RA Cheung C Tehrani K Assessing spirituality through personal goals: Implications for research on religion and subjective well being Social Indicators Res 1998 45 391 422 10.1023/A:1006926720976
Emmons RA Striving for the sacred: Personal goals, life meaning and religion J Social Issues 2004
Belschner W Tun und lassen – Ein komplementäres Konzept der Lebenskunst Transpers Psychol Psychother 2001 7 85 102
Koenig HG Larson DB Larson SS Religion and Coping with Serious Medical Illness Ann Pharmacother 2001 35 352 359 11261534 10.1345/aph.10215
Ostermann Th Büssing A Matthiessen PF Pilotstudie zur Entwicklung eines Fragebogens zur Erfassung spiritueller und religiöser Einstellung und des Umgangs mit Krankheit (SpREUK) Forsch Komplementärmed Klass Naturheilk 2004 11 346 353 10.1159/000082816
Büssing A Ostermann Th Matthiessen PF Role of Religion and Spirituality in Medical patients in Germany Journal of Religion and Health 2005 44 in press
Büssing A Ostermann Th Matthiessen PF Search for meaningful support and the meaning of illness in German cancer patients Anticancer Res 2005 25 in press
Büssing A Ostermann Th Matthiessen PF Fintelmann V „Spiritualität als Patientenbedürfnis" Onkologie auf anthroposophischer Grundlage 2005 Stuttgart: Johannes M. Mayer-Verlag Chapter 2.4.1
Rotter J Generalized expectations for internal versus external control reinforcement Psychological Monographs: General and Applied Psychology 1966 80 1 27
Levenson H Multidimensional locus of control in psychiatric patients J Consul Clini Psychol 1973 41 397 404
Pargament KI The psychology of religion and coping: Theory, research, practice 1997 New York, Guilford Press
Belschner W Die Skala Transpersonales Vertrauen. Manual Transpersonale Arbeitspapiere Nr 3 Oldenburg, Universität Oldenburg, AE Gesundheits- und Klinische Psychologie: BIS 1998
Albani C Bailer H Blaser G Geyer M Brähler E Grulke N Psychometrische Überprüfung der Skala "Transpersonales Vertrauen" (TPV) in einer repräsentativen Bevölkerungsstichprobe Transpers Psychol Psychother 2002 2 86 98
Wallberg B Michelson H Nystedt M Bolund C Degner L Wilking N The meaning of breast cancer Acta Onco 2003 42 30 35 10.1080/0891060310002203
Degner LF Hack T O'Neil J Kristjanson LJ A new approach to eliciting meaning in the context of breast cancer Cancer Nur 2003 26 169 78 10.1097/00002820-200306000-00001
Ferrell BR Smith SL Juarez G Melancon C Meaning of illness and spirituality in ovarian cancer survivors Oncol Nurs Forum 2003 30 249 257 12692659
Genia V Evaluation of the Spiritual Well-Being Scale in a Sample of College Students Int J Psychol Religion 2001 11 25 33 10.1207/S15327582IJPR1101_03
Little M Sayers EJ While there's life ... hope and the experience of cancer Soc Sci Med 2004 59 1329 1337 15210103 10.1016/j.socscimed.2004.01.014
| 15705195 | PMC550666 | CC BY | 2021-01-04 16:38:15 | no | Health Qual Life Outcomes. 2005 Feb 10; 3:10 | utf-8 | Health Qual Life Outcomes | 2,005 | 10.1186/1477-7525-3-10 | oa_comm |
==== Front
Cardiovasc UltrasoundCardiovascular Ultrasound1476-7120BioMed Central London 1476-7120-3-41571793610.1186/1476-7120-3-4ResearchAssessment of atrial regional and global electromechanical function by tissue velocity echocardiography: a feasibility study on healthy individuals Quintana Miguel [email protected] Peter [email protected] Samir K [email protected] Furia Francesca [email protected] Britta [email protected] Satish [email protected] Lars-Åke [email protected] Department of Clinical Physiology, Karolinska University Hospital, Huddinge. The Karolinska Institute, Stockholm, Sweden2 Department of Cardiology, Karolinska University Hospital, Huddinge. The Karolinska Institute, Stockholm, Sweden2005 18 2 2005 3 4 4 21 1 2005 18 2 2005 Copyright © 2005 Quintana et al; licensee BioMed Central Ltd.2005Quintana 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 appropriate evaluation of atrial electrical function is only possible by means of invasive electrophysiology techniques, which are expensive and therefore not suitable for widespread use. Mechanical atrial function is mainly determined from atrial volumes and volume-derived indices that are load-dependent, time-consuming and difficult to reproduce because they are observer-dependent.
Aims
To assess the feasibility of tissue velocity echocardiography (TVE) to evaluate atrial electromechanical function in young, healthy volunteers.
Subjects and methods
We studied 37 healthy individuals: 28 men and nine women with a mean age of 29 years (range 20–47). Standard two-dimensional (2-D) and Doppler echocardiograms with superimposed TVE images were performed. Standard echocardiographic images were digitized during three consecutive cardiac cycles in cine-loop format for off-line analysis. Several indices of regional atrial electrical and mechanical function were derived from both 2-D and TVE modalities.
Results
Some TVE-derived variables indirectly reflected the atrial electrical activation that follows the known activation process as revealed by invasive electrophysiology. Regionally, the atrium shows an upward movement of its walls at the region near the atrio-ventricular ring with a reduction of this movement towards the upper levels of the atrial walls. The atrial mechanical function as assessed by several TVE-derived indices was quite similar in all left atrium (LA) walls. However, all such indices were higher in the right (RA) than the LA. There were no correlations between the 2-D- and TVE-derived variables expressing atrial mechanical function. Values of measurement error and repeatability were good for atrial mechanical function, but only acceptable for atrial electrical function.
Conclusion
TVE may provide a simple, easy to obtain, reproducible, repeatable and potentially clinically useful tool for quantifying atrial electromechanical function.
atrial mechanical functionatrial electrical impulse spreadingleft ventricular functiontissue Doppler echocardiography
==== Body
Introduction
The enlargement of left atrial (LA) diameter is associated with cardiovascular disease and is a risk factor for atrial fibrillation, stroke and death. [1-6] LA function reliably predicts exercise capacity in patients with recent myocardial infarction[7] or non-ischemic dilated cardiomyopathy[8] and differs in patients with ischemic and dilated cardiomyopathy.[9] Moreover, LA volume is an independent prognostic factor in several subsets of patients. [10-12] Although commonly used, LA size assessed by M-mode echocardiography does not correlate well with LA volumes, so several methods to estimate LA volumes have been developed.[13,14] The LA reservoir, conduit and pump functions may be estimated from volume measurements. [15-19] However, the reliability and clinical usefulness of those methods have been poorly studied. Pulsed-wave Doppler interrogation of the blood flow velocity during atrial contraction, the peak mitral inflow A wave, and its velocity time integral have also been used as surrogate markers of atrial function. [20-22] These variables represent the diastolic properties of the LV [23-25] and do not accurately reflect atrial mechanical properties. Tissue velocity echocardiography (TVE) has now been developed as a valuable tool for the evaluation of left and right ventricular systolic and diastolic functions.[26,27] Furthermore, this technique has also been used to assess the regional functions of the left and right atrium.[28,29] Although atrial anatomy was described more than a century ago, a new interest in atrial anatomy and its relations with atrial electromechanical function has only recently emerged.[30,31] Conventionally, atrial electrical function has been evaluated from resting electrocardiography (ECG), and more accurately by invasive electrophysiology techniques. [32-34] The rapid development of these invasive techniques has improved not only diagnostic capabilities,[35,36] but also our understanding of how the electrical impulse spreads through atrial tissues,[37,38] and has led to improvements in the treatment of supra-ventricular tachy-arrhythmias. [38-40] However, the invasive nature and the high costs of these procedures limit their widespread use and repeatability. Therefore, the development of noninvasive, safe, accurate and repeatable methods that might provide similar information is necessary. We aimed here to find simple and repeatable methods to assess both electrical and mechanical regional atrial functions by means of TVE.
Methods
Population
We studied 37 healthy individuals: 28 men and nine women with a mean age of 29 years (range 20–47). The individuals were recruited from among hospital employees, cardiovascular technicians and medical students. None showed symptoms of cardiovascular disorders or were receiving pharmacological cardiovascular agents. All had normal standard two-dimensional (2-D) and Doppler echocardiograms. All subjects were on sinus rhythm and none hade A-V or intra-ventricular conduction defects. The Ethical Committee at the Karolinska University Hospital, Huddinge, approved the study. All individuals received written information and gave informed consent.
Echocardiography
A standard 2-D and Doppler echocardiogram with superimposed TVE images was performed using a 3.5 MHz transducer with commercially available equipment (System FiVe™, GE Vingmed, Horton, Norway). Standard parasternal short- and long-axis views as well as apical 2-, 3- and 4-chamber views acquired at expiratory apnea with at least 90 frames per second were digitized during three consecutive cardiac cycles in cine-loop format for off-line analysis.
Off-line analyses
All echocardiographic images were analyzed off-line using software (Echopac™ 6.3.4, GE Vingmed) for the calculation of standard 2-D and Doppler echocardiography as well as for the analysis of TVE variables.
Standard 2-D and Doppler echocardiography
Measurements of the left ventricular (LV) function comprised septum and posterior wall thickness; LV end-systolic and diastolic dimensions; LV fractional shortening, and LV ejection fraction (LVEF) according to international standards.[41] Measurements of atrial function comprised left atrial (LA) diameter measured from the parasternal long axis; right atrium (RA) and LA long and short axes; LA and RA maximal volume; LA and RA minimal volume, and RA and LA volumes at the beginning of the P-wave measured from the apical 4- and 2-chamber views. LA and RA ejection fractions were measured according to the formula: (maximal volume-minimal volume)/maximal volume. LA and RA active emptying values were calculated as (volume at P-wave-minimal volume)/volume at P-wave.[8,9,42]
Tissue velocity echocardiography
The RV and LV long axis functions were assessed from apical views. Six basal LV segments were identified as follows: the RV free wall; the LV postero-septal wall, and the LV lateral wall from the apical 4-chamber view; the LV inferior and anterior walls from the apical 2-chamber view, and the LV posterior wall from the apical 3-chamber view. A sample volume was positioned at the base of each ventricular wall excluding the A-V plane during the entire heart cycle to obtain a tissue velocity profile during three consecutive cardiac cycles. Both systolic and diastolic phases of the velocity profile were considered and the following parameters were analyzed (upper part of Fig. 1): peak systolic velocity (PSV, in centimeters per second), measured at the peak velocity during the ejection period; peak velocity at early diastole (E'-wave, in centimeters per second), measured at the peak velocity at early diastole, and peak velocity at late diastole (A'-wave, in centimeters per second), measured at the peak velocity at late diastole. The atrio-ventricular myocardial wall displacement (A'-V' disp., in millimeters) in the long axis was obtained by automated temporal integration of the PSV of the basal segments during the ejection period.
Figure 1 Assessment of atrial and ventricular mechanical function. The upper panel shows the systolic and diastolic velocities (a) and the A-V place displacement (b) measured at the basal level of the inter ventricular septum. The lower panel shows the atrial velocity (c), atrial displacement (d), atrial strain rate (e) and atrial strain (f) measured at the inter atrial septum below the mitral ring.
The different atrial walls were identified from the same apical views as follows: the right atrial wall (RA), the inter-atrial septum (IAS), and the left atrial lateral wall (LA-Lat) from the apical 4-chamber view; the left atrial inferior wall (LA-Inf) and the left atrial anterior wall (LA-Ant) from the apical 2-chamber view; and the left atrial posterior wall (LA-Post) from the apical 3-chamber view. Each atrial wall was studied at low and mid levels, placing a 2 mm sample volume at low atrial walls excluding the A-V plane during the entire cardiac cycle and at the mid portion of each atrial wall. The regional electromechanical function at each atrial wall was studied by the following time intervals (Figure 2). The PA-start interval (P-Aa' start) was defined as the time between the beginning of the P-wave on the monitor's ECG to the start of the A' wave on the TVE-curve profile. The PA-peak interval (P-Aa' peak) was the time between the beginning of the P-wave on the monitor's ECG to the peak of the A' wave on the TVE-curve profile. The A-wave duration (Aa'-dur.) was the time from the beginning to the end of the A'-wave on the TVE-curve profile. The total electromechanical activity (TEMA) was the time between the beginnings of the P-wave on the monitor ECG to the end of the A' wave on the TVE-curve profile.
Figure 2 Assessment of some time intervals and Aa' wave velocity at the low level of the inter atrial septum
The regional mechanical function of each atrial wall was assessed by the peak velocity during atrial contraction (Aa' peak vel.), the atrial displacement occurring during atrial contraction (Aa' disp.) and the ratio of atrial displacement measured at atrial level to the total LV myocardial displacement measured at ventricular level (Aa' cont.) (lower part of Figure 1). In addition, strain rate (Aa' SR) and strain (Aa' S) were assessed in each low atrial wall using a sample volume of 12 mm.
Statistical analyses
Data are presented as means ± standard deviations (SD). Analysis of variance (ANOVA) with repeated measures was used to test statistical significance of the studied variables at different atrial and ventricular walls. When ANOVA showed statistically significant differences among atrial and ventricular walls, post hoc analysis with Bonferroni's test was performed to assess differences among those walls. Correlation coefficients were calculated to assess the relationship among several markers of atrial mechanical function. The inter- and intra-observer repeatability and measurement errors for variables reflecting the atrial electromechanical function were assessed by the coefficient of variation and by the British Standards Institution method, the value below which the difference between two measurements will lie with a probability of 0.95. P < 0.05 was considered statistically significant.
Results
All demographic features and measures of standard 2-D and pulsed-wave Doppler echocardiography data are shown in Table 1. Of interest, no differences were found between measures of RA and LA functions, as assessed by short or long axes, or among volumes and volume-derived indices. TVE-derived variables assessing the RV and LV long-axis systolic and diastolic functions are shown in Table 2. No significant differences among LV walls were found for any index of systolic and diastolic function. TVE-derived variables obtained from the RV free wall were significantly different from each LV wall.
Table 1 Demographic features and resting echocardiographic data. Numbers are means ± SD.
Age, years 29 ± 7
Gender (M/F) 28/9
Height, cm 175 ± 8
Weight, kg 76 ± 14
Heart rate, bpm 66 ± 12
P-Q time, ms 166 ± 16
LA diameter, mm/m2 18.9 ± 1.5
Septal wall thickness, mm 9.6 ± 1.1
Posterior wall thickness, mm 9.5 ± 1.2
LV end diastolic diameter, mm/m2 26.5 ± 2.3
LV fractional shortening, % 35 ± 6
LV ejection fraction, % 72 ± 8
E-wave, cm 90 ± 17
A-wave, cm 56 ± 12
E/A ratio 1.68 ± 0.45
LA long axis, mm/m2 26 ± 2 RA long axis, mm/m2 25 ± 3
LA short axis, mm/m2 21 ± 3 RA short axis, mm/m2 22 ± 2
LA maximal volume, ml/m2 29 ± 5 RA maximal volume, ml/m2 31 ± 7
LA minimal volume, ml/m2 15 ± 3 RA minimal volume, ml/m2 17 ± 4
LA P-wave volume, ml/m2 18 ± 4 RA P-wave volume, ml/m2 19 ± 5
LA ejection fraction, % 49 ± 9 RA ejection fraction, % 46 ± 10
LA active emptying, % 17 ± 7 RA active emptying, % 15 ± 9
Table 2 Systolic and diastolic myocardial velocities measured at different right and left ventricular walls
Variable Ventricular walls
RV Post-sep Lateral Inferior Anterior Posterior
PSV, cm/s 10.5 ± 1.3 6.8 ± 0.9 8.5 ± 1.8 7.5 ± 1.1 7.9 ± 1.6 7.6 ± 1.3
E'-wave, cm/s 10.2 ± 2.3 9.9 ± 1.1 12.2 ± 1.5 10.6 ± 2.1 10.1 ± 2.1 12.4 ± 1.9
A'-wave, cm/s 8.4 ± 2.8 5.9 ± 1.3 4.7 ± 1.5 6.3 ± 2.0 4.7 ± 1.7 6.1 ± 2.1
E'/A' ratio 1.2 ± 0.1 1.7 ± 0.1 2.6 ± 0.1 1.7 ± 0.2 2.1 ± 0.2 2.0 ± 0.1
A'-V' disp., mm 21.5 ± 3.5 13.6 ± 1.5 13.7 ± 2.0 15.3 ± 1.6 13.7 ± 1.9 15.4 ± 1.8
Atrial disp., mm 5.8 ± 2.2 4.1 ± 1.3 2.5 ± 0.8 3.8 ± 1.5 2.9 ± 0.9 3.0 ± 1.1
Atrial cont., % 27 ± 8 30 ± 9 19 ± 7 25 ± 9 22 ± 8 20 ± 6
Abbreviations: cont., contribution; disp., displacement; PSV, peak systolic velocity
Table 3 shows several time intervals. The PA-start interval (P-Aa' start) was longer at low atrial levels in each atrial wall than at the mid atrial level and shorter at the RA than for all LA walls (Fig. 3). Some statistical significant differences among different LA walls were also found. The PA-peak interval (P-Aa' peak) was similar at low and mid atrial levels in almost all atrial walls with exceptions in the inferior and posterior LA walls. This interval was shorter for the IAS, inferior and posterior LA walls than for the lateral and anterior LA walls at mid and low levels (Fig. 4). The A-wave duration (Aa'-dur.) was shorter at the low atrial level than at the mid atrial level in each atrial wall, but not in the inferior and posterior LA walls. The total electromechanical activity (TEMA) was similar in all RA and LA atrial walls measured at low and mid levels, and no differences were found between any of the LA walls.
Table 3 Time intervals expressed in milliseconds measured at low and mid atrial levels in the myocardial walls of the RA and LA.
Variables Level RA IAS LA-Lat LA-Inf LA-Ant LA-Post P*
P-Aa' start Low 51 ± 11 59 ± 9 69 ± 11 62 ± 10 70 ± 10 62 ± 11 < 0.001
(ms) Mid 38 ± 9 47 ± 8 57 ± 9 51 ± 11 59 ± 10 52 ± 11 < 0.001
P < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
P-Aa' peak Low 117 ± 22 108 ± 14 119 ± 14 108 ± 12 123 ± 14 107 ± 11 <0.001
(ms) Mid 110 ± 20 104 ± 15 115 ± 15 98 ± 17 123 ± 16 99 ± 17 < 0.01
P 0.06 0.04 0.1 < 0.001 0.9 < 0.001
Aa'-dur. Low 135 ± 16 120 ± 16 106 ± 10 118 ± 13 113 ± 12 112 ± 12 < 0.001
(ms) Mid 145 ± 19 127 ± 16 117 ± 10 115 ± 12 121 ± 16 115 ± 12 < 0.01
P < 0.001 < 0.001 < 0.001 0.1 < 0.001 0.07
TEMA Low 186 ± 17 179 ± 18 175 ± 15 179 ± 15 183 ± 14 174 ± 12 0.07
(ms) Mid 183 ± 21 174 ± 16 173 ± 15 175 ± 15 180 ± 19 167 ± 17 0.2
P 0.3 0.1 0.5 0.1 0.3 0.06
Abbreviations: Aa'-dur., duration of the A wave; IAS, inter-atrial septum; LA-Ant, left atrial anterior wall; LA-Inf, inferior left atrial wall; LA-Lat, left lateral atrial wall; LA-Post, left posterior atrial wall; ms, milliseconds; P, by paired t test; P*, by analysis of variance; P-Aa' start, time from the beginning of the P-wave to the start of the A-wave; P-Aa' peak, time from the beginning of the P-wave to the peak of the A-wave; RA, right atrial wall; TEMA, total electromechanical activity.
P-Aa' start Low: RA vs all LA-walls (P < 0.001), IAS and LA-Inf vs LA-Lat (P < 0.001)
P-Aa' start Mid: RA vs all LA-walls (P < 0.01), IAS vs LA-Lat and LA-Ant (P < 0.001); LA-Inf vs LA-Lat and LA-Inf (P < 0.001); LA-Post vs LA-Ant (P < 0.01)
P-A'a peak Low and Mid: IAS, LA-Inf and LA-Post vs LA-Lat and LA-Ant (P < 0.001 for all comparisons)
Aa' dur. Low: RA, IAS and LA-Inf vs LA-Lat (P < 0.001)
Aa' dur. Mid: RA vs all walls (P < 0.001), IAS vs LA-Lat, LA-Inf and LA-Post (P < 001)
TEMA Low: RA vs LA-Lat, LA-Inf and LA-Post (P < 0.001). No differences among all LA-walls.
TEMA Mid: RA vs LA-Lat, LA-Inf, LA-Post (P < 0.001), LA-Ant vs LA-Inf and LA-Post (P < 0.001)
Figure 3 Assessment of the duration of the PA-start interval in all atrial walls. Comparisons were done with ANOVA with repeated measures and the Bonferroni's test.
Figure 4 Assessment of the duration of the PA-peak interval in all atrial walls. Comparisons were done with ANOVA with repeated measures and the Bonferroni's test.
Table 4 shows several velocities and velocity-derived variables: The peak velocity during atrial contraction (Aa' peak vel.) was higher at low than mid levels in each atrial wall, but no significant differences were found between any LA walls. This variable was higher in RA than in all LA walls. Similar results were found for the atrial displacement occurring during atrial contraction (Aa' disp.) and the ratio of atrial displacement measured at atrial level to the total LV myocardial displacement measured at ventricular level (Aa' cont.). The strain rate (Aa' SR) was higher in the RA than in all LA-walls, and lower in the IAS than in the lateral and posterior LA walls. The strain (Aa' S) was higher in the RA than in all LA-walls, and no differences were found between LA walls.
Table 4 Myocardial velocity and velocity-derived variables measured at right and left atrial myocardial walls.
Variables Level RA IAS LA-Lat LA-Inf LA-Ant LA-Post P*
Aa' peak vel. Low 8.1 ± 2.7 6.3 ± 1.4 6.2 ± 1.7 6.9 ± 1.8 6.3 ± 1.8 6.8 ± 1.8 NS
(cm/s) Mid 6.9 ± 2.4 5.2 ± 1.5 5.7 ± 1.4 5.1 ± 1.7 5.7 ± 1.8 5.4 ± 1.7 NS
P < 0.001 < 0.001 < 0.001 < 0.001 < 0.01 < 0.001
Aa' disp. Low 6.7 ± 2.3 4.5 ± 0.9 3.7 ± 1.0 4.2 ± 1.0 4.1 ± 1.2 4.0 ± 0.9 NS
(mm) Mid 5.8 ± 2.5 3.2 ± 0.9 3.4 ± 0.9 2.9 ± 0.7 3.1 ± 1.3 3.0 ± 0.8 NS
P < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
Aa' cont. Low 31 ± 10 33 ± 8 29 ± 10 28 ± 7 31 ± 12 26 ± 6 NS
(%) Mid 27 ± 11 24 ± 7 25 ± 9 19 ± 6 23 ± 11 19 ± 5 NS
P < 0.001 < 0.001 0.02 < 0.001 < 0.001 < 0.001
Aa' SR (s-1) -4.9 ± 0.8 -2.7 ± 0.7 -3.7 ± 0.9 -3.2 ± 0.8 -3.3 ± 1.0 -3.7 ± 1.0 0.001
Aa' S (%) 29 ± 7 14 ± 5 15 ± 4 16 ± 5 15 ± 6 16 ± 4 NS
Abbreviations: Aa' cont., atrial contribution; Aa' disp., atrial displacement; Aa' peak vel., atrial A-wave peak velocity; Aa' S, atrial strain; Aa' SR, atrial strain rate: P, paired t test; P*, analysis of variance; otherwise as in Table 3a.
Aa' peak vel Low: RA vs IAS, LA-Lat and LA-Ant (P < 0.001). No differences were found between any LA walls.
Aa' peak vel Mid: RA vs IAS, LA-Inf, LA-Post and LA-Ant (P < 0.001). No differences were found between any LA walls.
Aa' disp. Low and Mid: RA vs all walls (P < 0.001). No differences were found between any LA-walls.
Aa' cont. Low and Mid: No differences were found between walls
Aa' SR: RA vs all LA walls (P < 0.001), IAS vs LA-Lat and LA-Post (P < 0.001)
Aa' S: RA vs all LA walls (P < 0.001). No differences were found between any LA walls.
Tables 5 and 6 show the correlation coefficients between 2-D-derived and TVE-derived variables of LA and RA global mechanical function. There were no correlations between 2-D- and TVE-derived variables, apart from modest correlations between LA diameter and LA displacement, between RA long axis diameter and RA displacements, and between RA ejection fraction and strain rate.
Table 5 Correlation coefficients between 2-D- and TVE-derived variables of left global atrial mechanical function. Individual values of the inter-atrial septum, and the inferior, anterior, lateral and posterior LA walls were averaged.
LA-Aa' peak vel. LA-Aa' disp. LA-Aa' cont. LA-Aa' SR LA-Aa' S
LA-Diameter 0.33 0.57* 0.43 0.2 0.21
LA-Long axis 0.00 0.17 0.02 -0.21 -0.08
LA-Short axis -0.2 0.02 -0.08 -0.16 -0.07
LA-Area -0.12 0.01 -0.13 -0.12 0.12
LA-P-wave volume -0.15 0.02 -0.08 -0.14 -0.04
LA-Maximal volume 0.07 0.05 -0.1 -0.03 -0.21
LA-Minimal volume 0.2 0.15 0.13 -0.13 -0.21
LA-Ejection fraction -0.27 -0.23 -0.31 -0.03 0.19
LA-Active emptying -0.35 -0.22 -0.31 -0.03 0.19
Abbreviations: LA. Left atria; otherwise as in Table 3. * P < 0.01
Table 6 Correlation coefficients between 2-D- and TVE-derived variables of right global atrial mechanical function. Individual values of the inter-atrial septum, and the inferior, anterior, lateral and posterior LA walls were averaged.
RA-Aa' peak vel. RA-Aa' disp. RA-Aa' cont. RA-Aa' SR RA-Aa' S
RA-Long axis 0.35 0.58* 0.53 0.26 -0.07
RA-Short axis 0.06 0.30 0.20 -0.00 -0,14
RA-Area 0.10 0.31 0.20 -0.00 -0.14
RA-P-wave volume 0.09 0.31 0.21 0.08 -0.15
RA-Maximal volume -0.06 0.16 0.12 -0.04 -0.26
RA-Minimal volume 0.10 0.24 0.25 0.14 -0.18
RA-Ejection fraction -0.33 -0.24 -0.35 -0.41* -0.09
RA-Active emptying 0.07 0.27 0.12 -0.08 -0.08
Abbreviations: RA, right atrial; otherwise as in Table 3. * P < 0.01
The inter- and intra-observer measurement error and repeatability, as expressed by the British Standards Institution guidelines and coefficients of variation are presented in Table 7. The PA-start interval and the PA-peak interval, which mainly express atrial electrical function showed the largest inter- and intra-observer measurement errors and variability. However, the A-wave duration and the total electromechanical activity, which express a combination of the atrial electrical and mechanical functions, had better values of measurement error and repeatability. The same was true for all the TVE-derived variables that express regional and global atrial mechanical function.
Table 7 Assessment of inter- and intra-observer measurement error and repeatability according to the British Standards Institution guidelines and coefficients of variation
Inter-observer Intra-observer
Variable BSI CV (%) BSI CV (%)
P-Aa' start, ms 37 24 28 19
P-Aa' peak, ms 51 16 47 14
Aa' duration, ms 32 8.8 25 7.5
TEMA, ms 53 9.7 45 8.2
Aa' peak velocity, cm/s 2.14 10.1 1.78 8.7
Aa' displacement, mm 1.81 12.3 1.77 11.8
Aa' SR -0.79 9.4 -0.71 7.8
Aa' S, % 5.4 9.6 4.5 9.1
LA maximal volume, mL 22 18.4 18 14.7
Abbreviations: BSI, British Standards Institution; CV, coefficient of variation; otherwise as in Table 3.
Discussion
The main new findings of this study of healthy young individuals are as follows. (1) Some TVE-derived variables indirectly reflect the atrial electrical activation that follows the known activation process as revealed by invasive electrophysiology. (2) The regional and global atrial mechanical function is explained by an upward movement of the atrial walls at the region near the A-V ring with a continuous reduction of this movement towards the upper levels of atrial walls. (3) The atrial mechanical function is quite similar in all LA walls; however, all indices of mechanical function were higher in the RA than in the LA. (4) There were no correlations between the 2-D- and TVE-derived variables expressing atrial mechanical function. (5) Values of measurement error and repeatability were good for atrial mechanical function, but only acceptable for electrical function.
Atrial electrical activation, as assessed by the PA-start interval, began at the RA and followed through the IAS, to the inferior and posterior LA walls. This is the known normal electrical activation process, as obtained by invasive electrophysiology techniques.[32,37] In the present study, there were no statistical significant differences in the PA-start interval between IAS and the inferior and posterior LA walls, indicating that the activation process could indistinctly occur through any of these walls, as demonstrated by the presence of preferential conduction pathways nearby the IAS, the posterior LA wall and the coronary sinus.[33,34,37]. In a recent study, using M-mode color tissue Doppler registrations of the tricuspide and mitral rings, an abnormal time interval from the onset of P wave until the backward motion of the left atrio-ventricular ring was used to indirectly detect abnormal atrial electromechanical coupling in patients with paroxysmal atrial fibrillation.[43]
The relation between atrial anatomy and its mechanical function has been poorly studied. The present study showed that all atrial walls actively moved upwards from the region of the A-V ring at late diastole, with a reduction of this movement towards the upper parts, thus empting the atria and contributing towards the last part of filling of the LV. This longitudinal movement of the atrial walls is probably related to the longitudinal endocardial muscular fibers along the walls of the LA and RA. The more pronounced longitudinal movement in the RA may be explained in part by the larger pectinate muscles in the RA, but also by the lower pressures in the heart's right side. To what extent circumferential contraction of the atrial muscle fibers might contribute to atrial mechanical function is unknown. Anatomically, the large amount of circumferential muscle fibers present in the vestibules of the RA and LA[30,31,44] might imply some kind of circumferential or radial contraction of the atria. However, no movement of the posterior LA wall at late diastole can be observed by conventional M-mode echocardiography. Other circumferential fibers, such as Bachman's bundle located at the subepicardium joining the RA and LA, seem to play a critical role for electrical impulse spreading[37] rather than in circumferential atrial contraction. The assessment of circumferential atrial mechanical function by conventional echocardiography and TVE remains elusive.
No correlations were found between 2-D- and TVE-derived variables of atrial mechanical function, as was also found in a previous study[29]. Although 2-D-derived variables measure volumes and volume-derived indices that might indicate some kind of atrial mechanical force, it was surprising to find no correlations between the variables obtained by the two different techniques. This might indicate that the velocities and the displacements registered from all atrial walls by TVE are less dependent on volume loading conditions than 2-D-derived variables and therefore could be used as reliable measurement of pure atrial mechanical contraction or inotropism. In fact, Donald et al. showed that LA function assessed by TVE was relatively independent of LV function.[45] It should also be considered that movements of the heart not related to atrial contraction might partly contribute to the velocities and displacements registered from all atrial walls. Therefore, 2-D- and TVE-derived variables might not be used interchangeably to assess atrial mechanical function.
Some measures of atrial electrical function, for example the PA-start interval and the PA-peak interval, had only fair measures of repeatability and measurement error. However, most of the TVE-derived variables expressing atrial mechanical function had good values of repeatability and measurement error. Assessing atrial mechanical function by measuring volumes is time-consuming and depends on age, gender, and body surface area[14,19] In addition, atrial volume indices are also dependent on loading conditions[46,47] and are not necessarily more reproducible than TVE-derived variables.
Possible clinical implications
The identification of an abnormal electrical activation process could be of interest in some patients with atrial fibrillation or other supra-ventricular tachy-arrhythmias, in whom the premature atrial contraction acting as a triggering factor could be aggravated by local delayed conduction (reviewed in[48,49]). Further refinement of the TVE technique are necessary not only to identify the mechanical activation atrial sequence during normal sinus rhythm, but also to identify the origin and the activation sequence of supra-ventricular ectopic beats and in patients with RA, IAS or bi-atrial pacing. Thus, TVE could be an excellent adjunct to invasive electrophysiological techniques in selecting adequate patients and in the evaluation of atrial electromechanical consequences of RA, IAS or bi-atrial pacing.
The assessment of pure mechanical atrial function by means of atrial wall movements may give more concrete clues about the recovery process of atrial electromechanical function after conversion for atrial fibrillation and flutter and can give additional pathophysiological insights on the thromboembolic process that occur in some of those patients.[50] TVE-derived parameters may also give additional pathophysiological information on the process of atrial electromechanical remodeling that occurs in patients with sustained supra-ventricular tachy-arrhythmias.[51]
Several studies have shown the independent prognostic value of atrial function measurements in subsets of patients.[6,11,12] TVE-derived variables of atrial mechanical function may have an additional role for facilitating the assessment of atrial function and consequently in the process of risk stratification.
Study limitations
The results of the present study refer only to a group of young healthy individuals and the values for each of the studied variables are, therefore, only applicable to that population group. As discussed, the measures of atrial electrical function showed only fair values of repeatability and measurement error. There were two reasons: the image acquisition rate (less than 100 frames per second) means an implicit measurement error of 10 ms; it was also difficult to identify the beginning of the P-wave in the ECG from the monitor in the echocardiography machine. Improving temporal resolution by image acquisition at more than 200 frames per second, and improving and adjusting the ECG quality in the present equipment may help solve or decrease this problem. The velocities and displacements registered by atrial walls do not only represent the process of atrial contraction, but also the translational movement of the heart. Until now, no appropriate algorithms that correctly deal with this problem have been found. Presently, it is not possible by means of TVE to simultaneously record the electromechanical function of all atrial walls in one heartbeat. The development of three-dimensional TVE may help resolve this difficulty.
Conclusion
TVE is a noninvasive bedside tool that requires further refinements to provide reproducible, repeatable and potentially clinically useful data on atrial electromechanical function in health and disease.
==== Refs
Henry WL Morganroth J Pearlman AS Clark CE Redwood DR Itscoitz SB Epstein SE Relation between echocardiographically determined left atrial size and atrial fibrillation Circulation 1976 53 273 9 128423
Vaziri SM Larson MG Benjamin EJ Levy D Echocardiographic predictors of nonrheumatic atrial fibrillation. The Framingham Heart Study Circulation 1994 89 724 30 8313561
Benjamin EJ D'Agostino RB Belanger AJ Wolf PA Levy D Left atrial size and the risk of stroke and death. The Framingham Heart Study Circulation 1995 92 835 41 7641364
Predictors of thromboembolism in atrial fibrillation: II. Echocardiographic features of patients at risk. The Stroke Prevention in Atrial Fibrillation Investigators Ann Intern Med 1992 116 6 12 1727097
Mattioli AV Tarabini Castellani E Mattioli G Stroke in paced patients with sick sinus syndrome: influence of left atrial function and size Cardiology 1999 91 150 5 10516407 10.1159/000006902
Tsang TS Barnes ME Gersh BJ Takemoto Y Rosales AG Bailey KR Seward JB Prediction of risk for first age-related cardiovascular events in an elderly population: the incremental value of echocardiography J Am Coll Cardiol 2003 42 1199 205 14522480 10.1016/S0735-1097(03)00943-4
Jikuhara T Sumimoto T Tarumi N Yuasa F Hattori T Sugiura T Iwasaka T Left atrial function as a reliable predictor of exercise capacity in patients with recent myocardial infarction Chest 1997 111 922 8 9106570
Paraskevaidis IA Dodouras T Adamopoulos S Kremastinos DT Left atrial functional reserve in patients with nonischemic dilated cardiomyopathy: an echocardiographic dobutamine study Chest 2002 122 1340 7 12377862 10.1378/chest.122.4.1340
Triposkiadis F Moyssakis I Hadjinikolaou L Makris T Zioris H Hatzizaharias A Kyriakidis M Left atrial systolic function is depressed in idiopathic and preserved in ischemic dilated cardiomyopathy Eur J Clin Invest 1999 29 905 12 10583434 10.1046/j.1365-2362.1999.00563.x
Modena MG Muia N Sgura FA Molinari R Castella A Rossi R Left atrial size is the major predictor of cardiac death and overall clinical outcome in patients with dilated cardiomyopathy: a long-term follow-up study Clin Cardiol 1997 20 553 60 9181267
Beinart R Boyko V Schwammenthal E Kuperstein R Sagie A Hod H Matetzky S Behar S Eldar M Feinberg MS Long-term prognostic significance of left atrial volume in acute myocardial infarction J Am Coll Cardiol 2004 44 327 34 15261927 10.1016/j.jacc.2004.03.062
Moller JE Hillis GS Oh JK Seward JB Reeder GS Wright RS Park SW Bailey KR Pellikka PA Left atrial volume: a powerful predictor of survival after acute myocardial infarction Circulation 2003 107 2207 12 12695291 10.1161/01.CIR.0000066318.21784.43
Lester SJ Ryan EW Schiller NB Foster E Best method in clinical practice and in research studies to determine left atrial size Am J Cardiol 1999 84 829 32 10513783 10.1016/S0002-9149(99)00446-4
Pritchett AM Jacobsen SJ Mahoney DW Rodeheffer RJ Bailey KR Redfield MM Left atrial volume as an index of left atrial size: a population-based study J Am Coll Cardiol 2003 41 1036 43 12651054 10.1016/S0735-1097(02)02981-9
Payne RM Stone HL Engelken EJ Atrial function during volume loading J Appl Physiol 1971 31 326 31 5111850
Hitch DC Nolan SP Descriptive analysis of instantaneous left atrial volume – with special reference to left atrial function J Surg Res 1981 30 110 20 7464107 10.1016/0022-4804(81)90002-0
Tsang TS Barnes ME Gersh BJ Bailey KR Seward JB Left atrial volume as a morphophysiologic expression of left ventricular diastolic dysfunction and relation to cardiovascular risk burden Am J Cardiol 2002 90 1284 9 12480035 10.1016/S0002-9149(02)02864-3
Prioli A Marino P Lanzoni L Zardini P Increasing degrees of left ventricular filling impairment modulate left atrial function in humans Am J Cardiol 1998 82 756 61 9761086 10.1016/S0002-9149(98)00452-4
Nikitin NP Witte KK Thackray SD Goodge LJ Clark AL Cleland JG Effect of age and sex on left atrial morphology and function Eur J Echocardiogr 2003 4 36 42 12565061 10.1053/euje.2002.0611
Manning WJ Silverman DI Katz SE Riley MF Come PC Doherty RM Munson JT Douglas PS Impaired left atrial mechanical function after cardioversion: relation to the duration of atrial fibrillation J Am Coll Cardiol 1994 23 1535 40 8195510
Mattioli AV Castelli A Andria A Mattioli G Clinical and echocardiographic features influencing recovery of atrial function after cardioversion of atrial fibrillation Am J Cardiol 1998 82 1368 71 9856921 10.1016/S0002-9149(98)00643-2
Jones CJ Song GJ Gibson DG An echocardiographic assessment of atrial mechanical behaviour Br Heart J 1991 65 31 6 1825173
Nishimura RA Tajik AJ Evaluation of diastolic filling of left ventricle in health and disease: Doppler echocardiography is the clinician's Rosetta Stone J Am Coll Cardiol 1997 30 8 18 9207615 10.1016/S0735-1097(97)00144-7
Oh JK Appleton CP Hatle LK Nishimura RA Seward JB Tajik AJ The noninvasive assessment of left ventricular diastolic function with two-dimensional and Doppler echocardiography J Am Soc Echocardiogr 1997 10 246 70 9109691
Garcia MJ Thomas JD Klein AL New Doppler echocardiographic applications for the study of diastolic function J Am Coll Cardiol 1998 32 865 75 9768704 10.1016/S0735-1097(98)00345-3
Wilkenshoff UM Sovany A Wigstrom L Olstad B Lindstrom L Engvall J Janerot-Sjoberg B Wranne B Hatle L Sutherland GR Regional mean systolic myocardial velocity estimation by real-time color Doppler myocardial imaging: a new technique for quantifying regional systolic function J Am Soc Echocardiogr 1998 11 683 92 9692525
Kukulski T Jamal F D'Hooge J Bijnens B De Scheerder I Sutherland GR Acute changes in systolic and diastolic events during clinical coronary angioplasty: a comparison of regional velocity, strain rate, and strain measurement J Am Soc Echocardiogr 2002 15 1 12 11781548 10.1067/mje.2002.114844
Hesse B Schuele SU Thamilasaran M Thomas J Rodriguez L A rapid method to quantify left atrial contractile function: Doppler tissue imaging of the mitral annulus during atrial systole Eur J Echocardiogr 2004 5 86 92 15113019 10.1016/S1525-2167(03)00046-5
Thomas L Levett K Boyd A Leung DY Schiller NB Ross DL Changes in regional left atrial function with aging: evaluation by Doppler tissue imaging Eur J Echocardiogr 2003 4 92 100 12749870 10.1053/euje.2002.0622
Wang K Ho SY Gibson DG Anderson RH Architecture of atrial musculature in humans Br Heart J 1995 73 559 65 7626357
Ho SY Anderson RH Sanchez-Quintana D Atrial structure and fibres: morphologic bases of atrial conduction Cardiovasc Res 2002 54 325 36 12062338 10.1016/S0008-6363(02)00226-2
Markides V Schilling RJ Ho SY Chow AW Davies DW Peters NS Characterization of left atrial activation in the intact human heart Circulation 2003 107 733 9 12578877 10.1161/01.CIR.0000048140.31785.02
Antz M Otomo K Arruda M Scherlag BJ Pitha J Tondo C Lazzara R Jackman WM Electrical conduction between the right atrium and the left atrium via the musculature of the coronary sinus Circulation 1998 98 1790 5 9788835
Chauvin M Shah DC Haissaguerre M Marcellin L Brechenmacher C The anatomic basis of connections between the coronary sinus musculature and the left atrium in humans Circulation 2000 101 647 52 10673257
Gepstein L Hayam G Ben-Haim SA A novel method for nonfluoroscopic catheter-based electroanatomical mapping of the heart. In vitro and in vivo accuracy results Circulation 1997 95 1611 22 9118532
Smeets JL Ben-Haim SA Rodriguez LM Timmermans C Wellens HJ New method for nonfluoroscopic endocardial mapping in humans: accuracy assessment and first clinical results Circulation 1998 97 2426 32 9641695
Roithinger FX Cheng J SippensGroenewegen A Lee RJ Saxon LA Scheinman MM Lesh MD Use of electroanatomic mapping to delineate transseptal atrial conduction in humans Circulation 1999 100 1791 7 10534466
Haissaguerre M Jais P Shah DC Takahashi A Hocini M Quiniou G Garrigue S Le Mouroux A Le Metayer P Clementy J Spontaneous initiation of atrial fibrillation by ectopic beats originating in the pulmonary veins N Engl J Med 1998 339 659 66 9725923 10.1056/NEJM199809033391003
Cosio FG Lopez-Gil M Goicolea A Arribas F Barroso JL Radiofrequency ablation of the inferior vena cava-tricuspid valve isthmus in common atrial flutter Am J Cardiol 1993 71 705 9 8447269 10.1016/0002-9149(93)91014-9
Oral H Knight BP Tada H Ozaydin M Chugh A Hassan S Scharf C Lai SW Greenstein R Pelosi F JrStrickberger SA Morady F Pulmonary vein isolation for paroxysmal and persistent atrial fibrillation Circulation 2002 105 1077 81 11877358 10.1161/hc0902.104712
Schiller NB Shah PM Crawford M DeMaria A Devereux R Feigenbaum H Gutgesell H Reichek N Sahn D Schnittger I Recommendations for quantitation of the left ventricle by two-dimensional echocardiography. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms J Am Soc Echocardiogr 1989 2 358 67 2698218
Kircher B Abbott JA Pau S Gould RG Himelman RB Higgins CB Lipton MJ Schiller NB Left atrial volume determination by biplane two-dimensional echocardiography: validation by cine computed tomography Am Heart J 1991 121 864 71 2000754 10.1016/0002-8703(91)90200-2
Omi W Nagai H Takamura M Okura S Okajima M Furusho H Maruyama M Sakagami S Takata S Kaneko S Doppler tissue analysis of atrial electromechanical coupling in paroxysmal atrial fibrillation J Am Soc Echocardiogr 2005 18 39 44 15637487
Ho SY Sanchez-Quintana D Cabrera JA Anderson RH Anatomy of the left atrium: implications for radiofrequency ablation of atrial fibrillation J Cardiovasc Electrophysiol 1999 10 1525 33 10571372
Donal E Raud-Raynier P Racaud A Coisne D Herpin D Quantitative regional analysis of left atrial function by Doppler tissue imaging-derived parameters discriminates patients with posterior and anterior myocardial infarction J Am Soc Echocardiogr 2005 18 32 8 15637486
Appleton CP Galloway JM Gonzalez MS Gaballa M Basnight MA Estimation of left ventricular filling pressures using two-dimensional and Doppler echocardiography in adult patients with cardiac disease. Additional value of analyzing left atrial size, left atrial ejection fraction and the difference in duration of pulmonary venous and mitral flow velocity at atrial contraction J Am Coll Cardiol 1993 22 1972 82 8245357
Simek CL Feldman MD Haber HL Wu CC Jayaweera AR Kaul S Relationship between left ventricular wall thickness and left atrial size: comparison with other measures of diastolic function J Am Soc Echocardiogr 1995 8 37 47 7710749
Waldo AL Mechanisms of atrial flutter and atrial fibrillation: distinct entities or two sides of a coin? Cardiovasc Res 2002 54 217 29 12062328 10.1016/S0008-6363(01)00549-1
Nattel S Therapeutic implications of atrial fibrillation mechanisms: can mechanistic insights be used to improve AF management? Cardiovasc Res 2002 54 347 60 12062340 10.1016/S0008-6363(01)00562-4
Sparks PB Jayaprakash S Mond HG Vohra JK Grigg LE Kalman JM Left atrial mechanical function after brief duration atrial fibrillation J Am Coll Cardiol 1999 33 342 9 9973013 10.1016/S0735-1097(98)00585-3
Zipes DP Atrial fibrillation. A tachycardia-induced atrial cardiomyopathy Circulation 1997 95 562 4 9024138
| 15717936 | PMC550667 | CC BY | 2021-01-04 16:38:31 | no | Cardiovasc Ultrasound. 2005 Feb 18; 3:4 | utf-8 | Cardiovasc Ultrasound | 2,005 | 10.1186/1476-7120-3-4 | oa_comm |
==== Front
Harm Reduct JHarm Reduction Journal1477-7517BioMed Central London 1477-7517-2-21571322610.1186/1477-7517-2-2ResearchInitiation to heroin injecting among heroin users in Sydney, Australia: cross sectional survey Day Carolyn A [email protected] Joanne [email protected] Paul [email protected] Kate [email protected] National Drug and Alcohol Research Centre, University of New South Wales, SYDNEY NSW 2052, Australia2 Turning Point Alcohol and Drug Centre Inc. & Deakin University School of Health and Social Development, MELBOURNE VIC Australia2005 15 2 2005 2 2 2 26 3 2004 15 2 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
Heroin injection is associated with health and social problems including hepatitis C virus (HCV) transmission. Few studies have examined the circumstances surrounding initiation to heroin injecting, especially current users initiating others. The current study aimed to examine the age of first heroin use and injection; administration route of first heroin use; relationship to initiator; the initiation of others among a group of heroin users; and to examine these factors in relation to HCV status and risk.
Method
Heroin users in Sydney were recruited through needle and syringe programs, a methadone clinic and snowballing. Participants were interviewed about their own initiation to heroin use, blood-borne virus risk and knowledge, and whether they had initiated others to heroin injecting. Information on HCV status was collected via self-report. Data was analysed using univariate and multivariate statistical techniques for Normally distributed continuous and categorical data.
Results
The study recruited 399 heroin users, with a mean age of 31 years, 63% were male, 77% reported heroin as their primary drug and 59% were HCV positive (self-report). Mean age at first heroin use and injection was 19 and 21 years, respectively. The majority of heroin users commenced heroin use via injecting (65%), younger users (<25 years, 25–30 years) were less likely than older users (>30 years) to commence heroin use parenterally. Participants were initiated to injection mainly by friends (63%). Thirty-seven percent reported initiating others to heroin injection, but few factors were related to this behaviour. Those with longer heroin using careers were more likely to report initiating others to heroin injection, but were no more likely to have done so in the preceding 12 months. Participants who had initiated others were more likely to have shared injecting equipment (12 vs 23%), but were no more likely to be HCV positive (self-report) than those who did not.
Conclusion
Interventions to prevent heroin users initiating others to injecting are necessary. Peer groups may be well positioned to implement such interventions.
heroininjectinginitiationrisk behaviour
==== Body
Background
Heroin is one of the most commonly injected illicit substances worldwide [1,2]. Heroin injection is associated with a range of harms including poor health [3,4], poor psychosocial functioning [5] and increased risk of fatal and non-fatal heroin overdose [6,7]. Heroin injection is also a significant risk factor for blood-borne viral infections (BBVI) such as hepatitis C, hepatitis B and HIV [8-12]. Despite these harms little is known about initiation to heroin injecting.
Hepatitis C is probably the most prevalent health infection among injecting drug users (IDU) worldwide. The prevention of hepatitis C has proven difficult; unlike hepatitis B there is currently no vaccine available and programs which have been successful in reducing HIV have had only a small impact on the HCV epidemic [13,14]. It has therefore increasingly been acknowledged that prevention of initiation to drug injection is likely to be one of the most effective prevention strategies for blood-borne viral transmission [15]. Heroin injecting in particular is associated with increased HCV prevalence compared to injection of amphetamine – the two most commonly injected drugs in Australia [16].
A decrease in the age of initiation to drug use, including heroin use, across birth cohorts has been documented in both Australia [17,18] and the United States [19]. This decrease in the age of heroin initiation has been associated with greater poly drug use, unintentional overdose and criminal behaviour, independent of years of heroin use [17]. Better understanding of the circumstances surrounding initiation to injecting heroin use is needed if appropriate interventions are to be formulated.
Crofts et al. examined a number of factors surrounding initiation to injecting among a group of young, recently initiated IDUs [20]. They found that the majority of their sample had instigated the first injecting episode, but were assisted or injected by friends; only a very small proportion were injected by a family member. The research also found that more females than males reported being injected by their partner, a pattern consistent with the literature in that females are more likely than males to have an IDU sex partner and are less likely to be able to inject themselves [21-23]. A recent Canadian study of street youth also found that females were more likely than males to be injected by their partner or lover [24]. However, research from the United States found only 13% of IDU interviewed were initiated by their sex partner, with no difference between males and females; indeed females were more likely to be initiated by another female [25].
Little attention has been paid to the initiation of others into injecting [15]. One small study of a brief intervention to prevent initiation to injecting found that 40% of the 86 IDU interviewed had initiated a mean of two people to injecting [26]. Crofts et al. found that 47% of recently initiated IDUs had also initiated another into injecting and of those who had, few informed new initiates of BBVI risk [20]. The relationship between initiators' BBVI status and risk behaviour was not reported, so it was not possible to determine whether those engaging in risky behaviours were more or less likely to initiate others to injecting. Moreover, the study focused on young or recently initiated injectors, so the role of age and experience in initiating others to injecting was unable to be determined.
Crofts and colleagues found that the majority of recently initiated IDU were aware of HIV and hepatitis B and that the viruses could be transmitted via shared injecting equipment [20]. A much smaller proportion knew of HCV, though those who were aware of the virus were also aware that it could be transmitted via shared injecting equipment [20]. However, information on serostatus or hepatitis B vaccination status was not reported, nor was injecting risk behaviours such as needle and syringe sharing.
This study examined initiation to heroin injection. Specifically the study aimed to examine: 1) age of first heroin use and first heroin injection; 2) route of first heroin use; 3) relationship to initiator; and 4) the initiation of others among a group of heroin users. These factors were then examined in relation to demographic variables such as gender, ethnicity, level of education and also blood borne virus status.
Methods
Procedure
Heroin users in Sydney were volunteers recruited through needle and syringe programs and a methadone clinic. In order to sample a range of heroin users, snowballing, facilitated by a peer interviewer, was also used. Drug users were eligible to participate if they reported the use of heroin at least once a month in the preceding six months. All participants gave informed consent to be interviewed and received AUD20 for travel expenses.
Measures
A structured questionnaire was administered to participants by trained interviewers. Information was sought on blood-borne virus status and knowledge, and initiation to heroin use and included questions on age of first heroin use, age of first heroin injection, relationship to initiator and the number of times participants had initiated others to injecting.
Information on HCV status was collected via self-report. Self-reported HCV status has a concordance of approximately 80% for those who have been tested [27,28]. For the purposes of this study, self-reported HCV is relevant as risk behaviour is influenced by the belief about one's status, not actual status [29].
To determine ethnicity, participants were asked how they identified ethnically, those identified as being of Aboriginal or Torres Strait Islander decent were categorised as Indigenous, those not born in Australia and those born in Australia but who identified as belonging to another ethnic group were categorised as 'other Australian'.
The study was approved by four institutional ethics committees: University of New South Wales Human Research Ethics Committee, Central Sydney Area Health Service Ethics Review Committee, South Western Sydney Area Health Service Research Ethics Committee, and South Eastern Sydney Area Health Service Research Ethics Committee.
Analysis
Continuous variables were assessed using t-tests and one-way analysis of variance. Linear regression was employed to test the relationship between continuous variables. The chi square (χ2) statistic was used for univariate analysis of categorical data. Multiple logistic regression, using backward elimination, was used for multivariate analysis to examine independent relationships between dichotomous variables. All data were analysed using SPSS version 11.01.
Results
Sample characteristics
The sample consisted of 399 heroin users, of whom 63% were male. The mean age of participants was 31 years (SD 8.2, range 17, 58). The majority of participants were born in Australia (78%), a small proportion of which identified as another ethnic group. Participants were categorised as either 'Australian' (67%), Indigenous (17%) or 'other-Australian' (16%). The majority (81%) of participants had attended secondary school, but only 42 (11%) participants had tertiary education. Thirty-four (9%) participants had completed primary (elementary) school only. Sixty-one percent of participants had a history of incarceration.
Heroin was the primary drug used by 77% of the sample, with a median of 9.5 (range <1, 39) years of use and injecting. The majority (98%) of the sample injected heroin; 14% of the sample had been injecting heroin for three years or less and 80% for more than three years (18 cases had missing data for this variable). Those who did not inject heroin were excluded from analysis pertaining to injecting.
Age of initiation to heroin use and injection
The mean age at first heroin use was 19 years, (SD 6.0, range 9–43 years) and the mean age of first injection was 21 years (SD 6.3, range 13–43 years). The mean age of first heroin use and first heroin injection was similar for males (18 years for both) and females (18 and 19 years, respectively). There were no differences in terms of ethnicity for either age of first heroin use (18 years for all groups) or first heroin injection (18, 19 and 18 years for Australians, 'other Australians' and Indigenous Australians respectively).
For those who completed primary school only, the mean age of first heroin use was 17 (SD 5.4) years and 19 (SD 6.0) years for those who attended secondary school or above, the difference failed to reach significance (t395 = -1.89, p = .0.56). Participants who completed primary school only, were a mean age of 17 (SD 4.8) years when they first injected heroin and the mean age of first injection for those who attended secondary school and above was 20 (SD 6.3) years (t386 = -2.46, p = .014).
Linear regression was used to examine the relationship between current age and age of initiation to heroin use and injecting. There was a significant relationship between participants' current age and age of first heroin use (β = 0.41, p < .001). Similarly, current age and age of first injection were also significantly related (β = 0.39, p < .001).
Route of first heroin administration
Heroin was most commonly first administered by injection (65%). For the 141 participants who used non-parenteral routes of administration on initiation of heroin use, the most common method was smoking (burning and chasing), with 28% of the sample reporting initiating heroin use with this method. Only seven percent first used heroin intranasally, orally or by other means.
Participants who first injected heroin and those who first used heroin non-parenterally, commenced heroin use at similar ages (19 and 20 years, respectively). Those who initiated heroin use by injection had been using heroin for more years than those who initiated heroin use non-parenterally (13 years vs 9 years respectively, t379 = 4.36, p = 0.001).
Route of administration was also associated with ethnicity: 74% of Indigenous participants commenced heroin use via injection, 69% of 'Australians' and 36% of 'other Australians' (χ2 = 26.40, p < .001). There were no differences in the proportion of participants who commenced heroin use via injecting in terms of gender (males 64% and females 65%) or level of education (primary education 77%, secondary education 64%).
Logistic regression was used to determine the factors independently associated with initiation to heroin use via injecting. Variables entered into the model were age, gender, ethnicity and age at first injection (before or after 18 years). Level of education was not entered into the model as there were too few participants with only primary school level of education who commenced heroin use via a non-parenteral route of administration.
The final model was significant (χ2 = 43.92, 4df, p < .001) and the Hosmer and Lemeshow test indicated good fit (χ2= 1.5 6df, p = .96). Gender was not a characteristic independently associated with route of first heroin injection and was removed from the model. Participants aged less than 25 years and 25–30 years were less likely than those aged 30 years or more to initiate heroin use via injecting. 'Other Australians' were also less likely to reported initiating less injecting (Table 1).
Table 1 Characteristics of those who first injected heroin, using multivariate logistic regression
Characteristic No. participants %1st injected Adjusted odds ratio 95% CI P
Age
31+ 184 77 - -
25–30 106 60 0.48 0.28–0.83 .008
≤ 24 96 54 0.34 0.19–0.61 <.001
Ethnicity
Australian 269 67 - - -
ATSI 69 74 1.18 0.61–2.26 ns
Other Australians 61 36 0.30 0.16–0.55 <.001
Age 1st injected
≥ 18 164 77 - -
<18 234 56 0.63 0.39–1.02 .059
CI = Confidence interval
Relationship to initiator
Participants were usually taught to inject by a friend (63%), family member (14%) or their partner (11%). Ten percent of the sample reported 'other', which was typically self-taught. Males and females differed significantly in terms of who taught them to inject (χ2 = 24.75, df = 2, p < .001; Table 2). Specifically, a greater proportion of females were taught to inject by their partner than were males (χ2 = 23.96, df = 1, p < .001), while more males were taught to inject by a friend (or other) than females (χ2 = 11.77, df = 1, p < .001). The relationship between participants and their 'initiators' did not differ according age of first injection, ethnicity or level of education across those initiated by friends, family or their partner.
Initiating others
Over a third (37%) of participants reported having taught someone to inject drugs and 17% had done so in the preceding 12 months. Among those who had ever taught someone to inject (n = 149), the median numbers of people taught was three (range: 1–200) and two (range: 1–50) in the preceding 12 months.
Similar proportions of males and females reported teaching someone to inject ever (38% vs 37%) and in the preceding 12 months (16% vs 19%). There was also no significant difference in terms of ethnicity or mean age of those who had ever taught someone to inject ever or in the preceding 12 months. Those who had taught someone else to inject heroin had been injecting for a greater mean number of years than the remainder of the sample (13 v. 10 years; t = -3.08, df = 370, p = 0.002). However, this difference diminished for those who had recently (preceding 12 months) taught someone to inject (11 years for both groups).
Blood-borne viral infections
Ninety-two percent of participants reported being tested for HCV infection; 62% in the six months preceding interview and 78% in the 12 months preceding interview. The median number of weeks since the last test was 19 (range: 1–572 weeks). Two-hundred and thirty-five participants (59% of the total sample) self-reported being HCV positive.
Participants who reported being HCV positive initiated heroin injecting at a younger mean age (19 years, SD 5.7) than the remainder of the sample (21 years, SD 6.8). HCV positive participants were more likely to report initiating heroin use via injecting than those who were HCV negative (72% vs 58%, χ2 = 9.14, df = 1, p = 0.003). IDUs typically become infected with HCV early in their injecting career [30], thus analysis was stratified by years of injecting (≤ 3 years vs >3 years). The relationship remained only for those who had been injecting for more than three years (61% vs 76%, χ2 = 7.93, df = 1, p = 0.005). There was no difference in self reported HCV status between those who had initiated someone to injecting in the preceding 12 months and those who had not.
Eighty-eight percent of participants had reported being tested for HIV a median of six months prior to interview (range <1–144). Of these, seven (2% of the total sample) reported being HIV positive. One-hundred and forty-seven participants (37%) reported having been vaccinated against hepatitis B. There were no differences between those who reported being vaccinated against hepatitis B and those who had not in terms of age at first heroin injection, initial route of administration and initiating some one else to injecting in the preceding 12 months.
Only a small number of participants (9%) reported having used a needle or syringe after someone else in the month preceding interview. Sharing needles and syringes was not associated with mean age at initiation to heroin injection or initial route of heroin administration. More participants who had recently initiated someone to injecting (i.e. preceding 12 months) reported sharing needles and syringes (17%) than those who had not recently initiated someone (8%), though the difference failed to reach significance (χ2 = 3.51, df = 1, p = .061).
Sharing (borrowing or lending) injection paraphrenia in the month preceding interview was reported by just over half (52%) the sample. Participants who had engaged in this behaviour were more likely to have initiated heroin use via injecting (71% vs 61%, χ2 = 4.37, df = 1, p < .037) and to have recently initiated someone to injecting (12% vs 23%, χ2 = 7.95, df = 1, p < .005) compared to those who not shared injection paraphernalia. There was no difference in terms of age at first heroin injection.
Disscussion
This study has identified factors associated with initiation to heroin injecting. The majority of heroin users commenced heroin use via injecting, though those who initiated heroin use via this method had been using heroin for a longer period than those who initiated heroin use via non-parenteral methods. Participants were initiated to injection by a range of people, mainly friends. A large proportion of study participants also reported initiating another person to heroin injection, a practice that was associated with longer heroin use careers and recent sharing of injecting equipment.
Increasing attention has been paid to the role of IDUs in initiation to injecting drug use [15], though few studies have examined the prevalence of this behaviour. Crofts and colleagues postulate that the behaviour can be understood as communicable and modelled as in the case of infectious diseases [20]. Fewer participants in the current study reported initiating others to injecting than that reported by Crofts et al. (37% vs 47%) and only 17% reported doing so in the preceding 12 months. Crofts et al. sampled only young injecting drug users (i.e. 17–24 years) [20], whereas the injecting experience of participants sampled for this study ranged from less than one year to 39 years and younger participants were no more likely than older participants to report initiating others to injecting.
A third of the sample initially used heroin by non-parenteral routes of administration. Participants from ethnically diverse backgrounds (i.e. born outside Australia or identify other than Australian) were more likely than other participants to have first used heroin by means other than injecting, which is consistent with other Australian research which found smoking heroin to be more common among Indochinese than Caucasian heroin users [31].
Participants who commenced heroin use via injecting were also older than those who first used a non-injecting route of heroin administration, possibly indicating an overall shift toward non-injecting routes of administration. A number of interventions aimed at reducing the incidence of transition to injecting have received attention [15]. One study found that brief interventions delivered through existing drug services can have an impact on injecting drug users' behaviour, however, the study was modest and further research is warranted [26].
The current study found a direct significant relationship between current age and age of first heroin use and first heroin injection. This result is consistent with other Australian research [17,18]. Nevertheless the finding is subject to bias due to 'right censoring' of the data [17]. For example, an 18 year old recruited into the study cannot have an age of initiation above 18 years, while a 30 year old can have any age of initiation up to 30 years, and be included in the study even if they commenced injecting at 25 years [for a more detailed discussion see [17]]. There are numerous explanations for this decrease in initiation to heroin use; the rapid expansion of the Australian heroin market between 1996 and 2000, where the price of heroin decreased concomitant with an increase in purity and availability from 1996 to 2000 [32], may in part explain this phenomenon, though greater examination of the structural determinants of drug use are also warranted [33].
Although the proportion of participants reporting using a needle or syringe after another person (sharing) in the preceding month was low, it is consistent with other research examining transitions to injecting [31], but lower than Australian national estimates of this behaviour [16]. It is possible that the data is subject to a social desirability bias and thus under-reported. Participants reported comparatively high levels of 'indirect' sharing (sharing injecting paraphernalia other than needles and syringes), a phenomenon recently found to significantly and independently increase the risk of HCV transmission [34]. This behaviour was reported more often by those who reported teaching someone to inject in the 12 months preceding interview. That those who are more likely to engage in this behaviour are also teaching others to inject is cause for concern, as it has the potential to perpetuate the problem of injecting related risk taking behaviour among new recruits.
The majority of the sample believed themselves to be HCV positive and, not surprisingly, those who injected at a younger age were more likely to be HCV positive and to have first used heroin via injection. Though importantly, there was no difference in terms of HCV status between those who had recently taught another to inject and those who had not.
Conclusions
This study has confirmed that initiation to heroin use in Australia typically occurs via injection, though this is less apparent among younger heroin uses. The study has also found that more than a third of heroin users have initiated others into injecting, with close to a fifth having done so recently. Many of those who engaged in this behaviour tended to take greater injection related risks which has important implications for the transmission of blood-borne infections. A better understanding of the circumstances surrounding the initiation to heroin injection is needed. Peer-led interventions, which have been found to be effective in changing IDUs' attitudes and behaviours [26], may have a role to play in reducing the number of heroin users initiating others to injecting.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contribution
C Day coordinated the study, was responsible for the statistical analysis and writing the paper. P Dietze was responsible for study design at the national level, secured funding and provided comments on the manuscript. J Ross and K Dolan supervised all aspects of the work and provided extensive comments on the manuscript.
Acknowledgements
This research was funded by Australian Government Department of Health and Ageing and a postgraduate award held by the first author. The National Drug and Alcohol Research Centre is funded by the Australian Government Department of Health and Ageing.
==== Refs
Fernandez H Heroin 1998 Minnesota, Hazelden
United Nations Office for Drug Control and Crime Prevention World Drug Report 2000 2000 Great Britain, Oxford University Press
Ryan A 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
Darke S Ross J Teesson M Lynskey M Health service utilization and benzodiazepine use among heroin users: findings from the Australian Treatment Outcomes Study (ATOS) Addiction 2003 98 1129 1135 12873247 10.1046/j.1360-0443.2003.00430.x
Gossop M Marsden J Stewart D Lehmann P Edwards C Wilson A Segar G Substance use, health and social problems of service users at 54 drug treatment agencies. Intake data from the National Treatment Outcome Research Study British Journal of Psychiatry 1998 173 166 171 9850230
Darke S Ross J Hall W Overdose among heroin users in Sydney, Australia: I. Prevalence and correlates of non-fatal overdose Addiction 1996 91 405 411 8867202 10.1046/j.1360-0443.1996.9134059.x
Warner-Smith M Darke S Day C Morbidity associated with non-fatal heroin overdose Addiction 2002 97 963 967 12144598 10.1046/j.1360-0443.2002.00132.x
Stimson GV The global diffusion of injecting drug use: implications for human immunodeficiency virus infection Bulletin on Narcotics 1993 45 3 17 8305905
Des Jarlais DC Friedman SR HIV epidemiology and interventions among injecting drug users Int J STD AIDS 1996 7 157 161 10.1258/0956462961917654
Ray Kim W Global epidemiology and burden of hepatitis C Microbes and Infection 2002 4 1219 1225 12467763 10.1016/S1286-4579(02)01649-0
Lauer GM Walker BD Hepatitis C virus infection New England Journal of Medicine 2001 345 41 52 11439948 10.1056/NEJM200107053450107
Levine OS Vlahov D Nelson KE Epidemiology of hepatitis B virus infections among injecting drug user: seroprevalence, risk factors, and viral interactions Epidemiologic Reviews 1994 16 418 436 7713187
Kaldor JM Dore GJ Correll PKL Public health challenges in hepatitis C virus infection Journal of Gastroenterology and Hepatology 2000 15 (Suppl) E83 E90 10921388 10.1046/j.1440-1746.2000.02134.x
Commonwealth Department of Health and Aging Return on Investment in Needle and Syringe Exchange Programs in Australia 2002 Canberra, Commonwealth Department of Health and Aging
Hunt N Griffiths P Southwell M Stillwell G Strang J Preventing and curtailing injecting drug use: a review of the opportunities for developing and delivering 'route transition interventions' Drug and Alcohol Review 1999 18 441 451 10.1080/09595239996310
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
Lynskey M Hall W Cohort trends in age of initiation to heroin use Drug and Alcohol Review 1998 17 289 297 16203495
Degenhardt L Lynskey MT Hall W Cohort trends in the age of initiation of drug use in Australia Australian and New Zealand Journal of Public Health 2000 24 421 426 11011471
Johnson RA Gerstein DR Initiation of use of alcohol, cigarettes, marijuana, cocaine and other substances in US birth cohorts since 1919 Am J Public Health 1998 88 27 33 9584029
Crofts N Louies R Rosenthal D Jolley D The first hit: circumstances surrounding initiation into injecting Addiction 1996 91 1187 1196 8828246 10.1046/j.1360-0443.1996.918118710.x
Barnard MA Needle sharing in context: patterns of sharing among men and women injectors and HIV risks Addiction 1993 88 805 812 8329971
Dwyer R Richardson MA Ross MW Wodak A Miller ME Gold J A comparison of HIV risk between women and men who inject drugs AIDS Education and Prevention 1994 6 379 389 7818974
MacRae R Aalto E Gendered power dynamics and HIV risk in drug-using sexual relationships AIDS Care 2000 12 505 515 11091783 10.1080/09540120050123909
Roy E Haley N Leclerc P Cedras L J-F. B Drug injection among street youth: The first time Addiction 2002 97 1003 1009 12144603 10.1046/j.1360-0443.2002.00161.x
Doherty MC Garfein RS Monterroso E Latkin C Vlahov D Gender differences in the initiation of injection drug use among young adults J Urban Health 2000 77 396 414 10976613
Hunt N Stillwell G Taylor C Griffiths P Evaluation of a brief intervention to prevent initiation into injecting Drugs: Education, Prevention and Policy 1998 5 185 194
Loxley W Carruthers S Bevan J In the Same Vein: First report of the Australian Study of HIV and Injecting Drug Use (ASHIDU) 1995 Perth, Curtin University of Technology
Best D Noble A Finch E Gossop M Sidwell C Strang J Accuracy of perceptions of hepatitis B and C status: cross sectional investigation of opiate addicts in treatment British Medical Journal 1999
Best D Harris J Gossop M Farrell M Finch E Noble A Strang J Use of non-prescribed methadone and other illicit drugs during methadone maintenance treatment Drug and Alcohol Review 2000 19 9 16 10.1080/09595230096093
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
Swift W Maher L Sunjic S Transitions between routes of heroin administration: a study of Caucasian and Indochinese heroin users in south-west Sydney, Australia. Addiction 1999 94 71 82 10665099 10.1046/j.1360-0443.1999.941714.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
Spooner C Hall W Lynskey M Structural determinants of youth drug use 2001 Canberra, Australian National Council on Drugs
Thorpe LE Ouellet LJ Hershow R Bailey SL Williams IT Williamson J Monterroso ER Garfein RS Risk of hepatitis C virus infection among young adult injection drug users who share injection equipment American Journal of Epidemiology 2002 155 645 653 11914192 10.1093/aje/155.7.645
| 15713226 | PMC550668 | CC BY | 2021-01-04 16:37:10 | no | Harm Reduct J. 2005 Feb 15; 2:2 | utf-8 | Harm Reduct J | 2,005 | 10.1186/1477-7517-2-2 | oa_comm |
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-111570749310.1186/1475-2875-4-11ResearchAn AFLP-based genetic linkage map of Plasmodium chabaudi chabaudi Martinelli Axel [email protected] Paul [email protected] Richard [email protected] Pedro VL [email protected] David [email protected] Richard [email protected] Institute for Immunology and Infection Research, School of Biological Science, University of Edinburgh, Ashworth Laboratories, King's Buildings, West Mains Road, Edinburgh EH9 3JT, UK2 Centro de Malária e Outras Doenças Tropicais/IHMT/UEI Biologia Molecular/UNL, Rua da Junqueira, 96, 1349-008, Lisbon, Portugal2005 11 2 2005 4 11 11 15 12 2004 11 2 2005 Copyright © 2005 Martinelli et al; licensee BioMed Central Ltd.2005Martinelli 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
Plasmodium chabaudi chabaudi can be considered as a rodent model of human malaria parasites in the genetic analysis of important characters such as drug resistance and immunity. Despite the availability of some genome sequence data, an extensive genetic linkage map is needed for mapping the genes involved in certain traits.
Methods
The inheritance of 672 Amplified Fragment Length Polymorphism (AFLP) markers from two parental clones (AS and AJ) of P. c. chabaudi was determined in 28 independent recombinant progeny clones. These, AFLP markers and 42 previously mapped Restriction Fragment Length Polymorphism (RFLP) markers (used as chromosomal anchors) were organized into linkage groups using Map Manager software.
Results
614 AFLP markers formed linkage groups assigned to 10 of 14 chromosomes, and 12 other linkage groups not assigned to known chromosomes. The genetic length of the genome was estimated to be about 1676 centiMorgans (cM). The mean map unit size was estimated to be 13.7 kb/cM. This was slightly less then previous estimates for the human malaria parasite, Plasmodium falciparum
Conclusion
The P. c. chabaudi genetic linkage map presented here is the most extensive and highly resolved so far available for this species. It can be used in conjunction with the genome databases of P. c chabaudi, P. falciparum and Plasmodium yoelii to identify genes underlying important phenotypes such as drug resistance and strain-specific immunity.
==== Body
Background
Plasmodium chabaudi chabaudi is a malaria parasite of murine rodents. It has been widely used as a model to study various aspects of parasite biology and disease which are difficult to investigate using human malaria parasites. For instance, P. c. chabaudi is being used to study the genetic basis of drug resistance [1-4] and strain-specific immunity [5], because the execution and analysis of genetic crosses is relatively straightforward in this species [6]. The analysis of the genetic basis of aspects of malaria biology has been facilitated by recent developments in malaria genomics. Firstly, the Plasmodium falciparum genome has been fully sequenced and mapped [7] and there is also extensive sequence data now available for three of the four main malaria parasites of murine rodents [8]. Secondly, the degree of homology and conservation of gene synteny between the various species of malaria [4,9,10] allows the undertaking of comparative genomics and facilitates the elaboration of accurate genomic maps in these species.
However, a genetic linkage map of the 14 chromosomes of P. c. chabaudi is still important for the identification of loci which influence phenotypes such as drug resistance. A previous genetic linkage map of P. c. chabaudi was generated using over 40 RFLP markers [11]. However, due to the small number of markers available, this linkage map had limited usefulness. The authors have recently developed a large number of genome-wide polymorphic AFLP markers for P. c. chabaudi [11]. AFLP markers have previously been used to generate genetic linkage maps in another apicomplexan parasite, Eimeria tenella [12], as well as in Trypanosoma brucei [13].
This article presents a high-resolution genetic linkage map of P. c. chabaudi and an estimate of map unit size. The value of the genetic linkage map in the identification of genes determining selectable phenotypes is also described.
Methods
Mouse strains used in experiments
Inbred female CBA mice, obtained from the University of Edinburgh, were used for the growth of P. chabaudi parasites. Mice were housed in propylene cages with sawdust bedding and were fed on Harlan, SDS formula number I (Special Diet Services Ltd.) and drinking water supplemented with 0.05% paraminobenzoic acid (PABA) to aid parasite growth [14]. Temperature was maintained between 22 and 25°C with a 12 hour light/12 hour dark cycle.
Parasite lines
Clones of the genetically distinct isolates AS and AJ, originally isolated from wild thicket rats, Thamnomy rutilans [15] were used as parents in genetic crosses.
28 recombinant clones were analysed here. 20 clones originated from a cross between AJ and AS (3CQ) (a chloroquine-resistant clone derived from AS) [1], while 8 clones originated from a cross between AJ and AS (30CQ) (a clone with higher resistance to chloroquine, derived from AS (3CQ) [16].
Maintenance of parasites
For routine maintenance of parasites, parasitized red blood cells collected from the tail veins of infected mice were passaged in citrate saline into uninfected mice. Cryopreservation of infected blood was performed by exsanguination of mice anaesthetised with halothane. Blood was collected in a tube containing 2–3 volumes of citrate saline (0.9% NaCl, 1.5% tri sodium citrate dihydrate, adjusted to pH 7.2). The mixture was spun at 2000 rpm for five minutes, the supernatant discarded and the red cell pellet mixed with two volumes of a solution containing 28% (v/v) glycerol, 3% sorbitol and 0.65% NaCl. The mixture was then aliquoted into several glass capillaries, which were sealed by flame and deep-frozen in liquid nitrogen.
50 μl of cryopreserved blood was recovered by thawing capillaries into 10 μl of 12% NaCl and mixing for 3–5 mins. Nine volumes of 1.6% NaCl were then added dropwise and samples centrifuged at 2000 rpm for 3–5 mins. The supernatant was removed and nine volumes of 0.9% NaCl/ 0.2% dextrose solution were added dropwise. After mixing, the mixture was centrifuged again, supernatant removed and red blood cells resuspended in a 0.9% NaCl/ 0.2% dextrose solution for injection.
Estimation of parasitaemia
Parasitaemia was estimated by microscopic observation of thin blood smears taken 4–6 days after parasite injection and stained with 20% Giemsa staining solution (BDH) for 15 minutes. Parasitaemia was estimated by calculating the percentage of red blood cells infected in at least five microscopic fields.
Preparation of parasite DNA
Each parasite DNA preparation was obtained from five infected CBA female mice. Blood samples were taken from mice infected with AS, AJ and recombinant clones having high parasitaemias in the mid-afternoon, when parasites were trophozoites.
Host lymphocytes or nucleated cells present in the blood were removed as described previously [11]. Parasites were pelleted and stored at -70°C. DNA was extracted and purified as previously described [11] and stored at -20° for future use.
Amplified Fragment Length Polymorphism (AFLP) technique
The AFLP method was carried out according to the original protocol [17] with slight modifications, as described by Grech et al [11]. Briefly, parasite genomic DNA was digested with two enzymes, EcoRI and MseI, and ligated with adapters, to provide the complementary sequences for AFLP primers. The first round of amplification used primers containing the EcoRI or MseI recognition sequences at their 3' end. The second round of (selective) amplification use two additional (selective) bases (3' terminus) in both primers, one of which (the EcoRI primer) was radiolabeled with γ-[33P] ATP. PCR products were run on acrylamide gels and AFLP bands visualised on autoradiography films. Polymorphic bands between the two parental strains were used as markers for the genetic linkage map.
Organization of AFLP markers in a genetic linkage map
For every marker, the parental alleles identified in each of the progeny clones were entered in an Excel spreadsheet. The absence of a band in one parent was treated as the presence of the other parental allele at that locus. Data were then prepared for analysis with the Map Manager QTX software [18] according to the instruction manual. The dataset was designated as "Backcross" for the purpose of computer analysis.
Prior to linkage analysis, markers were tested for random assortment using a chi-square test, to exclude markers segregating in a non-random fashion from the initial analysis with Map Manager, and thus to avoid spurious linkage inferences between such markers. Because of the large number of tests for non -random assortment performed (n = 672), some markers showing apparent non-random assortment may have been falsely excluded from our analysis; i.e. some valid markers may indeed show non-random assortment and should be included. A Bonferroni correction was therefore applied to the chi-square test to decrease the stringency of the statistical test. The value of the Bonferroni correction represents the number of 'independent' comparisons and here was arbitrarily set at 24. This value was chosen as representing the likely number of chromosomal fragments at meiosis, and is supported by data presented in this paper. Markers within these fragments are not independently inherited. The chosen value (24) is a compromise between 672 (which assumes that all 672 markers are independently inherited) and 14 (the number of chromosomes, and which assumes that no pair of markers on one chromosome are inherited independently). Markers not following random assortment in the initial test were thus divided into two groups, i.e. those segregating in a non-random fashion before and after Bonferroni correction, and those segregating in a non-random fashion before but not after the Bonferroni correction. The markers in the latter group were added separately after linkage groups had been determined (see below).
Linkage groups using AFLP and RFLP markers were formed with an initial p-value of 0.0001 using the "Make Linkage Groups" command in Map Manager. p-values in Map Manager indicate the probability of a Type 1 error; that is, the probability of a false positive linkage. Following formation of linkage groups, the p-value was raised to 0.001. Linkage at p < 0.001 is considered significant. Using the command "Distribute", linkage groups were brought together. Then, other previously unlinked markers were allocated to these new linkage groups, again using the "Distribute" command. Markers with non-random assortment after statistical analysis without Bonferroni correction were added next, and those still segregating in a non-random fashion after Bonferroni correction were added last. The "Ripple" function was then used to position markers in an order which maximizes the total LOD (logarithmic odds) score for linkage. The software also estimated the optimum order and genetic distance between markers in centiMorgans (cM) by using the "Kosambi" function in the software. It was then possible to calculate a map unit size (i.e the physical distance corresponding to 1 cM).
The presence of 42 previously characterised RFLP markers which had been physically mapped onto P. c. chabaudi chromosomes [1] served as anchors for the placement of AFLP linkage groups onto specific chromosomes.
Results
The inheritance of 672 AFLP markers was determined in 28 progeny clones derived from two crosses between P. c. chabaudi AJ and either clone AS (3CQ) or AS (30CQ). The majority of the AFLP markers showed independent assortment in the 28 progeny clones, as illustrated previously [11]. However, 66 markers failed the chi-square test at 5%, 15 of which failed it after the Bonferroni correction. Markers were allocated to linkage groups using the Map Manager program and groups assigned to chromosomes using 42 previously mapped RFLP markers as anchors [1]. Estimated numbers of recombination events, genetic lengths of chromosomes and recombination frequencies were also determined for the identified chromosomes using Map Manager.
Allocation of markers to linkage groups
The 672 AFLP markers formed a total of 22 linkage groups with a final p-value of 0.001. Additional file 1 summarises the numbers of AFLP and RFLP markers assigned to each chromosome or to unassigned linkage groups, the estimated physical size of each chromosome [19] and the number of AFLP markers per Mb. 400 AFLP markers in 10 linkage groups could be assigned to P. c. chabaudi chromosomes 1 and 5–13, by virtue of their linkage to RFLP markers previously assigned to specific chromosomes by physical mapping [1]. 272 AFLP markers could not be assigned to a specific chromosome. 214 were placed in 12 unassigned linkage groups, each with between 2 and 51 AFLP markers. At least four of these linkage groups are likely to map to chromosomes 2, 3, 4 or 14. The failure to assign these linkage groups occurred because RFLP markers previously mapping to chromosomes 2, 3, 4 and 14 were not allocated to linkage groups. This was probably due to insufficient characterization of the inheritance patterns of these RFLP anchors which were determined in a small number of recombinant clones [1]. For instance, the inheritance of a RFLP marker assigned to chromosome 2, Ca-ATPase, was only determined for 7 out of the 28 recombinant clones. No independent physical mapping of unassigned linkage groups was attempted here.
58 AFLP markers, 21 of which segregated in a non-random fashion, could not be allocated to any linkage groups. Physical mapping of these markers would be required to assign them to specific chromosomes. Alternatively, unassigned linkage groups or unallocated markers might map to the small mitochondrial or apicoplast genomes, although these combined represent only 0.2% of the genome. Several RFLP markers could not be allocated to linkage groups by the Map Manager software, probably due to the small numbers of clones analysed for these markers. These markers were added to assigned linkage groups according to their chromosomal assignment, as previously determined by physical mapping [1].
With the exception of chromosomes 2, 3, 4 and 14 (discussed above), chromosomes 9 and 10 showed the lowest density per Mb of AFLP markers. Chromosome 7 showed the highest density. This may simply reflect natural variation in the frequency of AFLP polymorphisms on particular chromosomes. However, for chromosomes with a low apparent density of AFLP markers such as chromosomes 9 and 10, it is likely that some of the markers in unassigned linkage groups would physically map to these chromosomes. These unassigned groups may not show genetic linkage with (groups of) assigned markers because of factors such as a high rate of recombination between two linkage groups (one linked to the RFLP anchor) or because of an intervening region with a low density of AFLP markers. Both factors, or a combination of the two, may prevent two physically linked groups from being identified as genetically linked. Conversely an apparent unusually high frequency of AFLP markers (as in chromosome 7) may arise from strong linkage disequilibrium between loci on two different chromosomes. Some markers located on one chromosome may thus appear to be genetically linked to markers on another. This could arise where one locus exerts a strong constraint on another unlinked locus. For instance, the AJ allele of an enzyme in a metabolic pathway may only function with the presence of the product of the AJ allele encoding another enzyme in the same pathway. This constraint might be structural or functional. The same may be true of AS alleles of the same enzymes. In this case, the genes encoding these enzymes, and markers strongly linked to them, may appear in the same genetic linkage group.
Order of markers in the linkage groups
AFLP markers were initially ordered on the linkage groups as described in Materials and Methods. After inspection of the predicted marker order, occasional manual adjustments were made to correct markers which appeared to be inappropriately positioned.
Because Map Manager failed to allocate some RFLP markers to an assigned linkage group, these were positioned manually. The final distribution of the markers on the 10 linkage groups assigned to chromosomes is shown in Fig. 1, 2, 3 (See also Additional file 2, Additional file 3. and Additional file 4). 7 unassigned linkage groups containing 9 or more markers each are shown in Fig. 4 (see also Additional file 5).
Figure 1 Linkage map for chromosomes 1, 5, 6 and 7 of the P. c. chabaudi genome. The AFLP and RFLP markers assigned to chromosomes are displayed with genetic distances (in cM). AFLP markers were named as follows: the first two letters identify the clone to which a marker is specific, the next two letters indicate the EcoRI primer selective bases, the numbers identify the marker for that clone and primer combination in order of its molecular size, and the last two letters identify the MseI selective bases. RFLP markers were based on genes previously identified [1].
Figure 2 Linkage map for chromosomes 8–11 of the P. c. chabaudi genome. The AFLP and RFLP markers assigned to chromosomes are displayed with genetic distances (in cM).
Figure 3 Linkage map for chromosomes 12 and 13 of the P. c. chabaudi genome. The AFLP and RFLP markers assigned to chromosomes are displayed with genetic distances (in cM).
Figure 4 Unassigned linkage groups containing more than 8 markers. The AFLP markers assigned to unassigned linkage groups are displayed with genetic distances (in cM)
Number of recombination events per chromosome
If markers and recombination events were both uniformly distributed across the genome, then we would expect the number of predicted recombination events (totalled from 28 clones characterised here) in each chromosome to correlate with its physical size. The predicted total number of recombination events occuring in each linkage group is shown in the Additional file 1. Uniformity was evaluated by comparison of the frequency of recombination events (from the 28 clones) in each chromosome. This varies between 6.7/Mb (chromosome 10)) and 31.1/Mb (chromosome 7), with an overall value of 13.3/Mb. Chromosomes 1 and 11 also showed low frequencies. These differences may reflect natural variation in recombination rates across the genome. However, other factors may also contribute. For instance, there is likely to be a systematic underestimation of recombination frequency because the physical extent of the linkage groups assigned to particular chromosomes will be less than their actual size. For instance, if the chromosome 10 linkage group extends across only half of the chromosome, then the real density of markers and recombination events (per Mb) is likely to be about twice the apparent value given. Indeed, Additional file 1 shows that when data from the unassigned linkage groups are included, the recombination frequency (across the whole genome) increases to 15.9 events per Mb. Regardless of the number of markers assigned to each chromosome, we would expect the number of recombination events per AFLP marker to remain relatively constant, if both the frequency of polymorphism and the rate of recombination vary little between chromosomes. This is indeed the case. For the different chromosomes, the measure varies only between 0.4 and 0.9 recombination events per AFLP marker (see Additional file 1).
Assuming that the P. c. chabaudi genome consists of about 20 Mb [19] the value of 13.3 recombination events in 28 recombinant clones/Mb converts to about 9.5 recombination events/clone/genome which is very close to the value of 10 estimated for Plasmodium falciparum [21].
A significant number of double crossover events around a single AFLP marker were observed in many linkage groups. Any such occurrences were re-evaluated on the original X-ray film to detect any possible errors or ambiguous bands. Many distinct double crossover events were observed. The same phenomenon was also commonly observed in P. falciparum, and were interpreted as being due to non-reciprocal conversion events [20].
Genetic length of linkage groups
The apparent genetic length of each linkage group in cM was calculated by Map Manager based on the number of recombination events (Additional file 1). When added together, the linkage groups assigned to chromosomes combined to give a total genetic length of 1180 cM and the unassigned linkage groups a further 497 cM, giving a total for the genome of 1676 cM. Due to the limited number of clones available (28), the smallest genetic distance that could be measured between two markers was approximately 3.6 cM, corresponding to the presence of a single recombination event between two markers in 28 clones. In general, the estimated genetic lengths of the linkage groups assigned to chromosomes 1, 5–13 increased with the estimated physical sizes of the chromosomes (Figure 5). The Pearson correlation coefficient was 0.794 (p < 0.005). The estimated sizes of map unit for chromosomes 1 and 5–13 are shown in the Additional file 1. These values varied from 8.9 kb/cM (chromosome 5) to 24.1 kb/cM (chromosome 11) with an overall estimated mean of 15.1 kb/cM. It is likely that there is some overestimation of physical size of a map unit because individual identified linkage groups are unlikely to cover the full extent of any chromosome. Indeed, when the genetic lengths of the unassigned linkage groups are included in the analysis, the overall map unit size is reduced to 13.7 kb/cM. The inclusion of additional unallocated markers may reduce this value even further.
Figure 5 Relationship between genetic size and physical size of each chromosome. The relationship between the estimated genetic sizes and physical sizes of chromosomes 1, 5–13 (Table 1) is shown. The correlation between these two variables is 0.794.
However, some overestimation of genetic length in linkage groups is also possible, which leads to an underestimation of map unit size. For instance, Figure 2 shows that chromosomes 9 and 10 both show two abnormally large sections bounded by RFLP anchors without intervening AFLP markers. Specifically, chromosome 9 shows, at one end, an RFLP marker, ran, 40 cM from its nearest AFLP marker. At its other end, an RFLP marker, Ag 3027, lies about 35 cM from its nearest AFLP marker. Chromosome 10 has RFLP marker cDNA121 about 70 cM from its nearest AFLP marker and RFLP marker, 5S rRNA, a further 70 cM distant. These large gaps may be artefacts which arise for two reasons. Firstly, some unreliability in the typing of clones using RFLPs was previously noticed [4]. Secondly, the inheritance patterns of these markers were not determined in all 28 recombinant clones. Markers ran, Ag3027, cDNA121 and 5S rRNA were typed for only 9, 17, 9 and 16 clones, respectively. The characterisation of inheritance of RFLP markers in all of the 28 progeny clones, and the correction of possible mistakes may reduce the estimated genetic length. This would lead to an increase in map unit size. Nevertheless, it is notable that the value reported above (13.7 kb/cM) is close to estimates made for P. falciparum (15–30 kb/cM [21] or 17 kb/cM [20]), although slightly smaller. It is likely that the recombination rate may vary within as well as between chromosomes or genomic loci [22].
Estimate of potential alleles due to indel mutations
Of the 400 AFLP markers placed on chromosomes, 37 AS-AJ pairs (74 markers i.e. 18% of the total) shared the same selective bases at both primer ends and showed complementary segregation in the cross-progeny clones. Most of these markers also differed in size by only a few base pairs. They are likely to be alleles at the same loci. This was confirmed by sequencing two such pairs, namely ASTA01AC and AJTA01AC (chromosome 5), and ASTT02CA and AJTT02CA (chromosome 13) (sequence data not shown). This suggests that a significant proportion of the polymorphisms observed between AS and AJ may be due to small insertions or deletions. In fact, it was observed that small indels tend to occur in introns or intergenic regions (data not shown).
Reliability of the AFLP markers in the progeny clones
A few AFLP markers which were originally identified between clones AJ and AS [11] were not found in the progeny clones. Also, a few bands appeared in the progeny clones that were absent in the parents. All of these markers were ignored during the generation of the linkage map. It is possible that such markers could indicate genetic re-arrangements in the drug-resistant clones AS (3CQ) and AS (30CQ). However, they did not segregate with chloroquine resistance phenotype (data not shown). Some other markers were difficult to investigate because of their proximity to other bands or their location at the bottom of the gel, where bands tend to be fuzzier and more difficult to interpret.
Effect of typing mistakes in the markers
Ongoing work on linkage between chloroquine resistance and markers on chromosome 11 [4] suggested that AFLP and/or RFLP markers were occasionally incorrectly characterised in one ore more of the 28 clones. To test the effect of incorrect typing in AFLP markers, some deliberate mistakes were introduced by changing the parent from which a particular marker was inherited in a particular recombinant clone. The effect of such changes ranged from the appearance or disappearance of predicted double-crossover events and consequent change in the estimated genetic length, to a larger scale change in the order of markers within a linkage group. Occasionally markers were reallocated to a different linkage group. It was concluded that patterns of linkage may be sensitive to errors in genotyping individual clones.
Discussion
Genetic or physical linkage maps have been determined and reported for a number of apicomplexan parasites, including P. falciparum [20], P. c. chabaudi [11], Eimeria tenella [13], Toxoplasma gondii [23], Theileria parva [24] and Cryptosporidium parvum [25]. The map reported here is very extensive in terms of the numbers and density of markers included. Only the genetic map of P. falciparum exceeds its resolution
Of the AFLP markers analysed, most were assigned to 10 of the 14 chromosomes, while some were placed on 12 unassigned linkage groups, which probably include groups located on chromosomes 2, 3, 4 and 14. The remaining AFLP markers could not be allocated to any linkage group. Unallocated AFLP markers and unassigned linkage groups may arise in a number of possible ways, discussed in the Results section above, including mistakes or considerable gaps in the recorded inheritance pattern of RFLP markers and, to a lesser extent, AFLP markers. Other factors include variations in the density of AFLP markers or polymorphism in general, areas of the genome where the rate of recombination is particularly high, linkage disequilibrium between loci on different chromosomes and non-random representation of clones in our sample. The numbers of markers allocated, the approximate genetic lengths and the number of recombination events were estimated for each of the linkage groups. The genetic length of the entire genome was estimated to be 1684 cM and the overall size of map unit 13.7 kb/cM. The genetic length and number of recombination events were expected to increase with the size of chromosomes. This was generally found to be the case, although a number of factors may influence these data. These factors include the failure to assign some linkage groups, incomplete or incorrect inheritance data, particularly for RFLPs, variation in the frequency of AFLP markers across the genome, variation in the recombination rate across the genome, and incorrect assignation of some linkage groups due to linkage disequilibrium.
The presence and frequency of small indel mutations was confirmed. These markers could prove suitable for rapid typing of clones by size polymorphism and quantitative analysis by Real Time Quantitative PCR.
The generation of a complete AFLP genetic linkage map for P. chabaudi was originally conceived as an essential step towards the identification of loci linked to genes encoding important phenotypes, such as drug resistance. Indeed, the identification of a locus underlying chloroquine resistance in P. chabaudi within approximately 250 kb of chromosome 11 [4] relied upon elements of the present map in the analysis of linkage between phenotype (chloroquine resistance) and genotype (inheritance of parental AFLP markers) in individual recombinant clones. However we have also developed a novel strategy called Linkage Group Selection [5,26] which more rapidly identifies loci linked to genes underlying selectable phenotypes, such as drug resistance. For example, a drug resistant parasite is crossed with a genetically different drug sensitive parasite. The uncloned recombinant progeny are drug treated, and AFLP markers which are linked to loci underlying drug resistance may be identified as those reduced in their representation or intensity [27]. A genetic linkage map enables us to determine whether AFLP markers which are significantly reduced in intensity lie in the same linkage group, prior to further sequence analysis. Genome sequence data are now available for P. c. chabaudi (partial) [8] and P. falciparum (complete) [7], and sequenced AFLP markers can sometimes be mapped to the P. falciparum genome. Because of the extensive gene synteny between P. chabaudi (and other rodent malarias) and P. falciparum [4,9,10], markers closely linked in P. falciparum are likely to be closely linked in P. chabaudi too.
The correspondence between the genetic linkage map reported here and the mapping of AFLP markers to the P. falciparum genome in the studies discussed above [4,5,26] has increased our confidence both in the genetic linkage map reported here, and the extent of gene synteny between the P. c. chabaudi, Plsmodium yoelii and P. falciparum genomes [9,10]. The existence of a rodent malaria genome map, complete with syntenic relationships between it and the P. falciparum genome (Taco Kooij and Andy Waters, personal communication), will allow us to assign unallocated AFLP markers and unassigned linkage groups to particular chromosomes on the assumption that gene synteny is conserved.
Authors' Contributions
AM characterized the AFLP markers in the cross progeny, generated the genetic linkage map and drafted the article, PH helped in the generation of the linkage map, analysed genetic data from it and drafted the article, RF helped in the characterization of the AFLP markers in the cross progeny, PC and DW provided the recombinant clones and revised the article, RC designed and coordinated the study, revised the article and gave final approval. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
The file (.XLS) contains the following data for markers assigned to chromosomes and, where possible, for markers in unassigned linkage groups and the overall genome: – physical size of each chromosome (Mb), the number of markers (either RFLP or AFLP and total), the number of AFLP markers per Mb, the number of recombination events predicted for the 28 clones, the frequency of recombination events per Mb and per AFLP marker, the predicted genetic length of all linkage groups, and the estimated size of map unit (kb/cM).
Click here for file
Additional File 2
This file is the original PPT files from which figure 1 was derived.Figures 1-3 contain the linkage map for the chromosomes 1 and 5-13.
Click here for file
Additional File 3
This file is the original PPT files from which figure 2 was derived.Figures 1-3 contain the linkage map for the chromosomes 1 and 5-13.
Click here for file
Additional File 4
This file is the original PPT files from which figure 3 was derived.Figures 1-3 contain the linkage map for the chromosomes 1 and 5-13.
Click here for file
Additional File 5
This file is the original PPT files from which figure 4 was derived. Figure 4
contains various unassigned linkage groups.
Click here for file
Acknowledgments
We would like to thank Deborah Charlesworth and Peter Visscher for reading the manuscript and providing valuable comments and Les Steven for technical assistance. We acknowledge the support of the Medical Research Council (PH, RF, DW), the Wellcome Trust (RC) and the Ministério da Ciência e Ensino Superior/FCT of Portugal (PC). A special acknowledgement goes to Danilo Martinelli for supporting A. Martinelli.
==== Refs
Carlton JMR Mackinnon M Walliker D A chloroquine resistance locus in the rodent malaria parasite Plasmodium chabaudi Mol Biochem Parasitol 1998 93 57 72 9662028 10.1016/S0166-6851(98)00021-8
Cravo PV Carlton JM Hunt P Bisoni L Padua RA Walliker D Genetics of mefloquine resistance in the rodent malaria parasite Plasmodium chabaudi Antimicrob Agents Chemother 2003 47 709 18 12543682 10.1128/AAC.47.2.709-718.2003
Hayton K Ranford-Cartwright LC Walliker D Sulfadoxine-pyrimethamine resistance in the rodent malaria parasite Plasmodium chabaudi Antimicrob Agents Chemother 2002 46 2482 2489 12121922 10.1128/AAC.46.8.2482-2489.2002
Hunt P Martinelli A Fawcett R Carlton J Carter R Walliker D Gene synteny and chloroquine resistance in Plasmodium chabaudi Mol Biochem Parasitol 2004 136 157 164 15478795 10.1016/j.molbiopara.2004.03.008
Martinelli A Cheesman S Hunt P Culleton R Raza A Mackinnon M Carter R A genetic approach to the de novo identification of targets of strain-specific immunity in malaria parasites Proc Natl Acad Sci U S A
Carlton JM Hayton K Cravo PV Walliker D Of mice and malaria mutants: unravelling the genetics of drug resistance using rodent malaria models Trends Parasitol 2001 17 236 242 11323308 10.1016/S1471-4922(01)01899-2
Gardner MJ Hall N Fung E White O Berriman M Hyman RW Carlton JM Pain A Nelson KE Bowman S Paulsen IT James K Eisen JA Rutherford K Salzberg SL Craig A Kyes S Chan MS Nene V Shallom SJ Suh B Peterson J Angiuoli S Pertea M Allen J Selengut J Haft D Mather MW Vaidya AB Martin DM Fairlamb AH Fraunholz MJ Roos DS Ralph SA McFadden GI Cummings LM Subramanian GM Mungall C Venter JC Carucci DJ Hoffman SL Newbold C Davis RW Fraser CM Barrell B Genome sequence of the human malaria parasite Plasmodium falciparum Nature 2002 419 498 511 12368864 10.1038/nature01097
Plasmodium chabaudi Partial Genome Shotgun
Carlton JMR Vinkenoog R Waters AP Walliker D Gene synteny in species of Plasmodium Mol Biochem Parasitol 1998 93 285 294 9662712 10.1016/S0166-6851(98)00043-7
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
Grech K Martinelli A Pathirana S Walliker D Hunt P Carter R Numerous, robust genetic markers for Plasmodium chabaudi by the method of amplified fragment length polymorphism Mol Biochem Parasitol 2002 123 95 104 12270625 10.1016/S0166-6851(02)00142-1
Carlton JMR The genetics of chloroquine resistance in the rodent malaria parasite Plasmodium chabaudi PhD thesis University of Edinburgh, School of Biological Sciences 1995
Shirley MW Harvey DA A genetic linkage map of the apicomplexan protozoan parasite Eimeria tenella Genome Res 2000 10 1587 1593 11042156 10.1101/gr.149200
Tait A Masiga D Ouma J MacLeod A Sasse J Melville S Lindegard G McIntosh A Turner M Genetic analysis of phenotype in Trypanosoma brucei: a classical approach to potentially complex traits Philos Trans R Soc Lond B Biol Sci 2002 357 89 99 11839186 10.1098/rstb.2001.1050
Jacobs RL Role of p-aminobenzoic acid in P. berghei infection in the mouse Exp Parasitol 1964 15 213 215 14191322 10.1016/0014-4894(64)90017-7
Carter R Walliker D New observations on the malaria parasites of rodents of the Central African Republic -Plasmodium vinckei petteri subsp. nov. and Plasmodium chabaudi Landau, 1965 Ann Trop Med Parasitol 1975 69 187 196 1155987
Padua RA Plasmodium chabaudi: genetics of resistance to chloroquine Exp Parasitol 1981 52 419 426 6947894 10.1016/0014-4894(81)90101-6
Vos P Hogers R Bleeker M Reijans M van de Lee T Hornes M Frijters A Pot J Peleman J Kuiper M Zabeau M AFLP: a new technique for DNA fingerprinting Nucleic Acids Res 1995 23 4407 4414 7501463
Manly KF Cudmore RH JrMee JM Map Manager QTX, cross-platform software for genetic mapping Mamm Genome 2001 12 930 932 11707780 10.1007/s00335-001-1016-3
Su X Ferdig MT Huang Y Huynh CQ Liu A You J Wootton JC Wellems TE A genetic map and recombination parameters of the human malaria parasite Plasmodium falciparum Science 1999 286 1351 1353 10558988 10.1126/science.286.5443.1351
Walker-Jonah A Dolan SA Gwadz RW Panton LJ Wellems TE An RFLP map of the Plasmodium falciparum genome, recombination rates and favored linkage groups in a genetic cross Mol Biochem Parasitol 1992 51 313 320 1349423 10.1016/0166-6851(92)90081-T
Nachman MW Variation in recombination rate across the genome: evidence and implications Curr Opin Genet Dev 2002 12 657 663 12433578 10.1016/S0959-437X(02)00358-1
Sibley LD LeBlanc AJ Pfefferkorn ER Boothroyd JC Generation of a restriction fragment length polymorphism linkage map for Toxoplasma gondii Genetics 1992 132 1003 1015 1360931
Morzaria SP Young JR Restriction mapping of the genome of the protozoan parasite Theileria parva Proc Natl Acad Sci U S A 1992 89 5241 5245 1608931
Piper MB Bankier AT Dear PH A HAPPY map of Cryptosporidium parvum Genome Res 1998 8 1299 1307 9872984
Culleton R Martinelli A Hunt P Carter R Linkage group selection: Rapid gene discovery in malaria parasites Genome Res
Martinelli A Hunt P Cheesman SJ Carter R The application of Amplified Fragment Length Polymorphism to the genetic analysis of selectable phenotypes in malaria parasites Mol Biochem Parasitol 2004 136 117 122 15478791 10.1016/j.molbiopara.2004.02.011
| 15707493 | PMC550669 | CC BY | 2021-01-04 16:37:30 | no | Malar J. 2005 Feb 11; 4:11 | utf-8 | Malar J | 2,005 | 10.1186/1475-2875-4-11 | oa_comm |
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-121571590310.1186/1477-7819-3-12ResearchA case-control study to evaluate urinary tract complications in radical hysterectomy Behtash Nadereh [email protected] Fatemeh [email protected] Haleh [email protected] Hediyeh [email protected] Parviz [email protected] Department of Gynaecology Oncology, Tehran University of Medical Sciences, Tehran, Iran2 Vali-e-Asr Reproductive Health Research Center, Tehran University of Medical Sciences, Tehran, Iran3 Department of Gynaecology and Obstetrics, Tehran University of Medical Sciences, Tehran, Iran4 Department of Gynaecology Oncology, Rosenfeld Cancer Center, Abington Memorial Hospital, Abington, PA, USA2005 16 2 2005 3 12 12 13 8 2004 16 2 2005 Copyright © 2005 Behtash et al; licensee BioMed Central Ltd.2005Behtash 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
This study has evaluated urinary tract injuries and dysfunction after Radical Hysterectomy (RH) performed in patients with cervical cancer and has compared the cystometric parameters and urinary complications occurring in these patients with those occurring in patients who had undergone Simple Hysterectomy (SH).
Patients and methods
A prospective case-control study was conducted to evaluate urinary tract injuries (intra-operative and post-operative) and dysfunction in 50 patients undergoing RH for cervical cancer and to compare them with the same parameters in 50 patients who underwent SH for benign disease.
Results
Mean age in the RH group was 46.3 years and in the SH group was 50.1 (p = 0.63). There were no bladder and urethral injuries in either group of patients. There was one intra-operative ureteral injury in the RH patients but none in those who underwent SH. (p < 0.05). In the two weeks after surgery, 15% of RH patients and 11% of SH patients had experienced a urinary tract infection urinary tract infection (p = 0.61). Two week after surgery 62% of RH patients had no urinary symptoms, compared to 84% in the SH group who did (p < 0.02). Urinary residual volume, first urinary sensation and maximal bladder capacity were higher in the RH group, but this was not statistically significant. The only case of a urinary fistula appeared in a patient who received 5000 cGy radiation therapy pre-operatively, but this spontaneously healed after 3 weeks of catheterization.
Conclusions
Intra-operative and post-operative urinary tract complications are comparable in patients undergoing RH and SH and an expert gynaecological oncologist might be able to further decrease complications. However, radiation therapy before surgery may increase the risk of complications.
==== Body
Background
Although the incidence of lower urinary tract complications after RH has been reported with variable rates, up to one half of patients undergoing RH experience at least one lower urinary tract symptom that develops after surgery and at a variable period of time [1,2]. Several retrospective studies have examined lower urinary tract dysfunction and traumatic injuries in patients who have undergone RH [3,4].
In this study, we prospectively evaluated intra-operative urinary tract injuries in addition to post-operative urinary tract dysfunction and infection at 2, 6, and 14 weeks following surgery. We also compared these findings with those at the same times in patients who underwent SH for benign disease.
Patients and methods
Between October 2000 and December 2002, 50 women who underwent RH and bilateral lymph node dissection (BLND) were considered eligible for inclusion in the study. All patients had squamous cell carcinoma of cervix (SCC Cx.) and were staged as being Stages I and II. The operations were performed by the same gynaecological surgeons, using the same standard technique (class III Piver & Rutledge). Pre-operative management was standardised for all patients. Preoperatively a detailed medical history, physical examination, routine laboratory tests, pelvic CT-scan (with intravenous and oral contrast), urine analysis, and urine culture were carried out.
The exclusion criteria were; a history of voiding dysfunction, previous pelvic surgery, brain or spinal cord diseases, diabetes mellitus, and contraindications to urodynamic studies. The latter included a history of vesicoureteral reflux, hydronephrosis, frequent or recent urinary tract infection or urethral stricture. Patients received one pre-operative and three post-operative doses of a second generation cephalosporin (Cephazolin).
The duration of surgery, amount of intra-operative haemorrhage and the occurrence of any organ injuries were recorded. A Foley's catheter was inserted at the time of surgery and was left in place for two weeks after surgery. The patient's urinary catheter was removed when their post-voiding residual volume was less than 75 ml.
Water cystometry, urinalysis and urine culture were performed at 2, 6, 14 weeks after operation. The test for water cystometry was performed with the subject lying in a supine position. A 12F double-lumen catheter was introduced transurethrally into the bladder to withdraw residual urine. The pressure-volume relationship of the bladder was determined by filling the bladder with isotonic saline at a rate of 30–50 ml/min. The cystometry fill phase ceased when the patient experienced an urge to void urine, the first indication being leakage through the urethra, or a bladder volume of 600 ml (which ever occurred first). The volume at the termination of the fill-phase was designated as the maximum bladder capacity (MBC). We also assessed the bladder volume of each patient at their first desire to void (V desire, ml). Post-void residual urine volume (RU) was determined by transurethral catheterization after voiding had ceased. The presence or absence of any urinary symptoms was determined by both questionnaire and direct interview with the patient.
Fifty patients with benign disease who had underwent SH, were evaluated at the same time periods in the same way for comparison with the RH group of patients. In the SH group of patients, the Foley's catheter was inserted for 24 hours after operation. Data were analysed by SPSS statistical software using the chi-square and Student's 't' test for data analysis.
Results
During this study, 50 patients with early stage cervical cancer and who underwent RH for cervical cancer and 50 patients who had undergone SH for benign disease were evaluated. Two patients in the RH group and 3 from the SH group were lost during the study. The mean ages and their BMIs (Body Mass Index) in two groups of patients were not statistically different. However, parity in the RH group was higher (p < 0.05) (Table 1). In the RH group, the stages for the cervical cancer were 65.1%, 23.2% and 11.6% for I, IIA and IIB stages, respectively. Patients who had stage IB2 or higher stages of cervical cancer received 4500–5000 cGy of irradiation pre-operatively. None of these patients received adjuvant radiation during the interval between surgery and performance of urodynamic studies.
Table 1 Comparison of characteristics of Radical and Simple Hysterectomy groups of patients.
Characteristics RH group SH group p
Mean age 50.10 46.35 0.63
BMI 24.25 26.05 0.66
Parity 6 4 0.00
Blood loss (ml) 576 ml 416 ml 0.04
Mean operative time (min) 183. min 112. min 0.00
RH-Radical hysterectomy; SH-simple hyterectomy
In the SH group, the most common pathological conditions requiring hysterectomy were as follows; dysfunctional uterine bleeding (47.83%), uterine myoma (21.7%), ovarian cyst (10.8%), chronic pelvic pain (4.3%), adenomyosis (4.3%), endometrial cancer (4.35%), CIN (4.35%) and molar pregnancy (2.17%).
The average blood loss and mean operative time for both groups are shown in table 1. There were no bladder and urethral injuries occurring during the primary operation in either of the two groups of patients. One patient (with stage Ib1 cervical cancer) in the RH group had an intra-operative ureteral injury (which happened at the time of "unroofing" the distal part), and the ureteral anastomosis was carried out at that time. The urinary catheter and ureteral stent were removed four weeks after operation.
Another patient (with stage Ib2 cervical cancer) had received chemo-irradiation (5000 cGy) pre-operatively. She had a spontaneous urine leakage from the vagina approximately 2 months after surgery. A retrograde cystography revealed a minute vesico-vaginal fistula. After 3 weeks of continuous bladder drainage, the fistula resolved spontaneously and she had no urinary leakage at her follow-up visits.
Post-operative positive urine culture and urinary symptoms (dysuria, frequency, nocturia and dribbling) are showed in table 2. Urinary symptoms occurred more commonly in patients who had undergone pre-operative radiotherapy, but this difference was not statistically significant (table 3). The abnormal findings as regards water cystometry are shown in table 4.
Table 2 Postoperative urinary symptoms in RH and SH group of patients.
RH group SH group p
Positive U/C 1st visit 15% 11% 0.06
Positive U/C 2nd visit 31% 20% 0.00
Positive U/C 3rd visit 11% 9% NS
Urinary symptoms 1st visit 31% 20% 0.00
Urinary symptoms 2nd visit 40% 34% 0.07
Urinary symptoms 3rd visit 30% 33% NS
U/C-Urine Culture ; NS-Not significant; RH-Radical hysterectomy; SH-simple hyterectomy
Table 3 Comparison of urinary symptoms between patients undergoing RH but with or without pre-operative radiotherapy.
Urinary symptoms (Postoperatively) RH group XRT + RH p
2 weeks, 1st visit 39% 44% 0.76
6 weeks, 2nd visit 36% 44% 0.64
14 weeks, 3rd visit 26% 44% 0.27
XRT-History of pre-operative radiotherapy; RH-Radical hysterectomy
Table 4 Relative frequency of RV, MC, FS, SI and UTI in the patients.
Abnormal RV (%) Abnormal FS (%) Abnormal MC (%) Abnormal SI (%) UTI (%)
First visit* RH 4 66 68 22 31
SH 0 67 69 17 20
Second visit** RH 0 58 61 20 31
SH 0 66 70 16 20
Third visit*** RH 0 49† 65 17 11
SH 4 72† 65 18 9
*After discharging the drain
**Four weeks after discharging the drain
***Eight weeks after discharging the drain
†P-value = 0.02
Discussion
Modern surgical techniques have resulted in a decrease in the incidence of lower urinary tract complications occurring as a result of radical hysterectomy. In particular, in recent times, various surgical strategies have been developed to avoid damaging the inferior segment of the cardinal ligament as well as the terminal bundle in the uterosacral and pubocervicovesical ligaments. This has made it possible for patients' lower urinary functions to return more rapidly to their pre-operative states [2,5]. However, transient post-operative urinary dysfunction involving urinary storage and evacuation function continues to be of concern [6].
In a study by Zaino and colleagues intra-operative complications were reported to occur as being 4.5% urinary tract and 8.7% other organs (nervous, haemorrhage, intestinal) [7]. Ralph et al, reported a 6.6% rate of intra-operative urinary tract injuries during radical surgery for cervical carcinoma [2]. In our study, we had no intestinal or bladder injuries occurring during radical hysterectomy. The only ureteral injury (2%) occurred during "unroofing" of the distal ureter and this was recognised and repaired immediately.
Zaino et al, [7] reported a 20% risk of a post-operative urinary tract fistula after radical surgery [7] and this contrasts with a 4.4% risk of fistula in their series reported by Ralph [2]. In our study, the only fistula occurred in a patient who had received 5000 cGy radiotherapy before radical surgery, and with continuous bladder drainage for 3 weeks there was spontaneous healing of the fistula. The incidence of urinary tract infection (UTI) in our series was 11% by 14 weeks after surgery and this was comparable to that reported by Cardosi [8] but less than the figure of 20% documented by Abrao [9]. Also, Chen reported a 14% urinary tract infection rate following radical surgery [10].
In our study, urinary tract dysfunctions that followed radical surgery were that 4% had an abnormal post-voiding residual volume at the first post-operative visit. The first voiding sensation at the third visit was 49% and stress urinary incontinence was 17%. However, maximal capacity remained abnormal in 65% of cases by 14 weeks after surgery. Ralph et al reported that 67% of patients had impairment or absence of bladder sensation after a RH [2]. In the study from Chen et al, 84% of patients had an increased first desire to void and maximal capacity in the post-operative period [10].
Urinary symptoms in our study occurred in 20% 2 weeks after operation and which were higher in patients with pre-operative radiotherapy (although not statistically significant). Urinary symptoms remained high at the third post-operative visit, although they declined in patients who had undergone surgery alone.
In our study, the patients mean age; BMI, parity, operative time, and blood loss were higher in those undergoing RH. The mean age of our patients was higher than patients in the study by Vervest et al, (mean was 45 years) [11]. Also in this study [11] the patients BMI of 23.2 was lower than that of the patients in our study. Cystometric parameters and intra-operative and post-operative complications showed little difference between patients having either RH or SH. The small number of patients in our study could have biased the results. However, in spite of the different gravidity and days of bladder drainage in the two groups of patients, the overall complication rate is relatively low in the RH patients. The data in this study requires confirmation from future multicentric studies with greater numbers of patients.
In recent years, several studies support the role of a gynaecological oncologist who is specifically trained in such aspects of care and who can obtain optimal cytoreductive surgery in patients with ovarian carcinoma [12,13]. Therefore, it would seem that an experienced and appropriately trained gynaecological oncologist might achieve a complication rate for patients undergoing radical hysterectomy comparable with that reported by "general gynaecological surgeons..
Competing interests
The author(s) declare that they have no competing interests.
Authors' Contributions
NB carried out the surgery and participated in drafting the manuscript.
FG carried out follow-ups.
HA participated in the design of the study and helped to draft the manuscript.
HK helped in follow-ups and performed the statistical analyses.
PH participated in the design of the study and helped to draft the manuscript.
All authors read and approved the final manuscript.
Acknowledgements
The study was approved by the institutional review board
==== Refs
Kadar N Saliba N Nelson JH The frequency, causes and prevention of severe urinary dysfunction after hysterectomy Br J Obstet Gynaecol 1983 90 858 863 6615743
Ralph G Winter R Michelitsh L Tamussino K Radicality of the parametrial resection and dysfunction of lower urinary tract after radical hysterectomy Eur J Gynaecol Oncol 1991 12 27 30 2050156
Glahn BE The neurogenic factor in vesical dysfunction following radical hysterectomy for carcinoma of the cervix Scand J Urol Nephrol 1970 4 107 116 5518084
Scotti R Bergman A Bhatia NN Ostergard DR Urodynamic changes in urethrovesical function after radical hysterectomy Obstet Gynecol 1986 68 111 120 3725241
Fishman IJ Shabsigh R Kaplan AL Lower urinary tract dysfunction after radical hysterectomy for carcinoma of the cervix Urology 1986 28 462 468 3787918 10.1016/0090-4295(86)90144-5
Sekido N Kawai K Akaza T Lower urinary tract dysfunction as persistent complication of radical hysterectomy Int J Urol 1997 4 259 264 9255663
Zaoino L Albiero A Divirgilio G Radical hysterectomy. Operative complications Minerva Ginecol 1993 45 591 596 8139784
Cardosi RJ Cardosi RP Grendys EC Fiorica JV Hoffman MS Infectious urinary tract morbidity with prolonged bladder catheterization after radical hysterectomy Am J Obstet Gynecol 2003 189 380 384 14520200 10.1067/S0002-9378(03)00696-3
Abrao FS Breitbarg RC Oliveira AT Vasconcelos FA Complications of surgical treatment of cervical carcinoma Braz J Med Biol Res 1997 30 29 33 9222400
Chen GD Lin LY Wang PH Urinary tract dysfunction after radical hysterectomy for cervical cancer Gynecol Oncol 2002 85 292 297 11972390 10.1006/gyno.2002.6614
Vervest HA Barents JW Haspels AA Debruyne FM Radical hysterectomy and the function of the lower urinary tract. Urodynamic quantification of changes in storage and evacuation function Acta Obstet Gynecol Scand 1989 68 331 340 2618621
Olaitan A Weeks J Mocroft A Smith J Howe K Murdoch J The surgical management of women with ovarian cancer in the south west of England Br J Cancer 2001 85 1824 1830 11747321 10.1054/bjoc.2001.2196
Carney ME Lancaster JM Ford C Tsodikov A Wiggins CL A population-based study of patterns of care for ovarian cancer: who is seen by a gynecologic oncologist and who is not? Gynecol Oncol 2002 84 36 42 11748973 10.1006/gyno.2001.6460
Trimbos JB Franchi M Zanaboni F Velden J Vergote I 'State of the art' of radical hysterectomy; current practice in European oncology centres Eur J Cancer 2004 40 375 378 14746855 10.1016/j.ejca.2003.09.017
Sartori E Zanagnolo V Complications following radical hysterectomy for cervical carcinoma Recenti Prog Med 2003 94 562 567 14974153
Zullo MA Manci N Angioli R Muzii L Panici PB Vesical dysfunctions after radical hysterectomy for cervical cancer: a critical review Crit Rev Oncol Hematol 2003 48 287 293 14693341
| 15715903 | PMC550670 | CC BY | 2021-01-04 16:39:05 | no | World J Surg Oncol. 2005 Feb 16; 3:12 | utf-8 | World J Surg Oncol | 2,005 | 10.1186/1477-7819-3-12 | oa_comm |
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-51571004610.1186/1742-4682-2-5ResearchPropagated repolarization of simulated action potentials in cardiac muscle and smooth muscle Sperelakis Nicholas [email protected] Lakshminarayanan [email protected] Bijoy [email protected] Dept. of Molecular & Cellular Physiology University of Cincinnati College of Medicine Cincinnati, OH 45267-0576 USA2 Dept. of Electrical Computer Engineering and Computer Science University of Cincinnati College of Engineering Cincinnati, OH 45219 USA2005 14 2 2005 2 5 5 30 11 2004 14 2 2005 Copyright © 2005 Sperelakis et al; licensee BioMed Central Ltd.2005Sperelakis 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
Propagation of repolarization is a phenomenon that occurs in cardiac muscle. We wanted to test whether this phenomenon would also occur in our model of simulated action potentials (APs) of cardiac muscle (CM) and smooth muscle (SM) generated with the PSpice program.
Methods
A linear chain of 5 cells was used, with intracellular stimulation of cell #1 for the antegrade propagation and of cell #5 for the retrograde propagation. The hyperpolarizing stimulus parameters applied for termination of the AP in cell #5 were varied over a wide range in order to generate strength / duration (S/D) curves. Because it was not possible to insert a second "black box" (voltage-controlled current source) into the basic units representing segments of excitable membrane that would allow the cells to respond to small hyperpolarizing voltages, gap-junction (g.j.) channels had to be inserted between the cells, represented by inserting a resistor (Rgj) across the four cell junctions.
Results
Application of sufficient hyperpolarizing current to cell #5 to bring its membrane potential (Vm) to within the range of the sigmoidal curve of the Na+ conductance (CM) or Ca++ conductance (SM) terminated the AP in cell #5 in an all-or-none fashion. If there were no g.j. channels (Rgj = ∞), then only cell #5 repolarized to its stable resting potential (RP; -80 mV for CM and -55 mV for SM). The positive junctional cleft potential (VJC) produced only a small hyperpolarization of cell #4. However, if many g.j. channels were inserted, more hyperpolarizing current was required (for a constant duration) to repolarize cell #5, but repolarization then propagated into cells 4, 3, 2, and 1. When duration of the pulses was varied, a typical S/D curve, characteristic of excitable membranes, was produced. The chronaxie measured from the S/D curve was about 1.0 ms, similar to that obtained for muscle membranes.
Conclusions
These experiments demonstrate that normal antegrade propagation of excitation can occur in the complete absence of g.j. channels, and therefore no low-resistance pathways between cells, by the electric field (negative VJC) developed in the narrow junctional clefts. Because it was not possible to insert a second black-box into the basic units that would allow the cells to respond to small hyperpolarizing voltages, only cell #5 (the cell injected with hyperpolarizing pulses) repolarized in an all-or-none manner. But addition of many g.j. channels allowed repolarization to propagate in a retrograde direction over all 5 cells.
Propagated RepolarizationSimulated Action PotentialsPSpice simulationsElectric Field mechanismCardiac electrophysiology
==== Body
Introduction
There are no low-resistance connections between the cells in several different cardiac muscle and smooth muscle preparations [reviewed in refs. [1] and [2]]. In a computer simulation study of propagation in cardiac muscle, it was shown that the electric field (EF) that is generated in the narrow junctional clefts, when the prejunctional membrane fires an action potential (AP), depolarizes the postjunctional membrane to its threshold [3-5]. Propagation by mechanisms not requiring low-resistance connections have also been proposed by others [6-9]. This results in excitation of the postjunctional cell, after a brief junctional delay. The total propagation time consists primarily of the summed junctional delays. This results in a staircase-shaped propagation, the surface sarcolemma of each cell firing almost simultaneously [4]. Propagation has been demonstrated to be discontinuous (or saltatory) in cardiac muscle [10-13]. Fast Na+ channels are localized in the junctional membranes of the intercalated disks of cardiac muscle [5,14,15], a requirement for the EF mechanism to work [1-5].
We recently modeled propagation of APs of cardiac muscle and smooth muscle using the PSpice program for circuit design and analysis [16-18]. Like the mathematical simulation published in 1977 [3] and 1991 [4], the EF developed in the junctional clefts (negative VJC) was large and sufficient to allow transfer of excitation to the contiguous cell, without the requirement of gap-junction (g.j.) channels. Propagation of excitation can occur by the EF mechanism alone, even when the excitability of the cells was made low. In connexin-43 (heterozygous) and Cx40 knockout mice, propagation in the heart still occurs, but it is slowed [19-22] as predicted by our PSpice simulation study [18].
The present experiments were carried out to study propagated repolarization in this model of simulated action potentials (APs). Propagation of repolarization is a phenomenon that occurs in cardiac muscle [23]. It has been shown that propagation of vasodilation occurs in the microvasculature [24], and that the endothelial cells are involved in the conduction of hyperpolarization and vasodilation in an artery [25]. Therefore, our hypothesis was that propagated repolarization would also occur in our PSpice model.
Methods
The methods used and PSpice program (Cadence Co, Portland) have been described in detail previously, including the circuit [17,18]. In brief, each cell was represented by four basic excitable units, two for the long surface membrane of the cell (one upward-facing and one downward-facing) and one basic unit for each of the two junctional membranes (left end of cell and right end) (Fig 1). The radial (shunt) resistance of the junctional cleft (RJC) was placed in the junctions between adjoining cells. The basic units were connected internally by the intracellular longitudinal resistance (ri). The basic units were connected externally with the extracellular resistance (RO), broken down into a longitudinal component (Rol) and a transverse (radial) component (Ror). RO was connected to ground as depicted in Figure 1. The circuit used for each unit was kept as simple as possible, using only those ion channels that set the resting potential (RP) and predominate during the rising phase and plateau phase of the AP.
Figure 1 Cardiac Muscle and Smooth Muscle. Circuit diagram used for study of propagated repolarization in cardiac muscle and smooth muscle. A: 5-cell chain. A depolarizing stimulating pulse (IS1; 0.50 ms, 0.25 nA) was applied to the inside of the first cell (A1; left side). A hyperpolarizing pulse (IS2; variable intensity and duration) was applied to the inside of the fifth cell (A5; right side) a few milliseconds later when the action potentials (APs) initiated by IS1 were in their plateau phase (peak overshoot). B: Enlarged diagram to show a portion of the circuit for details of the basic units.
The myocardial cell was assumed to be a cylinder 150 μm long and 16 μm in diameter, and the smooth muscle cell a cylinder 200 μm long and 5 μm diameter. Since in vascular smooth muscle (VSM), the muscle fibers run in a circular direction, if transverse velocity is calculated, the fiber diameter should be used. The values of the capacitive and the resistive elements in each basic unit were set to reflect the input resistance (ca 20 MΩ) and input capacitance (ca 100 pF) of the individual cells, and the junctional units were prorated, with respect to the surface units, based on relative areas represented. At rest, the resistance of K+ compared to Na+ (cardiac muscle) or Ca++ (smooth muscle) were set to give resting potentials (RPs) of -80 mV for cardiac muscle and -55 mV for smooth muscle. During excitation, the action potentials (APs) overshot to +32 mV and +11 mV, respectively.
Electrical stimulations (IS1) were always applied internally to the first cell of the chain (cell A1). Rectangular depolarizing current pulses of 0.25 nA amplitude and 0.50 ms duration were applied. The delay time before the IS1 pulse was applied was usually set to 1.0 ms in SM. A second stimulus (IS2) that was hyperpolarizing was applied to the inside of the last cell (A5) of the chain when the APs of all 5 cells were in their plateau phase. The intensity and duration of the IS2 pulses were varied over a wide range in order to generate strength / duration (S/D) curves.
Because the PSpice program does not have a voltage-dependent resistance (to generate the increase in Na+ or Ca++ conductance during excitation), this function had to be done with a V-controlled current source (our "black-box"). The sigmoidal relationship between conductance and membrane potential (VM), over a relatively narrow VM range, was mimicked by the black-box. The Na+ or Ca++ current required for excitation had to be calculated for several VM values and inserted into the GTABLE function.
Experiments were done with a single chain of 5 cells or 2 cells. There were no gap junctions between the cells of the chain under initial conditions. The presence of gap junction connexons (tunnels) was represented by adding a variable shunt resistance (Rgj) across each cell-to-cell junction. This resistor connected the inner surface of the prejunctional membrane with the inner surface of the postjunctional membrane. This Rgj shunt resistance was varied between 10,000 MΩ (1 tunnel), 1000 MΩ (10 tunnels in parallel), 100 MΩ (100 tunnels), 10 MΩ (1,000 tunnels), and 1.0 MΩ (10,000 tunnels). Each tunnel was assumed to have a conductance of 100 pS.
Results
A. All-or- None Repolarization of Stimulated Cell A5
There was a sharp (all-or-none) repolarization of the stimulated cell (A5) of the 5-cell chain in both cardiac muscle (Fig. 2AB) and smooth muscle (Fig. 2CD). As shown, stimulation of cell A1 with a depolarizing current pulse (IS1) produced propagation of APs down the chain. At the plateau (peak) of the APs, a repolarizing pulse applied intracelluarly to cell A5, if of sufficient intensity (duration constant), produced a sudden repolarization of only cell A5 (Fig. 2B for cardiac muscle and D for smooth muscle). A slightly lower current intensity failed to produce a stable repolarization of cell A5 (Fig. 2A for cardiac muscle and 2C for smooth muscle). Note that the potential change (repolarizing) produced in neighboring call A4 was very small (< 1 mV). This emphasizes that there are indeed no low-resistance connections between the modeled cells under standard conditions. The hyperpolarizing pulse had to bring the Vm of cell A5 into the region of the GTABLE's sigmoidal curve. The transient repolarization is in agreement with the biological case [23-25].
Figure 2 Sharp Repolarization of Stimulated Cell (A5). Sharp repolarization of only the last cell (5th) of the 5-cell chain when a repolarizing IS2 pulse was applied in cardiac muscle (A-B) and in smooth muscle (C-D). Panels A and C illustrate the records obtained when the applied IS2 pulse was just not quite strong enough to produce a permanent repolarization of cell #5. In panels B and D, the IS2 intensity was slightly increased to produce an all-or-none repolarization. The membrane potential of adjacent cell #4 (A4) was only slightly changed when cell A5 underwent a very large change. The velocity of antegrade propagation (θa) was about 54 cm/sec in CM and 8.9 cm/sec in SM under that conditions.
B. Propagation of Repolarization
As indicated in the Methods section, it was not possible to insert a second black-box in the K+ leg of the basic circuit, because the PSpice program became erratic. Therefore, in order to achieve propagation of the repolarization of cell A5 in the retrograde direction, it was necessary to insert gap-junction channels between the cells of the chain (1, 10, 100, 1000, 10000 channels). This corresponded to adding resistive shunts between the cells across the junctions (Rgj) of 10000, 1000, 100, 10, and 1.0 MΩ (assuming each channel has a conductance of 100 pS).
The results of doing such an experiment are shown in Fig 3 for cardiac muscle (A – C) and for smooth muscle (D – F). When there were many channels (e.g. 10,000 in Fig. 3A and 3D or 1000 in Fig. 3B and 3E), the rising phase of the APs of all 5 cells were superimposed. This means that all 5 cells fired nearly simultaneously, as expected because of the high degree of low-resistance coupling. However, when a repolarizing current pulse was applied to cell A5, its repolarization spread to the neighboring cells. But the other cells did not repolarize simultaneously, as can be seen. Instead, there was a propagation of the repolarization at a certain velocity. This repolarization velocity became slower and slower as the number of channels was decreased. For example, with 100 channels (Fig. 3C and 3F), the propagated repolarization velocity was slower than with 1000 channels (Fig. 3B and 3E) or 10,000 channels (3A and 3D). With only 10 channels, the repolarization did not persist in either cardiac muscle or smooth muscle (not illustrated).
Figure 3 Insertion of gap junction channels. Propagation of the repolarization of cell A5 was produced when sufficient gap-junction (g.j) channels were inserted between the cardiac muscle cells and smooth muscle cells. A: Record obtained when 10,000 gj-channels were inserted (equivalent to a g.j. resistance (Rgj) of 1.0 MΩ). Note that the rising phase of the APs from all 5 cells were superimposed, indicating that they all fired simultaneously. Also note that repolarization propagated in a retrograde direction down the 5-cell chain. B: 1,000 gj-channels inserted (Rgj of 10 MΩ). Again, the rising phase of the APs of the 5 cells were nearly superimposed. C: 100 gj-channels (Rgj of 100 MΩ). With less coupling, the rising phase of the APs of the 5 cells were separated in time. The velocity of propagated repolarization (θr) was further slowed. D: Rgj = 1.0 MΩ(10,000 channels). The rising phase of the APs from all 5 cells were superimposed. Retrograde propagation of repolarization was very fast. E: Rgj = 10 MΩ(1000 channels). The rising phase of the 5 APs were still superimposed, but now the retrograde propagation velocity was slowed. F: Rgjj = 100 MΩ(100 channels). The rising phase of the 5 APs are now separated, indicating velocity of antegrade propagation (θa) was slowed. Velocity of retrograde propagation (θr) was slow.
C. Strength/Duration Curves
The intensity (strength) and duration of the rectangular hyperpolarizing current pulses (IS2) applied to cell A5 were varied over a wide range in order to generate strength / duration curves. This was done when Rgj was infinite (i.e., 0 channels) and when Rgj was 10 MΩ (1000 channels) for strong coupling. The pulse duration was initially constant at 1.0 ms (near the chronaxie value) and then lowered to 0.5 ms and to 0.25 ms. The current intensity was varied until the sharp endpoint occurred, namely the stable repolarization of all cells in the chain. These results are plotted in Figure 4 for cardiac muscle and smooth muscle. Panel A is the strength / duration (S / D) curve for when Rgj was infinite (0 channels), and Panel B is the S/D curve for when Rgj was 10 MΩ(1000 channels). Note that the IS2 intensity was about 8–10-fold greater when the cells were well-coupled, because the applied hyperpolarizing current had to spread to all 5 cells of the chain. Regardless, the chronaxie values were about the same (ca. 1.0 ms).
Figure 4 Strength/Duration Curves. Strength / duration (S/D) curves for cardiac muscle cells (filled circles) and smooth muscle cells (unfilled circles) (5-cell chains) when Rgj was ∞ (0 channels) (A) or when Rgj was 10 MΩ(1000 channels) (B). The S / D curves are rectangular hyperbolas. The time (pulse duration) it takes for a current intensity of twice the rheobasic intensity to produce the all-or-none repolarization is the chronaxie (σ) The rheobase is the asymptote of the data points extrapolated back to the ordinate, as shown. The chronaxie was about 1.0 ms, in both panels A and B. But the absolute current intensity required was about 8–10-fold greater in panel B compared to panel A. The membrane time constant (τm) is related to the chronaxie (σ) by the equation shown in panel A.
Discussion
In principle, the addition of a second black-box into the K+ leg of the basic circuit would allow the cell to repolarize in an all-or-none fashion to small repolarizing currents. When this was attempted, the program behaved erratically. So in the absence of g.j. channels, only the cell (A5) injected with repolarizing current (IS2) was able to repolarize in an all-or-none manner. The neighboring cell (A4) exhibited only a slight repolarization of <1 mV when cell A5 had repolarized completely back to the RP (-80 mV for CM and -55 mV for SM). This fact emphasized that there were no low-resistance connections between the cells under our initial conditions.
However, addition of 10,000, 1,000, or 100 g.j channels (corresponding to Rgj values of 1.0, 10, and 100 MΩ) did allow propagation of repolarization to occur. The borderline value was 10 g.j. channels (1000 MΩ Rgj), e.g., repolarization propagated part-way down the chain in SM and almost succeeded in CM. Of course, inserting the g.j. channels required that the IS2 repolarizing current applied be much greater. This is because the IS2 current had to spread down the entire chain, with the threshold current required to cause all cells to repolarize being determined by sufficient current entering distal cell A1 to repolarize it to the GTABLE sigmoidal region. Thus the proximal cells, like A5 and A4, became hyperpolarized beyond the level required for their repolarization.
The repolarizing IS2 current intensity required for the all-or-none repolarization was lower when the rectangular pulse duration was increased. This was true for both when only the injected cell A5 was repolarized (Rgj = ∞) and when all 5 cells repolarized (Rgj of 1.0, 10, and 100 MΩ). Plots of current intensity (ordinate) versus current duration (abscissa) gave the typical hyperbolic strength/ duration curve for excitable membranes. The chronaxie values were about 1.0 ms, for which a time constant τm of about 1.44 ms could be calculated. The S/D curves for the two conditions (Rgj = ∞ and Rgj = 10 MΩ) show that the current intensity required was about 8–10-fold greater when there were many gj-channels, in both CM and SM.
The calculated velocity for propagated repolarization (θr) varied with the number of gj-channels (Table 1), as expected. The more channels, the faster the velocity. For cardiac muscle, the θr was about 200 cm/s in the very well coupled case (10,000 channels) and about 50 cm/s in the less coupled case (100 channels) (Table 1). In all cases, the velocity for propagated repolarization (θr) was much lower than the velocity for antegrade propagation (θa.). In the 2-cell chain, the calculated velocities of propagated repolarization were similar to those for the 5-cell chain (Table 1).
Table 1 Calculated velocity of retrograde propagation (θr) as compared to that for antegrade propagation (θa) for cardiac muscle (CM) and smooth muscle (SM).
2-Cell Chain (Cardiac) 5-Cell Chain (cm/sec)
No. of GJ-Channels Rgj (MΩ) Threshold* (nA) θr (cm/s) CM SM
θr θa θr θa
0 ∞ # # ## 32 ## 3.7
1 10,000 # # ## 38 ## 6.8
10 1,000 95.4 7.1 ## 55 ## 13
100 100 15.8 30 50 115 73 42
1000 10 5.4 75 86 550 114 820
10,000 1 5.6 750 200 3000 400 3600
* Pulse duration was held constant at 1.0 ms.
# Second cell (A1) failed to repolarize.
# # Some cells failed to repolarize.
Velocities of antegrade propagation (θa) were taken from previously published data [17,18].
The present study provides some new and important information about the PSpice simulations. First, it verifies that propagation (orthodromic) can occur in the complete absence of gap-junction channels, as previously reported [3,4,16-18]. Second, it demonstrates for the first time that activation of Na+ (in CM) or Ca++ (in SM) channels is reversible, by bringing Vm back to the level of the sigmoidal activation curve (GTABLE). Third, it shows for the first time that, in the PSpice model, the membranes exhibit the characteristic strength/duration curves. Fourth, it shows that the PSpice program has some serious limitations.
In summary, because of technical difficulties with the PSpice program, it was necessary to insert gj-channels in order to produce propagation of repolarization. Otherwise, only the modeled cell injected (A5) with the repolarizing IS2 current was able to repolarize. Since the potential change in the neighboring cell was only about 1 mV or less, this emphasizes that there were no low-resistance connections between the simulated cells under initial conditions. Propagation in the orthodromic direction occurs by the electric field (EF) discussed in previous papers (1–4, 16–18). The repolarizing IS2 current gave S/D plots that were typical rectangular hyperbolic curves for excitable membranes, with chronaxie values of about 1.0 ms, both for CM and SM. The calculated velocity for propagated repolarization was greater when the number of gj-channels was increased. The antidromic (retrograde) propagation velocity was usually considerably slower than the orthodromic (antegrade) propagation velocity for depolarization. The present findings do not necessarily imply that, in biological tissue, gap junctions are required for propagated repolarization to occur.
Acknowledgements
The authors thank Cara Stevens for typing the manuscript.
==== Refs
Sperelakis N McConnell K Mohan RM An electric field mechanism for transmission of excitation from cell to cell in cardiac muscle and smooth muscles Research Advances in Biomedical Engineering 2001 2 Global Research Network 39 66
Sperelakis N McConnell K Electric field interactions between closely abutting excitable cells IEEE-EMB 2002 21 77 89
Sperelakis N Mann E Jr Evaluation of electric field changes in the cleft between excitable cells J Theor Biol 1997 64 71 96 836519 10.1016/0022-5193(77)90114-X
Picone J Sperelakis N Mann JE Expanded model of the electric field hypothesis for propagation in cardiac muscle Math & Computer Modeling 1991 15 17 35 10.1016/0895-7177(91)90079-M
Kucera JP Rohr S Rudy Y Localization of sodium channels in intercalated disks modulates cardiac conduction Circ Res 2002 91 1176 1182 12480819 10.1161/01.RES.0000046237.54156.0A
Pertsov AM Medvinski AB Electric coupling in cells without highly permeable cell contacts Biofizika 1976 21 698 700 1009155
Hogues H Leon LJ Roberge FA A model for study of electric field interactions between cardiac myocytes IEEE Trans Biomed Eng 1992 30 1232 1243 1487286 10.1109/10.184699
Suenson M Ephaptic impulse transmission between ventricular myocardial cells in vitro Acta Physiol Scand 1992 120 445 455 6331073
Barr RC Plonsey R Electrophysiological interaction through the interstitial space between adjacent unmyelinated parallel fibers Biophys J 1992 61 1164 1175 1600078
Spach MS Miller WT Geselowitz DB Barr R Kootsey JM Johnson EA The discontinuous nature of propagation in normal canine cardiac muscle. Evidence for recurrent discontinuity of intracellular resistance that affects the membrane currents 1981 48 39 54 7438345
Diaz PJ Rudy Y Plonsey R Intercalated discs as a cause for discontinuous propagation in cardiac muscle. A theoretical simulation Ann Biomed Eng 1983 11 177 189 6670783
Shaw RM Rudy Y Ionic mechanisms of propagation in cardiac tissue. Roles of the sodium and L-type calcium currents during reduced excitability and decreased gap junction coupling Circ Res 1997 81 727 741 9351447
Henriquez AP Vogel R Muller-Borer BJ Henriquez CS Weingart R Cascio WE Influence of dynamic gap junction resistance on impulse propagation in ventricular myocardium: a computer simulation study Biophys J 2001 81 2112 2121 11566782
Sperelakis N Cable properties and propagation of action potentials. CH 18 Cell Physiology Source Book 1995 1 Academic Press Publishers. San Diego 245 254
Cohen SA Immunocytochemical localization of rH1 sodium channel in adult rat heart atria and ventricle. Presence in terminal intercalated disks Circulation 1994 74 1071 1096
Sperelakis N Editorial An electric field mechanism for transmission of excitation between myocardial cells Circ Res 2002 91 985 987 12456483 10.1161/01.RES.0000045656.34731.6D
Sperelakis N Ramasamy L Propagation in cardiac muscle and smooth muscle based on electric field transmission at cell junctions: An analysis by PSpice IEEE-EMB 2002 21 130 143
Sperelakis N Combined electric field and gap junctions on propagation of action potentials in cardiac muscle and smooth muscle in PSpice simulation J Electrocard 2003 36 279 293 10.1016/j.jelectrocard.2003.08.001
Morley GE Vaidya D Samie FH Lo CW Taffet SM Delmar M Jalife J Characterization of conduction in the ventricles of normal and heterozygous Cx43 knockout mice using optical mapping J Cardiovasc Electrophysiol 1999 10 1361 1375 10515561
Tamaddon HS Vaidya D Simon AM Paul DL Jalife J Morley GE High resolution optical mapping of the right bundle branch in connexin40 knockout mice reveals low conduction in the specialized conduction system Circ Res 2000 87 929 936 11073890
Gutstein DE Morley GE Tamaddon H Vaidya D Schneider MD Chen J Chien KR Stuhlmann H Fishman GI Conduction slowing and sudden arrythmic death in mice with cardiac restricted inactivation of connexin43 Circ Res 2001 88 333 339 11179202
Vaidya D Tamaddon HS Lo CW Taffet SM Delmar M Morley GE Nullmutation of connexin43 causes slow propagation of ventricular activation in the late stages of mouse embryonic development Circ Res 2001 88 1196 1202 11397787
Hoffman BF Cranefield PF Electrophysiology of the Heart 1960 McGraw-Hill Publishers
Figueroa XF Paul DL Simon AM Central role of connexin40 in the propagation of electrically activated vasodilation in mouse cremasteric arterioles in vivo Circ Res 2003 92 793 800 12637364 10.1161/01.RES.0000065918.90271.9A
Emerson G Segal S Endothelial cell pathway for conduction of hyperpolarization and vasodilation along hamster feed artery Circ Res 2000 86 94 100 10625310
| 15710046 | PMC550671 | CC BY | 2021-01-04 16:39:26 | no | Theor Biol Med Model. 2005 Feb 14; 2:5 | utf-8 | Theor Biol Med Model | 2,005 | 10.1186/1742-4682-2-5 | oa_comm |
==== Front
Cardiovasc UltrasoundCardiovascular Ultrasound1476-7120BioMed Central London 1476-7120-3-31570520210.1186/1476-7120-3-3ReviewStress-echocardiography in idiopathic dilated cardiomyopathy: instructions for use Neskovic Aleksandar N [email protected] Petar [email protected] Cardiovascular Research Center, Dedinje Cardiovascular Institute, Belgrade, Serbia and Montenegro2 Belgrade University Medical School, Belgrade, Serbia and Montenegro2005 10 2 2005 3 3 3 8 1 2005 10 2 2005 Copyright © 2005 Neskovic and Otasevic; licensee BioMed Central Ltd.2005Neskovic and Otasevic; 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.
A number of studies have suggested that stress-echocardiography may be used for prognostic stratification in patients with idiopathic dilated cardiomyopathy. There is no consensus on which protocol or which measurements of left ventricular contractile reserve to use. The most frequently used protocol is low-dose dobutamine stress-echocardiography, and most commonly used measures of left ventricular systolic performance are ejection fraction, wall motion score index and cardiac power output.
Stress-echocardiography has been shown to predict improvement in cardiac function in patients with recently diagnosed dilated cardiomyopathy, as well as to predict which patients will benefit from the treatment with beta-blockers. Most importantly, stress-echocardiography can identify patients with worse prognosis in terms of cardiac death and need for transplantation. Additionally, contractile reserve is closely correlated with maximal oxygen consumption and can even be used for further stratification in patients with maximal oxygen consumption between 10 and 14 ml/kg/min.
Future studies are needed for head-to-head comparison of various protocols in an attempt to make standardization in the assessment of patients with dilated cardiomyopathy.
stress-echocardiographydilated cardiomyopathyprognosis
==== Body
Epidemiologic data from United States indicate that idiopathic dilated cardiomyopathy (DCM) is diagnosed in approximately 36/100.000 persons each year, and that it is responsible for more than 10.000 deaths per year [1]. Faced with the fact that the number of patients with DCM is constantly increasing [2], accurate assessment of patient's current status and prognosis is of the utmost importance for the implementation of optimal therapeutic algorithm as well as for the optimal utilization of resources.
Why stress-echocardiography?
There is a widespread belief that maximal oxygen consumption, assessed by cardiopulmonary testing, is one of most, if not the most important prognostic variables in DCM patients [3]. Maximal oxygen consumption is traditionally used for selection of patients for cardiac transplantation, with a values less than 12–14 ml/kg/min indicating poor prognosis and need for transplantation [4,5]. This approach is based upon assumption that maximal oxygen consumption during cardiopulmonary testing is determined exclusively by cardiovascular factors.
However, it appears that other factors have considerable influence on maximal oxygen consumption, since it has been shown that regular physical exercise can augment oxygen consumption with little or no impact on other parameters of cardiovascular function [6]. Additionally, a normal blood flow through lower limbs was found in patients who stopped cardiopulmonary testing because of fatigue, indicating skeletal muscle dysfunction rather than pump failure [7] Patient's age and sex, as well as muscle mass are also shown to have strong influence on performance during cardiopulmonary testing [8]. Furthermore, it appears that patients with various degrees of cardiovascular impairment may yield similar maximal oxygen consumption, suggesting that there is a role for other procedures in risk stratification of patients with DCM [9].
A number of studies have shown that assessment of ventricular contractile reserve by means of stress-echocardiography may refine prognosis in patients with left ventricular systolic dysfunction [10-12]. Nevertheless, stress-echocardiography is widely underused in routine work-up in patients with heart failure, probably because of the unfamiliarity with the technique in this clinical setting.
The aim of this review is to put stress-echocardiography in DCM patients in clinical context, to give practical tips how to perform it and what to measure, as well as to try to define its role in everyday clinical practice.
How to perform stress-echocardiography in DCM?
Unlike protocols for stress-echocardiography for coronary artery disease, there is no consensus about the protocol to be used in patients with left ventricular systolic dysfunction.
The majority of authors have used either low- or high-dose dobutamine echocardiography [13,14]. Low-dose is usually defined as 10 mcg/kg/min of dobutamine [13], although some authors also consider 20 mcg/kg/min as a low-dose infusion [15] High-dose is uniformly defined as 40 mcg/kg/min [10]. There is no consensus either on duration of each stage of dobutamine infusion, since some authors use 3-minute stage [16] while the others use 5-minute stage [10] Most authors withdraw beta-blockers prior to stress-echocardiography, but some do not [15] Atropine is generally not used to achieve submaximal heart rate.
Exercise testing has been used for the assessment of left ventricular contractile reserve but in conjuction with radionuclide angiography [17] and hemodynamic measurements [18]. The usual protocol is performed on supine bicycle in incremental stages of 25 W lasting three minutes each. There are no data about the value of exercise stress echocardiography in DCM patients. It can be assumed that exercise stress-echocardiography would have the same limitations as cardiopulmonary testing.
Dipyridamole stress-echocardiography has been proposed for stratification of patients with DCM. Standard high-dose dipyridamole protocol (0.84 mg over 10 minutes) has been used for the assessment of contractile reserve [19].
What to measure?
All studies on stress-echocardiography in DCM measured contractile reserve of the left ventricle. Contractile reserve is defined as the difference between values of an index of left ventricular contractility during peak stress and its baseline values. There is no consensus on what index to use.
Ejection fraction
This is the most frequently used index of left ventricular performance. However, it may not accurately reflect left ventricular contractility since it is heavily dependent on loading conditions [20] which is particularly important in patients with heart failure for the following reasons. First of all, mitral regurgitation is frequent in these patients, and can lead to overestimation of left ventricular contractility due to rise in ejection fraction caused by changes in loading conditions (higher preload, lower afterload) [21] Secondly, activation of neuroendocrine compensatory mechanisms may increase afterload, which in turn may subsequently decrease ejection fraction [22] Thirdly, left ventricular preload is dependant upon interventricular interaction which is exaggerated in cases of pulmonary hypertension [23] a frequent finding in DCM patients. Furthermore, dobutamine has variable influence on afterload, since it has been shown that it may decrease afterload by 10% in patients with mild heart failure, but may also increases afterload by 5% in patients with severe heart failure [24].
Despite all these potential drawbacks, the change in ejection fraction during stress has been shown to have crucial prognostic significance in patients with DCM. It is generally accepted that increase in ejection fraction by ≥ 5% or change from baseline ejection fraction by ≥ 20% during stress-echocardiography identifies patients with preserved left ventricular contractile reserve and better prognosis. Ejection fraction should be assessed by Simpson biplane formula.
Wall motion score index
Wall motion score index has been traditionally used in stress-echocardiography for the detection of coronary artery disease [25]. Only two reports used this index of left ventricular contractility to assess prognosis and functional recovery of DCM patients [10,15]. Wall motion score index was assessed in a standard manner, by using 16 segment model of the left ventricle according to the recommendations given by American Society of Echocardiography [26]. The major potential drawback for use of this index is semiquantitive assessment of wall motion, which is even more subjected to inter- and intraobserver variability in DCM patients due to preexisting wall motion abnormalities and substantial number of patients with left bundle branch. It has been suggested that dobutamine induced change in wall motion score index of ≥ 0.44 identifies patients who will do better during the follow-up.
Cardiac power output
This index is not sensitive to changes in afterload, and after optimization for preload accurately reflects contractile properties of the myocardium [27]. Noninvasive calculation of cardiac power output is relatively complex and requires special instrumentation [11]. The most practical formula to calculate cardiac power output in Watts was suggested by Cook and coworkers [28]:
Cardiac power output = (cardiac output × mean arterial pressure) × 2.22 × 10 -3
where cardiac output is calculated by multiplying aortic velocity-time integral by aortic valve area, and mean arterial pressure is calculated in a standard manner.
The major problem with this index is that its calculation is time consuming, requires skilled echocardiographer, and is subjected to numerous sources of error. Furthermore, very few cardiologist are familiar with this index which precludes its wider use. Suggested cut-off point between patients with respect to prognosis is dobutamine induced change in cardiac power output of ≥ 1 W.
Prognostic significance
There is no doubt that change in left ventricular contractility during stress has considerable prognostic significance and may have profound effect on therapeutic strategy. We will review available data according to the means how contractile response was elicited.
Low-dose dobutamine
Most authors prefer low-dose dobutamine stress-echocardiography, probably because it is considered safe and is not time consuming. A report by Paelinck and coworkers has suggested that low-dose dobutamine can identify patients with atrial fibrillation induced dilated cardiomyopathy who will improve following restoration of sinus rhythm [29]. These authors concluded that low-dose dobutamine may be used to identify patients with tachycardiomyopathy.
It has been shown, in a small number of patients, that changes in left ventricular wall motion score index and ejection fraction during low-dose dobutamine echocardiography are predictive of improvement of left ventricular systolic performance during medium term follow-up (Figure 1) [15]. Since the degree of beta-receptor downregulation and desenzititation is a marker of progressive deterioration of left ventricular systolic function [30], the authors hypothetized that improvement in contractility during dobutamine infusion is greater in patients with preserved beta-receptor function who will subsequently show improvement in systolic performance.
Figure 1 Change in ejection fraction during follow-up in patients wih preserved (group A) and diminished (group B) contractile reserve. Abreviations: LVEF, left ventricular ejction fraction. From: Kitaoka H, Takata T, Yabe N, Hitomi N, Furuno T, Doi YL: Low dose dobutamine stress echocardiography predicts the improvement of left ventricular systolic function in dilated cardiomyopathy. Heart 1999;81:523-27.
These findings are further extended, so it has been shown that the presence of myocardial contractile reserve identifies patients who will respond favorably to beta-blocker therapy [31]. Furthermore, Drozd and coworkers have demonstrated that the incidence of cardiac death or need for cardiac transplantation is lower in patients with preserved contractile reserve. In this paper, multivariate analysis identified left ventricular end-systolic volume of less than 150 ml after dobutamine infusion and no decrease of left ventricular end-diastolic volume after dobutamine infusion as significant predictors of combined end-point [32].
Contractile reserve has been shown to correlate well with peak oxygen consumption (Figure 2) [16]. At multivariate analysis in this report, only percentage change in end-systolic volume index was significantly associated with occurrence of cardiac death or hospitalization for worsening heart failure. The area under receiver-operating characteristic curve was similar for percentage change in end-systolic volume index and peak oxygen consumption (0.86 ± 0.04 vs. 0.80 ± 0.06). Additionally, a report by Paraskevidis and coworkers suggested that low-dose dobutamine may further refine prognosis in patients with maximal oxygen consumption between 10 and 14 ml/kg/min [13]. This finding may be used for prioritization of patients for cardiac transplantation.
Figure 2 Linear correlation with 95% confidence interval between peak oxygen consumption and percent change in end-systolic volume index. Abreviations: % ESVI, percent change in end-systolic volume index; peak VO2, peak oxygen consumption. From: Scrutinio D, Napoli V, Passantino A, Ricci A, Lagioia R, Rizzon P: Low-dose dobutamine responsivness in idiopathic dilated cardiomiopathy: relation to exercise capacity and clinical outcome. Eur Heart J 2000;21:927-34.
High-dose dobutamine
Use of high-dose dobutamine is not associated with serious complications in DCM patients, and has an overall feasibility of 88.7%. The most common adverse event requiring discontinuation of dobutamine infusion is occurrence of complex ventricular arrhythmias, which was noted in 8% of patients (frequent multifocal ventricular extrasystoles in 6.4% and nonsustained ventricular tachycardia in 1.6%) [10]. Although there are no data on the association of complex ventricular arrhythmias and serum potassium concentrations, it may be postulated that complex arrhythmias are more frequent in potassium depleted patients. Therefore, it appears prudent to check serum potassium level prior to high-dose dobutamine stress-echocardiography. Hypotension, defined as decrease in systolic blood pressure by more than 30 mmHg, is very rare in the absence of complex ventricular arrhythmias and, in authors experience, occurs in less than 1% of patients with angiographically documented idiopathic DCM. Potential advantage of high-dose, as compared to low-dose, dobutamine echocardiography is that it may evoke more complete contractile response.
Very intriguing finding is that early in the course of DCM, dobutamine induced change in left ventricular contractile response and geometry is able to predict late spontaneous recovery of left ventricular systolic performance [14]. It is interesting that this study confirmed previous findings that increased left ventricular mass is associated with better outcome in DCM [33], and suggested that the presence of left ventricular hypertrophy implies the presence of myocardial contractile reserve.
The largest study that studied prognostic significance of high-dose dobutamine included 186 DCM patients. The major findings of this study are that dobutamine induced change in wall motion score index is able to identify patients at greater risk for cardiac death during the follow-up (Figure 3), and that change in wall motion score index carries superior prognostic information than change in ejection fraction [10] Additionally, it has been reported that dobutamine induced change in ejection fraction by ≥ 8%, assessed by radionuclide ventriculography, is prognostically superior to maximal oxygen consumption in patients with severe DCM [34].
Figure 3 Kaplan-Meir survival curves (only cardiac deaths were considered) in patients stratified according to the dobutamine induced change in wall motion score index. Abbreviations: ΔWMSi, change in wall motion score index. From: Pratali L, Picano E, Otašević P, Vigna C, Palinkas A, Cortigiani L, Dodi C, Bojić D, Varga A, Csanady M, Landi : Prognostic significance of the dobutamine echocardiography test in idiopathic dilated cardiomyopathy. Am J Cardiol. 200;88:1374-8.
Recent data by our group demonstrate that contractile reserve indices assessed by high-dose dobutamine correlate with myocardial histomorphometric features, suggesting that contractile reserve is strongly related to the degree of hystological disruption in DCM patients. Myocyte diameter and interstitial fibrosis showed strongest correlation with change in wall motion score index (r = -0.667, p < 0.001, and r = -0.567, p = 0.004, respectively), followed by change in ejection fraction (r = -0.603, p = 0.002, and r = -0.467, p = 0.021, respectively) [35].
Dipyridamole
It appears that dipyridamole may be used instead of dobutamine to evoke contractile response, since it is less arrythmogenic [36] better tolerated and yields similar prognostic information in patients with coronary artery disease [37]. Ability of dipyridamole to recruit contractile reserve is mediated through increase in coronary blood flow and accumulation of endogenous adenosine [38,39]. Potential advantage of dipyridamole over dobutamine stress-echocardiography is that the former is not affected by the use beta-blocking agents which are frequently used in DCM patients. Only one study recently examined ability of dipyridamole to predict prognosis in DCM patients. The authors concluded that increase in wall motion score index ≥ 0.15 during dipyridamole stress identifies patients who are more likely to survive during the mean follow-up of more than three years [19]. Reported overall feasibility of dipyridamole stress-echocardiography in this study was 99.2%, which is significantly higher than previously reported feasibility of dobutamine stress-echocardiography.
Exercise
As previously said, there are no echocardiographic studies on exercise induced contractile response in DCM patients. However, Nagaoka and colleagues used radionuclide ventriculography to measure increase in ejection fraction during exercise in DCM patients with mild symptoms, and concluded that change in ejection fraction <4% identifies patients with worse prognosis [17]. Additionally, it has been suggested that variables, such as cardiac power output, obtained by direct hemodynamic measurements during exercise may have important prognostic implications in patients with systolic dysfunction [12].
What about the right ventricle?
Right ventricular contribution to global cardiac performance is minor in subjects with normal or mildly depressed left ventricular systolic function, but may become more important in patients with advanced left heart failure [40]. Previous studies have suggested that right ventricular enlargement is a strong marker for adverse prognosis in DCM patients [41], as well as that right ventricular long axis excursion is predictive of exercise tolerance [42].
However, there are only limited reports of the prognostic value of right ventricular contractile reserve in patients with DCM. DiSalvo and colleagues have demonstrated that an increase in RVEF to >35% during exercise is the only independent predictor of event-free survival in patients with advanced heart failure [43]. It has been also shown that preserved right ventricular contractile reserve (measured by pressure-area relations) induced by low-dose dobutamine infusion was associated with a good 30-day outcome in patients with NYHA class IV heart failure [44].
Data from our laboratory support prognostic significance of high-dose dobutamine induced change in right ventricular fractional area change [45]. It appears that fractional area change of >9% identifies patients with more favorable outcome. More importantly, these data suggest that patients in whom contractile reserve of both ventricles is preserved will most likely have good prognosis.
What role for stress-echocardiography?
Despite the wealth evidence that favor use of stress-echocardiography in patients with DCM, there is no clear-cut algorithm about its use in risk stratification and therapeutic strategy. The reasons for this are not clear, but probably reflect the lack of standardized protocol and measurements of left ventricular contractile reserve.
We strongly believe that stress-echocardiography should be used as a standard procedure, at least in centers which do not have access to cardiopulmonary testing, since data obtained when patients are subjected to some form of stress have far greater prognostic significance than data obtained at rest. Furthermore, stress-echocardiography should be used in patients who are not able to exercise or fail to achieve expected work load. Stress-echocardiography may also play an important role for detailed risk stratification in patients with maximal oxygen consumption of 10–14 ml/kg/min. The choice of stress-protocol, at least for the time being, should be based upon local expertise and preferences of attending physician.
Future directions
There is an obvious lack of studies that will contribute to standardization of stress-echocardiographic protocol. Head-to-head comparison of stressors, including low- and high-dose dobutamine, dipyridamole, and exercise, has to be performed in order to rank their ability to predict prognosis. Similar comparisons have to be made for various indices of left ventricular contractility. Additionally, it is not clear should the patients be tested with or without beta-blocker therapy, and how this therapy may affect our choice of stressor. Last but not the least, novel echocardiographic techniques that can easily assess regional and global contractility, like tissue Doppler imaging and strain-rate imaging, have not yet been tested in a prospective manner.
In conclusion, stress-echocardiography can be a valuable tool for the assessment of patients with DCM, but a lot of work has to be done before it becomes a part of a routine work-up.
==== Refs
Gillum RF Idiopathic cardiomyopathy in the United States, 1970–1982 Am Heart J 1986 111 752 5 3513505 10.1016/0002-8703(86)90111-0
Braunwald E Shattuck lecture – Cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities N Engl J Med 1997 337 1360 9 9358131 10.1056/NEJM199711063371906
Parameshwar J Keegan J Sparrow J Sutton GC Poole-Wilson PA Predictors of prognosis in severe chronic heart failure Am Heart J 1992 123 421 6 1736580 10.1016/0002-8703(92)90656-G
Mancini D Eisen H Kussmaul W Value of peak oxygen consumption for optimal timing of cardiac transplantation in ambulatory patients with heart failure Circulation 1991 83 778 86 1999029
Levine AB Levine TB Patient evaluation for cardiac transplantation Prog Cardiovasc Dis 1991 33 219 28 1994456 10.1016/0033-0620(91)90027-J
Coats AJS Adamopoulos S Radaelli A McCance A Meyer TE Bernardi L Solda PL Davey P Ormerod O Forfar C Controlled trial of physical training in chronic heart failure Circulation 1992 85 2119 31 1591831
Wilson JR Mancini DM Dunkman B Exertional fatigue due to skeletal muscle dysfunction in patients with heart failure Circulation 1993 87 470 5 8425294
Fleg JL Lakatta EG Role of muscle loss in the age-associated reduction in VO2 max J Appl Physiol 1988 65 1147 51 3182484
Wilson JR Rayos G Yeoh T Gothard P Dissociation between peak exercise consumption and hemodynamic dysfunction in potential heart transplant candidates J Am Coll Cardiol 1995 26 429 35 7608446 10.1016/0735-1097(95)80018-C
Pratali L Picano E Otašević P Vigna C Palinkas A Cortigiani L Dodi C Bojić D Varga A Csanady M Landi P Prognostic significance of the dobutamine echocardiography test in idiopathic dilated cardiomyopathy Am J Cardiol 2001 88 1374 8 2001 Dec 15 11741555 10.1016/S0002-9149(01)02116-6
Mannor A Shneeweiss A Prognostic value of noninvasively obtained left ventricular contractile reserve in patients with severe heart failure J Am Coll Cardiol 1997 29 422 28 9014999 10.1016/S0735-1097(96)00493-7
Griffin BP Shah PK Ferguson J Rubin SA Incremental prognostic value of exercise hemodynamic variables in chronic congestive heart failure secondary to coronary artery disease or to dilated cardiomyopathy Am J Cardiol 1991 67 848 53 1901438 10.1016/0002-9149(91)90618-U
Paraskevidis IA Adamopoulos S Kremastinos Th Dobutamine echocardiographic study in patients with nonischemic dilated catdiomyopathy and prognosticall borderline values of peak exercise oxygen consumption: 18-month follow-up study J Am Coll Cardiol 2001 37 1685 91 11345385 10.1016/S0735-1097(01)01194-9
Naqvi TS Goel RK Forrester JS Siegel RJ Myocardial contractile reserve on dobutamine echocardiography predicts late spontaneous improvement in cardiac function in patients with recent onset idiopathic dilated cardiomyopathy J Am Coll Cardiol 1999 34 1537 44 10551704 10.1016/S0735-1097(99)00371-X
Kitaoka H Takata T Yabe N Hitomi N Furuno T Doi YL Low dose dobutamine stress echocardiography predicts the improvement of left ventricular systolic function in dilated cardiomyopathy Heart 1999 81 523 27 10212172
Scrutinio D Napoli V Passantino A Ricci A Lagioia R Rizzon P Low-dose dobutamine responsivness in idiopathic dilated cardiomiopathy: relation to exercise capacity and clinical outcome Eur Heart J 2000 21 927 34 10806017 10.1053/euhj.1999.1937
Nagaoka H Isobe N Kubota S Iizuka Imai S Suzuki T Nagai R Myocardial contractile reserve as prognostic determinant in patients with idiopathic dilated cardiomyopathy without overt heart failure Chest 1997 111 344 50 9041980
Griffin BP Shah PK Ferguson J Rubin SA Incremental prognostic value of exercise hemodynamic variables in chronic congestive heart failure secondary to coronary artery disease or to dilated cardiomyopathy Am J Cardiol 1991 67 848 53 1901438 10.1016/0002-9149(91)90618-U
Pratali L Prognostic value of contractile reserve during dipyridamole stress-echocardiography in idiopathic dilated cardiomyoptahy Eur J Heart Fail 2005
Rankin LS Moos S Grossman W Alterations in preload and ejection phase indices of left ventricular performance Circulation 1975 51 910 9 1122594
Grossman W Baim DS, Grossman W Evaluation of systolic and diastolic function of the myocardium Cardiac catheterization, angiography and intervention 1996 Williams and Wilkins, Baltimore 333 58
Viquerat CE Daly P Swedberg K Endogenous cateholamine levels in chronic heart failure: relation to the severity of hemodynamic abnormalities Am J Med 1985 78 455 60 3976704 10.1016/0002-9343(85)90338-9
Carrol JD Lang RM Neumann A Borow KM Rajfer SI The differential effects of positive inotropic and vasodilator therapy in patients with congestive cardiomyopathy Circulation 1986 74 815 22 3757193
Borow KM Lang RM Neumann A Carrol JD Rajfer SI Physiologic mechanisms governing hemodynamic responses to positive inotropic therapy in patients with dilated cardiomyopathy Circulation 1988 77 625 37 3342493
Beleslin BD Ostojić M Stepanović J Djordjevic-Dikic A Stojkovic S Nedeljkovic M Stankovic G Petrasinovic Z Gojkovic L Vasiljevic-Pokrajcic Z Stress echocardiography in the detection of myocardial ischemia: head-to-head comparison of exercise, dobutamine, and dipyridamole tests Circulation 1994 90 1168 76 7916274
Schiller NB Shah PM Crawford M DeMaria A Devereaux R Feingebaum H Gutgesell H Reichek N Sahn D Schnittger I Recommendations for quantitation of the left ventricle by two-dimensional echocardiography J Am Soc Echocardiography 1989 2 358 67
Kass DA Beyar R Evaluation of contractile state by maximal ventricular power divided by the square end-diastolic volume Circulation 1991 84 1698 708 1914109
Cooke GA Marshall P Al-Timman JK Physiological cardiac reserve: development of a non-invasive method and first estimates in man Heart 1998 79 289 94 9602665
Paelinck B Vermeersch P Stockman D Convens C Vaerenbeg M Usefulness of low-dose dobutamine stress echocardiography in predicting recovery of poor left ventricular function in atrial fibrillation dilated cardiomyopathy Am J Cardiol 1999 83 1668 70 10392875 10.1016/S0002-9149(99)00177-0
Fowler MB Laser JA Hopkins GL Minobe W Bristow MR Assessment of beta-adrenergic receptor pathway in the intact failing human heart: progressive receptor down-regulation and subsensitivity to agonist response Circulation 1986 74 1290 1302 3022962
Jourdain P Funck F Fulla Y Hagege A Bellorini M Guillard N Loiret J Thebault B Desnos M Myocardial contractile reserve under low doses of dobutamine and improvement of keft ventricular ejection fraction with treatment by carvedilol Eur J Heart Fail 2002 4 269 76 12034151 10.1016/S1388-9842(01)00239-2
Drozd J Krzeminska-Pakula M Plewka M Ciesielczyk M Kasprzak JD Prognostic value of low-dose dobutamine echocardiography in patients with dilated cardiomyopathy Chest 2002 121 1216 22 11948056 10.1378/chest.121.4.1216
Ida K Sersu MF Fujieda K Pathologig significance of left ventricular hypertrophy in dilated cardiomyopathy Clin Cardiol 1996 19 704 8 8874989
Ramahi TM Longo MD Cadariu AR Rohlfs K Slade M Carolan S Vallejo E Wackers FJTh Dobutamine-iduced augmentation of left ventricular ejection fraction predicts survival of heart failure patients with severe non-ischemic cardiomyopathy Eur Heart J 2001 22 849 56 11350094 10.1053/euhj.2001.2654
Otašević P Popović ZB Vasiljević JD Vidaković R Pratali L Vlahović A Nešković AN Relation of myocardial histomorphometric features and left ventricular contractile reserve assessed by high-dose dobutamine stress echocardiography in patients with idiopathic dilated cardiomyopathy Eur J Heart Fail 2005 7 49 56 15642531 10.1016/j.ejheart.2004.01.017
Picano E Marzullo P Gigli G Reisenhofer B Parodi O Distante A L'Abbate A Identification of viable myocardium by dipyridamole-induced improvement in regional left ventricular function assessed by echocardiography in myocardial infarction and comparison with thallium scintigraphy at rest Am J Cardiol 1992 70 703 10 1519518 10.1016/0002-9149(92)90545-A
Pingitore A Picano E Varga A Gigli G Cortigiani L Previtali M Minardi G Colosso MQ Lowenstein J Mathias W JrLandi P Prognostic value of pharmacological stress echocardiography in patients with known or suspected coronary artery disease: a prospective, large-scale, multicenter, head-to-head comparison between dipyridamole and dobutamine test. Echo-Persantine International Cooperative (EPIC) and Echo-Dobutamine International Cooperative (EDIC) Study Groups J Am Coll Cardiol 1999 34 1769 77 10577568 10.1016/S0735-1097(99)00423-4
Biaggioni I Olafsson B Robertson RM Hollister AS Robertson D Cardiovascular and respiratory effects of adenosine in conscious man. Evidence for chemoreceptor activation Circ Res 1987 61 779 86 3677336
Lucarini AR Picano E Marini C Favilla S Salvetti A Distante A Activation of sympathetic tone during dipyridamole test Chest 1992 102 444 7 1643930
Bernard D Alpert JS Right ventricular function in health and disease Curr Probl Cardiol 1987 13 423 49
Lewis JF Webber JD Sutton LL Chesoni C Currz CL Discordance in degree of right and left ventricular dilatation in patients with dilated cardiomyopathy: recognition and clinical implications J Am Coll Cardiol 1993 21 649 54 8436746
Webb-Peploe KM Henein MY Coats AJ Gibson DG Echo derived variables predicting exercise tolerance in patients with dilated and poorly functioning left ventricle Heart 1998 80 565 69 10065024
DiSalvo TG Mathier M Semigran MJ Dec GW Preserved right ventricular ejection fraction predicts exercise capacity and survival in advanced heart failure J Am Coll Cardiol 1995 25 1143 53 7897128 10.1016/0735-1097(94)00511-N
Gorcsan J Murali S Counihan PJ Mandarino WA Kormos RL Right ventricular performance and contractile reserve in patients with severe heart failure: assessment by pressure-area relations and association with outcome Circulation 1996 94 3190 7 8989128
Otasevic P Popovic Z Pratali L Vlahovic A Vasiljevic JD Neskovic AN Right versus left ventricular contractile reserve in one-year prognosis of patients with idiopathic dilated cardiomiopathy: assessment by dobutamine stress-echocardiography Eur J Echocar 2005
| 15705202 | PMC550672 | CC BY | 2021-01-04 16:38:31 | no | Cardiovasc Ultrasound. 2005 Feb 10; 3:3 | utf-8 | Cardiovasc Ultrasound | 2,005 | 10.1186/1476-7120-3-3 | oa_comm |
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-41570307510.1186/1742-4682-2-4ResearchSingle cell studies of the cell cycle and some models Mitchison JM [email protected] Institute for Cell, Animal and Population Biology, University of Edinburgh, Edinburgh EH9 3JT, UK2005 9 2 2005 2 4 4 14 1 2005 9 2 2005 Copyright © 2005 Mitchison; licensee BioMed Central Ltd.2005Mitchison; 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.
Analysis of growth and division often involves measurements made on cell populations, which tend to average data. The value of single cell analysis needs to be appreciated, and models based on findings from single cells should be taken into greater consideration in our understanding of the way in which cell size and division are co-ordinated. Examples are given of some single cell analyses in mammalian cells, yeast and other microorganisms. There is also a short discussion on how far the results are in accord with simple models.
==== Body
Introduction
What is the point of single cell studies of the cell cycle? The simple answer is that they provide extra information that is not available from studies of cell populations. Without them a cell biologist can be misled.
It is easiest for me to start with the theme of the extensive results on single cells of the fission yeast Schizosaccharomyces pombe with which I have worked since the mid-1950s. It was then a fairly obscure organism for physiological studies though it had a good genetic background found by U. Leupold in Bern [1]. Since then it has flourished and quite large international meetings are now devoted entirely to it. For those unfamiliar with it, it is like a scaled-up bacterial rod with division at a medial septum, unlike budding yeasts.
One the early results on its growth came from a single cell study by Bayne-Jones and Adolph [2]. Here I need to make a small digression about references. They will be given in this article but there are much longer accounts of nearly all the topics in my recent 100-page review [3]. When I took up fission yeast in the mid-fifties, I used a new microscopic technique, which gave by optical interferometry the total dry mass of single growing cells as well as their volume [4]. Volume increased, approximately in an exponential curve, through the first three quarters of the cycle but then stayed constant for the last quarter between mitosis and division. But total dry mass increased approximately linearly through the whole cycle. This was the first demonstration of linear growth, and I was surprised.
Early synchrony techniques by induction
This period of the fifties was when attention in this field was largely focused on the successful synchronisation of Tetrahymena and Chlorella by periodic changes in their environment. Good synchronous cultures would mean that powerful biochemical techniques, often enzyme activity assays at that time, could be applied in a cell cycle context. In the next 15 years, induction synchrony was somewhat improved but the cell cycles were always and inevitably distorted. Methods were also developed to select out a fraction of an asynchronous culture in one stage of the cycle and grow it up separately (for example," membrane elution", where cells growing on a membrane come away at division). They produce less distortion but a much lower yield than induction.
Because of what can be measured in synchronous cultures, they are the natural choice for the molecular biologist. But it is as well to remember their limitations. The distortions after induction have been mentioned, but even with selection synchrony there are problems. The main one is that they are, in practice, not all that synchronous. The selected cells come from more than a very narrow region of the cycle. Some of the variation can be reduced by a correction for asynchrony [5] but there is still cell-to-cell variation in cycle stage and this can obscure the fine detail of the cycle. Single cell measurements may help here.
Single cell analysis in yeast
Returning to single cell analyses of fission yeast, volume growth was followed in finer detail by Mitchison and Nurse [6]. One part of this analysis, on films taken previously by Fantes [7], showed that increase in volume was not a simple exponential during the growth phase in the first three quarters of the cycle but rather two linear segments with a rate change point (RCP) between them. The position of the RCP showed a large cell-to-cell variation. An important moral here is that these two linear segments vanished into an apparent exponential increase in a "well synchronised" culture made by selection. Such a culture scarcely showed the plateau in growth during the last quarter of the cycle. This distinction between single cells and synchronous cultures does of course depend on the frequency and accuracy of the data points. If the points have too much scatter, the fine detail of the single cell linear patterns is lost. There is also a second RCP at the end of the cycle.
A much more detailed analysis of populations of single cells followed on films was made by Sveiczer et al. [8] on fission yeast. A plot of extension growth against birth size has a strong negative slope. So also does a plot of cycle time against birth size. This has important implications for the definitions of "size control", discussed in that paper.
Problems of single cell analysis
Single cell studies have their problems. We have been lucky in using yeasts that are not apparently affected by growing on warm agar pads under a coverslip. They show "balanced growth", a property in which there is no change in extensive properties between successive cycles [9] and that should always be checked. Useful deductions can often be made with unbalanced growth but it will be a distortion of the normal cycle.
The cells also have to keep still or be followed, a problem discussed below. We have not found ways of sticking yeast to glass (e.g. with lectins) that permit "normal" growth. Cells may also need a continuous supply of fresh medium, probably for oxygenation. Various types of microscopic mounting chambers have been described in the last 50 years or so, e.g. [10], but few seem to have been stringently tested.
Many experimental studies on cell growth kinetics can be tedious; single cell studies are no exception. Here, however, modern automation is beginning to have very promising prospects. Anyone who has spent a day on a yeast film re-focusing the microscope every 5 min will welcome auto-focusing devices that are now available. Analysis has also become much easier with electronic imaging followed by image analysis programmes, and perhaps presentation on spreadsheets. It is now possible to have a programme that requires some hand work in the initial setting up under the microscope but will then run automatically, measuring cell length and diameter. This has been done for fifty or more single cells of fission yeast – a long way from the early days of using a ruler to measure the length of yeast cells on projected film images. Another point that should be raised here is that the new technology could profitably be applied to the growth of Escherichia coli. The limitations of synchronous cultures in hiding the fine detail of increases in volume or area could well mean that single cell studies might reveal more than an exponential increase. There might even be something like the two linear patterns that were popular models in earlier work with this bacterium [11].
What to measure
Volume and area of a rod-shaped organism are two of the parameters that can be measured in single growing cells. So is dry mass by interferometry. But there others, of which one of the most interesting is the use of the Cartesian diver, which was originally developed some fifty years ago at the Carlsberg Laboratory in Copenhagen. It requires technical skills and very tightly controlled temperature in water baths, but it is exquisitely sensitive. It can be used in at least two ways. One is as a diver balance, which measures "reduced weight" or weight in water. Providing there are not major changes in chemical composition, this is proportional to total dry mass. It was used on single cells of Amoeba proteus in an important classic paper by Prescott [12] mentioned below. It can also be used with minute divers as a respirometer. Hamburger [13] measured oxygen uptake in Acanthamoeba and CO2 production in fission yeast (Hamburger et al. [14]), in both cases over several cell cycles starting with single cells – a remarkable achievement. In both cases, the results were elegant linear patterns with an RCP at division.
Another interesting single cell method was the colorimetric enzyme assay of single yeast cells in microdrops [15]. This might have been developed with promise, but was not followed up, partly perhaps because the results differed from similar assays in synchronous cultures.
One of the advantages of single cell work with yeasts is that they stay still on an agar pad so they can be followed for a couple of cycles before overlapping spoils the image. This is not true of many mammalian cells, which move around on the substrate. One solution to this problem comes in the work on fibroblasts (mouse L cells) described in Zetterberg [16], Killander and Zetterberg [17], Zetterberg and Killander, [16]. These are part of an impressive body of work initiated using optical machinery gathered by Trigvar Caspersson, along with a great deal of skill and hard work. In one set of experiments on single cells [17], they made a measurement of the dry mass of single cells by interferometry and then placed it in the cycle by following it as it moved about until it divided. The difference in timing between the measurement and cell division gave the timing in the cycle. A second set of experiments used frequency analysis to set the cycle stages. This is a method widely used to determine G1, S and G2 in flow cytometry but is less suitable for the slow and imprecise doublings in something like dry mass. I therefore regard the single cell analyses as more reliable and they are not the same as those from the second method.
What are needed now are techniques that combine the subtlety and precision of single cell measurements with the new techniques of automation. A promising start was made by Zicha and Dunn [19], and the development is being actively pursued elsewhere.
Organisms which tend to be forgotten about these days are those lower eukaryotes that make poor material for molecular biologists because of inadequate genetic backgrounds. One important set of results are those from the early pioneer work of Prescott [12] on Amoeba proteus mentioned above. The results showed that the increase of single cell "dry mass" fell in a reverse exponential, with a rapid increase at the start of the cycle falling to zero towards the end. This, of course, is lethal for anyone who believes that a rising exponential is the paradigm for the cell cycle.
Tetrahymena pyriformis has a long and distinguished history in the cell cycle with its early induction synchrony. But in the 1960s there was a burst of studies on selected single cells or small groups. The growth patterns were often not well defined but it seems that absolute measurements of volume and of respiration rate were a better fit to linear growth (Prescott, [20]). Such analyses might now be checked using some of the semi-automated procedures referred to above.
Growth in syncytia
Physarum polycephalum is a myxomycete of considerable importance in some earlier work on cell cycle control. It is effectively a big single multinucleate cell with complete natural synchrony in nuclear division. It does not show exponential increase in macromolecular synthesis. For instance, there are two peaks in the rate of protein synthesis, one in the S period and the other in G2 (Mittermayer et al, [21]).
General conclusion
It would appear that there are no universal patterns of growth in these lower eukaryotes.
Models
My title makes mention of "some models". Let me be clear that there are two quite different types of cell cycle models. One type includes detailed mathematical and molecular models dealing with discrete periodic events like mitosis (e.g. [22]). These are complex and can illustrate the relations between many components of a network at the event, on reasonable assumptions. They are important aids in understanding the events and are a fairly recent development in the cell cycle world. There are certain limitations at present. With mitosis, the models have problems with the starting event (a size control?), with location in cellular compartments, and with the final mechanical events. However, such models will certainly develop.
However, what I am concerned with here are much earlier and much simpler models, not of periodic events in the cycle like DNA synthesis, but of continuous growth. Here the two dominant models were, for simplicity, an exponential pattern of increase and a linear one. My own view [3] of the earlier experiments is that, on the whole, they favour linear increase but it was also clear that some patterns, e.g. volume in fission yeast, are more complex. Linear increases with rate change points have certainly survived in fission yeast where there are no exponential increases (Table 1 in [3]) and this has revived for me an old hypothesis of "gene dosage". What, for instance, happens to synthesis rates between G1 and G2? But one thing is clear – that a single unifying dream of exponential synthesis is not in accord with the facts. It is really useless to wave Occam's Razor around. The end of his razor blade is "without necessity". In all reasonable judgements, the necessity is there. Beyond that is prejudice.
Figure 1 Modes of growth in cell length of wild-type and wee1 mutant cells of fhe fission yeast Schizosaccharomyces pombe, after Sweiczer et al. (1996)
Figure 2 Growth through one cycle of individual Amoeba, after Prescott (1976)
==== Refs
Munz P Wolf K Kohli J Leupold U Nasim A, Young P, Johnson BF Genetics Overview Molecular Biology of the Fission Yeast 1989 San Diego: Academic Press 1 30
Bayne-Jones S Adolph EF Growth in size of micro-organisms measured from motion pictures J Cell Comp Physiol 1932 1 387 407 10.1002/jcp.1030010306
Mitchison JM Growth during the cell cycle Int Rev Cytol 2003 222 165 258 12921238
Mitchison JM The growth of single cells. 1. Schizosaccharomyces pombe Exp Cell Res 1957 13 244 262 13480293 10.1016/0014-4827(57)90005-8
Creanor J Mitchison JM The kinetics of H1 histone activation during the cell cycle of wild type and wee mutants of the fission yeast Schizosaccharomyces pombe J Cell Sci 1994 107 1197 1204 7929629
Mitchison JM Nurse P Growth in cell length in the fission yeast Schizosaccharomyces pombe J Cell Sci 1985 75 357 376 4044680
Fantes PA Control of cell size and cycle time in Schizosaccharomyces pombe J Cell Sci 1977 24 51 67 893551
Sveiczer A Novak B Mitchison JM The size control of fission yeast revisited J Cell Sci 1996 109 2947 2957 9013342
Campbell A Synchronization of cell division Bacteriol Rev 1957 21 263 272 13488884
Harris NK Powell EO A culture chamber for the microscopical study of living bacteria, with some observations on the spore-bearing aerobes J R Microsc Soc 1951 71 407 420 14939270
Ward CB Glaser DA Correlation between rate of cell growth and rate of DNA synthesis in Escherichia coli B/r Proc Natl Acad Sci USA 1971 68 1061 1064 4930241
Prescott DM Relations between cell growth and cell division. 1. Reduced weight, cell volume, protein content and nuclear volume of Amoeba proteus from division to division Exp Cell Res 1955 9 328 337 13262045 10.1016/0014-4827(55)90106-3
Hamburger K Respiratory rate through the growth-division cycle of Acanthamoeba Sp C R Trav Lab Carlsberg 1975 40 175 185
Hamburger K Kramhaft B Nissen SB Zeuthen E Linear increases in glycolytic activity through the cell cycle of Schizosaccharomyces pombe J Cell Sci 1977 24 69 79 893552
Yashpe J Halvorson HO β-D-galactosidase activity in single yeast cells during the cell cycle of Saccharomyces lactis Science 1976 191 1283 1284 1257751
Zetterberg A Nuclear and cytoplasmic growth during interphase Almqvist Wicksells Boktrykeri AB, Uppsala (Sweden) 1966
Killander D Zetterberg A Quantitative cytochemical studies on interphase growth. I. Determination of DNA, RNA and mass content of age determined mouse fibroblasts in vitro and of intercellular variation in generation time Exp Cell Res 1965 38 272 284 14284508 10.1016/0014-4827(65)90403-9
Zetterberg A Killander D Quntitative cytochemical studies on interphase growth. II. Derivation of synthesis curves from the distribution of DNA, RNA and mass values of individual mouse fibroblasts in vitro Exp Cell Res 1965 39 22 32 5831240 10.1016/0014-4827(65)90003-0
Zicha D Dunn GA An image processing system for cell behaviour studies in subconfluent cultures J Microscop 1995 179 11 21
Prescott DM Reproduction of eukaryotic cells 1976 Academic Press, New York
Mittermayer C Chayka TG Braun R Rusch HP Polysome patterns and protein synthesis during the mitotic cycle of Physarum polycephalum Nature 1966 210 1133 1137 6007180
Chen KC Csikasz-Nagy A Gyorffy B Val J Novak B Tyson JJ Kinetic analysis of a molecular model of the budding yeast cell cycle Mol Biol Cell 2000 11 369 391 10637314
| 15703075 | PMC550673 | CC BY | 2021-01-04 16:39:26 | no | Theor Biol Med Model. 2005 Feb 9; 2:4 | utf-8 | Theor Biol Med Model | 2,005 | 10.1186/1742-4682-2-4 | oa_comm |
==== Front
Med ImmunolMedical Immunology1476-9433BioMed Central London 1476-9433-4-11571792910.1186/1476-9433-4-1ReviewFADD adaptor in cancer Tourneur Léa [email protected] Agnès [email protected] Gilles [email protected] Département d'Immunologie, Institut Cochin, INSERM U 567, CNRS UMR 8104, IFR 116, Université René Descartes, Paris V, Paris, France2 Service d'Hématologie Adultes, Hôpital Necker-Enfants Malades, Paris, France2005 17 2 2005 4 1 1 31 1 2005 17 2 2005 Copyright © 2005 Tourneur et al; licensee BioMed Central Ltd.2005Tourneur 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.
FADD (Fas Associated protein with Death Domain) is a key adaptor molecule transmitting the death signal mediated by death receptors. In addition, this multiple functional protein is implicated in survival/proliferation and cell cycle progression. FADD functions are regulated via cellular sublocalization, protein phosphorylation, and inhibitory molecules. In the present review, we focus on the role of the FADD adaptor in cancer. Increasing evidence shows that defects in FADD protein expression are associated with tumor progression both in mice and humans. Better knowledge of the mechanisms leading to regulation of FADD functions will improve understanding of tumor growth and the immune escape mechanisms, and could open a new field for therapeutic interventions.
==== Body
The FADD molecule
The FADD gene is located on chromosome 11q13.3 in humans and 7 in mice [1]. Mutations in the FADD gene containing locus are frequently observed in human malignancies [2]. For instance, the 11q13 region contains the fibroblast growth factor 3 and 4 genes which are coamplified in melanoma. It also includes the multiple endocrine neoplasia I gene whose mutation leads to tumor development of several endocrine glands including thyroid. Moreover, two genes implicated in leukemia are found in this locus: NUMA1 which is translocated in acute promyelocytic leukemia, and BCL1 which is located very close to the FADD gene and is mutated in B-cell leukemia/lymphoma. Although FADD has a central role in multiple receptor-induced cell death as discussed hereafter, no mutation of the FADD gene itself has been reported so far.
Human and mouse FADD genes have the same quite simple organization consisting of two exons (286 bp and 341 bp in humans; 332 bp and 286 bp in mice) separated by a unique intron of approximately 2 kb. Interestingly, no cap site was reported on the human FADD mRNA [1], suggesting a particular regulation of FADD mRNA translation, although this topic has not been further investigated.
Human and mouse FADD proteins are very similar (Figure 1). They consist of 208 and 205 amino acids (AA) respectively, and share 80% similarity and 68% identity [3]. FADD mRNA and protein are almost ubiquitously expressed in fetal and adult tissues, both in humans and mice [4]. Two domains are particularly well conserved between species: the death domain (DD) at the COOH-terminus of the protein, and the death effector domain (DED) at the NH2-terminus of the protein [5,6]. Both domains play a crucial role in transducing the apoptotic signal mediated by death receptors. Furthermore, a single serine (Ser) phosphorylation site essential for determining cell cycle progression is conserved in both species (human Ser 194 [7] and mouse Ser 191 [8]).
Figure 1 Human and mouse FADD protein. Amino acids (AA) corresponding to the human FADD protein are marked in black, whereas AA corresponding to the mouse FADD protein are marked in grey. The death domain (DD) and death effector domain (DED) are essential for interaction with death receptors and transmission of the apoptotic signal. Human nuclear export sequence (NES in red) and nuclear localization sequence (NLS in blue) determine localization of the protein in the cytoplasm and the nucleus, which are associated with cell death and survival functions of the FADD protein, respectively. Human Ser 194 and mouse Ser 191 phosphorylation site (in purple) have a crucial role in survival/proliferation and cell cycle progression.
Since the first role ascribed to FADD was to transmit apoptotic signals through its interaction with death receptors expressed at the cell membrane, it was assumed that FADD protein was exclusively localized in the cytoplasm of the cell. However, a nuclear localization sequence (NLS) and a nuclear export sequence (NES) were recently identified in the human FADD protein (Figure 1), and account for FADD protein expression in the nucleus and the cytoplasm of the cell, respectively [9]. The vast majority of the reports on FADD focused on the cytoplasmic FADD protein because of its pro-apoptotic function. In contrast, the role of the nuclear FADD is much more mysterious. It was recently reported that FADD expression in the nucleus protects cells from apoptosis, but the mechanism implicated in this survival function has not been investigated [9]. On the other hand, it has been shown that FADD could interact within the nucleus of adherent cells with the methyl-CpG binding domain protein 4 (MBD4) [10]. MBD4 is a GT mismatch repairing protein. Association between MBD4 and FADD within the nucleus could couple MBD4-mediated genome surveillance with FADD-mediated cell death. Thus, nuclear FADD could perhaps have a pro-apoptotic function, at least in response to DNA-damaging agents.
Functions of the FADD protein
An essential molecule for embryonic development
The essential role of the FADD molecule was highlighted by generating FADD mutant null mice [11,12]. Indeed, FADD knockout mice were not viable. FADD null embryos died in utero at day 12.5 of development, due to underdevelopment, abdominal hemorrhage, and cardiac failure. Moreover, FADD loss of function did not result in a lymphoproliferative disorder as observed in viable Fas mutant mice [13,14]. These results indicated that in addition to its well known role in cell death, FADD was also implicated in survival/proliferation of some cell types.
A main death transducer for DD-containing receptors
FADD is the main signal transducing intermediate adaptor molecule of several death receptors including Fas, TNF-R1 (tumor necrosis factor receptor 1), DR3 (death receptor 3), TRAIL-R1 (TNF-related apoptosis-inducing ligand, DR4), and TRAIL-R2 (DR5) [4,11,12,15]. All these receptors possess, in their intra-cytoplasmic tail, a DD homologous to the DD of FADD allowing FADD recruitment to the activated receptor. FADD can be recruited either directly to Fas and TRAIL-Rs (Figure 2A) or indirectly to TNF-R1 (Figure 2B). In the latter case, FADD is recruited via another DD-containing adaptor molecule (TRADD, TNF receptor-associated protein with DD). Next, FADD recruits DED-containing initiator pro-caspase 8 or 10 through DED/DED interactions [16-18], thus forming the death-inducing signaling complex (DISC) [19]. Autoprocessing of initiator pro-caspase leads to activation of effective caspases which cleave intracellular substrates, causing the apoptotic death of the cell [20,21].
Figure 2 Apoptosis mediated by death receptors requires the FADD adaptor molecule. (A) Engagement of a Fas ligand trimer on a trimer of Fas leads to FADD adaptor molecule recruitment through homotypic DD interactions. FADD next binds initiator pro-caspase through DED interactions. This Fas/FADD/pro-caspase complex forms the Fas death-inducing signaling complex (DISC) since initiator pro-caspase activates a caspase cascade resulting in apoptotic death of the cell. Alternatively, c-FLIPs can promote cell survival by interacting with FADD through their respective DED, thus hindering recruitment and activation of initiator pro-caspase. (B) Signaling mediated by TNF-R1 implicates formation of two sequential complexes [61]. The complex I (in blue) contains the TNF-R1, the adaptor TRADD, the receptor interacting kinase (RIP), and the TNF-receptor associated factor 2 (TRAF2). It assemblies rapidly following TNF-α stimulation and activates the NF-kB pathway which in turn induces expression of survival genes, including c-FLIP. Later on, complex I dissociates from the TNF-R1 and is internalized. FADD can then bind the liberated DD of TRADD and recruits initiator pro-caspase, forming complex II (in red) which is cytoplasmic. Activation of initiator pro-caspase 8/10 in complex II results in apoptosis of the cell. Red box: DD; hatched red box: DED.
Control of FADD recruitment to the DISC can occur following several mechanisms depending on the cell type and the death receptor [22]. The best characterized death receptor signaling inhibitors are DED-containing viral and cellular FLICE-inhibitory proteins (v-FLIPs and c-FLIPs, respectively) [23,24]. Inhibition of Fas-, TNF-R1-, and TRAIL-Rs-induced apoptosis by endogenous FLIPs results from binding of the c-FLIPs to the DED of FADD, thus hindering pro-caspase 8 activation (Figure 2A). Similarly, v-FLIPs inhibit apoptosis mediated by death receptors either by binding to FADD and blocking pro-caspase 8 processing, or by binding to pro-caspase 8 and inhibiting FADD interaction. Therefore, equilibrium between FADD and the expression of its inhibitors determines the outcome of the death receptor-stimulated cell, i.e. apoptosis or survival.
All the main death receptors described up to now require FADD adaptor for transmitting their apoptotic signal. Consequently, FADD is a central protein that controls multiple essential cellular processes including cellular homeostasis and elimination of pathological cells, particularly during the course of an immune response.
Death receptor independent FADD induced apoptosis
Formation of cytoplasmic death effector filaments (DEF) by oligomerization of DED-containing proteins, including FADD, is responsible for death receptor independent cellular apoptosis [25]. Indeed, FADD over-expression by itself is known to induce cell death through DEF formation that recruits and activates pro-caspase 8. However, the existence of DEF has not been established in vivo, and increasing evidences suggest that DEF could be artefactual structures resulting from protein over-expression. As a consequence, the ability of endogenous FADD to aggregate and form DEF in normal situation should be reconsidered.
Functions in proliferation and cell cycle progression
Beside being a main death adaptor molecule, FADD is also required for T cell proliferation. The first evidence of this property of FADD came from observations made in chimeric FADD knockout mice. Five-week-old chimeric FADD-/- mice presented a lack of thymocytes compared to wild type animals, with few or no CD4+ CD8+ double positive thymocytes remaining [12]. Moreover, several groups have demonstrated that FADD deficiency in peripheral T lymphocytes resulted in an inhibition of mitogen-induced T cell proliferation [12,26-30]. The mechanism leading to FADD-dependent T cell proliferation did not involve the early events associated with cell proliferation since expression level and functionality of the IL-2 receptor, level of IL-2 secretion, mobilization of intracellular calcium, and activation of NF-kB, p38-MAPK, and p44/42-MAPK appeared normal in FADD-/- T lymphocytes [12,29,30]. Recent data showed that FADD-/- T lymphocytes entered the cell cycle upon mitogenic stimulation, but died during progression through the cell cycle [30]. Therefore, lack of proliferation of FADD-deficient T cells results from defective survival associated with progression through the cell cycle rather than defective activation. Up to now, the molecular pathway implicated in FADD-mediated survival of lymphocytes has not been described.
In addition to impairing survival during cell division, FADD deficiency also leads to a dysregulation of the cell cycle machinery [31]. The pattern of expression of molecules implicated in both G1/S and G2/M transitions was aberrant in FADD-/- lymphocytes, resulting in spontaneous entry and progression through the cell cycle of 10% of freshly isolated FADD-/- T cells (as compared to less than 2% of wild type T cells) [31].
The mechanisms responsible for FADD regulation of cell cycle progression are not fully understood. However, the phosphorylation of the 194 human and 191 mouse Ser of the protein (Figure 1) has recently drawn attention. Indeed, human FADD was phosphorylated at Ser 194, by a still unidentified 70 kDa protein kinase, in cells arrested in G2/M, whereas it was unphosphorylated in G1/S [7]. Generation of FADD Ser 191 mutant mice confirmed that FADD phosphorylation is involved in proliferation in vivo [8]. Replacement of Ser 191 by an aspartic acid resulting in constitutive phosphorylation of the FADD protein led to abnormal development of FADD mutant mice that shared the same phenotype as FADD deficient mice [8], including few CD4+ CD8+ thymocytes and defective progression through the cell cycle [12,31].
Tumor cells are constantly cycling cells. Although it does not seem to directly affect the cell cycle progression, FADD phosphorylation at Ser 194 sensitizes these cells to reagents that induce G2/M arrest such as the Taxol anticancer drug [32]. In human prostate cancer cell lines, treatment with Taxol resulted in Ser 194 FADD phosphorylation and G2/M arrest [33]. Moreover, etoposide or cisplatin chemotherapeutic drug-induced apoptosis of these cells was enhanced by pretreatment with Taxol, a process that was inhibited by cellular over-expression of an unphosphorylable FADD mutant [33]. Therefore, tumor cells that express a Ser 194 FADD mutant that cannot be phosphorylated or are unable to phosphorylate FADD at this Ser position are expected to resist apoptosis induced by anticancer drugs that induce G2/M arrest, and to be insensitive to the synergistic effect of chemotherapy. Obviously, a lack of FADD expression will have the same consequences.
Role in tumor development
FADD as a tumor suppressor
The role of the FADD adaptor in cancer was initially demonstrated by generating RAG-1 deficient transgenic mice that target expression of a FADD dominant negative mutant in lymphocytes. With age these mice developed thymic lymphoblastic lymphoma, whereas FADD+/+ RAG-1-/- mice did not [34]. Moreover, thymic lymphoblastic lymphomas were never observed in FADD-/- RAG-1+/+ or FADD-/- RAG-1+/- mice, demonstrating that absence of FADD expression was necessary but not sufficient to induce tumor development in this model [34]. These results were the first demonstration that FADD adaptor can act as a tumor suppressor in vivo.
Using a mouse model of thyroid adenoma/adenocarcinoma, we showed spontaneous disappearance of FADD protein expression during the course of tumor development [35]. The so called gsp transgenic mice expressed an oncogene specifically in thyroid follicular cells (TFC), and developed thyroid hyperplasia that eventually transformed into hyperfunctioning adenomas or adenocarcinomas around the age of 8 months [36]. The fact that gsp mice developed hyperfunctioning adenomas or adenocarcinomas belatedly suggested that oncogene expression conferred a predisposition but that an additional event was necessary for thyroid tumor development. We found that FADD protein was highly expressed in all non-pathological and in almost all hyperplastic thyroid glands from gsp mice. In contrast, thyroid adenoma/adenocarcinoma expressed low or no FADD protein [35]. These results raised the possibility that loss of FADD protein expression could be an additional event contributing to tumorigenesis, and suggested that FADD plays a role as a tumor suppressor.
We recently showed that absence of FADD protein expression in cancer cells is also a relevant phenomenon in human malignancies [37]. We looked for FADD protein expression in human acute myeloid leukemia (AML) cells. Leukemic cells of most AML patients are resistant to Fas-mediated cell death despite expressing the Fas receptor [38] and/or the FasL molecule [39]. Moreover, chemotherapeutic drugs used for AML treatment can kill target cells via several mechanisms, including death receptor-induced apoptosis [40-43]. We performed a retrospective study of 70 consecutive patients with de novo AML treated homogeneously, and found that leukemic cells of 2/3 of patients at diagnosis expressed low or no FADD protein [37]. Moreover, in this cohort of patients, we showed that absent/low FADD protein expression in AML cells at diagnosis was a new independent prognostic factor for poor response to chemotherapy (in terms of complete remission rate, event free and overall survivals) [37]. Importantly, absent/low FADD protein expression in AML cells at diagnosis was a prognostic factor even for patients classified as standard- or good-risk AML cases by cytogenetic and molecular criteria [37]. As a consequence, this new prognostic factor is of clinical importance since it will allow early identification of patients with chemoresistant AML who could benefit from more intensive post-remission therapy.
Absence of FADD confers numerous advantages on cancer cells
The fact that absence of FADD expression was found in different types of tumor cells both in mice and humans strongly suggested that absence of FADD contributed to tumor development. Indeed, lack of FADD protein can confer numerous advantages on pathological cells, which predominantly result in tumor survival/growth gain (Figure 3).
Figure 3 Lack of FADD expression confers survival/growth advantages on tumor cells. Absence of FADD protein confers multiple death receptor-mediated apoptosis resistance and allows tumor cells to co-express death receptors and ligands without committing cell death. (A) Lack of FADD contributes to immune escape and resistance to chemotherapy. FADD deficient tumor cells resist death receptor-mediated apoptosis induced by TIL and chemotherapeutic drugs. Anthracyclines increase Fas and FasL expression on tumor cells. Etoposide induces Fas receptor trimerization, leading to Fas-mediated cell death independently of FasL expression. Both drugs enhance TRAIL-R2-mediated apoptosis. (B) Lack of FADD contributes to tumor counter-attack. Lack of FADD expression allows many types of tumor cells to express innocuously functional FasL that can kill TIL. Secretion of TNF-α by AML cells can have cytotoxic effects on TIL. (C) Lack of FADD contributes to tumor growth. In the absence of FADD, Fas signaling leads to a proliferative signal instead of an apoptotic one. Concomitant death receptor and ligand expression, in the absence of FADD, allows autocrine (in red) and paracrine (in orange) proliferation of tumor cells. Activated TIL can contribute to paracrine (in purple) proliferation of FADD-deficient tumor cells.
Immune escape and resistance to chemotherapy
Fas, TRAIL-Rs, TNF-R1, DR3, and potentially other receptors use FADD adaptor for transmitting the death signal. Thus, absence of FADD expression in tumor cells must confer multiple resistance of these cells to death receptor cytotoxicity. In agreement with this assumption, we observed that FADD-/- TFC as well as FADD-/- AML cells were resistant to Fas- and TNF-α-mediated apoptosis [35], and unpublished data). Since cytotoxic tumor infiltrating lymphocytes (TIL) use, among other mechanisms, death receptor-mediated cell death to kill pathological cells, tumor cells lacking FADD molecule expression may partially avoid immune attack (Figure 3A). On the other hand, some anticancer drugs exert their cytotoxic effect by inducing death receptor and/or death ligand expression on tumor cells, thereby inducing suicidal/fratricidal apoptosis of the cells. For instance, anthracyclines and etoposide, two chemotherapeutic drugs used for AML treatment, enhance Fas- and TRAIL-R2-mediated cell death in vitro, a process requiring FADD molecule expression [40-42,44,45]. As a consequence, absence of FADD expression in AML cells of our patients may contribute to chemoresistance of leukemic cells (Figure 3A).
Tumor counter-attack
As described above, absence of FADD allows co-expression of death receptors and ligands without inducing autocrine or paracrine apoptosis of tumor cells. For instance, although the Fas receptor was expressed at all stages of thyroid tumor development in gsp mice, FasL expression was gained with a high expression in adenomatous/adenocarcinomatous glands [35]. Using the same method, we found that leukemic cells of most AML patients expressed FasL [39] and secreted TNF-α despite expressing Fas and TNF-R1 [37], and unpublished data). Moreover, co-expression of Fas and FasL molecules that did not cause cell death was also observed in lymphoma [46], melanoma [47], astrocytoma [48], cancers of the colon [49], liver [50], lung [51], human thyroid [52]. The role of death ligand expression in tumor cells is still a controversial issue [53-55], but it is now well accepted that it allows at least some cancer cells to kill TIL that express death receptors, a process called the "tumor counter-attack" (Figure 3B).
Proliferative advantage
Since the tumor counter-attack hypothesis cannot apply to all types of cancer cells, one could wonder whether other benefits of death ligand expression by tumor cells exists. Previous reports demonstrated that Fas signaling could lead to proliferation instead of apoptosis, depending on the cell type and the environmental conditions [56-58]. For instance, agonistic anti-Fas antibody-induced proliferation of hematological and non-hematological tumors has been described [59]. Moreover, we have shown that stimulation of FADD lacking thyrocytes by an agonistic anti-Fas antibody resulted in accelerated growth of TFC, via a particular Daxx adaptor-mediated pathway [35]. Therefore, Fas signaling, particularly in the absence of FADD, can confer proliferative advantage on tumor cells (Figure 3C). Thus, FasL expression on Fas+ FADD- adenomatous/adenocarcinomatous thyroid from gsp mice, as well as on human AML cells, may allow both autocrine and paracrine proliferation of these cells [35,37] (Figure 3C). Furthermore, we can formulate the same hypothesis for TNF-α secretion by TNF-R1 expressing leukemic cells (Figure 3C). If our hypotheses are correct, then activated TIL that express death ligands could contribute to FADD-deficient tumor proliferation (Figure 3C).
Restoring FADD protein expression- a new therapeutic issue
Absence of FADD expression could confer multiple growth advantages on cancer cells (Figure 3), and is expected to contribute to disease progression. As a consequence, finding how to restore FADD protein expression in FADD-negative tumor cells represents a research field with potentially direct clinical applications. In fact, it is possible that some of the current cancer treatments act, at least partially, through this mechanism. For instance, carboplatin is a cytotoxic drug potentially effective at reestablishing functional FADD protein expression. Indeed, the human tongue carcinoma cell lines SCC-9 and SCC-25 express very low levels of FADD protein. Moreover, treatment with carboplatin enhances FADD protein expression, thus rendering cancer cells sensitive to Fas-mediated apoptosis [60]. The combination of carboplatin with chemotherapeutic drugs that induce death receptor-mediated cell death may result in improved treatment.
Molecules implicated in FADD regulation of expression also represent new therapeutic targets. However, very little is known about the mechanisms leading to absent/low FADD protein expression in tumor cells. In SCC-9 and SCC-25 carcinoma cell lines, carboplatin upregulated FADD protein expression by increasing FADD mRNA [60]. However, FADD mRNA was normally expressed in mouse thyroid adenoma/adenocarcinoma and in human AML cells, and lack of FADD mRNA could not account for poor FADD protein expression in these cancer cells [35,37]. These results suggest that several mechanisms could be implicated in loss of FADD protein, depending on the type of cell and the environmental pressure. Besides, docking FADD protein away from the death receptor, in the nucleus for example, would have the same consequences. Understanding such mechanisms is the first necessary step towards development of new anticancer drugs targeting molecules that regulate FADD protein expression.
Conclusion
FADD is mainly known as a key adaptor molecule for numerous death receptors. However, increasing evidences have shown that FADD is a much more complex molecule implicated in apoptosis, survival, cell cycle progression, and proliferation of the cells. Therefore, FADD plays a central role in the frightening control of cell death and life. As a consequence, a defect in the FADD molecule can contribute to the development of diseases, and particularly cancer. Absence of FADD protein expression is a marker of tumor development in the mouse, and a prognostic factor for poor response to chemotherapy in humans. Since FADD deficiency could contribute to several malignancies, in view of its almost ubiquitous pattern of expression, study of the role of FADD in tumor development, growth, and resistance to treatment, and understanding how the expression of this puzzling molecule is regulated, are targets that merit further investigation.
Authors' Contributions
LT conceived the review and drafted the manuscript. GC participated in conceiving the review. AB and GC participated in the preparation of the manuscript. All authors read and approved the final manuscript.
Acknowledgments
LT is a recipient of a "Société Française d'Hématologie" (SFH) post-doctoral training fellowship. This work was supported by the Institut National de la Santé Et de la Recherche Médicale (INSERM).
==== Refs
Kim PK Dutra AS Chandrasekharappa SC Puck JM Genomic structure and mapping of human FADD, an intracellular mediator of lymphocyte apoptosis J Immunol 1996 157 5461 5466 8955195
Katoh M FLJ10261 gene, located within the CCND1-EMS1 locus on human chromosome 11q13, encodes the eight-transmembrane protein homologous to C12orf3, C11orf25 and FLJ34272 gene products Int J Oncol 2003 22 1375 1381 12739008
Zhang J Winoto A A mouse Fas-associated protein with homology to the human Mort1/FADD protein is essential for Fas-induced apoptosis Mol Cell Biol 1996 16 2756 2763 8649383
Chinnaiyan AM O'Rourke K Tewari M Dixit VM FADD, a novel death domain-containing protein, interacts with the death domain of Fas and initiates apoptosis Cell 1995 81 505 512 7538907 10.1016/0092-8674(95)90071-3
Weber CH Vincenz C The death domain superfamily: a tale of two interfaces? Trends Biochem Sci 2001 26 475 481 11504623 10.1016/S0968-0004(01)01905-3
Eberstadt M Huang B Chen Z Meadows RP Ng SC Zheng L Lenardo MJ Fesik SW NMR structure and mutagenesis of the FADD (Mort1) death-effector domain Nature 1998 392 941 945 9582077 10.1038/31972
Scaffidi C Volkland J Blomberg I Hoffmann I Krammer PH Peter ME Phosphorylation of FADD/ MORT1 at serine 194 and association with a 70-kDa cell cycle-regulated protein kinase J Immunol 2000 164 1236 1242 10640736
Hua ZC Sohn SJ Kang C Cado D Winoto A A function of Fas-associated death domain protein in cell cycle progression localized to a single amino acid at its C-terminal region Immunity 2003 18 513 521 12705854 10.1016/S1074-7613(03)00083-9
Gomez-Angelats M Cidlowski JA Molecular evidence for the nuclear localization of FADD Cell Death Differ 2003 10 791 797 12815462 10.1038/sj.cdd.4401237
Screaton RA Kiessling S Sansom OJ Millar CB Maddison K Bird A Clarke AR Frisch SM Fas-associated death domain protein interacts with methyl-CpG binding domain protein 4: a potential link between genome surveillance and apoptosis Proc Natl Acad Sci U S A 2003 100 5211 5216 12702765 10.1073/pnas.0431215100
Yeh WC Pompa JL McCurrach ME Shu HB Elia AJ Shahinian A Ng M Wakeham A Khoo W Mitchell K El-Deiry WS Lowe SW Goeddel DV Mak TW FADD: essential for embryo development and signaling from some, but not all, inducers of apoptosis Science 1998 279 1954 1958 9506948 10.1126/science.279.5358.1954
Zhang J Cado D Chen A Kabra NH Winoto A Fas-mediated apoptosis and activation-induced T-cell proliferation are defective in mice lacking FADD/Mort1 Nature 1998 392 296 300 9521326 10.1038/32681
Watanabe-Fukunaga R Brannan CI Copeland NG Jenkins NA Nagata S Lymphoproliferation disorder in mice explained by defects in Fas antigen that mediates apoptosis Nature 1992 356 314 317 1372394 10.1038/356314a0
Adachi M Suematsu S Kondo T Ogasawara J Tanaka T Yoshida N Nagata S Targeted mutation in the Fas gene causes hyperplasia in peripheral lymphoid organs and liver Nat Genet 1995 11 294 300 7581453 10.1038/ng1195-294
Kuang AA Diehl GE Zhang J Winoto A FADD is required for DR4- and DR5-mediated apoptosis: lack of trail-induced apoptosis in FADD-deficient mouse embryonic fibroblasts J Biol Chem 2000 275 25065 25068 10862756 10.1074/jbc.C000284200
Nagata S Apoptosis by death factor Cell 1997 88 355 365 9039262 10.1016/S0092-8674(00)81874-7
Kischkel FC Lawrence DA Tinel A LeBlanc H Virmani A Schow P Gazdar A Blenis J Arnott D Ashkenazi A Death receptor recruitment of endogenous caspase-10 and apoptosis initiation in the absence of caspase-8 J Biol Chem 2001 276 46639 46646 11583996 10.1074/jbc.M105102200
Wang J Chun HJ Wong W Spencer DM Lenardo MJ Caspase-10 is an initiator caspase in death receptor signaling Proc Natl Acad Sci U S A 2001 98 13884 13888 11717445 10.1073/pnas.241358198
Kischkel FC Hellbardt S Behrmann I Germer M Pawlita M Krammer PH Peter ME Cytotoxicity-dependent APO-1 (Fas/CD95)-associated proteins form a death-inducing signaling complex (DISC) with the receptor Embo J 1995 14 5579 5588 8521815
Muzio M Chinnaiyan AM Kischkel FC O'Rourke K Shevchenko A Ni J Scaffidi C Bretz JD Zhang M Gentz R Mann M Krammer PH Peter ME Dixit VM FLICE, a novel FADD-homologous ICE/CED-3-like protease, is recruited to the CD95 (Fas/APO-1) death--inducing signaling complex Cell 1996 85 817 827 8681377 10.1016/S0092-8674(00)81266-0
Martin DA Siegel RM Zheng L Lenardo MJ Membrane oligomerization and cleavage activates the caspase-8 (FLICE/MACHalpha1) death signal J Biol Chem 1998 273 4345 4349 9468483 10.1074/jbc.273.8.4345
Tibbetts MD Zheng L Lenardo MJ The death effector domain protein family: regulators of cellular homeostasis Nat Immunol 2003 4 404 409 12719729 10.1038/ni0503-404
Thome M Schneider P Hofmann K Fickenscher H Meinl E Neipel F Mattmann C Burns K Bodmer JL Schroter M Scaffidi C Krammer PH Peter ME Tschopp J Viral FLICE-inhibitory proteins (FLIPs) prevent apoptosis induced by death receptors Nature 1997 386 517 521 9087414 10.1038/386517a0
Irmler M Thome M Hahne M Schneider P Hofmann K Steiner V Bodmer JL Schroter M Burns K Mattmann C Rimoldi D French LE Tschopp J Inhibition of death receptor signals by cellular FLIP Nature 1997 388 190 195 9217161 10.1038/40657
Siegel RM Martin DA Zheng L Ng SY Bertin J Cohen J Lenardo MJ Death-effector filaments: novel cytoplasmic structures that recruit caspases and trigger apoptosis J Cell Biol 1998 141 1243 1253 9606215 10.1083/jcb.141.5.1243
Newton K Harris AW Bath ML Smith KG Strasser A A dominant interfering mutant of FADD/MORT1 enhances deletion of autoreactive thymocytes and inhibits proliferation of mature T lymphocytes Embo J 1998 17 706 718 9450996 10.1093/emboj/17.3.706
Kabra NH Kang C Hsing LC Zhang J Winoto A T cell-specific FADD-deficient mice: FADD is required for early T cell development Proc Natl Acad Sci U S A 2001 98 6307 6312 11353862 10.1073/pnas.111158698
Strasser A Newton K FADD/MORT1, a signal transducer that can promote cell death or cell growth Int J Biochem Cell Biol 1999 31 533 537 10399313 10.1016/S1357-2725(99)00003-5
Newton K Kurts C Harris AW Strasser A Effects of a dominant interfering mutant of FADD on signal transduction in activated T cells Curr Biol 2001 11 273 276 11250157 10.1016/S0960-9822(01)00067-7
Beisner DR Chu IH Arechiga AF Hedrick SM Walsh CM The requirements for Fas-associated death domain signaling in mature T cell activation and survival J Immunol 2003 171 247 256 12817005
Zhang J Kabra NH Cado D Kang C Winoto A FADD-deficient T cells exhibit a disaccord in regulation of the cell cycle machinery J Biol Chem 2001 276 29815 29818 11390402 10.1074/jbc.M103838200
Alappat EC Volkland J Peter ME Cell cycle effects by C-FADD depend on its C-terminal phosphorylation site J Biol Chem 2003 278 41585 41588 12954630 10.1074/jbc.C300385200
Shimada K Matsuyoshi S Nakamura M Ishida E Kishi M Konishi N Phosphorylation of FADD is critical for sensitivity to anticancer drug-induced apoptosis Carcinogenesis 2004 25 1089 1097 15001534 10.1093/carcin/bgh130
Newton K Harris AW Strasser A FADD/MORT1 regulates the pre-TCR checkpoint and can function as a tumour suppressor Embo J 2000 19 931 941 10698935 10.1093/emboj/19.5.931
Tourneur L Mistou S Michiels FM Devauchelle V Renia L Feunteun J Chiocchia G Loss of FADD protein expression results in a biased Fas-signaling pathway and correlates with the development of tumoral status in thyroid follicular cells Oncogene 2003 22 2795 2804 12743602 10.1038/sj.onc.1206399
Michiels FM Caillou B Talbot M Dessarps-Freichey F Maunoury MT Schlumberger M Mercken L Monier R Feunteun J Oncogenic potential of guanine nucleotide stimulatory factor alpha subunit in thyroid glands of transgenic mice Proc Natl Acad Sci U S A 1994 91 10488 10492 7937980
Tourneur L Delluc S Levy V Valensi F Radford-Weiss I Legrand O Vargaftig J Boix C Macintyre EA Varet B Chiocchia G Buzyn A Absence or Low Expression of Fas-Associated Protein with Death Domain in Acute Myeloid Leukemia Cells Predicts Resistance to Chemotherapy and Poor Outcome Cancer Res 2004 64 8101 8108 15520222
Iijima N Miyamura K Itou T Tanimoto M Sobue R Saito H Functional expression of Fas (CD95) in acute myeloid leukemia cells in the context of CD34 and CD38 expression: possible correlation with sensitivity to chemotherapy Blood 1997 90 4901 4909 9389707
Buzyn A Petit F Ostankovitch M Figueiredo S Varet B Guillet JG Ameisen JC Estaquier J Membrane-bound Fas (Apo-1/CD95) ligand on leukemic cells: A mechanism of tumor immune escape in leukemia patients Blood 1999 94 3135 3140 10556200
Laurent G Jaffrezou JP Signaling pathways activated by daunorubicin Blood 2001 98 913 924 11493433 10.1182/blood.V98.4.913
Friesen C Fulda S Debatin KM Cytotoxic drugs and the CD95 pathway Leukemia 1999 13 1854 1858 10557062 10.1038/sj/leu/2401333
Wen J Ramadevi N Nguyen D Perkins C Worthington E Bhalla K Antileukemic drugs increase death receptor 5 levels and enhance Apo-2L-induced apoptosis of human acute leukemia cells Blood 2000 96 3900 3906 11090076
Altucci L Rossin A Raffelsberger W Reitmair A Chomienne C Gronemeyer H Retinoic acid-induced apoptosis in leukemia cells is mediated by paracrine action of tumor-selective death ligand TRAIL Nat Med 2001 7 680 686 11385504 10.1038/89050
Micheau O Solary E Hammann A Dimanche-Boitrel MT Fas ligand-independent, FADD-mediated activation of the Fas death pathway by anticancer drugs J Biol Chem 1999 274 7987 7992 10075697 10.1074/jbc.274.12.7987
Micheau O Solary E Hammann A Martin F Dimanche-Boitrel MT Sensitization of cancer cells treated with cytotoxic drugs to fas-mediated cytotoxicity J Natl Cancer Inst 1997 89 783 789 9182976 10.1093/jnci/89.11.783
Mullauer L Mosberger I Chott A Fas ligand expression in nodal non-Hodgkin's lymphoma Mod Pathol 1998 11 369 375 9578088
Hahne M Rimoldi D Schroter M Romero P Schreier M French LE Schneider P Bornand T Fontana A Lienard D Cerottini J Tschopp J Melanoma cell expression of Fas(Apo-1/CD95) ligand: implications for tumor immune escape Science 1996 274 1363 1366 8910274 10.1126/science.274.5291.1363
Saas P Walker PR Hahne M Quiquerez AL Schnuriger V Perrin G French L Van Meir EG de Tribolet N Tschopp J Dietrich PY Fas ligand expression by astrocytoma in vivo: maintaining immune privilege in the brain? J Clin Invest 1997 99 1173 1178 9077524
O'Connell J O'Sullivan GC Collins JK Shanahan F The Fas counterattack: Fas-mediated T cell killing by colon cancer cells expressing Fas ligand J Exp Med 1996 184 1075 1082 9064324 10.1084/jem.184.3.1075
Strand S Hofmann WJ Hug H Muller M Otto G Strand D Mariani SM Stremmel W Krammer PH Galle PR Lymphocyte apoptosis induced by CD95 (APO-1/Fas) ligand-expressing tumor cells--a mechanism of immune evasion? Nat Med 1996 2 1361 1366 8946836 10.1038/nm1296-1361
Niehans GA Brunner T Frizelle SP Liston JC Salerno CT Knapp DJ Green DR Kratzke RA Human lung carcinomas express Fas ligand Cancer Res 1997 57 1007 1012 9067260
Mitsiades N Poulaki V Mastorakos G Tseleni-Balafouta ST Kotoula V Koutras DA Tsokos M Fas ligand expression in thyroid carcinomas: a potential mechanism of immune evasion J Clin Endocrinol Metab 1999 84 2924 2932 10443700 10.1210/jc.84.8.2924
Restifo NP Not so Fas: Re-evaluating the mechanisms of immune privilege and tumor escape Nat Med 2000 6 493 495 10802692 10.1038/74955
Restifo NP Countering the 'counterattack' hypothesis Nat Med 2001 7 259 11231598 10.1038/85357
Favre-Felix N Fromentin A Hammann A Solary E Martin F Bonnotte B Cutting edge: the tumor counterattack hypothesis revisited: colon cancer cells do not induce T cell apoptosis via the Fas (CD95, APO-1) pathway J Immunol 2000 164 5023 5027 10799856
Desbarats J Wade T Wade WF Newell MK Dichotomy between naive and memory CD4(+) T cell responses to Fas engagement Proc Natl Acad Sci U S A 1999 96 8104 8109 10393955 10.1073/pnas.96.14.8104
Desbarats J Newell MK Fas engagement accelerates liver regeneration after partial hepatectomy Nat Med 2000 6 920 923 10932231 10.1038/78688
Freiberg RA Spencer DM Choate KA Duh HJ Schreiber SL Crabtree GR Khavari PA Fas signal transduction triggers either proliferation or apoptosis in human fibroblasts J Invest Dermatol 1997 108 215 219 9008237 10.1111/1523-1747.ep12334273
Owen-Schaub LB Meterissian S Ford RJ Fas/APO-1 expression and function on malignant cells of hematologic and nonhematologic origin J Immunother 1993 14 234 241 7507710
Mishima K Nariai Y Yoshimura Y Carboplatin induces Fas (APO-1/CD95)-dependent apoptosis of human tongue carcinoma cells: sensitization for apoptosis by upregulation of FADD expression Int J Cancer 2003 105 593 600 12740905 10.1002/ijc.11133
Micheau O Tschopp J Induction of TNF receptor I-mediated apoptosis via two sequential signaling complexes Cell 2003 114 181 190 12887920 10.1016/S0092-8674(03)00521-X
| 15717929 | PMC550674 | CC BY | 2021-01-04 16:39:09 | no | Med Immunol. 2005 Feb 17; 4:1 | utf-8 | Med Immunol | 2,005 | 10.1186/1476-9433-4-1 | oa_comm |
==== Front
Biomed Eng OnlineBioMedical Engineering OnLine1475-925XBioMed Central London 1475-925X-4-111572071210.1186/1475-925X-4-11ResearchDispersion of cardiac action potential duration and the initiation of re-entry: A computational study Clayton Richard H [email protected] Arun V [email protected] Department of Computer Science, University of Sheffield, UK2 School of Biomedical Sciences, University of Leeds UK2005 18 2 2005 4 11 11 11 11 2004 18 2 2005 Copyright © 2005 Clayton and Holden; licensee BioMed Central Ltd.2005Clayton and Holden; 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 initiation of re-entrant cardiac arrhythmias is associated with increased dispersion of repolarisation, but the details are difficult to investigate either experimentally or clinically. We used a computational model of cardiac tissue to study systematically the association between action potential duration (APD) dispersion and susceptibility to re-entry.
Methods
We simulated a 60 × 60 mm 2 D sheet of cardiac ventricular tissue using the Luo-Rudy phase 1 model, with maximal conductance of the K+ channel gKmax set to 0.004 mS mm-2. Within the central 40 × 40 mm region we introduced square regions with prolonged APD by reducing gKmax to between 0.001 and 0.003 mS mm-2. We varied (i) the spatial scale of these regions, (ii) the magnitude of gKmax in these regions, and (iii) cell-to-cell coupling.
Results
Changing spatial scale from 5 to 20 mm increased APD dispersion from 49 to 102 ms, and the susceptible window from 31 to 86 ms. Decreasing gKmax in regions with prolonged APD from 0.003 to 0.001 mS mm-2 increased APD dispersion from 22 to 70 ms, and the susceptible window from <1 to 56 ms. Decreasing cell-to-cell coupling by changing the diffusion coefficient from 0.2 to 0.05 mm2 ms-1 increased APD dispersion from 57 to 88 ms, and increased the susceptible window from 41 to 74 ms.
Conclusion
We found a close association between increased APD dispersion and susceptibility to re-entrant arrhythmias, when APD dispersion is increased by larger spatial scale of heterogeneity, greater electrophysiological heterogeneity, and weaker cell-to-cell coupling.
==== Body
1. Background
Cardiac disease remains an important cause of sudden death in the industrialised world, and in many cases the lethal events are the cardiac arrhythmias called ventricular tachycardia (VT) and ventricular fibrillation (VF). Spontaneous episodes of VT and VF occur in patients where cardiac disease or congenital abnormality has remodelled either the structure or function of cardiac cells and tissue. There is abundant experimental evidence to support the idea that VT and VF are sustained by re-entry [1,2], but the initiation of re-entry in a particular individual is not well understood, and so is difficult to either predict or prevent.
Slow conduction and unidirectional block have long been known to facilitate re-entry [3], and experimental studies have established a link between regional differences in repolarisation, and an increased vulnerability to re-entrant arrhythmias following one or more premature stimuli [4-7]. One of the earliest computer models of activation in cardiac tissue was used to demonstrate that regional differences in repolarisation can allow fibrillation to develop following a premature stimulus [8]. Further experimental studies have found that steep gradients in repolarisation correlate with arcs of conduction block around which re-entry circulates [9-12], and have suggested that regions with longer refractory period must be of a critical size for sustained re-entry to occur [13]. Computational and theoretical studies [14-16] have also shown how a region with prolonged repolarisation can block a premature excitation resulting in initiation of re-entry, and that the size of the inhomogeneity determines the characteristics and persistence of re-entry.
Regional differences in repolarisation are often described as action potential duration (APD) dispersion. The difference between the longest and shortest observed APD is a conceptually simple and easily obtained quantity and has widely been used to measure APD dispersion, although other indices have been proposed [17]. Some experimental studies have established critical values of APD dispersion above which re-entry is initiated consistently [18]. In others the gradient of APD has been measured, and spatial gradients of between 2 and 12.5 ms mm-1 were associated with block and re-entry [9,19,20]. Spatial APD gradients arise from regional differences in ion channel function, but their magnitude depends on electrotonic current flow during repolarisation. APD dispersion can be produced by the spatial scale of regional differences and the magnitude of functional heterogeneity, and is modulated by electrotonic current flow which depends on the strength of cell-to-cell coupling [16,21,22]. The relative effect of these three quantities on APD dispersion and vulnerability to re-entry is important because both disease and congenital abnormalities can result in changes to one or more of them. However, it is difficult to control these tissue properties independently in experiments.
Computational models offer a powerful research tool for addressing these questions, because the properties of a virtual tissue can be controlled precisely and independently in a way that would be extremely difficult to achieve experimentally. The purpose of this study was therefore to investigate systematically how measured APD dispersion and vulnerability to re-entry in a computational model of ventricular tissue are related to: (i) the spatial scale of heterogeneity, (ii) the magnitude of differences in K+ channel conductance between regions with short and long APD, and (iii) strength of cell-to-cell coupling.
2. Methods
2.1 Computational model of electrical activation
We simulated electrical activation in a 2 D isotropic monodomain virtual tissue [23]
where Vm is membrane voltage, Cm specific membrane capacitance, D a diffusion coefficient and Iion current flow through the cell membrane per unit area. We used the Luo-Rudy phase 1 (LR1) model [24] to give Iion,
where INa, ICa (described as Isi in the original model) and IK are time and voltage dependent currents flowing through Na+, Ca2+, and K+ channels, IK1 a time-independent K+ current, IKp a plateau K+ current, and Ib a background current. We changed two parameters from the original Luo and Rudy paper [24]. We reduced maximum Na+ conductance from 0.23 mS mm-2 to 0.16 mS mm-2 as in the later version of the model [25], and we reduced the maximum conductance of the slow inward current from 0.0009 mS mm-2 to 0.0005 mS mm-2 to produce an APD comparable to that in the canine ventricle. We controlled repolarisation by varying maximum K+ conductance (gKmax) from the default value of 0.00282 mS mm-2 to a value between 0.001 mS mm-2 and 0.004 mS mm-2 (see below).
2.2 Numerical methods
We solved equation 1 and the LR1 equations using an explicit Euler method, with both a lookup table of the voltage dependent parameters in the LR1 model, and an adaptive operator splitting technique [26]. We applied no-flux boundary conditions, set Cm to 0.001 μF mm-2, and set D to between 0.05 and 0.2 mm2 ms-1. We used an adaptive timestep of either 0.02 or 0.1 ms depending on the magnitude of dVm/dt at each grid point [26]. With a space step of 0.2 mm, and D of 0.1 mm2 ms-1 we obtained a conduction velocity (CV) for a stable plane wave of 0.56 m s-1, with a speedup of about two times and an error in CV of 2.5 % compared to computations with a fixed timestep of 0.01 ms. Simulations with smaller fixed and adaptive timesteps yielded plane waves with a comparable CV. Changing the space step to 0.25 mm and 0.15 mm resulted in a change of CV for a plane wave of <5 % compared to the CV computed with a space step of 0.2 mm. These findings indicated the stability of our numerical method.
Figure 1 shows action potentials, APD restitution curves, and CV restitution for virtual tissue with different values of gKmax. In each case a propagating action potential could be elicited with a minimum diastolic interval of about 10 ms, indicating that the refractory period was very close to the APD.
Figure 1 (a) Action potentials recorded during steady pacing at 500 ms intervals for different values of gKmax. (b) Action potential duration restitution and (c) conduction velocity restitution for different values of diastolic interval measured with a premature stimulus during steady pacing at 500 ms intervals. All measurements obtained from a narrow strip of virtual tissue 10 mm long and with uniform gKmax of 0.001, 0.002, 0.003, and 0.004 mS mm-2 as indicated.
2.3 Virtual tissue and heterogeneity
Each of the virtual tissues in this study represented a 60 × 60 mm 2 D sheet with a 10 mm border around each edge with gKmax set to 0.004 mS mm-2. The central 40 × 40 mm region was heterogeneous, and was subdivided into squares. Within alternate squares gKmax was set to either 0.004 mS mm-2 or to a specific lower value (see below). Thus the virtual tissues had a border region with short APD, and a central heterogeneous region divided into a chequerboard with alternating squares of either short or long APD (Figure 2).
Figure 2 Configuration of 60 × 60 mm virtual tissues used in the study where (a) spatial scale, (b) ΔgKmax, and (c) strength of cell-to-cell coupling were varied as indicated on the figure. Greyscale shows gKmax, in each region, with light grey, mid grey, dark gray and black corresponding to gKmax values of 0.004, 0.003, 0.002, and 0.001 mS mm-2 respectively.
In our reference virtual tissue, the 40 × 40 mm heterogeneous region was divided into 16 squares giving heterogeneity with a spatial scale of 10 mm. In half of the 16 squares gKmax was set to 0.001 mS mm-2, giving a functional heterogeneity with a difference in gKmax (Δgkmax) of 0.003 mS mm-2. The diffusion coefficient in the whole virtual tissue was set to 0.1 mm2 ms-1.
We varied the spatial scale of heterogeneity by changing the size of heterogeneity in the reference virtual tissue from 10 mm to 20 mm and 5 mm (Figure 2a), varied Δgkmax from 0.003 to 0.002 and 0.001 mS mm-2 by changing gKmax in the alternate squares from 0.001 to 0.002 and 0.003 mS mm-2 respectively (Figure 2b), and varied the strength of cell-to-cell coupling by changing the diffusion coefficient from 0.1 to 0.05 and 0.2 mm2 ms-1 (Figure 2c).
2.4 APD dispersion
Action potentials were initiated in each virtual tissue by holding the membrane voltage along one edge at 0 mV for 2 ms. We measured APD to 90% recovery (APD90) at every grid point. We estimated APD dispersion during steady pacing with three measured that have been used in experimental studies [17]. First we measured APD across the whole virtual tissue, and determined the difference between the maximum and minimum APD (APDdiff). Second we measured the standard deviation of APD (APDSD) across the whole virtual tissue. Finally we measured the APD difference between each grid point and its neighbours 1 mm above, below, left and right, and determined the maximum value of measurements throughout the whole virtual tissue (maxLD) [17].
2.5 Vulnerability to re-entry
A spatially homogenous control virtual tissue with uniform gKmax supported propagating plane waves following S1 stimulation along one edge. These propagating plane waves had a depolarising wavefront and a repolarising wave back both aligned parallel to the edge that was stimulated. A premature S2 stimulus delivered to the same edge as the S1 stimulus therefore resulted either in block or in a propagating plane wave.
In the heterogeneous virtual tissues the wave back was not a plane wave because some regions repolarised more quickly than others. A premature S2 stimulus could produce a wavefront that would encounter a mixture of recovered and refractory tissue, and hence elicit wavebreak and re-entry. We therefore assessed vulnerability to re-entry in the heterogeneous virtual tissues by delivering two S1 stimuli to one edge at 500 ms intervals, and then a premature S2 stimulus to the same edge. We varied the timing of the S2 stimulus in steps of 1 ms. The virtual tissue response was characterised as either block if the S2 stimulus failed to propagate, re-entry if the S2 stimulus elicited re-entry that completed more than one cycle, wavebreak if the S2 stimulus elicited a wave that broke but did not re-enter, or propagation if the S2 stimulus elicited a wave that propagated without wavebreak.
Vulnerability to re-entry is typically estimated by applying a local premature stimulus that interacts with repolarising tissue, and the vulnerable window is the range of stimulus strength and timing that elicits re-entry. In this study, we investigated the initiation of re-entry by S1 S2 stimulation from the same stimulus site, and we estimated the vulnerability of each virtual tissue to re-entry from the range of S2 intervals that resulted in either wavebreak or re-entry. To avoid confusion we refer to our estimate of vulnerability as susceptibility to re-entry, and to the range of S2 intervals as the width of the susceptible window.
2.6 Potential antiarrhythmic strategies
We made a preliminary investigation into two candidate mechanisms for reducing susceptibility, inactivation of the Na+ channel and conductance of the time independent K+ channel gK1.
Block of an action potential occurs if there is insufficient Na+ current to support a propagating wavefront [27]; when Na+ channels have not recovered from inactivation, then a propagating wave blocks and dissipates [28]. Na+ channel inactivation is controlled by the j-gate in the LR1 model [24]. We prolonged recovery of Na+ channels from inactivation throughout the virtual tissue by multiplying the time constant of Na+ channel inactivation τj by 10 [29].
The time independent K+ current iK1 is a voltage dependent current that holds the membrane at its resting potential. It is activated during repolarisation and at rest, and is also activated close to the core of re-entrant waves [30,31]. We investigated the effect of doubling the conductance of iK1 throughout the virtual tissue.
3. Results
3.1 Propagation and APD dispersion during pacing
Figure 3 shows the spatial distribution of APD90 in variants of the virtual tissue during pacing at a cycle length of 500 ms from the bottom edge. The three columns show the effect of changing spatial scales (Figure 3a), Δgkmax (Figure 3b), and strength of cell-to-cell coupling (Figure 3c). The range of colours and number of contours on each figure indicates the range of APD. Hence the leftmost column shows that when Δgkmax and the strength of cell-to-cell coupling are held constant, increasing spatial scale increases the range of APD and decreasing spatial scale decreases the range of APD. Overall, Figure 3 shows that small spatial scales, small Δgkmax, and strong cell-to-cell coupling act to reduce the range of observed APD. This effect can also be seen in Figure 4, which summarises how the three measures of APD dispersion were affected by spatial scale, Δgkmax, and strength of cell-to-cell coupling. Each of these measures changed monotonically within the range of spatial scale, Δgkmax and strength of cell-to-cell coupling that we studied. APDdiff was halved by reducing spatial scale from 20 to 5 mm or by increasing the diffusion coefficient from 0.05 to 0.1 mm2 ms-1.
Figure 3 Spatial distribution of APD during from pacing at 500 ms intervals in each variant of the 60 × 60 mm virtual tissue. (a) Effect of changing spatial scale, with ΔgKmax fixed at 0.003 mS mm-2, and diffusion coefficient fixed at 0.1 mm2 ms-1. (b) Effect of changing ΔgKmax, with spatial scale fixed at 10 mm, and diffusion coefficient fixed at 0.1 mm2 ms-1. (c) Effect of changing strength of cell-to-cell coupling by changing the diffusion coefficient, with spatial scale fixed at 10 mm, and ΔgKmax fixed at 0.003 mS mm-2.
Figure 4 APDdiff, APDSD, and MaxLD measured in virtual tissues where (a) the spatial scale of heterogeneity, (b) magnitude of functional heterogeneity, and (c) strength of cell-to-cell coupling were changed.
When re-entry was initiated, we observed break-up into multiple re-entrant wavelets with up to 18 phase singularities. The mechanism of instability was likely to be a combination of the spatial heterogeneity leading to localised conduction block combined with dynamical instability resulting from steep APD restitution [32] (Figure 1), but was not investigated explicitly. In some simulations the re-entrant waves coalesced and re-entry spontaneously terminated. However, there was no clear association between this observation and S2 timing, spatial scale, functional heterogeneity, or coupling.
3.2 Susceptibility to re-entry
Figure 5 shows examples of re-entry, wavebreak and propagation in the reference virtual tissue. In each case the pacing (S1), and premature (S2) stimuli were delivered to the bottom edge. In the top row (Figure 5a – see also the movie in additional file 1), the premature S2 activation was blocked at each of the regions with prolonged repolarisation, and curled round to give figure-of-8 re-entry. In the second row (Figure 5b – see also the movie in additional file 2) the S2 stimulus was 15 ms later, and although the S2 activation was partially blocked, re-entry was prevented by collision of the wavebreak with antegrade activation of the regions with prolonged APD. With a later S2 stimulus (Figure 5c – see also the movie in additional file 3), the regions with prolonged APD had recovered enough to conduct the S2 activation, although the activation wave was delayed slightly by each region with prolonged APD.
Figure 5 Example responses to premature S2 stimulus in virtual tissue with heterogeneity on spatial scale of 10 mm, ΔgKmax of 0.003 mS mm-2, and diffusion coefficient set to 0.1 mm2 ms-1. Blue lines outline regions with prolonged repolarisation.. (a) S2 at 230 ms and induction of re-entry. (b) S2 at 250 ms, wavebreaks are indicated with a star. (c) S2 at 265 ms and delayed propagation with no wavebreak. In (a-c) snapshots 50, 100 and 150 ms after the S2 stimulus are included, with propagation from bottom to top of the figure. Isochrones at intervals of 5 ms are shown, and colour coding indicates the progress of the wavefront. Movies of the simulations shown in this figure are available as additional files Figure 5a Figure 5b, and Figure 5c.
Figure 6 shows the detailed response of each virtual tissue for a range of S2 intervals, and indicates how each of the three interventions affects the response of the tissue to a premature stimulus. Increasing spatial scale, increasing Δgkmax, and decreasing the strength of cell to cell coupling all resulted in an greater range of S2 intervals that resulted in wavebreak (red) or re-entry (orange), and hence an increase in the width of the susceptible window The lower bound of the susceptible window where S2 was blocked depended on the refractory period of the border tissue with short APD, and was not greatly affected by changes in spatial scale, Δgkmax or strength of cell-to-cell coupling. The upper bound showed a similar trend to the measures of APD dispersion shown in Figure 4, with a large spatial scale, large Δgkmax and weak cell-to-cell coupling associated with a wide susceptible window.
Figure 6 Response of virtual tissues to premature S2 stimulus. Blue indicates block, orange re-entry, red wavebreak, and purple propagation. (a) Results for changing in spatial scale, (b) changes in ΔgKmax, and (c) changes in strength of cell-to-cell coupling, where D is the diffusion coefficient.
The virtual tissue with a spatial scale of 20 mm showed a different pattern of susceptibility compared to the others, with two ranges of S2 that initiated wavebreak and two ranges of S2 that initiated re-entry. This behaviour is illustrated in Figure 7. For S2 delivered between 198 and 226 ms, the regions with prolonged APD blocked the premature activation, and re-entry was initiated by retrograde activation through the isthmus between these two regions (Figure 7a – see also the movie in additional file 4). For values of S2 between 227 and 249 ms, the isthmus conducted the S2 activation resulting in wavebreak, but the regions with prolonged APD remained refractory (Figure 7b – see also the movie in additional file 5). For values of S2 between 250 and 263 ms, re-entry was initiated by retrograde activation of the regions with prolonged APD (Figure 7c – see also the movie in additional file 6).
Figure 7 (See text for details) Example responses to premature S2 stimulus in virtual tissue with heterogeneity on spatial scale of 20 mm and ΔgKmax of 0.003 mS mm-2. Arrows show direction of propagation. (a) S2 at 200 ms and induction of re-entry with two phase singularities. (b) S2 at 245 ms and broken wave, retrograde activation is blocked. (c) S2 at 255 ms with retrograde activation and re-entry with four phase singularities i.e. two systems of figure-of-eight re-entry In each figure isochrones at intervals of 5 ms are shown, and the activation wavefront at S2+150 ms (a and c) and S2+155 ms (b) is shown as a thick black line. Movies of the simulations shown in this figure are available as additional files Figure 7a, Figure 7b, and Figure 7c.
Figure 8a shows the width of the susceptible window plotted against APDdiff and reveals an approximately linear relationship. The correlation coefficient R2 = 0.99, which indicates a strong association between APDdiff and susceptibility. The interception of the line with the APDdiff axis also suggests that for the S1 S2 configuration used in this study, the susceptible window falls to zero for APD dispersion less than 20 ms. The association between the other measures of APD dispersion (APDSD figure 8b, maxLD Figure 8c) and susceptibility is also monotonic, but less well correlated than for APDdiff.
Figure 8 Association between width of the susceptible window and (a) APDdiff, (b) APDSD, and (c) MaxLD, for each of the three interventions In each case the grey circle indicates the reference virtual tissue with spatial scale of 10 mm, ΔgKmax of 0.003 mS mm-2 and diffusion coefficient set to of 0.1 mm2 ms-1.
3.3 Potential antiarrhythmic strategies
Although prolonging τj by a factor of 10 had only a small (< 2 ms) effect on maximum and minimum APD, the width of the susceptible window was decreased from 56 ms to 39 ms. The lower bound of the susceptible window moved from 198 to 216 ms, reflecting an increase in the refractory period of the virtual tissue as well as more prominent conduction velocity restitution.
Doubling gK1 increased current flow across the membrane during repolarisation, shortening APD by about 10% and decreasing APDdiff from 70 ms to 49 ms. The width of the susceptible window was also reduced from 56 ms to 42 ms when gK1 was doubled. The lower bound of the susceptible window moved from 198 ms to 175 ms, as a result of the shorter APD.
4. Discussion
In this study we have used a computational model of cardiac tissue to dissect out the effects of spatial scale, Δgkmax, and strength of cell-to-cell coupling on APD dispersion and susceptibility to re-entry. Wavebreaks and re-entry can be created in cardiac tissue when an activation wavefront encounters a gradient of recovery. Experimental studies have therefore found an association between increased APD dispersion and greater susceptibility to re-entry because a premature stimulus is more likely to be blocked in tissue with regions of prolonged APD. In this study we found that large spatial scale heterogeneity, large Δgkmax, and reduced strength of cell-to-cell coupling all increased both APD dispersion and susceptibility to re-entry. Tissue heterogeneities produce APD dispersion, and APD dispersion is modulated by electrotonic current flow [21]. In this study, spatial scale and Δgkmax affected APD dispersion directly, whereas changing the strength of cell-to-cell coupling affected electrotonic current flow. This study indicates that each of these factors could be an important component in arrhythmogenesis, and that the susceptibility of a heterogenous tissue to re-entry can be estimated from simple measures of APD dispersion.
4.1 Relation to other work
Although the spatial scale of heterogeneity has been identified as potentially important in other studies [13], there is little information in the experimental literature to indicate how spatial scale affects susceptibility to re-entry. In one experimental study a ~1 cm2 region of thin layer of rabbit ventricular epicardium was cooled to produce a small region with prolonged APD, producing a dispersion of refractory periods ranging from 27 and 45 ms [10]. Re-entry could be initiated in this preparation using 4 increasingly premature stimuli. The spatial scale of heterogeneity in this experimental study was comparable to the reference virtual tissue used in our present study, but the effects of Δgkmax and strength of cell-to-cell coupling in the experimental study are difficult to establish. Nevertheless the initiation of re-entry by block and retrograde activation in the region with prolonged APD followed a broadly similar pattern to our simulations, although the activation pathways in the experimental study were more complex, presumably due to anisotropic conduction in the rabbit ventricle. A decrease in strength of cell-to-cell coupling results in slowed conduction, and this is a common finding in tissue damaged by ischaemia, infarction [33], and other pathology [34]. Several studies have shown that decreasing cell-to-cell coupling can expose ionic heterogeneities [22,35-37].
Recent computational studies have addressed the influence of heterogeneous acetylcholine distribution on the vulnerability to and stability of re-entry in the atria [15,32]. The findings of these studies are broadly similar to the present study, although the effects of cell-to-cell coupling were not explicitly addressed, and initiation of re-entry was by either crossfield stimulation [32] or by S1 and S2 stimulation at different sites [15]. Both of these protocols would be expected to induce re-entry even in uniform tissue. In the present study both S1 and S2 stimuli were delivered from the same location, which does not initiate re-entry in uniform tissue, allowing us to examine the effect of heterogeneity on the initiation of re-entry in isolation.
The dynamical behaviour of APD is recognised as important not only for the stability of re-entry [2] but also in the development of alternans [38]. Recent experimental [39] and computational [38,40] studies have shown that APD dispersion can arise dynamically leading to discordant alternans, wavebreak, and re-entry in tissue that is either homogeneous or in which the ionic properties vary smoothly[38,41,42]. In the present study we measured APD dispersion at a fixed cycle length of 500 ms. The APD restitution curves given in Figure 1 indicate that APD dispersion could have been affected by pacing at shorter cycle lengths. This observation raises the possibility that heterogenous APD restitution could act to amplify APD dispersion.
The effect of heterogeneity on the stability of spiral waves has been investigated by Xie et al [43]. This study found that the amount of heterogeneity required to destabilise re-entry decreased as the degree of dynamical instability resulting from a steep APD restitution curve increased. In the present study we were interested in the initiation of re-entry rather than the stability of re-entry once initiated.
4.2 APD dispersion and susceptibility to arrhythmias
Normal ventricular tissue is remarkably resistant to the initiation of re-entry, but this robustness is greatly reduced by actions that increase the spatial dispersion of refractoriness. In this computational study we have shown that regional differences in repolarisation have an interlinked effect not only on the initiation of re-entry but also on measures of APD dispersion. Measures of APD dispersion are valuable in clinical practice because they could provide an estimate of arrhythmia risk, and various indices have been developed in experimental studies [17]. In our present study we have found that relatively simple measures of APD dispersion obtained from the tissue were related to the width of the susceptible period for re-entry.
4.3 Potential antiarrhythmic strategies
These preliminary investigations suggest that, in our model, susceptibility to re-entry could be reduced if recovery of Na+ channels from inactivation can be prolonged, or if the conductance of the iK1 channel can be increased. The effect of this kind of intervention in the intact heart may however be more complex. Other computational studies have shown that modifying the kinetics of the Na+ channel can have a pro-arrhythmic effect. Delaying recovery of Na+ channels from inactivation can increase the slope of the APD restitution curve and hence the likelihood of alternans and re-entry [29], and reducing Na+ channel conductance increases the vulnerable window [44]. Differences in the spatio-temporal complexity of VF between left and right ventricles have been attributed to differences in the current density of the iK1 channel in experimental studies [30]. Although this experimental finding is not directly connected to the effects of the iK1 channel conductance on susceptibility to re-entry investigated in the present study, it does highlight the potential importance of this channel for the mechanisms of re-entry.
The influence of individual ion channel currents on the initiation and subsequent behaviour of re-entry is an important direction for future research, but will require more biophysically detailed cell models than the LR1 model used in this study.
4.4 Limitations of the study
The electrical behaviour of cardiac tissue is complex, and depends on processes that act at tissue, cell, sub-cellular, and molecular levels. Computational models of electrical activation and conduction in the heart aim to simulate processes that are relevant to the research question, and simplifications are made accordingly. This study involved a large number of computations to establish susceptibility to re-entry, and so we chose to use a model that was a compromise between fidelity to real cardiac tissue and computational requirements.
More detailed versions of the LR model and others incorporating a fuller description of ion channels, pumps, exchangers, as well as Ca2+ storage and release have been developed [22,25,45,46]. In tissue with regional ischaemia, Ca2+ handling may become heterogeneous in addition to APD, and so it is possible that susceptibility to re-entry could also be modified if this additional feature is taken into account.
In the present study we chose to use an idealised geometrical heterogeneity based on square regions because this approach allowed us to assess the initiation of re-entry under well controlled conditions. In real cardiac tissue we would expect the heterogeneities to be much more irregular in shape and gradient, and the conditions that favour re-entry to be dependent on the relative location of the heterogenous region and the stimulus site.
The behaviour of re-entry in 3 D tissue is more complex than in 2 D, especially when the effects of rotational anisotropy and transmural differences in action potential shape and duration are taken into account [47]. Studies relating APD dispersion and susceptibility to re-entry in anatomically detailed 3 D tissue are another important project for the future. Recent computational studies indicate that the mechanical properties can not only modify the behaviour of re-entrant waves [48], but also that stretch activated channels in the cell membrane can contribute to susceptibility to re-entry if the tissue is stretched during repolarisation [49,50]. Since electrical repolarisation occurs at the same time as force generation in cardiac cells, the effect of cardiac mechanics on susceptibility to re-entry remains an important research question.
Authors' contributions
RHC conceived and designed the study, wrote the simulation code, and ran the simulations. AVH participated in the study design, and helped draft the manuscript. Both authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Movie relating to Figure 5a
Click here for file
Additional File 2
Movie relating to Figure 5b
Click here for file
Additional File 3
Movie relating to Figure 5c
Click here for file
Additional File 4
Movie relating to Figure 7a
Click here for file
Additional File 5
Movie relating to Figure 7b
Click here for file
Additional File 6
Movie relating to Figure 7c
Click here for file
Acknowledgements
This work was supported by the British Heart Foundation through the award of Basic Science Lectureship BS98001, and project grant PG/03/102/15852 to RHC. We are also grateful to the United Kingdom Medical Research Council and Engineering and Physical Sciences Research Council for additional financial support.
==== Refs
Jalife J Ventricular fibrillation: Mechanisms of initiation and maintenance Annu Rev Physiol 2000 62 25 50 10845083 10.1146/annurev.physiol.62.1.25
Chen PS Wu TJ Ting CT Karagueuzian HS Garfinkel A Lin SF Weiss JN A tale of two fibrillations. Circulation 2003 108 2298 2203 14609997 10.1161/01.CIR.0000094404.26004.07
Mines GR On circulating excitations in heart muscles and their possible relation to tachycardia and fibrillation Transactions of the Royal Society of Canada 1914 4 43 53
Han J Moe GK Nonuniform recovery of excitability in ventricular muscle Circulation Research 1964 14 44 60 14104163
Han J Garcia DeJalon PD Moe GK Adrenergic effects on ventricular vulnerability Circulation Research 1964 14 516 524 14169970
Behrens S Li C Franz MR Effects of myocardial ischaemia on ventricular fibrillation inducibility Journal of the American College of Cardiology 1997 29 17 24
Kirchhof PF Fabritz CL Zabel M Franz MR The vulnerable period for low and high energy T-wave shocks: Role of dispersion of repolarization and effect of d-sotalol Cardiovascular Research 1996 31 953 962 8759252 10.1016/0008-6363(96)00058-2
Moe GK Rheinboldt WC Abildskov JA A computer model of atrial fibrillation American Heart Journal 1964 67 200 220 14118488 10.1016/0002-8703(64)90371-0
Gough WB Mehra R Restivo M Zeiler RH El-Sherif N Reentrant ventricular arrhythmias in the late myocardial infarction period in the dog. 13. Correlation of activation and refractory maps. Circulation Research 1985 57 432 442 4028346
Boersma L Zetelaki Z Brugada J Allessie MA Polymorphic re-entrant ventricular tachycardia in the isolated rabbit heart studied by high density mapping Circulation 2002 105 3053 3061 12082002 10.1161/01.CIR.0000019407.35848.AF
Robert E Aya AGM De La Coussaye JE Peray P Juan JM Brugada J Davy JM Eledjam JJ Dispersion-based reentry: mechanism of initiation of ventricular tachycardia in isolated rabbit hearts Americal Journal of Physiology (Heart and Circulatory Physiology) 1999 45 H413 H423
Wolk R Cobbe SM Kane KA Hicks MN Relevance of inter- and intraventricular electrical dispersion to arrhythmogenesis in normal and ischaemic rabbit myocardium: A study with Cromalkim, 5-Hydroxydecanoate and Glibenclamide Journal of Cardiovascular Pharmacology 1999 33 323 334 10028944 10.1097/00005344-199902000-00022
Allessie MA Bonke FI Schopmann FTG Circus movement in rabbit atrial muscle as a mechanism of tachycardia II. The role of nonuniform recovery of excitability in the occurrence of unidirectional block studied with multiple microelectrodes Circulation Research 1976 39 168 177 939001
Panfilov A Vasiev BN Vortex initiation in a heterogeneous excitable medium Physica D 1991 49 107 113
Vigmond E Tsoi V Kuo S Arevalo H Kneller J Nattel S Trayanova N The effect of vagally induced dispersion of action potential duration on atrial arrhythmogenesis Heart Rhythm 2004 1 334 344 15851180 10.1016/j.hrthm.2004.03.077
Lesh MD Pring M Spear JF Cellular uncoupling can unmask dispersion of action potential duration in ventricular myocardium. A computer modeling study. Circulation Research 1989 65 1426 1440 2805251
Burton FL Cobbe SM Dispersion of ventricular repolarization and refractory period Cardiovascular Research 2001 50 10 23 11282074 10.1016/S0008-6363(01)00197-3
Kuo CS Munkata K Reddy P Surawicz B Characteristics and possible mechanism of ventricular arrhythmia dependent on the dispersion of action potential durations Circulation 1983 67 1356 1367 6851031
Restivo M Gough WB El-Sherif N Ventricular arrhythmias in the subacute myocardial infarction period. High resolution activation and refractory patterns of re-entrant rhythms Circulation Research 1990 66 1310 1327 2335029
Osaka T Kodama I Tsuboi N Toyama J Yamada K Effects of activation sequence and anisotropic cellular geometry on the repolarization phase of action potential in the dog ventricles Circulation 1987 76 226 236 3594771
Clayton RH Holden AV Propagation of normal beats and re-entry in a computational model of ventricular cardiac tissue with regional differences in action potential shape and duration Progress in Biophysics & Molecular Biology 2004 85 473 499 15142758 10.1016/j.pbiomolbio.2003.12.002
Viswanathan PC Shaw RM Rudy Y Effects of I-Kr and I-Ks heterogeneity on action potential duration and its rate dependence - A simulation study Circulation 1999 99 2466 2474 10318671
Clayton RH Computational models of normal and abnormal action potential propagation in cardiac tissue: Linking experimental and clinical cardiology Physiological Measurement 2001 22 R15 R34 11556683 10.1088/0967-3334/22/3/201
Luo CH Rudy Y A model of the ventricular cardiac action potential. Depolarization, repolarization and their interaction. Circulation 1991 68 1501 1526
Luo CH Rudy Y A Dynamic-Model of the Cardiac Ventricular Action-Potential .1. Simulations of Ionic Currents and Concentration Changes CircRes 1994 74 1071 1096
Qu ZL Garfinkel A An advanced algorithm for solving partial differential equation in cardiac conduction IEEE Trans Biomed Eng 1999 46 1166 1168 10493080 10.1109/10.784149
Shaw RM Rudy Y The vulnerable window for unidirectional block in cardiac tissue: Characterisation and dependence on membrane excitability and intercellular coupling Journal of Cardiovascular Electrophysiology 1995 6 115 131 7780627
Biktashev VN Dissipation of the excitation wave fronts Physical Review Letters 2002 89 168102 12398758 10.1103/PhysRevLett.89.168102
Qu ZL H.S. K A. G Weiss J Effects of Na+ channel and cell coupling abnormalities on vulnerability to re-entry: a simulation study American Journal of Physiology (Heart and Circulatory Physiology) 2004 286 H1310 H1321 14630634 10.1152/ajpheart.00561.2003
Samie FH Berenfeld O Anumono J Mironov S Udassi S Beaumont J Taffet S Pertsov A Jalife J Rectification of the background potassium current. A determinant of rotor dynamics in ventricular fibrillation Circulation Research 2001 89 1216 1223 11739288
Beaumont J Davidenko N Davidenko JM Jalife J Spiral waves in two-dimensional models of ventricular muscle: Formation of a stationary core Biophys J 1998 75 1 14 9649363 10.1016/S0301-4622(98)00194-X
Kneller J Zou R Vigmond E Wang Z Leon LJ Nattel S Cholinergic atrial fibrillation in a computer model of a two-dimensional sheet of canine atrial cells with realistic ionic properties Circulation Research 2002 90 e73 e87 12016272 10.1161/01.RES.0000019783.88094.BA
Wit AL Janse MJ The ventricular arrhythmias of ischaemia and infarction 1993 New York, Futura
Saumarez RC Camm AJ Panagos A Gill JS Stewart JT Belder MAD Simpson IA McKenna WJ Ventricular fibrillation in hypertrophic cardiomyopathy is associated with increased fractionation of paced right ventricular electrograms. Circulation 1992 86 467 474 1638716
Conrath CE Wilders R Coronel R De Bakker JMT Taggart P De Groot JR Opthof T Intercellular coupling through gap junctions masks M cells in the human heart Cardiovascular Research 2004 62
Sampson KJ Henriquez CS Simulation and prediction of functional block in the presence of structural and functional ionic heterogeneity American Journal of Physiology (Heart and Circulatory Physiology) 2001 281 H2597 H2603 11709428
Sampson KJ Henriquez CS Interplay of ionic and structural heterogeneity on functional action potential duration gradients: Implications for arrhythmogenesis Chaos 2002 12 819 828 12779610 10.1063/1.1497735
Qu ZL Garfinkel A Chen PS Weiss JN Mechanisms of discordant alternans and induction of reentry in simulated cardiac tissue Circulation 2000 102 1664 1670 11015345
Euler DE Cardiac alternans: Mechanisms and pathophysiological significance Cardiovascular Research 1999 42 583 590 10533597 10.1016/S0008-6363(99)00011-5
Watanabe M Fenton F Evans SJ Hastings HM Karma A Mechanism for discordant alternans Journal of Cardiovascular Electrophysiology 2001 12 196 206 11232619 10.1046/j.1540-8167.2001.00196.x
Qu ZL Weiss JN Garfinkel A Cardiac electrical restitution properties and stability of reentrant spiral waves: a simulation study Am J Physiol-Heart Circul Physiol 1999 276 H269 H283
Yuuki K Hosoya Y Kubota I Yamaki M Dynamic and not static change in ventricular repolarisation is a substrate of ventricular ischaemia on chronic ischaemic myocardium Cardiovascular Research 2004 63 645 652 15306220 10.1016/j.cardiores.2004.04.017
Xie FG Qu ZL Garfinkel A Weiss JN Electrophysiological heterogeneity and stability of reentry in simulated cardiac tissue Am J Physiol-Heart Circul Physiol 2001 280 H535 H545
Starmer CF Romashko DN Reddy RS Zilberter YI Starobin J Grant AO Krinsky VI Proarrhythmic response to potassium channel blockade. Numerical studies of polymorphic tachyarrhythmias Circulation 1995 92 595 605 7634474
Noble D Rudy Y Models of cardiac ventricular action potentials: iterative interaction between experiment and simulation Philos Trans R Soc Lond Ser A-Math Phys Eng Sci 2001 359 1127 1142
Faber GM Rudy Y Action potential and contractility changes in Na+ (i) overloaded cardiac myocytes: A simulation study Biophys J 2000 78 2392 2404 10777735
Clayton RH Holden AV Effect of regional differences in cardiac cellular electrophysiology in the stability of ventricular arrhythmias: A computational study Physics in Medicine and Biology 2003 48 95 111 12564503 10.1088/0031-9155/48/1/307
Nash MP Panfilov AV Electromechanical model of excitable tissue to study reentrant cardiac arrhythmias Progress in Biophysics & Molecular Biology 2004 85 501 522 15142759 10.1016/j.pbiomolbio.2004.01.016
Rice JJ Winslow RL Dekanski J McVeigh E Model studies of the role of mechano-sensitive currents in the generation of cardiac arrhythmias Journal of Theoretical Biology 1998 190
Garny A Kohl P Mechanical induction of arrhythmias during ventricular repolarization. Modeling cellular mechanisms and their interaction in two dimensions Annals of the New York Academy of Sciences 2004 1015 133 143 15201155 10.1196/annals.1302.011
| 15720712 | PMC550675 | CC BY | 2021-01-04 16:37:34 | no | Biomed Eng Online. 2005 Feb 18; 4:11 | utf-8 | Biomed Eng Online | 2,005 | 10.1186/1475-925X-4-11 | oa_comm |
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-121572535410.1186/1742-4690-2-12EditorialJacov Tal (1940 – 2005): remembrances of a friend Kaufmann Gabriel [email protected] Kuan-Teh [email protected] Department of Biochemistry, Tel Aviv University, Tel Aviv 69978, Israel2 Laboratory of Molecular Microbiology, NIAID, NIH Bethesda, Maryland 20892, USA2005 22 2 2005 2 12 12 22 2 2005 22 2 2005 Copyright © 2005 Kaufmann and Jeang; licensee BioMed Central Ltd.2005Kaufmann and Jeang; 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.
An obituary commemorates the life and works of Jacov Tal.
==== Body
A friend's passing elicits a set of emotions. Collecting and sharing remembrances are steps of closure in bidding farewell. Here, we honor and remember Jacov Tal (Fig. 1) who passed away on February 8th, 2005. At the time of his passing, Jacov was the Head of Virology at Ben-Gurion University Medical School, Israel. In brief, four of us, who befriended Jacov in different capacities, write our remembrances. It is appropriate to recall the words of a colleague who on another occasion upon the passing of a giant in American science remarked, "Well, ghosts can't make men do anything!" Thus, the true reflection of a person is not what (s)he through his/her prestige, wealth, and position makes others do in life, but how (s)he is remembered by others in death. Jacov started as a young retrovirologist in J. M. Bishop and H. Varmus' laboratory; and it is fitting that he is remembered by friends in Retrovirology.
Figure 1 Jacov Tal, circa 2004.
"I remember Jacov as my actual mentor in my first year as a graduate student in the laboratory of Uri Littauer that Jacov had joined a year earlier. Jacov took similar care of other new comers interested in nucleic acids and molecular biology including Hiroshi Inouye, Inder Verma and Jacque Beckman, even to the point of neglecting his own work. Ibelieve that Jacov's selfless involvement in public affairs and later decision to become a virologist date back to a childhood experience; namely, only a ruling elite and diplomatic corpse in a country that Jacov's parents were stationed in, knew about the polio epidemic, considered then an ideological insult.
Jacov Tal studied at the Hebrew University in Jerusalem where he obtained the Bachelor degree in Biochemistry and Microbiology in 1964 and the Master degree in Biochemistry in 1966. He continued his studies at the Weizmann Institute of Science where he obtained his PhD degree in Biochemistry in 1971. Jacov Tal did his postdoctoral training first with Hershel Raskas in Washington University where he investigated adenovirus gene expression and later with Harold Varmus and Michael Bishop at USCF where he studied the relation between the retroviral genome and the genome of its cellular host. After these formative years as a molecular virologist Jacov Tal joined the newly founded Ben Gurion University in the Israeli desert city Beer Sheva where he initiated and led the Medical School's Virology Department." (Gabriel Kaufmann)
"Professor Jacov Tal will be best remembered by the scientific community for his extensive studies of the parvoviruses adeno-associated virus (AAV) and minute virus of mice (MVM). One of his significant achievements was the determination of parameters of site-specific integration of AAV, leading to development of potential vectors for gene therapy. Other important contributions were his insights into MVM's ability to kill cancerous cells, while leaving normal cells unaffected. The students of Professor Tal will also remember him for his dedication to training them to be rigorous and discerning scientists, and his concern for their well being. His colleagues will miss his sharp wit and analytic acumen." (Maureen Friedman)
"Jacov Tal and I collaborated closely to his last days. Together, we found that during embryonic development MVM somehow senses a differentiation signal, and we suggested a relation between this observation and the MVM's anti tumor cell activity. I recall Jacov's 'freshman' enthusiasm in private, and his public posture as one who speaks up his mind without hesitation; standing up against any perceived injustice." (Claytus Davis)
"I met Jacov towards the latter years of his career. In 1993, Jacov came to the US to do a sabbatical in Peter Chiang's laboratory. By and by, he drifted into my laboratory and actually spent the entire year working with me. Jacov by then, already a senior scientist for many years, not unexpectedly struggled heroically (and largely unsuccessfully) at the bench; and certainly it was not his bench-skills that impressed me. What did impress me was Jacov's common sense and his very human and generous attitudes. I recall an incident during Jacov's first week when he had not yet gotten to know all the members of my lab. At that time, there was a tall, curly-haired, darkly-handsome and academically gifted young man, working as a post-doc with me, who had graduated from Yale, obtained his MD degree from Duke, and received house staff and infectious diseases training from the University of Virginia. This person also has four siblings who are MDs. Jacov upon meeting this young man, whispered to me excitedly, 'Now here is a nice Jewish boy who is going to make the mother of a Jewish daughter very happy!' Surprise, surprise...that person turned out not to be Jewish, but a Catholic Lebanese-American of Arabic descent. Afterwards, a sheepish Jacov explained to me that it is very difficult to nearly impossible to tell between an Arab and a Jew in Israel; and as far as he was concerned, it made no difference whether Arab-American or Jewish-American. I was struck by his frankness and openness. In his typically thoughtful and 'dovish' ways, over the next many years, Jacov would email me from Israel his periodic 'roadmaps' for peace in the Middle East, accompanied by his incisive commentaries. I will deeply miss my friend's common sense advice and humor." (Kuan-Teh Jeang)
Acknowledgements
These vignettes honor our memories of Jacov Tal. In aiming for a timely closure, we apologize to Jacov's many other friends and colleagues who would have additional valuable remembrances.
| 15725354 | PMC550676 | CC BY | 2021-01-04 16:36:39 | no | Retrovirology. 2005 Feb 22; 2:12 | utf-8 | Retrovirology | 2,005 | 10.1186/1742-4690-2-12 | oa_comm |
==== Front
BMC Palliat CareBMC Palliative Care1472-684XBioMed Central London 1472-684X-4-11567606910.1186/1472-684X-4-1Research ArticleHome visits by family physicians during the end-of-life: Does patient income or residence play a role? Burge Frederick I [email protected] Beverley [email protected] Grace [email protected] Department of Family Medicine, Dalhousie University, Halifax, NS, Canada2 School of Health Services Administration, Dalhousie University; and Cancer Care Nova Scotia, Halifax, NS, Canada2005 27 1 2005 4 1 1 1 9 2004 27 1 2005 Copyright © 2005 Burge et al; licensee BioMed Central Ltd.2005Burge 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 a growing trend for those with advanced cancer to die at home, there is a corresponding increase in need for primary medical care in that setting. Yet those with lower incomes and in rural regions are often challenged to have their health care needs met. This study examined the association between patient income and residence and the receipt of Family Physician (FP) home visits during the end-of-life among patients with cancer.
Methods
Data Sources/Study Setting. Secondary analysis of linked population-based data. Information pertaining to all patients who died due to lung, colorectal, breast or prostate cancer between 1992 and 1997 (N = 7,212) in the Canadian province of Nova Scotia (NS) was extracted from three administrative health databases and from Statistics Canada census records. Study Design. An ecological measure of income ('neighbourhood' median household income) was developed using census information. Multivariate logistic regression was then used to assess the association of income with the receipt of at least one home visit from a FP among all subjects and by region of residency during the end-of-life. Covariates in the initial multivariate model included patient demographics and alternative health services information such as total days spent as a hospital inpatient. Data Extraction Methods. Encrypted patient health card numbers were used to link all administrative health databases whereas the postal code was the link to Statistics Canada census information.
Results
Over 45% of all subjects received at least one home visit (n = 3265). Compared to those from low income areas, the log odds of receiving at least one home visit was significantly greater among subjects who reside in middle to high income neighbourhoods (for the highest income quintile, adjusted odds ratio [OR] = 1.37, 95% confidence interval [CI] = 1.15, 1.64; for upper-middle income, adjusted OR = 1.19, 95%CI = 1.02, 1.39; for middle income, adjusted OR = 1.33, 95%CI = 1.15, 1.54). This association was found to be primarily associated with residency outside of the largest metropolitan region of the province.
Conclusion
The likelihood of receiving a FP home visit during the end-of-life is associated with neighbourhood income particularly among patients living outside of a major metropolitan region.
==== Body
Background
In the last ten years, more and more of those dying of cancer in Canada are doing so out of hospital[1]. In Nova Scotia, a Canadian Maritime province with a total population of approximately 950,000 people, the proportion of cancer deaths occurring out of hospital has recently grown by fifty per cent[2]. This trend appears to be associated with a number of factors. More individuals with cancer are choosing to remain in the home setting, hospitals have down-sized thus reducing the number of beds available for end-of-life care [3-5], and there is growing availability of services in the community such as homecare and community based palliative care programs [6-8].
As this trend has developed, it has become even more important for patients and families to have access to medical care in the community. Such first line medical care, in Canada, is generally provided by a family physician, usually previously known to the patient. Initially those with terminal illness will obtain their medical care in the office or out patient setting. As they become sicker, however, there will come a time when getting from the home to the clinic office will be too difficult. At such a time, access to home visiting by a physician becomes very important [9-11]. Research has shown that access to a supportive family physician willing to make home visits is associated with a greater likelihood of a home death [12-14], as is access to a comprehensive palliative care program (PCP)[2].
There is evidence that those better off financially live longer and are in better health than poorer individuals[15,16]. Reasons for this have been postulated to include the fact that those with higher incomes have higher educational achievement, better living circumstances, and less risky health behaviours. Such better health may also be due, in part, to better access to services for those with fewer financial barriers. In Canada, such an association should not be the result of inadequacy of health services provided to those with lower incomes as our federal government has committed to the provision of a universal, accessible, comprehensive publicly administered health insurance system which aims to ensure that all residents have access to necessary hospital and physician services on a prepaid basis[17].
We wondered if access to terminal care home visiting by family physicians is better for those with higher incomes even in our publicly-funded health system. Therefore, the purpose of this study was to examine the association between income and the likelihood of receiving home visits by family physicians during the end-of-life among those with cancer. In addition, we examined the effect of regional residency, specifically residency in a major urban centre versus all other regions of NS (which are much smaller in size).
Methods
Data
Data for this retrospective, population-based study were obtained through the linkage of individual-level information extracted from four administrative health databases: (1) the Nova Scotia Cancer Centre Oncology Information System (OPIS) which includes the Nova Scotia Cancer Registry (NSCR) and provincial vital statistics information, (2) the Nova Scotia Medical Services Insurance Physician Services (MSIPS), (3) the Nova Scotia Hospital Admissions / Separations (HAS) file, and (4) the Queen Elizabeth II Health Sciences Center Palliative Care Program (PCP). The MSIPS provides a record of all services provided by physicians to residents of Nova Scotia whereas the HAS contains information relating to all hospital inpatient and outpatient stays and procedures. Because individual-level income information is not available from these sources, the Postal Code Conversion File (PCCF) and 1996 Statistics Canada census data were used to develop an 'ecologic-level' proxy for household income, enumeration area median (EAM) income quintiles or 'neighbourhood income'. These aggregate measures are derived from census information grouped by provincial enumeration areas or 'neighbourhoods'. The resulting quintiles are based on the median income value of each enumeration area. Evidence suggests the use of such proxies in population-based studies is a valid alternative in situations where household level information in not available[18]. It is, however, important to recognize that ecologic measures represent conceptually distinct measures of SES even when individual measures are available [19-21].
Encrypted patient health care numbers were used to link all four administrative health databases whereas the postal code was the link to the PCCF and Statistics Canada census information. Ethics approval for this project was provided by the Queen Elizabeth II Health Sciences Centre research ethics committee.
Subjects
All adults identified on death certificates in the NSCR database as having died due to lung, colorectal, breast or prostate cancer death (International Classification of Diseases, 9th revision [ICD9-CM]) from 1992 to 1997 were included as subjects. These four cancers represent the most common causes of cancer death in Nova Scotia.
Measures
Patient characteristics included sex, date of birth, region of residency (Halifax regional municipality [HRM], all regions outside of HRM), date of initial cancer diagnosis, year of death (1992–1997) cancer cause of death (lung, colorectal, breast, prostate), and neighbourhood income categorized as provincial quintiles (lower, lower middle, middle, upper middle, upper). Almost 40% (39.5%) of Nova Scotia's population resides in the HRM which spans a primarily urban geographical region. Within HRM's boundaries are all of Nova Scotia's major tertiary health care centres, several community hospitals and many specialized care programs. Although regions outside of HRM also encompass many towns with regional and community health care facilities, they span a much larger, diverse, geographical area and may be considered to be relatively more rural than HRM.
Health services information was limited to each subject's 'survival time'. In our end-of-life research we have defined 'end-of-life' as the last 180 days of life, or if of shorter duration, from the date of initial cancer diagnosis to death. This six month time period is commonly used in end-of-life studies [22-24].
Total family physician home visits received during this survival time were counted and, due to the highly skewed distribution evidenced, also dichotomized to represent at least one home visit received or none. Additional health services of interest and potential covariates included the total number of ambulatory visits made to a family physicians by each subject, the total number of visits made by the subject to a specialist, the total number of days spent as a hospital inpatient, receipt of palliative radiotherapy, and whether or not the patient had been admitted to the PCP, a comprehensive palliative care program which has been operating since 1992. As an indirect measure of whether the patient was a resident of a long term care (LTC) facility during the end-of-life, a flag was created indicating whether a patient had received at least one family physician within a LTC centre.
Analysis
Following descriptive statistics and the application of nonparametric tests to assess median differences and cross-tabulations with chi-square analyses for association, regression techniques were employed to estimate the effect of neighbourhood income (EAM income quintile) on the receipt of family physician home visits. Our initial regression analysis retained the total number of home visits as a continuous dependent variable and involved negative binomial regression where differences in survival time were accounted for as an offset variable. This was followed by logistic regression techniques where the probability of receiving at least one FP home visit versus no home visits was assessed. For both forms of regression, unadjusted analyses were followed by multivariate where the initial model included neighbourhood income in addition to sex, year of death, age, cancer cause of death, region of residency, the number of visits made to a medical specialist, the receipt of palliative radiotherapy and admission to the PCP as covariates. To account for the possibility that the subject may not have been 'at home' during their 'survival time' and hence unable to receive a home visit, our LTC residency flag and the total days spent as a hospital inpatient were added to the model.
To control for differences in 'survival time', the number of days from the initial cancer diagnosis date to death were categorized and added to the model. Subsequent modeling involved the sequential elimination of covariates and confounders found to no longer be significantly associated with the receipt of FP home visits in the multivariate model at the p = 0.05 level of significance.
Since our administrative data do not provide the ability to make adjustments for regional differences such as the availability of alternative health services, physician density or community resources, as an alternative, we stratified each analysis by region of residency.
All analyses were performed using SAS software[25].
Results
In total, 7212 adults were identified from death certificate information as having died from lung, colorectal, prostate or breast cancer between 1992 and 1997 in Nova Scotia. Over 94% of these advanced cancer patients had seen a family physician at least once during the end-of-life. In total, home visits accounted for 29% of all ambulatory visits provided by family physicians with 3265 (45.3%) patients receiving at least one FP home visit. The total number of FP home visits received varied widely, from 0 to 89 with an average number of 2 (standard deviation [SD] 4.2) and median of 0. Table 1 records the number of home visits received within each neighbourhood income quintile and by region of residency. Although patients from upper and middle income neighbourhoods across all of Nova Scotia appear to receive a greater number of home visits than those from lower income neighbourhoods, examination by region of residency reveal that this gradient by income is primarily associated with residency outside of HRM. Furthermore, patients residing outside of HRM tend to receive fewer home visits in general (mean 1.75, SD 4.1; median 0, range 0–89) than those living within the metropolitan region (mean 2.53, SD 4.4; median 1, range 0–56). The differences between the mean and median number of visits by region of residency were significant at the p < 0.0001 level. Results were similar in the examination of home visits as a dichotomy. A greater proportion of patients residing in middle to upper income neighbourhoods received at least one home visit than those from lower income areas (Table 2). Again, after controlling for region of residency, this association was found only to apply to those residing in regions outside of HRM (p < 0.0001).
Table 1 Family physician home visits by neighbourhood income quintiles and region of residency
Family physician home visits Mean (standard deviation); Median (range)
Region of residency
Neighbourhood income quintile All adult Nova Scotians Halifax regional municipality All other regions
Lower 1.67 (4.2); 0 (0–89) 2.30 (4.1); 0 (0–30) 1.53 (4.2); 0 (0–89)
Lower middle 1.77 (4.1); 0 (0–69) 2.60 (4.6); 1 (0–31) 1.60 (4.0); 0 (0–69)
Middle 2.25 (4.7); 0 (0–58) 2.96 (5.1); 1 (0–45) 2.02 (4.9); 0 (0–58)
Upper middle 2.15 (4.0); 0 (0–56) 2.44 (4.4); 1 (0–56) 1.95 (3.6); 0 (0–24)
Upper 2.42 (3.9); 1 (0–35) 2.36 (3.7); 1 (0–23) 2.65 (4.6); 0 (0–35)
All 2.00 (4.2); 0 (0–89) 2.53 (4.4); 1 (0–56) 1.75 (4.1); 0 (0–89)
Note: Mean and median visits differ significantly at the p < 0.0001 level
Table 2 Characteristics of Nova Scotians by receipt of home visits during the end-of-life, 1992–1997
Characteristic Home visit receipt; No. (and %) of adult Nova Scotians*
No home visit n = 3947 At least one home visit n = 3265
Neighbourhood income quintile†
Lower 1029 (60.0) 685 (40.0)
Lower middle 866 (58.4) 618 (41.6)
Middle 747 (51.5) 704 (48.5)
Upper middle 676 (52.3) 610 (47.4)
Upper 399 (45.2) 483 (54.8)
Sex†
Male 2323 (56.9) 1763 (43.2)
Female 1624 (52.0) 1502 (48.1)
Year of death‡
1992 643 (56.3) 499 (43.7)
1993 637 (54.8) 525 (45.2)
1994 655 (52.7) 588 (47.3)
1995 648 (52.1) 597 (48.0)
1996 679 (54.7) 562 (45.3)
1997 685 (58.1) 494 (41.9)
Age group§
< 65 years 968 (55.6) 772 (44.4)
65–74 years 1165 (54.2) 986 (45.8)
75+ years 1814 (54.6) 1507 (45.4)
Cancer case of death†
Lung 2094 (57.0) 1580 (43.0)
Colorectal 674 (55.1) 549 (44.9)
Breast 629 (50.6) 614 (49.4)
Prostate 550 (51.3) 522 (48.7)
Survival time†
<61 days 935 (78.4) 258 (21.6)
61–120 days 307 (49.8) 310 (50.2)
121–180+ days 2705 (50.1) 2697 (49.9)
Region of residency†
Halifax regional municipality 1082 (47.2) 1211 (52.8)
All other regions of Nova Scotia 2855 (58.2) 2049 (41.8)
Visit within a long term care center (LTC)†
None 3421 (53.6) 2968 (46.5)
At least one LTC visit 526 (63.9) 297 (36.1)
Specialty visits‡
0–2 1066 (57.5) 789 (42.5)
3–6 1025 (54.4) 859 (45.6)
7–13 736 (54.8) 608 (45.2)
14+ 1120 (52.6) 1009 (47.4)
Total days as hospital inpatient‡
0 568 (55.4) 458 (44.6)
1–12 1247 (55.3) 1007 (44.7)
13–31 1132 (52.0) 1044 (48.0)
32+ 1000 (57.0) 756 (43.1)
Admission to palliative care program†
No 3342 (59.2) 2301 (40.8)
Yes 605 (38.6) 964 (61.4)
Received palliative radiation†
No 3085 (56.8) 2346 (43.2)
Yes 862 (48.4) 919 (51.6)
* Total number of patients by characteristic may vary due to missing values. Proportions are row percentages and may total more than 100 due to rounding.
†Characteristic is associated with receipt of home visit (p < 0.001)
‡Characteristic is associated with receipt of home visits (p < 0.05)
§No significant association demonstrated
Subject characteristics and health service utilization by receipt of at least one home visit are displayed in Table 2. In addition to patients who received at least one FP home visit tending to reside in higher income neighbourhoods, they also were more likely to be female, have a breast cancer cause of death, survived at least 61 days from their initial cancer diagnosis date, did not receive a FP visit within a LTC facility, made more than 14 specialty visits, spent 13–31 days as a hospital inpatient during their survival time, received palliative radiotherapy, and were admitted into the PCP.
Regression results incorporating the total number of home visits using negative binomial regression and those derived from logistic techniques assessing the log odds of receiving at least one home visit compared to none proved similar. Therefore, for ease of presentation, we present the logistic regression results only.
Displayed in Table 3 are the adjusted multivariate logistic regression results examining the effect of 'neighbourhood' income and additional predictors on the receipt of at least one FP home visit during the end-of-life among all advanced cancer patients and by region of residency. Examination of the crude odds ratios (OR) and related confidence intervals (CI) indicate the log odds of receiving at least one home visit was significantly greater among subjects who reside in middle to high income neighbourhoods compared to those from low income. Following adjustments for all other significant predictors retained in the model, this significant association remained, although less strongly. Compared to advanced cancer patients from lower income neighbourhoods, those from upper income neighbourhoods were 37% more likely to receive at least one FP home visit (adjusted odds ratio [OR] 1.37, 95% confidence interval [CI] 1.15, 1.64). Cancer patients from the upper-middle and middle income neighbourhoods were also significantly more likely to have received at least one family physician home visit than those from the lowest income area (for upper-middle income adjusted OR 1.19, 95%CI 1.02, 1.39; for middle income adjusted OR 1.33, 95%CI 1.15, 1.54). However, this association is not experienced equally across the province. Although patients residing in the large metropolitan region of HRM tend to receive more FP home visits in general, the receipt of such visits are not associated with neighbourhood income. In contrast, among patients living outside the HRM, those from upper income neighbourhoods were more than twice as likely to receive a FP home visit than others residing in lower neighbourhood income areas (adjusted OR 2.23; 95%CI 1.63, 3.07).
Table 3 Odds of receiving a home visit by income and other characteristics for Nova Scotia overall and by region
Predictor Nova Scotia overall Halifax Regional Municipality [HRM] All regions outside of HRM
Crude OR Adjusted* OR (and 95% CI) Adjusted* OR (and 95% CI) Adjusted* OR (and 95% CI)
Neighbourhood income quintile
Low 1.0 1.0 (-) 1.0 (-) 1.0 (-)
Lower middle 1.07 1.04 (0.90, 1.21) 1.42 (0.99, 2.03) 0.98 (0.84, 1.16)
Middle 1.42 1.33 (1.15, 1.54) 1.40 (1.00, 1.94) 1.30 (1.10, 1.53)
Upper middle 1.36 1.19 (1.02, 1.39) 1.21 (0.90, 1.64) 1.20 (0.99, 1.44)
Upper 1.82 1.37 (1.15, 1.64) 1.18 (0.88, 1.56) 2.23 (1.63, 3.07)
Survival time
<61 days 1.0 1.0 (-) 1.0 (-) 1.0 (-)
61–120 days 3.66 3.83 (3.08, 4.77) 3.91 (2.62, 5.82) 3.69 (2.81, 4.84)
121–180+ days 3.61 4.04 (3.45, 4.72) 3.64 (2.76, 4.81) 4.21 (3.47, 5.11)
Admission to palliative care program
No 1.0 1.0 (-) 1.0 (-) 1.0 (-)
Yes 2.30 2.25 (1.97, 2.56) 3.05 (2.49, 3.74) 1.47 (1.15, 1.86)
Visit within a long term care center (LTC)
None 1.0 1.0 (-) 1.0 (-) 1.0 (-)
At least one LTC visit 0.65 0.55 (0.46, 0.65) 0.47 (0.34, 0.64) 0.60 (0.48, 0.75)
Age group
< 65 years 1.0 1.0 (-) 1.0 (-) 1.0 (-)
65–74 years 1.06 1.25 (1.09, 1.44) 1.24 (0.97, 1.59) 1.28 (1.08, 1.51)
75+ years 1.04 1.41 (1.24, 1.61) 1.94 (1.52, 2.49) 1.29 (1.09, 1.51)
Total days as a hospital inpatient
0 1.0 1.0 (-) 1.0 (-) 1.0 (-)
1–12 1.0 1.06 (0.90, 1.26) 1.02 (0.76, 1.37) 1.07 (0.87, 1.30)
13–31 1.14 1.07 (0.91, 1.27) 1.04 (0.77, 1.40) 1.08 (0.87, 1.32)
32+ 0.94 0.80 (0.67, 0.95) 0.87 (0.64, 1.20) 0.78 (0.63, 0.97)
Year of death
1992 1.0 1.0 (-) 1.0 (-) 1.0 (-)
1993 1.06 1.05 (0.88, 1.25) 1.07 (0.78, 1.48) 1.02 (0.82, 1.27)
1994 1.16 1.17 (0.98, 1.40) 1.13 (0.81, 1.57) 1.16 (0.94, 1.43)
1995 1.19 1.10 (0.92, 1.31) 0.88 (0.63, 1.22) 1.18 (0.95, 1.45)
1996 1.07 0.94 (0.79, 1.13) 0.76 (0.55, 1.05) 1.02 (0.83, 1.27)
1997 0.93 0.82 (0.69, 0.98) 0.80 (0.57, 1.11) 0.82 (0.66, 1.01)
Sex
Male 1.0 1.0 1.0 (-) 1.0 (-)
Female 1.22 1.15 (1.04, 1.28) 1.24 (1.03, 1.50) 1.11 (0.98, 1.25)
Note: OR = odds ratio, CI = confidence interval
*Adjusted for all other listed predictors
Among all advanced cancer patients, additional factors predictive of receiving at least one FP home visit included a longer length of survival (for 121 to more than 180 days survival: adjusted OR 4.04; 95%CI 3.45, 4.72), admission to the QEII Palliative Care Program (PCP) (adjusted OR 2.25, 95%CI 1.97, 2.56), older age (for those 75 years and older: adjusted OR 1.41; 95%CI 1.24, 1.61) and being female (adjusted OR 1.15; 95%CI 1.04, 1.28). Patients who were in LTC at some point during their survival period (adjusted OR 0.55; 95%CI 0.46, 0.65) and those who spent 32 or more days in hospital compared to none (adjusted OR 0.80; 95%CI 0.67, 0.95) tended to be less likely to receive at least one FP home visit. Over time, the likelihood of receiving a home visit tended to decline. However, after accounting for all other predictors in the model, year of death was not a major factor.
Cancer cause of death, receipt of physician specialty visits and undergoing palliative radiation were not associated with home visit receipt in the final multivariate model.
Discussion
The neighbourhood income of those dying of cancer is associated with the likelihood of receiving a home visit during the end-of-life by a family physician in Nova Scotia. The association found, however, appears to be modified by region of residence for those who died of cancer. It appears that income plays less of a role in predicting home visits by a family physician for those who live in the larger, urban centre of Halifax Regional Municipality. Given the finding that patients followed by the QEII Palliative Care Program are also more likely to receive family physician home visits, we speculate that the urban centre may provide a collaborative 'team-care' advantage to cancer patients. The publicly-funded PCP may act to equalize the opportunity to stay at home and facilitate family physician home involvement in ways rural locations may not be able to. In previous work, income was not associated with location of death but region of residency was[2]. We have also found that those who live in higher income areas tend to use the emergency department less[24].
In the United Kingdom, Aylin and colleagues[26] found, for the general population, those in social class 1 (highest income) received the fewest home visits. Their study also revealed a dose-response relationship in that as one moves to lower income class, the more likely one is to have received a home visit. Our study shows some gradient element, albeit in the opposite direction, but not as clearly. McNiece and Majeed found home visiting rates among patients with highest income to be half that of those with the lowest income[27]. Their study results and ours were adjusted for age and sex, such that the relationships between income, age and sex cannot be confounding the results. Aylin postulates that the reason for greater home visiting among those with lower income may be due to a number of factors including increased morbidity, poorer access to a car, and differing expectations of the services supplied by their general practitioners[26]. Such factors should also hold true for cancer patients. Nevertheless, our findings are opposite to those of Aylin.
We hypothesize that when it comes to routine home visits for brief, episodic illnesses, the home visiting trend may be as Aylin suggests. However, for those who are at home and looking to stay at home with advanced cancer, there are more substantial financial issues driving whether this is likely to happen or not. In Nova Scotia, as in many Canadian settings, there is access to home visiting nurses and some other health professionals through the publicly-funded health system. However, as disease progresses, the ability of a family to support death at home depends on many other factors. These include the presence of a family member who can stay at home, the ability of a family member to manage the medications and symptom assessment along with health professionals, the cost of drugs which are paid for in hospital but not in the home (unless the patient has private health insurance or is 65 or more years old), the cost of equipment in the home, and possibly the cost of additional nursing or personal care workers in the home (variably covered by the public system, and only sometimes covered by private insurance)[17]. All of these factors point to the fact that those with greater income would be more likely to succeed at staying at home[7]. In addition, in more rural settings there is less access to specialized services such as palliative care programs.
The home visits provided by family physicians may therefore be a direct response to the other capacities of patients and families to stay at home rather than being the critically enabling feature. Thus, the home visit is essential but not sufficient and without the rest of the support required, the patient will not be able to stay at home, thereby resulting in fewer home visits.
Another interpretation is that those with lower incomes may actually make different choices about how they wish to receive health care and where they would like to spend their last days. Depending on the study cited, up to 80% of those with advanced cancer wish to spend their last days at home[7,28-30]. Grande and colleagues reported that those who lived in higher socioeconomic areas were more likely to die at home than their counterparts[31]. Sims found that those with cancer from social class IV and V (semi-skilled and unskilled occupations) were under-represented in deaths that occurred in hospices and homes when compared to those in social class I and II (professional occupations and managerial/technical occupations)[32]. All of this may be supported by any one or a combination of factors such as less desire to remain at home, less capacity (financial or otherwise) to remain at home or bias in the delivery of health service by professionals. Our study is the first we know of to show that the number of home visits made by family physicians to those at the end-of-life is also less for these individuals.
Home visiting has long been an element of continuity of care across settings (office, hospital, home, nursing home) provided by family physicians. Some would argue that home visits may be influenced by the geography in which the physician operates daily. As a result, physician travel patterns to and from the office (when home visiting often occurs) may not take them through low income neighbourhoods, thus reducing the likelihood of a visit. The work of Aylin[26] and McNiece[27], however, does not support this.
New initiatives are underway in Canada which may provide more opportunities for the enhanced presence of a range of health professionals in the homes of the dying. In response to the Romanow Commission[33], the federal government of Canada has initiated agreement with the provinces for them to provide coverage for enhanced home-based end-of-life care. In the future, we may see more nurses or nurse-practitioners making home visits as part of the community-based care team along with family physicians[34]. In rural or remote areas where there is a scarcity of family physicians, nurses and nurse practitioners with advanced assessment skills may play an even greater role.
Our study has limitations. As we are using routinely collected data used for administrative and billing purposes there may be biases operating. The data reflect those family physicians who bill for the services they provide. It should also reflect the "shadow-billing" of those on alternate payment mechanisms (estimated to be less than five per cent of family physicians at the time of the study) but in reality, these physicians may have less incentive to capture these fee codes and so may under-report home visiting slightly.
The data file used in this study was originally created for an alternate project looking at health service utilization among patients who died due to lung, colorectal, breast or prostate cancer. We were therefore limited to examining home visits provided to these patients only and are not able to report whether the use of family physician home services among those who died due to all other cancer causes is similar or different.
We are unable to adjust for homecare utilization (data did not exist for the study years), family member caregiving status (no data available) or account for additional insurance coverage (above provincial) which may have covered additional costs associated with drugs, home nursing, home equipment, etc. Our attempt to account for service availability by region is crude. HRM is more homogeneous with respect to services than our combination all other regions outside of HRM; however, the effect evidenced may, therefore, be a conservative estimate.
Conclusions
And so, in conclusion, it appears that even in death those with fewer financial resources may be less likely to achieve the same access to health services as those better off. What does this mean for our care of the dying? We must examine carefully which elements are the cause of this inequality. If there is less desire among those with more financial barriers, we need to examine the origins of these desires. Is it fear of caring for those at home? Is it an established culture of caring to move loved ones to hospital at the end-of-life? These issues need identification if we are to support such families. If there truly are financial barriers to such things as drugs, equipment and personnel then we must redefine health policy to make these more accessible to those with fewer financial resources. And finally, we should attempt to understand whether or not there is any bias operating on the part of health professionals. All of these need further research if we are to ensure patients receive the care where they wish to as they approach death.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
FB and BL participated in the conceptualization and design of the project, the analysis and interpretation of the data, first-drafted the majority of the article, and incorporated co-author comments into the final draft. GJ participated in the interpretation of the data and the drafting and revising of the manuscript. All authors gave approval to this final version.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors wish to thank Natalie Dawson for preparation of the manuscript. Dr. Burge is supported by a Senior Clinical Research Scholar Career Award from the Faculty of Medicine, Dalhousie University. This research was supported by a grant from the Canadian Institutes of Health Research (Grant #44617).
==== Refs
Wilson DM Anderson JC Fainsinger RL al. Social and health care trends influencing palliative care and the location of death in Twentieth-Century Canada, Final NHRDP Report 1998 Edmonton, University of Alberta
Burge F Lawson B Johnston G Trends in the place of death of cancer patients, 1992-1997 C M A J 2003 168 265 270
Tully P Saint-Pierre E Downsizing Canada's hospitals, 1986/87 to 1994/95 Health Reports 1997 8 33 39
Health NSD Transitions in care: Nova Scotia Department of Health Facilities Review 2000
Wilson DM Truman CD Does the availability of hospital beds affect utilization patterns? The case of end-of-life care Health Serv Manage Res 2001 14 229 239 11725590 10.1258/0951484011912735
Nova Scotia Department of Health Home care Nova Scotia: Update 1997 1 14
Rural Palliative Home Care Staff and Consultants A rural palliative home care model: The development and evaluation of an integrated palliative care program in Nova Scotia and Prince Edward Island. A Federal Health Transition Fund Project Report 2001 Nova Scotia, Communications Nova Scotia
Burge F Johnston G Lawson B Dewar R Cummings I Population based trends in referral of the elderly to a comprehensive palliative care program Palliat Med 2002 16 255 256 12047004 10.1191/0269216302pm550xx
Association CM Strengthening the foundation: The role of the physician in primary health care in Canada 1994 Ottawa, Canadian Medical Association 1 20
McWhinney IR Caring for patients with cancer: Family physicians' role Can Fam Physician 1994 40 16 17 8312750
McWhinney IR Stewart MA Home care of dying patients: Family physicians' experience with a palliative care support team Can Fam Physician 1994 40 240 246 7510562
McWhinney IR Bass MJ Orr V Factors associated with location of death (home or hospital) of patients referred to a palliative care team C M A J 1995 152 361 367
Cantwell P Turco S Brenneis C Hanson J Neumann CM Bruera E Predictors of home death in palliative care cancer patients J Palliat Care 2000 16 23 28 10802960
Brazil K Bedard M Willison K Factors associated with home death for individuals who receive home support services: a retrospective cohort study BMC Palliat Care 2002 1 2 11911767 10.1186/1472-684X-1-2
Lynch JW Smith GD Kaplan GA House JS Income inequality and mortality: Importance to health of individual income, psychosocial environment, or material conditions BMJ 2000 320 1200 1204 10784551 10.1136/bmj.320.7243.1200
Adler NE Ostrove JM Socioeconomic status and health: What we know and what we don't Ann NY Acad Sci 1999 896 3 15 10681884
Canada H Canada Health Act: Annual Report 2000-2001 2001 Ottawa (ON), Health Canada
Mustard CA Derksen S Berthelot JM Wolfson M Assessing ecologic proxies for household income: A comparison of household and neighbourhood level income measures in the study of population health status Health & Place 1999 5 157 171 10670997 10.1016/S1353-8292(99)00008-8
Geronimus AT Bound J Neidert IJ On the validity of using census geocode characteristics to proxy individual socioeconomic statistics J Am Stat Assoc 1996 91 529 537
Geronimus AT Bound J Use of census-based aggregate variables to proxy for socioeconomic group: Evidence from national samples Am J Epidemiol 1998 148 475 486 9737560
Kaplan GA People and places: Contrasting perspectives on the association between social class and health Int J Health Serv 1996 26 507 519 8840199
Somogyi-Zalud E Zhong Z Hamel MB Lynn J The use of life-sustaining treatments in hospitalized persons aged 80 and older JAGS 2002 50 930 934 10.1046/j.1532-5415.2002.50222.x
Latimer EA Verrilli D Welch WP Utilization of physician services at the end of life: Differences between the United States and Canada Inquiry 1999 36 90 100 10335314
Burge F Lawson B Johnston G Family physician continuity of care and emergency department use in end-of-life cancer care Med Care 2003 41 992 1001 12886178 10.1097/00005650-200308000-00012
SAS Institute Inc SAS/STAT Version 8.2 1999 Cary (NC), SAS Institute Inc
Aylin P Majeed FA Cook DG Home visiting by general practitioners in England and Wales BMJ 1996 313 207 210 8696199
McNiece R Majeed A Socioeconomic differences in general practice consultation rates in patients aged 65 and over: prospective cohort study BMJ 1999 319 26 28 10390456
Dunlop RJ Davies RJ Hockley JM Preferred versus actual place of death: A hospital palliative care support team experience Palliat Med 1989 3 197 201
Townsend J Frank AO Fermont D Dyer S Karran O Walgrove A Piper M Terminal cancer care and patients' preference for place of death: A prospective study BMJ 1990 301 415 417 1967134
Higginson IJ Sen-Gupta GJA Place of care in advanced cancer: A qualitative systematic literature review of patient preferences J Palliat Med 2000 3 287 300 15859670 10.1089/10966210050085359
Grande GE Addington-Hall JM Todd CJ Place of death and access to home care services: Are certain patient groups at a disadvantage? Soc Sci Med 1998 47 565 579 9690840 10.1016/S0277-9536(98)00115-4
Sims A Radford J Doran K Page H Social class variation in place of cancer death Palliat Med 1997 11 369 373 9472593
Romanow RJ Building on values: The future of health care in Canada 2002 Saskatoon (SK), Commission on the Future of Health Care in Canada
Canada H A 10-year plan to strengthen health care 2004
| 15676069 | PMC551519 | CC BY | 2021-01-04 16:30:52 | no | BMC Palliat Care. 2005 Jan 27; 4:1 | utf-8 | BMC Palliat Care | 2,005 | 10.1186/1472-684X-4-1 | oa_comm |
==== Front
J BiolJournal of Biology1478-58541475-4924BioMed Central London jbiol201572072410.1186/jbiol20Research ArticleNuclear localization is required for Dishevelled function in Wnt/β-catenin signaling Itoh Keiji 1Brott Barbara K 1Bae Gyu-Un 1Ratcliffe Marianne J 1Sokol Sergei Y [email protected] Department of Microbiology and Molecular Genetics, Harvard Medical School, and Beth Israel Deaconess Medical Center, Boston, MA 02215, USA2 Current address: Department of Molecular, Cell and Developmental Biology, Mount Sinai School of Medicine, Box 1020, One Gustave L. Levy Place, New York, NY 10029, USA2005 15 2 2005 4 1 3 3 29 6 2004 30 11 2004 22 12 2004 Copyright © 2005 Itoh 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
Dishevelled (Dsh) is a key component of multiple signaling pathways that are initiated by Wnt secreted ligands and Frizzled receptors during embryonic development. Although Dsh has been detected in a number of cellular compartments, the importance of its subcellular distribution for signaling remains to be determined.
Results
We report that Dsh protein accumulates in cell nuclei when Xenopus embryonic explants or mammalian cells are incubated with inhibitors of nuclear export or when a specific nuclear-export signal (NES) in Dsh is disrupted by mutagenesis. Dsh protein with a mutated NES, while predominantly nuclear, remains fully active in its ability to stimulate canonical Wnt signaling. Conversely, point mutations in conserved amino-acid residues that are essential for the nuclear localization of Dsh impair the ability of Dsh to activate downstream targets of Wnt signaling. When these conserved residues of Dsh are replaced with an unrelated SV40 nuclear localization signal, full Dsh activity is restored. Consistent with a signaling function for Dsh in the nucleus, treatment of cultured mammalian cells with medium containing Wnt3a results in nuclear accumulation of endogenous Dsh protein.
Conclusions
These findings suggest that nuclear localization of Dsh is required for its function in the canonical Wnt/β-catenin signaling pathway. We discuss the relevance of these findings to existing models of Wnt signal transduction to the nucleus.
==== Body
Background
The specification of cell fates during embryonic development frequently depends on inductive interactions, which involve transmission of extracellular signals from the cell surface to the nucleus. In the transforming growth factor β (TGFβ) signal transduction pathway, Smad proteins that are initially associated with TGFβ receptors move to the nucleus to regulate target genes [1]. Another example of a direct link between the cell surface and the nucleus during embryonic development is the proteolytic cleavage and nuclear translocation of the cytoplasmic fragment of the Notch receptor [2]. In contrast, multiple steps appear to be required for a Wnt signal to reach the nucleus. In this molecular pathway, signals from Frizzled receptors are transduced to Dishevelled (Dsh), followed by inactivation of the β-catenin degradation complex that includes the adenomatous polyposis coli protein (APC), Axin and glycogen synthase kinase 3 (GSK3) [3,4]. Stabilization of β-catenin is thought to promote its association with members of the T-cell factor (Tcf) transcription factor family in the nucleus, resulting in the activation of target genes [5,6]. As well as the canonical β-catenin-dependent pathway, Frizzled receptors also activate small GTPases of the Rho family, protein kinase C and Jun-N-terminal kinases (JNKs) to regulate planar cell polarity in Drosophila and convergent extension cell movements and tissue separation in Xenopus [7-12]. Thus, the Wnt/Frizzled pathway serves as a model for molecular target selection during signal transduction.
Dsh is a common intracellular mediator of several pathways activated by Frizzled receptors and is composed of three conserved regions that are known as the DIX, PDZ and DEP domains [13]. Different domains of Dsh are engaged in specific interactions with different proteins, thereby leading to distinct signaling outcomes [13]. Daam, a formin-related protein, promotes RhoA activation by Dsh [9], whereas Frodo, another Dsh-binding protein, is required for Wnt/β-catenin signaling in the nucleus [14]. These interactions may take place in various cellular compartments, linking specific activities of Dsh to its distribution inside the cell. Dsh is found in a complex with microtubules and with the actin cytoskeleton [15-17]. Dsh is also associated with cytoplasmic lipid vesicles, and this localization was shown to require the DIX domain [7,16,18]. Overexpressed Frizzled receptors can recruit Dsh to the cell membrane in Xenopus ectoderm, and this redistribution requires the DEP domain [7,18,19]. The DIX domain is essential for the Wnt/β-catenin pathway, whereas the DEP domain plays a role in the planar cell polarity pathway [7,8,16,18,20,21]. Thus, the specific subcellular localization of Dsh may be crucial for local signaling events.
The current study was based on our initial observation that a Dsh construct lacking the carboxy-terminal DEP domain was found in cell nuclei. We have now identified a nuclear export signal in the deleted region and also discovered that Dsh proteins accumulate in the nuclei of Xenopus ectodermal cells and mammalian cells upon inhibition of nuclear export. Dsh also accumulated in the nuclei after stimulation of mammalian cells with Wnt3a-containing culture medium. By analyzing various mutant Dsh constructs in Xenopus ectoderm, we show that the signals responsible for Dsh nuclear localization reside in a novel domain and that the nuclear translocation of Dsh is essential for its ability to activate Wnt/β-catenin signaling.
Results and discussion
A nuclear export signal in Dsh is responsible for the cytoplasmic localization of Dsh
We studied the subcellular distribution of fusions of Dsh with green fluorescent protein (GFP) in Xenopus ectodermal cells. In contrast to Dsh-GFP, which is localized in punctate structures within the cytoplasm [7,18], the Ds2 construct, lacking the carboxy-terminal region, accumulates in the nucleus (Figure 1a–c), indicating that the carboxyl terminus contains sequences for nuclear export. Indeed, we found a potential leucine-rich nuclear export signal (NES) in Dsh at positions 510–515, corresponding to the conserved consensus M/LxxLxL (single letter amino-acid code, where x is any amino acid) [22,23]. When leucines 513 and 515 in this putative NES were substituted with alanines, the mutated Dsh fusion construct, DsNESm, was localized predominantly in the nucleus (Figure 1a,d), demonstrating that the sequence is a functional nuclear export signal.
To examine whether inhibition of nuclear export abrogates Dsh activity, we compared the abilities of DsNESm and wild-type Dsh-GFP to induce secondary axes in frog embryos. Although the molecular mechanism operating during axis induction remains to be elucidated, this assay faithfully reflects the biological activity of Dsh in the canonical Wnt/β-catenin pathway [14,16,18,24]. DsNESm, which was expressed at similar levels to the wild-type Dsh-GFP (data not shown), induced secondary axes at least as efficiently as Dsh-GFP (Table 1). Induced axes contained pronounced head structures with eyes and cement glands (Figure 1e–g). These results suggest that Dsh may function in the nucleus to trigger dorsal axial development.
Nuclear localization of Dsh in cells treated with nuclear export inhibitors
Accumulation of DsNESm in the nucleus implies that the wild-type Dsh shuttles between the nucleus and the cytoplasm. We therefore studied the subcellular distribution of Dsh in Xenopus embryonic cells under conditions in which nuclear export is blocked. When ectodermal cells expressing Dsh-GFP were incubated with N-ethylmaleimide (NEM), an inhibitor of the nuclear export receptor CRM1/exportin [25,26], Dsh-GFP was detected predominantly in the nucleus, compared to the punctate cytoplasmic pattern of Dsh-GFP in untreated cells (Figure 2a,b). This effect was specific to full-length Dsh-GFP, as Ds3, a Dsh construct that lacks 48 amino acids adjacent to the PDZ domain (Figure 1a), did not accumulate in the nucleus after NEM treatment (Figure 2e,f). The nuclear retention of Dsh-GFP was also observed using leptomycin B (LMB), another inhibitor of CRM1-dependent nuclear export [22,23] (Figure 2c,d). These results indicate that Dsh shuttles between the cytoplasm and the nucleus, and that its abundance in the cytoplasm is due to highly efficient nuclear export.
To ensure that the Dsh-GFP fusion behaves similarly to the endogenous Dsh protein, we examined the localization of endogenous Dvl2, a mammalian homolog of Dsh, in human and rat tissue culture cells. Human embryonic kidney (HEK) 293 cells treated with LMB accumulated Dvl2 in the nucleus, contrasting with the cytoplasmic localization of Dvl2 in untreated cells (Figure 3a–c). We also evaluated the subcellular localization of endogenous Dvl2 in Rat-1 fibroblasts, which are known to respond to Wnt signaling. Fractionation of cells into nuclear and cytoplasmic protein pools revealed only a small amount of endogenous Dvl2 in intact nuclei, whereas after NEM treatment, Dvl2 was localized predominantly in the nuclear fraction (Figure 3d). The efficiency of subcellular fractionation was controlled for by staining with antibodies to glyceraldehyde phosphate dehydrogenase (GAPDH) and nuclear lamins. These proteins remained exclusively cytoplasmic or nuclear, respectively, in both untreated and NEM-treated cells (Figure 3d). Thus, our data show that Dsh translocates into the nucleus and is actively exported into the cytoplasm of both Xenopus ectodermal cells and mammalian fibroblasts.
Identification of sequences responsible for Dsh nuclear localization
To identify specific amino-acid sequences that direct the transport of Dsh to the nucleus, we studied the subcellular distribution of mutated Dsh-GFP fusion constructs (Figure 4a). The removal of the DIX and PDZ domains (Ds1) did not eliminate nuclear translocation in response to NEM or LMB (Figure 4a–d), indicating that these two domains are not required for the nuclear import. Similarly, the DEP domain is not required for Dsh nuclear localization (Ds2; Figure 1a,c). Comparison of Ds1 and Ds2 (see Figure 4a), both capable of nuclear localization, reveals a short stretch of shared amino acids located between the PDZ and DEP domains. Strikingly, the removal of just this 48 amino-acid region abrogated nuclear import of Dsh in the presence of NEM or LMB (Ds3; Figures 2e,f and 4a). Together these experiments identify amino acids 333–381 as the region required for nuclear localization of Dsh.
Although this short sequence is highly conserved in all Dsh homologs from Hydra to humans (Figure 4j), it does not bear detectable similarity to nuclear localization signals characterized in other proteins [27]. This sequence may interact directly with components of the nuclear import machinery or bind to a protein that itself binds a karyopherin/importin and mediates the nuclear import of Dsh by a piggyback mechanism. Interestingly, this region overlaps a novel proline-rich domain identified by mutational analysis of Dsh in Drosophila [28]. To define further the specific amino acids necessary for nuclear localization, a panel of Dsh constructs with point mutations spanning the conserved region was examined (data not shown). Nuclear import was eliminated with the substitution of three amino acids, converting IVLT into AVGA (DsNLSm; Figure 4a,e–g,j), indicating that these three amino acids are critical.
Dsh nuclear translocation is crucial for its function in the β-catenin pathway
To determine whether nuclear localization of Dsh is required for its activity, we compared the abilities of DsNLSm and wild-type Dsh to induce secondary axes in frog embryos. We also assessed activation of a luciferase reporter construct for Siamois [29], an immediate target of Wnt/β-catenin signaling. DsNLSm had impaired ability to induce secondary axes and to activate the Siamois reporter when compared with wild-type Dsh (Figure 5a,b; Table 1). Furthermore, DsNLSm failed to stabilize β-catenin (Figure 5c). This difference was not due to differences in protein expression, as both constructs were present in embryo lysates at similar levels (Figure 5c). Thus, these findings indicate that the nuclear localization of Dsh is critical for its functional activity in the β-catenin pathway.
Not only was the function of DsNLSm in the β-catenin pathway impaired, but we found that this construct behaved as a dominant inhibitor of Wnt signaling and prevented the activation of the Siamois reporter by Xwnt3a and Xwnt8 RNAs (Figure 6a,b). Consistent with these observations, another construct lacking the region responsible for the nuclear localization (Ds3; see Figure 4a) also suppressed Wnt signaling (Figure 6b). Despite these inhibitory properties, dorsally injected DsNLSm RNA, like Xdd1, a dominant negative deletion mutant [24], did not suppress primary axis formation (data not shown).
Impaired activity of the DsNLSm construct may be due to its inability to translocate to the nucleus, or due to a coincidental elimination of a binding site for an essential cofactor that functions together with Dsh in the cytoplasm. To exclude the latter possibility, the IVLT sequence of Dsh NLS was replaced with KKKRK, an unrelated NLS from SV40 T antigen [27]. This construct, DsSNLS, relocated to the nucleus even in the absence of nuclear export inhibitors (Figure 4a,i). Notably, all activities of wild-type Dsh, including induction of complete secondary axes, activation of the Siamois promoter and β-catenin stabilization were significantly restored in DsSNLS (Figure 5a–c; Table 1). In contrast to DsNLSm, DsSNLS did not inhibit the ability of Wnt ligands to activate pSia-Luc (Figure 6b), consistent with its being a positive regulator of the Wnt pathway. We note that the signaling activity of DsSNLS was not enhanced compared to wild-type Dsh, suggesting that the rate of the nuclear translocation of Dsh rather than its steady state levels in the nucleus is critical for target gene activation. It is also possible that other nuclear components, rather than Dsh, become rate-limiting for signaling. Overall, the simplest interpretation of our data is that the nuclear import of Dsh is essential for its activity.
We next examined the ability of DsNLSm to bind critical Wnt signaling components, such as casein kinase 1ε (CK1ε), a positive regulator of the β-catenin pathway [30,31], or Axin, a negative regulator [20,32-36], both of which are known to bind Dsh. Both DsSNLS, enriched in the nucleus, and DsNLSm and Ds3, which do not enter the nucleus, bound CK1ε and XARP, a Xenopus Axin-related protein [20] (Figure 7). Thus, these mutated Dsh constructs retain the ability to associate with critical components of the Wnt/β-catenin pathway, arguing that defective nuclear translocation of DsNLSm is likely to be responsible for its inability to activate β-catenin signaling.
Suppression of Dsh nuclear import does not affect noncanonical signaling
Besides the β-catenin pathway, Dsh also functions in a planar cell polarity (PCP) pathway, which involves Rho GTPase and JNK and controls morphogenetic movements in early embryos [8,9,37-39]. We asked whether mutations in DsNLSm influence the β-catenin pathway exclusively or affect the PCP pathway as well. First, we observed that both Dsh-GFP and DsNLSm-GFP were efficiently recruited to the cell membrane by overexpressed Xfz8, a Frizzled family member [40] (Figure 8a). As Dsh relocalization to the cell membrane in response to Frizzled is associated with its ability to signal in the PCP pathway [7,8], this observation suggests that DsNLSm can respond to Frizzled signaling independent of β-catenin.
In Xenopus, the PCP pathway involving Dsh is implicated in the control of convergent extension movements [24,41,42]. Overexpression of the Xdd1 deletion mutant leads to the development of short embryos when expressed in dorsal marginal cells ([24]; Figure 8b). Severe convergent extension defects (Figure 8b) were observed in 22%, and mild defects were observed in 28% of the embryos injected with Xdd1 RNA (N = 35). In contrast, only mild morphogenetic defects were observed in embryos coinjected with Dsh (15%; N = 40) or DsNLSm RNA (18%; N = 39), indicating that both Dsh and DsNLSm partially rescued the effect of Xdd1. This indicates that DsNLSm is active in noncanonical PCP-like signaling. We also examined whether DsNLSm activates c-Jun N-terminal kinase (JNK), which is thought to function downstream of Dsh in the PCP pathway [8,37-39]. Both DsNLSm and Dsh activated JNK at equivalent levels (Figure 8c), suggesting that nuclear localization of Dsh is not required for its function in noncanonical signaling.
Nuclear accumulation of Dsh following Wnt3a stimulation
Our findings are consistent with a scenario in which Wnt signaling may cause nuclear translocation of Dsh followed by formation of a stable β-catenin/Tcf3 complex and transcriptional activation of target genes. In support of this hypothesis, Dsh was reported to move to the nucleus in response to Wnt signaling in primary embryonic kidney cells [17]. In Rat-1 cells, we did not detect a significant change in Dsh distribution in response to Wnt signals (data not shown), possibly due to highly efficient nuclear export of Dsh in these cells. But immunofluorescence staining for Dvl2 revealed the nuclear accumulation of the protein in HEK293 and MCF7 cells after 3–6 h stimulation with Wnt3a-containing medium (Figure 9a, and data not shown). The effect was quantified by measuring nuclear to cytoplasmic (N/C) ratios of fluorescence intensity. The N/C ratio averaged 28% after 6 h treatment with the control medium, but increased to 91% after stimulation with Wnt3a-conditioned medium (Figure 9b). These observations are consistent with the view that Dsh regulates Wnt-dependent gene targets in the nucleus.
A role for Dsh in the nucleus
In the current view, Wnt signaling causes inactivation of the β-catenin degradation complex, leading to stabilization and nuclear translocation of β-catenin [3]. Given that Dsh is genetically upstream of the β-catenin degradation complex [3,4] and that β-catenin degradation is thought to occur in the cytoplasm [43], Dsh nuclear import is unexpected. Nevertheless, our data demonstrate that Dsh shuttles between the cytoplasm and the nucleus and that its presence in the nucleus is critical for signaling. One explanation of these results is that β-catenin degradation may occur in the nucleus. Consistent with this possibility, APC, Axin and GSK3, components of the β-catenin degradation complex, have also recently been found to shuttle between the cytoplasm and the nucleus [22,23,44-47]. Moreover, Frat/GBP, a positive regulator of β-catenin, has been reported to expel GSK3 from the nucleus [47]. We show that the ability of Dsh constructs to enter the nucleus correlates with their ability to stabilize β-catenin (Figure 5c). These observations indicate that Wnt/β-catenin signaling may depend on the nuclear localization of pathway components.
Alternatively, nuclear localization of Dsh may affect β-catenin stability indirectly, by regulating protein interactions that sequester β-catenin in the nucleus, thereby preventing its cytoplasmic degradation [48]. Although we did not detect a significant change in nuclear import of β-catenin-GFP in Xenopus ectoderm cells overexpressing Dsh (data not shown), this process may be cell-context-dependent. On the other hand, we recently showed that Frodo, a nuclear Dsh-interacting protein, associates with Tcf3 and influences Tcf3-dependent transcription [49]. It is thus possible that Frodo links Tcf3 and Dsh to regulate Wnt target genes. Future studies should examine molecular components critical for the nuclear function of Dsh.
Materials and methods
DNA constructs
GFP-tagged Dsh constructs were all derived from the DshGFP-RN3 plasmid that encodes the Xdsh protein fused at amino acid 724 to the first amino acid of GFP (Figures 1a, 4a). Ds1 lacks the first 332 amino-terminal amino acids. Ds2 is the carboxy-terminal deletion of Xdsh, starting with amino acid 383. Ds3 lacks amino acids 334–381. In DsNLSm, the IVLT residues at positions 334–337 were replaced with AVGA, whereas in DsSNLS the same region is replaced with KKKRK, the SV40 T antigen NLS [27]. In DsNESm, L513 and L515 were substituted for alanines.
To generate these constructs, DshGFP-pRN3 was used as a template. The in-frame deletion in Ds3 was made by PCR. Other GFP fusion constructs were synthesized with specific primers and PfuI DNA polymerase followed by DpnI digestion of the template [50]. The following primers were used: 5'-GTCCATAAACCGGGGCCCGCAGTCGGCGCCGTGGCCAAATGCTGG-3' for DsNLSm; 5'-ACACTAGGCCGCAGAATGCCCATTGTCCTGACCGTG-3' for Ds1; 5'-TCCATAAACCGGGGCCAAAGAAGAAGCGAAAGGTGGCCAAATGCTGGGA-3' for DsSNLS; 5'-TTCCCAGTGTACCCCGGGGCCATGGTGAGCAAGGGC-3' for Ds2, and 5'-GAGAACTATGACCAACGCTAGCGCGAATGACAACGATGGAT-3' for DsNESm. All constructs were verified by sequencing. Myc-tagged Dsh mutant constructs were made by replacing mutated regions with corresponding regions of Myc-Dsh [24]. Cloning details are available as an Additional data file with the online version of this article.
Embryo culture, axis-induction assay and axis-extension assay
In vitro fertilization, culture and microinjections of Xenopus eggs were essentially as described previously [24]. Stages were determined according to Nieuwkoop and Faber [51]. Axis induction was carried out by injecting mRNAs encoding different Dsh constructs (1 ng) into a single vegetal ventral blastomere at the 4–8-cell stage and assessed when the injected embryos reached stage 36–40. To monitor axis extension defects, 0.6 ng of Xdd1 RNA was injected alone or with 2 ng of Dsh or DsNLSm RNA into two dorsovegetal blastomeres of 4-cell embryos and the injected embryos were allowed to develop until sibling embryos reached stage 32.
GFP fluorescence and luciferase assay
For subcellular localization of Dsh-GFP constructs, mRNAs were injected into the animal pole region of 2–4-cell embryos. Animal cap explants were dissected at stages 9–10.5, incubated for 60 min in 10 mM N-ethylmaleimide (NEM; Sigma, St Louis USA) in 0.8 × MMR (Marc's Modified Ringer's solution, 1 × MMR: 100 mM NaCl, 2 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 5 mM HEPES, pH 7.4), or in control (0.8 × MMR), then fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 30–45 min, washed three times in PBS, and mounted in 70% glycerol, 30% PBS containing 25 mg/ml of diazabicyclo(2,2,2)-octane (Sigma). Leptomycin B was used at 50 ng/ml in low-calcium medium (76 mM NaCl, 1.4 mM KCl, 0.2 mM CaCl2, 0.1 mM MgCl2, 0.5 mM Hepes, 1.2 mM sodium phosphate, (pH 7.5), 0.6 mM NaHCO3 and 0.06 mM EDTA) for one hour prior to fixation. In some experiments, nuclei were stained by addition of 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI) to the final PBS wash. For membrane localization studies, Xfz8 RNA was coinjected with RNAs encoding the Dsh constructs in the animal-pole region; animal-cap explants were dissected at stage 9–9.5 and mounted for observation. Fluorescence was visualized using a Zeiss Axiophot microscope.
For luciferase assays, pSiaLuc reporter plasmid (20–40 pg) was coinjected with mRNAs encoding Xwnt3a [52] or Xwnt8 [53] and different Dsh constructs into one or two animal-ventral blastomeres or into one ventral-vegetal blastomere at the 4–8-cell stage. Luciferase activity was measured as described [29].
Tissue culture, immunocytochemistry and subcellular fractionation
Rat-1 fibroblasts, human embryonic kidney (HEK) 293 cells and MCF7 human breast carcinoma cells were cultured in 1 × Dulbecco's Modified Eagle Medium (DMEM; Gibco/Invitro-gen, Carlsbad, USA) supplemented with 10% fetal calf serum and 5 μg/ml gentamicin. Conditioned medium was prepared from L cells stably transfected with Wnt3a as described [54], with the medium from untransfected L cells serving as a control.
For immunocytochemistry, HEK293 cells were treated with 50 ng/ml LMB for 14 h while MCF7 cells were treated with Wnt3a or control conditioned medium for 1, 3, 6 or 8 h. Cells were fixed with 4% paraformaldehyde, immersed in methanol, and incubated with anti-Dvl2 antibodies and then Cy3-conjugated anti-rabbit IgG. Nuclei were stained by addition of 1 μg/ml DAPI as described for animal-cap cells. Fluorescence was observed under the Zeiss Axiophot microscope; 10–15 cells from each group were randomly picked up for measurement of the nuclear and cytoplasmic staining intensity using Image-Gauge software (Fuji Film, Tokyo, Japan).
For subcellular fractionation, confluent cultures of Rat-1 cells were harvested by scraping plates and resuspended in hypotonic lysis buffer containing 1 mM EGTA, 1 mM EDTA, 2 mM MgCl2, 10 mM KCl, 1 mM DTT, 10 mM β-glycerophosphate, 1 mM sodium orthovanadate, 1 μg/ml leupeptin, 1 μg/ml aprotinin, and 1 μg/ml pepstatin. Cells were swollen for 30 min, and broken open with 25 strokes in a tight fitting Dounce homogenizer. Lysates were layered into tubes containing 1 M sucrose in hypotonic lysis buffer, and spun at 1600 × g for 10 min. Supernatant remaining above the sucrose cushion was used as the cytoplasmic fraction. The pellet, containing nuclei, was resuspended in an equivalent volume of hypotonic lysis buffer.
Immunoprecipitation and western blotting
Immunoprecipitation and western analysis were carried out with cell and embryo lysates as described [14]. To prepare embryo lysates at stage 10+, four animal blastomeres of 4–8-cell embryos were injected with RNAs encoding different forms of Dsh, ΔRGS-Axin [32], Flag-β-catenin [55], CK1ε [30] and HA-XARP [20]. To generate anti-Xdsh polyclonal antibodies, rabbits were immunized with a carboxy-terminal half of Xdsh (amino acids 301–736) fused to GST. First, GST beads were used for purification of anti-GST antibodies. Subsequently anti-Xdsh antibodies were affinity-purified on GST-Xdsh (301–736) beads. Polyclonal anti-Dvl2 antibody was generated in rabbits and affinity-purified on PVDF membrane blotted with human Dvl2 (79–249) [56]. A small aliquot of anti-human Dvl2 was obtained from M. Snyder (Yale University, New Haven, USA). Anti-GAPDH antibody was a gift from A. Stuart-Tilley and S. Alper (Beth Israel Deaconess Medical Center, Boston, USA), anti-lamin antibody was from F. McKeon (Harvard Medical School, Boston, USA). Anti-β-tubulin antibodies were from BioGenex (San Ramon, USA), anti-Flag M2 antibody was from Sigma and anti-CK1ε antibodies were from BD Biosciences (Palo Alto, USA). Anti-Myc and anti-HA monoclonal antibodies are hybridoma supernatants of 9E10 and 12CA5 cells (Roche Applied Science, Indianapolis, USA).
JNK assay
Four-cell embryos were injected with 4 ng Dsh or DsNLSm RNA into four animal blastomeres. Embryo lysates were prepared at stage 10.5 and in vitro kinase assays were carried out essentially as described [57], except that phosphorylated c-Jun-GST was detected with anti-phospho-c-Jun-specific antibodies (Cell Signaling Technology, Beverly, USA) by western blotting rather than with autoradiography.
Additional data files
The following is provided as an additional data file with the online version of this article. Additional data file 1, containing cloning details of Dsh mutant constructs.
Supplementary Material
Additional data file 1
Cloning details of Dsh mutant constructs
Click here for additional data file
Acknowledgements
We thank S. Alper, F. McKeon and M. Snyder for antibodies, and X. He, F. Costantini and J. Graff for plasmids, J. Kitajewsky for Rat-1 cells, R. Nusse for L cells transfected with Wnt3a, and J. Martinez, Y. Yoneda and M. Yoshida for leptomycin B. We also thank V. Krupnik and M. Lisovsky for help with the generation of anti-Xdsh and anti-Dvl2 antibodies. We are grateful to J. Green, V. Krupnik, B. Neel, N. Perrimon and members of this laboratory for reading of the manuscript and useful discussions. This work was supported by NIH grants to S.Y.S.
Figures and Tables
Figure 1 Nuclear export of Dsh is not critical for its activity. (a) The Dsh constructs used to analyze nuclear export. (b-d) RNAs encoding Dsh-GFP, Ds2 and DsNESm (0.5 ng each) were injected into two animal blastomeres of 4–8-cell embryos. Animal-cap explants were excised at stage 10, fixed and examined for GFP fluorescence. (b) Wild type Dsh-GFP localized in punctate structures of the cytoplasm, whereas (c) Ds2 and (d) DsNESm accumulated in the nucleus of animal pole cells. (e,f) One ventral vegetal blastomere of 8-cell embryos was injected with 1 ng Dsh-GFP or DsNESm RNA as indicated. Complete secondary axes were induced in both cases. (g) Uninjected sibling embryos.
Figure 2 Accumulation of Dsh in the nucleus in the absence of nuclear export. (a-d) Dsh-GFP RNA (0.7 ng) was injected into two animal blastomeres of 4–8 cell embryos. Animal caps were excised at stage 10 and then left (a) untreated or (b) treated with 10 mM NEM or (c,d) 50 ng/ml leptomycin B (LMB), fixed and examined for GFP fluorescence. (a) Dsh-GFP is mainly localized to vesicular structures in the cytoplasm. In the presence of (b) NEM or (c) LMB, Dsh-GFP accumulates in the nucleus, as supported by (d) DAPI staining of nuclei in the same field as in (c). Nuclear staining is marked by arrowheads (c,d). (e,f) The Ds3 construct, lacking amino acids 334–381, remained in the cytoplasm in the (e) absence or (f) presence of NEM.
Figure 3 Endogenous Dsh shuttles between the cytoplasm and nucleus. Immunofluorescent staining of HEK293 cells with anti-Dvl2 antibodies reveals different subcellular localization of Dvl2 (a) without or (b) with LMB treatment. (c) DAPI staining shows the location of nuclei in the same field as (b); the arrowheads indicate corresponding nuclei in (b) and (c). (d) Distribution of endogenous Dvl2 recognized by anti-Dvl2 antibodies in the nuclear and the cytoplasmic fractions of Rat-1 fibroblasts. In the absence of NEM, Dvl2 is localized mainly in the cytoplasm (C), while after NEM treatment Dvl2 is exclusively localized in the nuclei (N). W, whole cell lysate. Antibodies to lamin and GAPDH show the separation of the nuclear and cytoplasmic fractions.
Figure 4 Mapping nuclear localization signals in Dsh. (a) The Dsh constructs used to study nuclear transport and their localization to the nucleus after NEM or LMB treatment; the DIX, PDZ and DEP domains are shown as in Figure 1a; B is the basic region and nd denotes not done. (b-i) Subcellular localization of Dsh-GFP constructs in the absence or presence of NEM or LMB. Embryos were injected with 0.5 ng of each mRNA, and GFP analysis was carried out as in Figure 1b-d. (b-d) Ds1, (e-g) DsNLSm, (h) Dsh, (i) DsSNLS. (b,e,i) no NEM treatment; (c,f) after NEM treatment; (d,g,h) after LMB treatment. (j) Comparison of conserved amino-acid sequences that are required for Dsh nuclear localization; X denotes the Xenopus protein, m the mouse and h the human. Amino-acid residues mutated in DsNLSm are indicated by asterisks.
Figure 5 Activation of the Wnt/β-catenin pathway requires nuclear localization of Dsh. (a) Axis-inducing activity of Dsh constructs. One ventral vegetal blastomere of 8-cell embryos was injected with 1 ng Dsh-GFP, DsNLSm, or DsSNLS mRNA as indicated. Uninjected sibling embryos are also shown. (b) Activation of the Siamois reporter gene. The reporter -833pSia-Luc plasmid (20 pg) was coinjected with Dsh-GFP, DsNLSm or DsSNLS mRNA (0.5 ng each) into a single animal ventral blastomere of 8-cell embryos. Injected embryos were lysed at stage 10+ for luciferase activity determination. Results are shown in relative light units as the mean +/- standard deviation from triplicate samples. (c) Requirement for Dsh NLS for the stabilization of β-catenin. Flag-β-catenin mRNA (0.4 ng) was coinjected with Dsh, DsNLSm, DsSNLS or ΔRGS-Axin mRNA (2 ng each) into four animal blastomeres of 4–8-cell embryos. Levels of β-catenin and Dsh constructs were assessed in stage 10 embryo lysates with anti-Flag antibodies and anti-Xdsh antibodies; β-tubulin serves as a loading control. Dsh and DsSNLS, but not DsNLSm, are able to stabilize β-catenin. ΔRGS-Axin was used as a control for an activator of the Wnt pathway.
Figure 6 Dominant inhibition of Wnt-dependent transcription by Dsh mutants. Eight-cell embryos were injected (a) in one animal ventral blastomere or (b) in one vegetal ventral blastomere with -833pSia-Luc DNA (20 pg), mRNAs encoding Xwnt3a (5 pg) or Xwnt8 (2 pg), and Dsh-GFP, DsNLSm, Ds3 or DsSNLS mRNA (0.5 ng) as indicated. Luciferase activity was measured as described in Figure 5b.
Figure 7 Dsh mutants retain the ability to bind CK1ε and XARP. Four-cell embryos were injected in four sites in the animal hemisphere with CK1ε, HA-XARP, Myc-tagged Dsh, DsNLSm, Ds3 or DsSNLS RNA alone (2 ng each) or in combinations as indicated. The embryonic lysates were collected at stage 10.5 for immunoprecipitation with anti-Myc antibodies. Co-immunoprecipitated (a) CK1ε or (b) HA-XARP was probed with anti-CK1ε or anti-HA antibodies; β-tubulin served as a loading control.
Figure 8 DsNLSm, defective in the β-catenin pathway, is active in noncanonical signaling. (a) Fz8-dependent recruitment of Dsh-GFP constructs to the cell membrane. Dsh-GFP or DsNLSm RNA (0.5 ng) was injected alone or with Fz8 RNA (1 ng) into two animal blastomeres at the 4–8-cell stage. GFP fluorescence was assessed in animal cap explants as in Figure 1b-d. Both Dsh and DsNLSm are efficiently recruited to the cell membrane by Fz8. Arrowheads point to cell membranes. (b) DsNLSm can rescue convergent extension defects caused by Xdd1. Four-cell embryos were injected with 0.6 ng Xdd1 RNA alone or together with 2 ng Dsh-GFP or DsNLSm RNA into two vegetal dorsal blastomeres. The injected embryos were allowed to develop until the sibling embryos reached stage 32. (c) Activation of JNK by the Dsh nuclear import mutant. Four animal blastomeres of four-cell embryos were each injected with 1 ng of RNAs encoding Dsh-GFP or DsNLSm. Embryonic lysates were collected at stage 10.5 for in vitro JNK activity assay using anti-phospho-specific c-Jun antibodies. Total GST-c-Jun levels were assessed with anti-GST antibodies. Dsh-GFP and DsNLSm were equally expressed, as monitored with anti-Dvl2 antibodies; β-tubulin served as a loading control.
Figure 9 Nuclear translocation of Dvl2 upon Wnt3a treatment. (a) MCF7 cells were treated either with Wnt3a-conditioned or control medium for 6 h, fixed and immunostained with anti-Dvl2 antibodies. In control cells, cytoplasmic and perinuclear staining is visible. Wnt3a-conditioned, but not control, medium enhanced nuclear translocation of Dvl2. DAPI staining indicates the position of cell nuclei. Corresponding cells are shown by arrowheads. (b) Nuclear/cytoplasmic (N/C) ratios of fluorescence were calculated for each panel in (a) as the mean +/- standard deviation.
Table 1 Axis induction by Dsh constructs
Injected RNA Total number of injected embryos Complete secondary axes (%) Partial secondary axes (%)
Experiment 1
Dsh-GFP 150 46.6 25.3
DsNESm-GFP 194 54.6 30.4
Experiment 2
Dsh-GFP 144 28.5 45.1
DsNLSm-GFP 149 0.7 39.5
DsSNLS-GFP 137 24.0 42.3
Embryos were injected as described in Figure 1e,f. Partial secondary axes are defined by a morphologically visible ectopic neural tube up to the hindbrain level. Complete axes are defined by the presence of the secondary head structures, including eyes and cement glands. The frequency of secondary axes in uninjected embryos was less than 1%. Data pooled from several independent experiments are shown.
==== Refs
Massagué J Wotton D Transcriptional control by the TGF-β/Smad signaling system EMBO J 2000 19 1745 1754 10775259 10.1093/emboj/19.8.1745
Struhl G Adachi A Nuclear access and action of Notch in vivo Cell 1998 93 649 660 9604939 10.1016/S0092-8674(00)81193-9
Peifer M Polakis P Wnt signaling in oncogenesis and embryogenesis - a look outside the nucleus Science 2000 287 1606 1609 10733430 10.1126/science.287.5458.1606
Wodarz A Nusse R Mechanisms of Wnt signaling in development Annu Rev Cell Dev Biol 1998 14 59 88 9891778 10.1146/annurev.cellbio.14.1.59
Bienz M Clevers H Linking colorectal cancer to Wnt signaling Cell 2000 103 311 320 11057903 10.1016/S0092-8674(00)00122-7
Gumbiner BM Carcinogenesis: a balance between β-catenin and APC Curr Biol 1997 7 R443 R446 9210368 10.1016/S0960-9822(06)00214-4
Axelrod JD Miller JR Shulman JM Moon RT Perrimon N Differential recruitment of Dishevelled provides signaling specificity in the planar cell polarity and Wingless signaling pathways Genes Dev 1998 12 2610 2622 9716412
Boutros M Paricio N Strutt DI Mlodzik M Dishevelled activates JNK and discriminates between JNK pathways in planar polarity and wingless signaling Cell 1998 94 109 118 9674432 10.1016/S0092-8674(00)81226-X
Habas R Kato Y He X Wnt/Frizzled activation of Rho regulates vertebrate gastrulation and requires a novel Formin homology protein Daam1 Cell 2001 107 843 854 11779461 10.1016/S0092-8674(01)00614-6
Sheldahl LC Park M Malbon CC Moon RT Protein kinase C is differentially stimulated by Wnt and Frizzled homologs in a G-protein-dependent manner Curr Biol 1999 9 695 698 10395542 10.1016/S0960-9822(99)80310-8
Sokol SY A role for Wnts in morphogenesis and tissue polarity Nat Cell Biol 2000 2 E124 E126 10878822 10.1038/35017136
Winklbauer R Medina A Swain RK Steinbeisser H Frizzled-7 signalling controls tissue separation during Xenopus gastrulation Nature 2001 413 856 860 11677610 10.1038/35101621
Boutros M Mlodzik M Dishevelled: at the crossroads of divergent intracellular signaling pathways Mech Dev 1999 83 27 37 10507837 10.1016/S0925-4773(99)00046-5
Gloy J Hikasa H Sokol SY Frodo interacts with Dishevelled to transduce Wnt signals Nat Cell Biol 2002 4 351 357 11941372
Ciani L Krylova O Smalley MJ Dale TC Salinas PC A divergent canonical WNT-signaling pathway regulates microtubule dynamics: Dishevelled signals locally to stabilize microtubules J Cell Biol 2003 164 243 253 10.1083/jcb.200309096
Capelluto DGS Kutateladze TG Habas R Finkielstein CV He X Overduin M The DIX domain targets dishevelled to actin stress fibres and vesicular membranes Nature 2002 419 726 729 12384700 10.1038/nature01056
Torres MA Nelson WJ Colocalization and redistribution of Dishevelled and Actin during Wnt-induced mesenchymal morphogenesis J Cell Biol 2000 149 1433 1442 10871283 10.1083/jcb.149.7.1433
Rothbächer U Laurent MN Deardorff MA Klein PS Cho KWY Fraser SE Dishevelled phosphorylation, subcellular localization and multimerization regulate its role in early embryogenesis EMBO J 2000 19 1010 1022 10698942 10.1093/emboj/19.5.1010
Yang-Snyder J Miller JR Brown JD Lai C-J Moon RT A frizzled homolog functions in a vertebrate Wnt signaling pathway Curr Biol 1996 6 1302 1306 8939578 10.1016/S0960-9822(02)70716-1
Itoh K Antipova A Ratcliffe MJ Sokol S Interaction of Dishevelled and Xenopus Axin-related protein is required for Wnt signal transduction Mol Cell Biol 2000 20 2228 2238 10688669 10.1128/MCB.20.6.2228-2238.2000
Yanagawa S van Leeuwen F Wodarz A Klingensmith J Nusse R The Dishevelled protein is modified by Wingless signaling in Drosophila Genes Dev 1995 9 1087 1097 7744250
Henderson BR Nuclear-cytoplasmic shuttling of APC regulates β-catenin subcellular localization and turnover Nat Cell Biol 2000 2 653 660 10980707 10.1038/35023605
Rosin-Arbesfeld R Townsley F Bienz M The APC tumour suppressor has a nuclear export function Nature 2000 406 1009 1012 10984057 10.1038/35023016
Sokol SY Analysis of Dishevelled signalling pathways during Xenopus development Curr Biol 1996 6 1456 1467 8939601 10.1016/S0960-9822(96)00750-6
Kudo N Matsumori N Taoka H Fujiwara D Schreiner EP Wolff B Yoshida M Horinouchi S Leptomycin B inactivates CRM1/exportin 1 by covalent modification at a cysteine residue in the central conserved region Proc Natl Acad Sci USA 1999 96 9112 9117 10430904 10.1073/pnas.96.16.9112
Holaska JM Paschal BM A cytosolic activity distinct from Crm1 mediates nuclear export of protein kinase inhibitor in permeabilized cells Proc Natl Acad Sci USA 1998 95 14739 14744 9843959 10.1073/pnas.95.25.14739
Jans DA Xiao C-Y Lam MHC Nuclear targeting signal recognition: a key control point in nuclear transport? BioEssays 2000 22 532 544 10842307 10.1002/(SICI)1521-1878(200006)22:6<532::AID-BIES6>3.0.CO;2-O
Penton A Wodarz A Nusse R A mutational analysis of dishevelled in Drosophila defines novel domains in the Dishevelled protein as well as novel suppressing alleles of axin Genetics 2002 161 747 762 12072470
Fan MJ Grüning W Walz G Sokol SY Wnt signaling and transcriptional control of Siamois in Xenopus embryos Proc Natl Acad Sci USA 1998 95 5626 5631 9576934 10.1073/pnas.95.10.5626
Peters JM McKay RM McKay JP Graff JM Casein kinase 1 transduces Wnt signals Nature 1999 401 345 350 10517632 10.1038/43830
Sakanaka C Leong P Xu L Harrison SD Williams LT Casein kinase 1ε in the Wnt pathway: regulation of β-catenin function Proc Natl Acad Sci USA 1999 96 12548 12552 10535959 10.1073/pnas.96.22.12548
Zeng L Fagotto F Zhang T Hsu W Vasicek TJ Perry WL Lee JJ Tilghman SM Gumbiner BM Costantini F The mouse Fused locus encodes Axin, an inhibitor of the Wnt signaling pathway that regulates embryonic axis formation Cell 1997 90 181 192 9230313 10.1016/S0092-8674(00)80324-4
Kishida S Yamamoto H Hino S Ikeda S Kishida M Kikuchi A DIX domains of Dvl and Axin are necessary for protein interactions and their ability to regulate β-catenin stability Mol Cell Biol 1999 19 4414 4422 10330181
Smalley MJ Sara E Paterson H Naylor S Cook D Jayatilake H Fryer LG Hutchinson L Fry MJ Dale TC Interaction of Axin and Dvl2 proteins regulates Dvl-2-stimulated TCF-dependent transcription EMBO J 1999 18 2823 2835 10329628 10.1093/emboj/18.10.2823
Li L Yuan H Weaver CD Mao J Far III GH Sussman DJ Jonkers J Kimelman D Wu D Axin and Frat1 interact with Dvl and GSK, bridging Dvl to GSK in Wnt-mediated regulation of LEF-1 EMBO J 1999 18 4233 4240 10428961 10.1093/emboj/18.15.4233
Salic A Lee E Mayer L Kirschner MW Control of β-catenin stability: reconstitution of the cytoplasmic steps of the Wnt pathway in Xenopus egg extracts Mol Cell 2000 5 523 532 10882137 10.1016/S1097-2765(00)80446-3
Li L Yuan H Xie W Mao J Caruso AM McMahon A Sussman DJ Wu D Dishevelled proteins lead to two signaling pathways. Regulation of LEF-1 and c-Jun N-terminal kinase in mammalian cells J Biol Chem 1999 274 129 134 9867820 10.1074/jbc.274.1.129
Moriguchi T Kawachi K Kamakura S Masuyama N Yamanaka H Matsumoto K Kikuchi A Nishida E Distinct domains of mouse Dishevelled are responsible for the c-Jun N-terminal kinase/stress-activated protein kinase activation and the axis formation in vertebrates J Biol Chem 1999 274 30957 30962 10521491 10.1074/jbc.274.43.30957
Habas R Dawid IB He X Coactivation of Rac and Rho by Wnt/Frizzled signaling is required for vertebrate gastrulation Genes Dev 2003 17 295 309 12533515 10.1101/gad.1022203
Itoh K Jacob J Sokol SY A role for Xenopus Frizzled 8 in dorsal development Mech Dev 1998 74 145 157 9651509 10.1016/S0925-4773(98)00076-8
Tada M Smith JC Xwnt11 is a target of Xenopus Brachyury: regulation of gastrulation movements via Dishevelled, but not through the canonical Wnt pathway Development 2000 127 2227 2238 10769246
Wallingford JB Rowning BA Vogeli KM Rothbächer U Fraser SE Harland RM Dishevelled controls cell polarity during Xenopus gastrulation Nature 2000 405 81 85 10811222 10.1038/35011077
Wiechens N Fagotto F CRM1- and Ran-independent nuclear export of β-catenin Curr Biol 2001 11 18 27 11166175 10.1016/S0960-9822(00)00045-2
Wiechens N Heinle K Englmeier L Schohl A Fagotto F Nucleocytoplasmic shuttling of Axin, a negative regulator of the Wnt-β-catenin pathway J Biol Chem 2004 279 5263 5267 14630927 10.1074/jbc.M307253200
Cong F Varmus H Nuclear-cytoplasmic shuttling of Axin regulates subcellular localization of β-catenin Proc Natl Acad Sci USA 2004 101 2882 2887 14981260 10.1073/pnas.0307344101
Diehl JA Cheng M Roussel MF Sherr CJ Glycogen synthase kinase-3β regulates cyclin D1 proteolysis and subcellular localization Genes Dev 1998 12 3499 3511 9832503
Franca-Koh J Yeo M Fraser E Young N Dale TC The regulation of glycogen synthase kinase-3 nuclear export by Frat/GBP J Biol Chem 2002 277 43844 43848 12223487 10.1074/jbc.M207265200
Lee E Salic A Kirschner MW Physiological regulation of β-catenin stability by Tcf3 and CK1ε J Cell Biol 2001 154 983 993 11524435 10.1083/jcb.200102074
Hikasa H Sokol SY The involvement of Frodo in TCF-dependent signaling and neural tissue development Development 2004 131 4725 4734 15329348 10.1242/dev.01369
Makarova O Kamberov E Margolis B Generation of deletion and point mutations with one primer in a single cloning step BioTechniques 2000 29 970 972 11084856
Nieuwkoop PD Faber J Normal table of Xenopus laevis (Daudin) 1967 2 Amsterdam: North Holland
Wolda SL Moody CJ Moon RT Overlapping expression of Xwnt3A and Xwnt1 in neural tissue of Xenopus laevis embryos Dev Biol 1993 155 46 57 8416844 10.1006/dbio.1993.1005
Sokol S Christian JL Moon RT Melton DA Injected Wnt RNA induces a complete body axis in Xenopus embryos Cell 1991 67 741 752 1834344 10.1016/0092-8674(91)90069-B
Willert K Shibamoto S Nusse R Wnt-induced dephosphorylation of Axin releases β-catenin from the Axin complex Genes Dev 1999 13 1768 1773 10421629
Liu C Kato Y Zhang Z Do VM Yanker BA He X β-Trcp couples β-catenin phosphorylation-degradation and regulates Xenopus axis formation Proc Natl Acad Sci USA 1999 96 6273 6278 10339577 10.1073/pnas.96.11.6273
Semënov MV Snyder M Human Dishevelled genes constitute a DHR-containing multigene family Genomics 1997 42 302 310 9192851 10.1006/geno.1997.4713
Lysovsky M Itoh K Sokol SY Frizzled receptors activate a novel JNK-dependent pathway that may lead to apoptosis Curr Biol 2002 12 53 58 11790303 10.1016/S0960-9822(01)00628-5
Sokol SY Klingensmith J Perrimon N Itoh K Dorsalizing and neuralizing properties of Xdsh, a maternally expressed Xenopus homolog of dishevelled Development 1995 121 1637 1647 7600981
Lemaire P Garrett N Gurdon JB Expression cloning of Siamois, a Xenopus homeobox gene expressed in dorsal-vegetal cells of blastulae and able to induce a complete secondary axis Cell 1995 81 85 94 7720076 10.1016/0092-8674(95)90373-9
| 15720724 | PMC551520 | CC BY | 2021-01-04 16:05:45 | no | J Biol. 2005 Feb 15; 4(1):3 | utf-8 | J Biol | 2,005 | 10.1186/jbiol20 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r111569394010.1186/gb-2005-6-2-r11ResearchComparison of the oxidative phosphorylation (OXPHOS) nuclear genes in the genomes of Drosophila melanogaster, Drosophila pseudoobscura and Anopheles gambiae Tripoli Gaetano [email protected]'Elia Domenica [email protected] Paolo [email protected] Corrado [email protected] University of Bari, DAPEG Section of Genetics, via Amendola 165/A, 70126 Bari, Italy2 CNR, Institute of Biomedical Technology, Section of Bari, via Amendola 122/D, 70126 Bari, Italy2005 31 1 2005 6 2 R11 R11 24 9 2004 8 12 2004 7 1 2005 Copyright © 2005 Tripoli 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.
An analysis of nuclear-encoded oxidative phosphorylation genes in Drosophila and Anopheles reveals that pairs of duplicated genes have strikingly different expression patterns.
Background
In eukaryotic cells, oxidative phosphorylation (OXPHOS) uses the products of both nuclear and mitochondrial genes to generate cellular ATP. Interspecies comparative analysis of these genes, which appear to be under strong functional constraints, may shed light on the evolutionary mechanisms that act on a set of genes correlated by function and subcellular localization of their products.
Results
We have identified and annotated the Drosophila melanogaster, D. pseudoobscura and Anopheles gambiae orthologs of 78 nuclear genes encoding mitochondrial proteins involved in oxidative phosphorylation by a comparative analysis of their genomic sequences and organization. We have also identified 47 genes in these three dipteran species each of which shares significant sequence homology with one of the above-mentioned OXPHOS orthologs, and which are likely to have originated by duplication during evolution. Gene structure and intron length are essentially conserved in the three species, although gain or loss of introns is common in A. gambiae. In most tissues of D. melanogaster and A. gambiae the expression level of the duplicate gene is much lower than that of the original gene, and in D. melanogaster at least, its expression is almost always strongly testis-biased, in contrast to the soma-biased expression of the parent gene.
Conclusions
Quickly achieving an expression pattern different from the parent genes may be required for new OXPHOS gene duplicates to be maintained in the genome. This may be a general evolutionary mechanism for originating phenotypic changes that could lead to species differentiation.
==== Body
Background
The accessibility of whole-genome sequence data for several organisms, together with the development of efficient computer-based search tools, has revolutionized modern biology, allowing in-depth comparative analysis of genomes [1-4]. In many cases, comparisons among species at various levels of divergence have helped to define protein-coding genes, recognize nonfunctional genes, and find regulatory sequences and other functional elements in the genome. When applied to a set of genes correlated by function and/or subcellular localization of their products, intra- and interspecies comparative analyses can be especially efficient tools to obtain information on the functional constraints acting on the evolution of the gene set and on the mechanisms regulating its coordinate expression.
A set of genes present in all eukaryotic genomes and expected to be subject to peculiar evolutionary constraints is represented by the genes involved in oxidative phosphorylation (OXPHOS), the primary energy-producing process in all aerobic organisms [5]. To generate cellular ATP, OXPHOS uses the products of both nuclear and mitochondrial genes, organized in five large complexes embedded in the lipid bilayer of the inner mitochondrial membrane. Except for complex II, which is formed by four proteins encoded by nuclear genes, the other respiratory complexes depend on both mitochondrial and nuclear genomes; so, assembling the OXPHOS complexes and fine tuning their activity to satisfy cell- and tissue-specific energy demands requires specialized regulatory mechanisms and evolutionary strategies to optimize the cross-talk between the two genomes and ensure the coordinated expression of their relevant products.
Analysis of co-regulated mitochondrial and nuclear genes, and of the transcription factors regulating the functional network they constitute, might also be a useful approach to investigate the origin of mitochondrial dysfunction in humans. Disorders of mitochondrial oxidative phosphorylation are now recognized as the most common inborn errors of metabolism, affecting at least one in 5,000 newborn children [6]. In this context, the expanding spectrum of identified mitochondrial proteins provides an opportunity to test a whole new range of candidate genes whose mutations may be responsible for common human diseases. For example, a recent study by Mootha et al. [7] suggests a promising strategy for clarifying the molecular etiology of mitochondrial pathologies by profiling the tissue-specific expression pattern of candidate mitochondrial proteins.
Despite the long evolutionary divergence time, many key pathways that control development and physiology are conserved between Drosophila and humans, and about 70% of the genes associated with human disease have direct counterparts in the Drosophila genome [8,9]. For example, the potential role of Drosophila as a model system for understanding the molecular mechanisms involved in human genetic disease is validated by the recent identification of a Drosophila mutation causing a necrotic phenotype that mimics in detail the diseases that arise from serpin mutations in humans [10].
It has been suggested that comparisons between D. melanogaster and other species of the genus Drosophila could provide a model system for developing and testing new algorithms and strategies for the functional annotation of complex genomes [3]. To obtain new information on the evolution of a set of genes that control a basic biological function by encoding products targeted to a specific cellular compartment, we have performed a comparative analysis of the OXPHOS genes of D. melanogaster and D. pseudoobscura; the complete genome of the latter was recently made available by the Baylor Human Genome Sequencing Center. These two species are the only species of the Drosophila genus for which whole-genome sequence data exist at present [11-13]. We also took advantage of the complete sequence of the A. gambiae genome [14] to compare the Drosophila OXPHOS genes with those of this more distantly related dipteran (the divergence time between D. melanogaster and A. gambiae is thought to be approximately 250 million years, as compared to 46 million years between D. melanogaster and D. pseudoobscura [15,16]). Although extensive reshuffling within and between chromosomal regions is known to have occurred since the divergence of Anopheles from Drosophila [4,17,18], we show that in these organisms the conservation of the OXPHOS genes is still sufficient to permit their meaningful comparison.
Here we report the identification of 78 D. pseudoobscura and 78 A. gambiae genes representing the counterparts of D. melanogaster OXPHOS genes which, in turn, were previously identified as putative orthologs of human OXPHOS genes [19]. We have annotated these genes, taking into account conservation in amino-acid sequence, intron-exon structure, intron length, and the presence of duplications in the genome. The conservation of genomic organization and evidence from evolutionary trees based on sequence similarity suggest that these genes are one-to-one orthologs in the three species, and that in many cases they originated (produced?) duplicates by transpositional and/or recombinational events during evolution. We have identified in the three dipteran genomes a total of 47 genes that probably originated by duplication of the above-mentioned genes, and we show that the duplicate gene has usually acquired a pattern of expression strikingly different from that of the gene from which it derived. Moreover, when the comparison is possible, the gene duplicate almost always shows a strongly testis-biased expression, in contrast to the soma-biased expression of its parent gene.
Results and discussion
Identification and comparative annotation of D. pseudoobscura and A. gambiae OXPHOS genes
We have previously reported [19] the identification of 285 D. melanogaster nuclear genes encoding mitochondrial proteins that represent the counterparts of human peptides annotated in the Swiss-Prot database as mitochondrial [20]. On the basis of comparative evidence obtained by BLASTP analysis, 78 of these genes are involved in the OXPHOS system, encoding 66 proteins known to be components of the five large respiratory complexes and 12 proteins involved in oxidative phosphorylation as accessory proteins. To identify the putative counterparts of the D. melanogaster OXPHOS genes in D. pseudoobscura and A. gambiae we performed a TBLASTN search [13,21] on the whole genome sequences of these species using the amino-acid sequences of the 78 D. melanogaster peptides as queries. Sequences giving the best reciprocal BLAST hits were tentatively assumed to identify functional counterparts in two species if they could be aligned over at least 60% of the gene length and the BLAST E-score was less than 10-30. By these criteria, all the 78 D. melanogaster OXPHOS genes investigated have a counterpart both in D. pseudoobscura and in A. gambiae. To better compare the structure of the OXPHOS genes in the three dipteran species, we used the predicted coding sequences as queries for a search of expressed sequence tags (EST) [21], and used the retrieved sequences to annotate the transcribed noncoding sequences of the A. gambiae genes investigated. Although little EST information is available for D. pseudoobscura, it was still possible to predict unambiguously the exon-intron gene structure of the OXPHOS genes in this species, as well as the amino-acid sequence of their full-length products, by exploiting the high level of similarity with D. melanogaster. The results of BLAST analysis, together with the construction of phylogenetic trees that also include other genes that show lesser but still significant sequence similarity to the 78 genes assumed to be one-to-one orthologs in the three species investigated (see below), strongly suggest that the newly identified D. pseudoobscura and A. gambiae genes are the functional counterparts of the 78 D. melanogaster genes used as probes.
Table 1 lists the 78 putative orthologous OXPHOS genes in the three dipteran genomes and their cytological location. For each gene, a record showing the gene map and reporting the annotated genomic sequences as well as the mRNA and protein sequences is available and can be queried at the MitoComp website [22] (see also Additional data files). MitoComp also compares the structure of the D. melanogaster, D. pseudoobscura and A. gambiae putative orthologous genes and their duplications when present (see below), and aligns the orthologous coding sequences (CDS), and also aligns their deduced amino-acid products with the corresponding human protein.
Amino-acid sequence comparison
For the products of the OXPHOS genes investigated, the D. melanogaster/D. pseudoobscura average amino-acid sequence identity is 88%, compared to 64% between D. melanogaster and A. gambiae. Figure 1 shows the frequency distribution of sequence identities, and Additional data file 1 lists all pairwise identity values between the products of the 78 OXPHOS genes when orthologous D. melanogaster/D. pseudoobscura, D. melanogaster/A. gambiae and D. melanogaster/human gene products are compared. A multiple alignment of each cluster of homologous proteins is shown at the MitoComp website [22].
It should be kept in mind that identity values reported in Figure 1 and in the table in Additional data file 1 were calculated on the whole sequence of the predicted unprocessed proteins; they are much higher if the putative amino-terminal pre-sequences are excluded, since such sequences, possessed by most mitochondrion-targeted products, show little amino-acid sequence conservation [23,24], although they do share specific physicochemical properties [25,26]. When only the predicted mature protein is considered, the average percentage identity increases to 90% between D. melanogaster and D. pseudoobscura, and to 70% between D. melanogaster and A. gambiae.
A striking example of evolutionary conservation is provided by the genes encoding cytochrome c (an essential and ubiquitous protein found in all organisms) in the three dipteran species: the amino-acid sequences of the gene products are identical in D. melanogaster and D. pseudoobscura, whereas 96% identity is preserved between Drosophila and Anopheles. Coding sequences are also extremely conserved, suggesting that the nucleotide sequence itself is subject to strong evolutionary constraints, maybe due to codon usage bias. Only synonymous substitutions (21 out of 108 codons) were found on comparing D. melanogaster and D. pseudoobscura cytochrome c coding sequences, whereas 28 synonymous substitutions and only four nonsynonymous substitutions were observed between D. melanogaster and A. gambiae (see MitoComp website [22]).
Gene structure comparisons
It is well known that a given function may be supplied in different species by genes that are not directly derived from a common ancestor, that is, by paralogous, not orthologous, genes. Therefore, we thought it would be interesting to compare the structural organization of the OXPHOS genes in the three species investigated, on the principle that it should be possible to infer derivation from a common ancestor, that is, 'structural orthology', if an identical or very similar overall structure was preserved. As the introns of the putative orthologous OXPHOS genes in the three species are, as expected, too divergent in DNA sequence to be aligned, we used conservation of number of introns, conservation of their location in the coding sequence, and preservation of the reading frame with respect to the flanking exons as our primary criteria.
With the only exception of Dpse\CG5037, putatively encoding protoheme IX farnesyltransferase, whose 5' genomic sequence was impossible to find in the relevant contig assembly, all other investigated D. pseudoobscura genes show a structural organization almost identical to that of their D. melanogaster counterparts. Of the 78 Anopheles genes studied, 39 maintain the structural organization observed in Drosophila, whereas gain or loss of introns occurred in 33, and in six the location of introns is not preserved at all. In agreement with a previous report [4], the intron-exon structure of the gene appears to be conserved in all three dipteran species when splicing of alternative coding exons occurs: the alternative splice forms of both the Drosophila NADH-ubiquinone oxidoreductase acyl carrier protein (mtacp1, CG9160) [27] and the Drosophila ATP synthase epsilon chain (sun, CG9032) [19] have very similar counterparts in Anopheles, as shown by genomic structure comparison, alignment of splice variants and EST mapping (Figure 2).
Genes encoding the acyl carrier protein (mtacp1) in the three species are characterized by the mutually exclusive use of homologous exons that are repeated in tandem (Figure 2a). The duplicate exons occur at the same location in the aligned amino-acid sequences, and are flanked on both sides by a phase 1 intron. When the sequences of the duplicated exons are compared, they show the expected divergence pattern (that is, the similarity between duplicate exons within a gene is less than the similarity of each exon to its equivalent in the orthologous gene). Evidence from genomic and transcribed sequences (GenBank accession numbers BI510891 and BI508135) shows that the duplicated mtacp1exons are also preserved in the more distantly related insect Apis mellifera (honeybee) (Figure 2c,d), indicating a specific adaptive benefit for this gene structure, as also suggested by the evolutionary convergence leading to the occurrence of alternative splicing in members of three different ion-channel gene families from Drosophila to humans [28]. However, there is no evidence from ESTs that duplicated mtacp1 exons undergo alternative splicing in vertebrates and nematodes.
Analysis of intron length
Interspecies comparison of the introns of putative orthologous genes indicates that there is little constraint on their nucleotide sequence, which undergoes nucleotide substitutions at a rate comparable to that of pseudogenes [29]. However, several observations suggest that intron size is subject to natural selection. For example, in D. melanogaster and several other organisms the distribution of intron length has been shown to be asymmetrical, with a large group of introns falling into a narrow distribution around a 'minimal' length and the remaining showing a much broader length distribution, ranging from hundreds to thousands of base-pairs [30-32].
Of the introns that interrupt the coding sequence in the 78 OXPHOS genes investigated in the present study, 88 (64.7%) of 136 in D. melanogaster, 96 (70.5%) of 136 in D. pseudoobscura and 87 (67.9%) of 128 in A. gambiae fall into the short-size class (Figure 3a). However, in A. gambiae the length distribution of these introns appears slightly broader (62-150 bp, compared with 51-100 bp in both Drosophila species). The remaining introns show a broad length distribution, ranging from 151 to 4,702 bp with no clear boundary between classes.
A comparison of the length of introns in corresponding positions in the putative D. melanogaster, D. pseudoobscura and A. gambiae orthologs suggests that changes from the short-size to the long-size (more than 300 bp) intron class, or the converse, have been rare in the evolutionary history of these species: only seven class changes were observed comparing D. melanogaster and D. pseudoobscura introns, and six between D. melanogaster and A. gambiae (Figure 3b). On the whole, our data confirm the highly asymmetrical intron length distribution in D. melanogaster and extend this finding to the introns of the D. pseudoobscura and A. gambiae OXPHOS genes.
OXPHOS gene duplications
It is generally accepted that gene duplication is the basic process that underlies the diversification of genes and the origination of novel gene functions [33]; however, many features of this process are still elusive. To obtain more information on the molecular evolution of the genes involved in the OXPHOS system, we searched the genomes of D. melanogaster, D. pseudoobscura and A. gambiae for duplications of the 78 OXPHOS genes whose orthologs we have identified in the three species.
Duplicate gene pairs were tentatively identified within each genome as best reciprocal hits with an E-value of less than 10-20 in both directions in a TBLASTN search using the default parameters. Deciding whether two proteins may be considered homologous becomes difficult when their sequence identity is within the 20-30% range (the so-called 'twilight zone' [34]), and so the following additional criteria were used: first, the two sequences could be aligned over more than 60% of their length; second, the putative processed proteins encoded had to have more than 40% identity; and third, amino-acid percentage similarity had to be larger than percentage identity [35]. Even if meeting these criteria and reported as different genes in the ENSEMBL database [36], identical Anopheles nucleotide sequences were excluded from further analysis, as they are likely to reflect annotation artifacts.
Duplications, or in some instances triplications, of 24 OXPHOS genes were found. Overall, we identified 47 genes (20 in D. melanogaster, 19 in D. pseudoobscura and eight in A. gambiae) each of which shows significant similarity with one of the 78 OXPHOS genes reported above. When the structure of a member of a paralogous gene set indicates that it has been produced by retroposition, it seems reasonable to assume that it is derived from a pre-existing 'parent' gene. For duplicates not clearly originating by retroposition, we also assume, on the basis of the much higher level of conservation and expression, that the genes we find to be the structural orthologs in all three species are the parent ones, and in this case also we will henceforth refer to their paralogs as OXPHOS gene duplicates. The amino-acid percentage identity between the products of duplicate gene pairs ranges from 40% to 85%. For each of the OXPHOS gene duplicates, cytological localization, number of exons interrupting the coding sequence, and number of ESTs found in the D. melanogaster and A. gambiae EST databases are reported in Table 2. Neighbor-joining trees derived from distance matrix analysis and showing the inferred evolutionary relationship between members of each gene cluster are available at the MitoComp website [22].
Duplications (or triplications) of 16 of the 78 OXPHOS genes investigated were found in both D. melanogaster and D. pseudoobscura. In such cases, to assign pairwise orthology, besides taking into account conservation of structural organization, given the general conservation of microsyntenic gene order in the two species, we used the products of D. melanogaster genes flanking the duplicate loci to search for homologous sequences also flanking the same genes in the D. pseudoobscura genome.
The genomic organization of many OXPHOS duplicates shows that they were originated by retropositional events, because they are intronless, or have only very few introns that are likely to have been inserted into the coding sequence after the duplication event. In other cases, duplication apparently resulted from transposition of genomic DNA sequences or from recombinational events, as duplicate genes maintain an identical or very similar structural organization.
On the basis of the presence of the duplication in both species, supported by evidence from evolutionary trees and conservation of microsyntenic gene order, it can be inferred that 15 of the duplications identified occurred before the D. melanogaster/D. pseudoobscura divergence (about 46 million years ago). On the other hand, five duplications were found only in D. melanogaster and four only in D. pseudoobscura; in these instances, if the duplication occurred before the divergence of the two species, it has been followed by loss of one of the copies in the lineage leading to the species in which the gene is no longer duplicated. On the assumption that the rate of gene duplication is constant over time, this translates to approximately 0.0014 duplications per gene per million years (4 or 5 duplications per 78 genes per 46 million years) that achieved fixation and long-term preservation in the genome. This value is about twofold lower than the 0.0023 value calculated by Lynch and Conery [37] for the 13,601 genes of the whole genome of D. melanogaster. However, it can be argued that the rate of long-term preservation in the genome of OXPHOS gene duplicates cannot be meaningfully compared with the general rate of preservation of duplicates in the whole genome since, while recent data suggest that in eukaryotic genomes there is preferential duplication of conserved proteins [38], duplicates of genes that encode subunits of multiprotein complexes, as most of the genes we have investigated do, negatively influence the fitness of an organism [39], and are therefore unlikely to become fixed in the population. In summary, it appears reasonable to assume that the preservation in the genome of OXPHOS gene duplicates should occur very infrequently, unless special mechanisms allowing their fixation in the population are present (see the next section).
In A. gambiae we found only four duplications and two triplications of the OXPHOS genes analyzed; of these, four involve genes also duplicated in one or both Drosophila species (Table 2). Pairwise orthology could not be assigned between Drosophila and Anopheles gene duplicates as neither microsynteny nor evolutionary trees provide sufficient evidence for the origin of the gene pairs from a single-copy gene before the Drosophila/Anopheles divergence.
Expression pattern of OXPHOS gene duplicates
The relative abundance of ESTs in a EST library may be assumed roughly to reflect the level of expression of each mRNA in the tissues from which the library was prepared. We therefore used the mRNA sequences predicted in silico to be transcribed from the OXPHOS duplicate genes investigated in this work as queries in a search of the public D. melanogaster and A. gambiae EST databases to infer the relative abundance of the mRNA copies from the hits scored. For each gene, the number of ESTs found in the databases is detailed in Table 2. With the exception of one of the paralogs of the A. gambiae gene encoding ubiquinol-cytochrome c reductase core protein 1, in all cases the search found the number of ESTs originating from the duplicate gene was strikingly lower than that originating from the putative parent gene, in both D. melanogaster and A. gambiae (in total, 100 versus 1,747 in D. melanogaster and 60 versus 687 in A. gambiae). A smaller number of ESTs originating from the OXPHOS gene duplicates was observed even in A. gambiae EST libraries that are normalized. Remarkably, and regardless of the mechanism of the duplication, in D. melanogaster, in which several organ-specific or developmental stage specific libraries are available, the search showed that the expression of the OXPHOS gene duplicates is strongly testis-biased, as 97 out of the 100 ESTs originating from them were found in testis-derived libraries, while only 27 out of the 1,769 ESTs originating from the parent genes were found in such libraries, the bulk of them being instead found in libraries derived from embryos or somatic tissues.
Our finding that the expression of the OXPHOS gene originated by duplication is strongly testis-biased is validated by the data obtained by Parisi et al. [40] using the FlyGEM microarray to identify D. melanogaster genes showing ovary-, testis- or soma-biased expression. With the exception of CG7349, CG30354, CG30093 and CG12810, for which no data were presented by Parisi et al. [40], all other genes reported in this work as OXPHOS gene duplicates were found in the genomic fraction showing testis-biased expression, whereas all the parent genes present in the dataset showed soma-biased expression. Additional data file 2 summarizes the relevant data extracted from Parisi et al. [40].
The pattern of strongly testis-biased expression of OXPHOS gene duplicates holds for a further sample of 40 duplications of genes annotated in the MitoDrome database [19] as encoding products that are mitochondrion-targeted but not involved in the OXPHOS system. For 15 of these no data are provided by Parisi et al. [40], but all the remaining 25 genes show a testes-biased expression (data not shown).
Duplications of genes encoding OXPHOS subunits, for which stoichiometry is important, are likely to be strongly deleterious owing to the negative consequences of an imbalance in the concentration of the respiratory complex constituents, unless, as proposed by Lynch and Force [41], 'subfunctionalization' and/or a differential expression pattern of duplicate copies occurs. In this case, the duplicate OXPHOS genes would have a reduced or absent capacity to functionally complement mutations in their parent genes, in contrast to what is generally assumed to be the main short-term advantage of gene duplication. In D. melanogaster at least there is evidence for this, as FlyBase [42] and BDGP P-Element Gene Disruption Project [43] searches for P-insertion mutants in the D. melanogaster OXPHOS genes found that lethal alleles for 11 out of 19 D. melanogaster parent genes are known (see the MitoComp website [22]), indicating that loss-of-function of the parent gene cannot be compensated for by the presence of the gene duplicate. P-insertion mutants with an abnormal phenotype, indicating a functional divergence, are known for only one of the D. melanogaster OXPHOS gene duplicates - Cyt-c-d, encoding cytochrome c). Interestingly, although Cyt-c-d is adjacent to its putative parent gene, Cyt-c-p, it shows a different pattern of expression, suggesting that the two genes must be regulated at individual gene level and not at chromatin domain level (see Table 2).
A systematic investigation of the expression pattern of other D. melanogaster duplicate genes will be necessary to answer the question of whether the testis-biased expression pattern reported here is specific to the duplicates of genes encoding mitochondrial proteins, or is a more general phenomenon. According to the balance hypothesis, validated by experimental results obtained on yeast [39], single gene duplications involving genes encoding components of multiprotein complexes are expected to severely affect fitness. Therefore, the expression pattern we have observed could be a necessary condition to maintain some gene duplicates in the D. melanogaster genome, at least until they evolve a new useful function. Finally, as nothing is known about the tissue-specific pattern of expression of the genes investigated in D. pseudoobscura and Anopheles, it also remains unclear whether the testis-biased expression of gene copies originated by duplication is specific to D. melanogaster, or is also to be found in other dipterans, and possibly in other organisms.
Codon usage in the OXPHOS genes
Because of the preferential use of codons ending in C or G, the D. melanogaster coding sequences have an average GC content higher than the genomic average [44,45]. This is also true for the 78 D. melanogaster OXPHOS coding sequences reported in this work and for their D. pseudoobscura and A. gambiae counterparts (68% of the codons in the OXPHOS genes end in C or G in D. pseudoobscura and 77% in A. gambiae, compared to 74% in D. melanogaster). In all three species, the coding sequences of OXPHOS gene duplicates show a lower percentage of codons ending in C or G, when compared to both the entire set of 78 orthologous OXPHOS genes and the gene subset including only their parent genes. In samples including all the OXPHOS gene duplicates annotated in this paper the aggregate percentage of C- or G-ending codons is 63%, 46% and 73% in D. melanogaster, D. pseudobscura and A. gambiae respectively, as compared with 70%, 64% and 88% in their parent genes. In D. pseudoobscura, the shift toward a higher percentage of A- or T-ending codons is also detected in the pattern of synonymous codon usage; for 12 of the 18 amino acids that are encoded by more than one codon, the most frequently used codon in the D. pseudoobscura gene duplicates is different from the one used in their parent genes (see Additional data file 3).
Chromosomal arm location, interarm homology and microsynteny
It has been reported that in many eukaryotes including yeast [46], C. elegans [47], D. melanogaster [48,49] and humans [50], genes with related functions and similar expression patterns tend to be clustered, suggesting that they share aspects of transcriptional regulation depending on their inclusion in the same chromatin domain. In particular, Boutanaev et al. [48] reported that in D. melanogaster clusters of three or more testis-specific genes are much more frequent than expected by chance. Therefore, we investigated the chromosomal distribution of the OXPHOS genes to determine whether clustering could be detected. In all three dipteran species considered, the 78 OXPHOS orthologous genes are randomly distributed on all chromosomal arms (Table 1). Two D. melanogaster genes (Ucrh, encoding the 11 kDa subunit of ubiquinol-cytochrome c reductase, and CG40002, encoding the AGGG subunit of NADH-ubiquinone oxidoreductase) have a heterochromatic location.
No evidence of OXPHOS gene duplicate clustering was found either, despite the common testis-biased expression of such genes. Moreover, no evidence of clustering with other testis-specific genes was found when an EST database search for such genes was performed in the regions flanking the investigated gene duplicates.
However, in accord with two studies reporting a significant deficit of genes with a male-biased expression on the D. melanogaster X chromosome [51,52], only one out of the 20 D. melanogaster OXPHOS gene duplicates, two out of 19 in D. pseudoobscura and none (out of eight) in A. gambiae were found to be X-linked (Table 2). It may be that duplications of X-linked genes encoding OXPHOS subunits would be especially deleterious because of the male X chromosome transcriptional hyperactivity, which allows dosage compensation.
In all three dipteran species, a disproportionately high fraction of OXPHOS gene duplicates appears to be constituted of autosomal genes derived from parent genes located on the X chromosome (Table 2). As suggested by recent work on the generation and preservation of functional genes produced by retroposition both in Drosophila [53] and in the human and mouse genomes [54], this may be explained by a selective advantage for duplicates of X-linked genes that move to an autosomal location and so escape the X inactivation in early spermatogenesis that occurs both in Drosophila [55] and in mammals [56].
We would like to speculate that such selective advantage may be especially significant for duplicates of OXPHOS genes, given the heavy reliance of sperm on mitochondrial function. In fact, the excess of autosomal duplicates of X-linked genes is not observed for MitoDrome annotated genes not involved in the OXPHOS system (see above). However, as the general pattern of much lower, testis-biased expression holds even for OXPHOS and other mitochondrial gene duplicates that apparently derive from autosomal parental genes, and even for X-linked duplicates, this pattern (and the explanation of the evolutionary preservation of such genes) cannot only be due to the selective advantage of escaping X inactivation during spermatogenesis.
With the exception of CG9603, all euchromatic D. melanogaster orthologs maintain their localization on the homologuos D. pseudoobscura chromosomal arm (Table 3). CG9603, encoding the VIIa polypeptide of cytochrome c oxidase, is located on the 3R chromosomal arm in D. melanogaster, whereas Dpse\CG9603, its counterpart in D. pseudoobscura, is located on XR; microsyntenic gene order with the flanking genes is conserved in both species, suggesting that a chromosomal rearrangement occurred after their divergence.
OXPHOS gene duplicates also almost always maintain the same chromosomal location and microsyntenic gene order in D. melanogaster and in D. pseudoobscura. However, a more complex situation was observed with regard to the gene encoding subunit IV of cytochrome c oxidase, which is duplicated in D. melanogaster and triplicated in D. pseudoobscura (Table 2). On the basis of identical genomic organization, conserved chromosomal location and mycrosyntenic gene order Dpse\CG10664 is inferred to be the ortholog of D. melanogaster CG10664. Dm CG10396, Dpse\CG10396.1 and Dpse\CG10396.2 are intronless, and neither interarm homology nor microsyntenic order offer any clue to their phylogenetic relationship. The dendrogram based on sequence divergence (see the MitoComp website [22], complex IV, subunit IV) suggests, however, that a duplication event occurred before the D. melanogaster/D. pseudobscura speciation, originating the CG10664-CG10396 gene pair (Dpse\CG10664-Dpse\CG10396 in D. pseudoobscura). A further duplication event, occurring in the D. pseudoobscura lineage after the D. melanogaster/D. pseudoobscura divergence, probably created the Dpse\CG10396.1-Dpse\CG10396.2 gene pair.
In contrast to the maintained location of almost all investigated genes on homologous chromosomal arms in the two Drosophila species, when D. melanogaster and A. gambiae are compared the only meaningful correspondence found concerns the genes on the D. melanogaster 2L and the A. gambiae 3R chromosomal arms (Table 3). This result is consistent with previous reports that compared the location of homologous genes in D. melanogaster and A. gambiae, concluding that extensive reshuffling both within and between chromosomal regions has occurred since the divergence of the two species [4,17].
Conclusions
We have catalogued 78 nuclear genes that control oxidative phosphorylation in three dipteran species and compiled a web-based dataset, MitoComp [22], that contains all the data on which this article is based and which is available with the online version of this article. We have conducted only some basic comparative analyses of the many which are possible using such a dataset, and it is our hope that it will provide a valuable resource for those looking for information about nuclear genes encoding mitochondrion-targeted products in the context of functional genomics and proteomics. Future studies based on this information, especially if the comparative analysis is extended to other species, will surely allow a better understanding of the evolutionary history of a set of genes that control a basic biological function, and also offer interesting insights into the mechanisms of their coordinated expression. In fact, a first in silico analysis of the D. melanogaster and D. pseudoobscura nuclear energy gene sequences suggests that a genetic regulatory circuit, based on a single regulatory element, coordinates the expression of the whole set of energy-producing genes in Drosophila [57].
The comparative analysis of the 78 OXPHOS genes in the three dipteran species shows a high level of amino-acid sequence identity, as well as a substantial conservation of intron-exon structure, indicating that these genes are under strong selective constraints. An unexpected and intriguing result of this study is that in D. melanogaster, duplication-originated OXPHOS genes are expressed at a much lower level (or possibly not expressed at all) in most or all the tissues where their parent genes are expressed, as judged by the abundance of ESTs derived from their transcripts in all libraries other than those derived from testis. On the other hand, OXPHOS gene duplicates have a strongly testis-biased pattern of expression, a finding validated by other authors with a different approach based on the use of microarrays [40]. In A. gambiae, although no testis-specific ESTs databases are available, a pattern of expression of almost all duplicate OXPHOS genes different from that of the gene from which they originated, and possibly limited to specific tissues, is suggested by the fact that in all EST libraries available the abundance of the sequences originated from the duplicate genes is very low when compared with that of the sequences derived from their respective parent genes.
We suggest that, at least in D. melanogaster, the acquisition of a new, testis-biased pattern of expression may be required to maintain duplicates of certain genes in the genome. This may also allow rapid acquisition of new functions by the gene product(s), as it has recently been shown that proteins encoded by duplicated genes with a changed expression pattern often show accelerated evolution [58,59]. Subfunctionalization could then further favor the preservation of multiple paralogous genes.
No data are at present available to support the possibility that our findings could be extrapolated to other gene sets or even to the whole genome. However, we propose that duplication of the genes encoding products that are part of multiprotein complexes may be especially deleterious, unless sequence divergence allowing only testis-specific expression of one of the duplicate copies occurs. In turn, this could facilitate the development of novel functions, which is usually assumed to be the main evolutionary advantage of gene duplication, providing a general mechanism for originating phenotypic changes that might also lead to species differentiation.
Materials and methods
To identify orthologous OXPHOS genes and their duplications in D. pseudoobscura and A. gambiae, contigs from BCM [13] and scaffolds from AnoBase [21] were searched using TBLASTN with the D. melanogaster OXPHOS peptides listed in the MitoDrome database [19] as queries.
Amino-acid sequence identity and similarity values were obtained from pairwise alignments using the Needleman-Wunsch global alignment algorithm at the EMBL-EBI server [60]. Multiple sequence alignments of the OXPHOS amino-acid and coding sequences and visualization of the dendrograms were obtained using the MultAlin 5.4.1 software [61] from MultAlin server [62].
The genomic sequence of each gene was manually searched for intron-exon boundaries and the predicted mRNA sequence reconstructed in silico. A. gambiae mRNAs were assembled by overlapping ESTs extracted from AnoBase [21].
We have named each newly identified A. gambiae gene with the four-letter code 'agEG' followed by the last four or five digits of its Ensembl [36] gene number, excluding the multiple zeros of the prefix; the D. pseudoobscura genes were named with the code 'Dpse\CG' followed by the Celera number of their D. melanogaster counterparts.
The D. pseudoobscura OXPHOS genes investigated here were assigned a chromosomal location where possible, using the putative chromosomal assignments available at BCM [13] for the majority of the large D. pseudoobscura contigs. We also utilized the Ensembl mosquito genome server [36] to identify and visualize the chromosomal location of the A. gambiae annotated OXPHOS DNA sequences.
The D. melanogaster EST database, available from the National Center for Biotechnology Information (NCBI) contains ESTs from cDNA libraries obtained from different developmental stages and body parts. The relative abundance of the transcripts of duplicate or triplicate D. melanogaster OXPHOS genes was defined by counting their cognate ESTs in non-normalized cDNA libraries generated by the Berkeley Drosophila Genome Project (BDGP) [43] from embryos (LD), larvae/pupae (LP), and adult ovary (GM), head (GH) and testes (AT), and also the ESTs from adult testes generated at the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) [63]. ESTs from BDGP normalized EST libraries generated from head (RH) and embryos (RE) were also considered. The relative abundance of the transcripts of duplicate or triplicate A. gambiae OXPHOS genes was defined by counting their cognate ESTs in all libraries recovered from the Anobase server [21]. Since the number of sequences in the EST databases changes as new EST sequences are added, our values are calculated on the EST sequences present in the databases as of July 2004.
The list of D. melanogaster P-insertion OXPHOS mutants is reported in the MitoComp website [22] and was mostly compiled using information from FlyBase [42] and from the BDGP P-Element Gene Disruption Project [43].
Additional data files
A web-based dataset, MitoComp, contains all data on which this work is based and is available at [22]. It includes information on the cytological location of each gene, its genomic organization and the structure of its transcript(s). The genomic structures of the D. melanogaster, D. pseudoobscura and A. gambiae putative OXPHOS orthologs are shown and compared, and their deduced amino-acid products are aligned with the corresponding human protein. When paralogs of the gene exist, neighbor-joining trees derived from distance matrix analysis are also shown to visualize the evolutionary relationships between them. Additional data files available with the online version of this article are as follows. Additional data file 1 contains a table that reports pairwise amino-acid sequence conservation values between the D. melanogaster OXPHOS genes investigated and their D. pseudoobscura, A. gambiae and human counterparts. Additional data file 2 contains data extracted from the Parisi et al. dataset [40]. Additional data file 3 reports the codon usage in the orthologous and duplicate OXPHOS genes of D. melanogaster, D. pseudoobscura and A. gambiae.
Supplementary Material
Additional data file 1
A table that reports pairwise amino-acid sequence conservation values between the D. melanogaster OXPHOS genes investigated and their D. pseudoobscura, A. gambiae and human counterparts
Click here for additional data file
Additional data file 2
Data extracted from the Parisi et al. dataset
Click here for additional data file
Additional data file 3
The codon usage in the orthologous and duplicate OXPHOS genes of D. melanogaster, D. pseudoobscura and A. gambiae
Click here for additional data file
Acknowledgements
This work was supported by grants from Centro Eccellenza (CE) and Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR). We thank Cecilia Saccone and Graziano Pesole for critical reading of the manuscript.
Figures and Tables
Figure 1 Histogram of pairwise sequence identities between the unprocessed products of 78 orthologous D. melanogaster, D. pseudoobscura, A. gambiae and human OXPHOS genes.
Figure 2 Conservation of alternative splice variants of two OXPHOS genes in D. melanogaster, D. pseudoobscura and A. gambiae. (a,b) Schematic representation and comparison of intron-exon structure of the genes encoding the NADH ubiquinone-oxidoreductase acyl carrier protein and the ATP synthase epsilon chain in D. pseudoobscura (Dp), D. melanogaster (Dm) and A. gambiae (Ag). Coding exons are represented by red boxes and untranslated UTRs by blue boxes. Introns are not drawn to scale. Because no sufficient information is available about the transcribed non coding sequences of D. pseudoobscura, only the coding exons of the D. pseudoobscura genes are shown. mtacp1 exons duplicated in tandem are labelled 'a' and 'b'. (c) alignment of the amino-acid sequences encoded by the duplicate a and b exons of the mtacp1 gene in D. melanogaster (Dm), D. pseudoobscura (Dp), A. gambiae (Ag) and A. mellifera (Am). Residues conserved in both exons are shown in white on a black background. (d) Dendrogram showing the phylogenetic relationships between the duplicated exon DNA sequences used for the alignment shown in (c). The neighbor-joining tree derived from distance matrix analysis was constructed using MultAlin [62]. Other tree-construction methods produced similar results. PAM, percent point accepted mutations.
Figure 3 Length distribution of OXPHOS gene introns. (a) Length distribution of the 400 introns interrupting the coding sequence in the 78 D. melanogaster, D. pseudoobscura and A. gambiae OXPHOS genes investigated. (b) Comparison of the orthologous introns in the three species. Length of 138 D. melanogaster introns plotted in ascending length order was compared with the length of the 138 D. pseudoobscura orthologous introns and with the length of 98 orthologous A. gambiae introns. Note that length class shifts are rare.
Table 1 Number of exons and chromosomal localization of the 78 orthologous D. melanogaster, D pseudoobscura and A. gambiae OXPHOS genes
Cluster ID* Protein name D. melanogaster gene name Number of exons† Map position FlyBase ID D. pseudoobscura gene name Number of exons† Map position A. gambiae gene name Number of exons† Map position
Complex I: NADH:ubiquinone oxidoreductase
NUMM 13 kDa A subunit CG8680 3 2L;25C6 FBgn0031684 Dpse\CG8680 3 4 agEG14117 3 3R;33B
NUFM 13 kDa B subunit CG6463 3 3L;67E7 FBgn0036100 Dpse\CG6463 3 XR agEG15380 3 2L22E
NIPM 15 kDa subunit CG11455 2 2L;21B1-2 FBgn0031228 Dpse\CG11455 2 4 agEG13302 2 3R;35C-D
NUYM 18 kDa subunit CG12203 3 X;18C7 FBgn0031021 Dpse\CG12203 3 XL agEG18985 4 2L;27A
NUPM 19 kDa subunit CG3683 4 2R;60D13 FBgn0035046 Dpse\CG3683 4 3 agEG19249 3 2L;26B
NUKM 20 kDa subunit CG9172 1 X; 14A5 FBgn0030718 Dpse\CG9172 1 ND agEG16939 1 X;4A
NUIM 23 kDa subunit ND23 3 3R;89A5 FBgn0017567 Dpse\CG3944 3 2 agEG9698 2 2R;9A
NUHM 24 kDa subunit CG5703 3 X; 16B10 FBgn0030853 Dpse\CG5703 3 XL agEG16953 5 2R;11A
NUGM 30 kDa subunit CG12079 3 3L;63B7 FBgn0035404 Dpse\CG12079 3 XR agEG11610 3 2L;24D
NUEM 39 kDa subunit CG6020 4 3L;77C6 FBgn0037001 Dpse\CG6020 4 XR agEG18760 3 3L;40A
NUDM 42 kDa subunit ND42 2 3R;94A1 FBgn0019957 Dpse\CG6343 2 2 agEG10090 2 3L;41C
NUCM 49 kDa subunit CG1970 6 4;102C2 FBgn0039909 Dpse\CG1970 6 ND agEG18856 1 X;1B
NUBM 51 kDa subunit CG9140 4 2L;26B6-7 FBgn0031771 Dpse\CG9140 4 4 agEG9927 4 3R;36D
NUAM 75 kDa subunit ND75 5 X;7E1 FBgn0017566 Dpse\CG2286 5 XL agEG19681 4 2R;8D
NI8M B8 subunit CG15434 3 2L;24F3 FBgn0040705 Dpse\CG15434 3 4 agEG16251 3 2R;15B
NB2M B12 subunit CG10320 2 2R;57F6 FBgn0034645 Dpse\CG10320 2 3 agEG9277 1 3L;46D
NB4M B14 subunit CG7712 3 2R;47C6 FBgn0033570 Dpse\CG7712 3 3 agEG12033 2 2R;15A
N4AM B14.5A subunit CG3621 2 X; 2D6-E1 FBgn0025839 Dpse\CG3621 2 XL agEG14707 4 2R;17A
N4BM B14.5B subunit CG12400 3 2L:23D3 FBgn0031505 Dpse\CG12400 3 4 agEG16232 3 2R;13C
NB5M B15 subunit CG12859 2 2R;51C2 FBgn0033961 Dpse\CG12859 2 3 agEG17759 2 3L;44C
NB6M B16.6 subunit CG3446 2 X;5F2 FBgn0029868 Dpse\CG3446 2 XL agEG7829 3 3R;35A
NB7M B17 subunit l(2)35Di 3 2L;35D FBgn0001989 Dpse\CG13240 3 4 agEG18567 3 3R;34D
N7BM B17.2 subunit CG3214 4 2L;23A1 FBgn0031436 Dpse\CG3214 4 4 agEG10758 4 3R;31A
NB8M B18 subunit CG5548 1 X;13A8 FBgn0030605 Dpse\CG5548 1 XL agEG8436 3 2L;28C
NI2M B22 subunit CG9306 3 2L;34B8 FBgn0032511 Dpse\CG9306 3 4 agEG12344 3 3R;35C-D
ACPM Acyl carrier mtacp1 4 3L;61F6 FBgn0011361 Dpse\CG9190 4 XR agEG11237 5 3L;38B
NIAM ASHI subunit CG3192 3 X;6C5 FBgn0029888 Dpse\CG3192 3 XL agEG8821 3 2R;10A
NUML MLRQ subunit CG32230 3 3L;80E2 FBgn0052230 Dpse\CG32230 3 XR agEG12063 3 2R;15A
NINM MNLL subunit CG18624 1 X;7C FBgn0029971 Dpse\CG18624 1 XL agEG22692 1 X;5A
NIDM PDSW subunit Pdsw 3 2L;23F3 FBgn0021967 Dpse\CG8844 3 4 agEG7887 4 3R;29A
NISM SGDH subunit l(3)neo18 4 3L;68F5 FBgn0011455 Dpse\CG9762 4 XR agEG13573 2 2L;27D
NIGM AGGG subunit CG40002 3 ND FBgn0058002 Dpse\CG40002 3 XR agEG18653 2R;12D
Complex II: Succinate dehydrogenase
DHSA Flavoprotein subunit Scs-fp 4 2R;56D3 FBgn0017539 Dpse\CG17246 4 3 agEG7754 3 3L;38B
DHSB Iron-sulfur protein SdhB 3 2R;42D3-4 FBgn0014028 Dpse\CG3283 3 3 agEG13539 4 2L;27D
C560 Cytochrome B560 subunit CG6666 2 3R;86D7-8 FBgn0037873 Dpse\CG6666 2 2 agEG14929 2 3L;39B
DHSD Cytochrome b small subunit CG10219 4 3R;95B1 Fbgn0039112 Dpse\CG10219 4 XR agEG16772 3 X;1C
Complex III: Ubiquinol-cytochrome c reductase
UCRY 6.4 kDa protein CG14482 2 2R;54C9 FBgn0034245 Dpse\CG14482 2 3 agEG12505 2 3L;43B
UCRX 7.2 kDa protein ox 2 2R;49C2 FBgn0011227 Dpse\CG8764 2 3 agEG15210 2 2L;20C
UCRH 11 kDa protein Ucrh 2 3R FBgn0066066 Dpse\Ucrh 2 2 agEG19398 2 2R;11B
UCR6 14 kDa protein CG3560 3 X;14B10 FBgn0030733 Dpse\CG3560 3 XL agEG11611 3 3L;46A
UCRI Iron-sulfur subunit RFeSP 3 2L;22A3 FBgn0021906 Dpse\CG7361 3 4 agEG16975 4 3R;32C
CY1 Cytochrome c1, heme protein CG4769 6 3L;64C13 FBgn0035600 Dpse\CG4769 6 XR agEG19223 4 2L;26C
UCR1 Core protein 1 CG3731 6 3R;88D6 FBgn0038271 Dpse\CG3731 6 2 agEG21302 3 X;5C
UCR2 Core protein 2 CG4169 4 3L;73A10 FBgn0036642 Dpse\CG4169_1 4 XR agEG17930 4 2L;24A
UCRQ Ubiquinone-binding protein QP- CG7580 2 3L;74C3 FBgn0036728 Dpse\CG7580 2 XR agEG20223 2 3L;38C
Complex IV: Cytochrome c oxidase
CX41 Polypeptide IV CG10664 2 2L;38A8 FBgn0032833 Dpse\CG10664 2 4 agEG13327 2 3R;31C
COXA Polypeptide Va CoVa 1 3R;86F9 FBgn0019624 Dpse\CG14724 1 2 agEG19581 1 3L;41D
COXB Polypeptide Vb CG11015 3 2L;26E3 FBgn0031830 Dpse\CG11015 3 4 agEG8633 4 3R;31C
COXD Polypeptide VIa CG17280 2 2R;59E3 FBgn0034877 Dpse\CG17280 2 3 agEG7821 2 X;5A
COXG Polypeptide VIb CG18809 1 X;18E5 FBgn0042132 Dpse\CG18809 1 XL agEG11043 1 2L;25A
COXH Polypeptide VIc cype 2 2L;25D6 FBgn0015031 Dpse\CG14028 2 4 EST357342 2 3R;29A
COXK Polypeptide VIIa CG9603 2 3R;84F13 FBgn0040529 Dpse\CG9603 2 XR agEG17423 3 X;4B
COXO Polypeptide VIIc CG2249 2 2R;46D8-9 FBgn0040773 Dpse\CG2249 2 3 agEG22887 2 2L;28C
Complex V: ATP synthase
ATPA Alpha chain blw 4 2R;59B1-2 FBgn0011211 Dpse\CG3612 4 3 agEG7500 4 2L;21E
ATPB Beta chain ATPsyn-beta 3 4;102D1 FBgn0010217 Dpse\CG11154 3 ND agEG14379 1 3L;45C
ATPG Gamma chain ATPsyn-gamma 1 3R;99B10 FBgn0020235 Dpse\CG7610 1 2 agEG7678 2 3R;29C
ATPD Delta chain CG2968 3 X;9B4 FBgn0030184 Dpse\CG2968 3 ND agEG16076 1 3R;29B
ATPE Epsilon chain sun 4 X;13F12 FBgn0014391 Dpse\CG9032 4 ND agEG10095 4 X;3D
ATPF B chain ATPsyn-b 3 3L;67C5 FBgn0019644 Dpse\CG8189 3 XR agEG9580 3 2R;7A
ATPQ D chain ATPsyn-d 1 3R;91F FBgn0016120 Dpse\CG6030 1 ND agEG10180 3 3L;41C
ATPJ E chain CG3321 1 3R;88B4 FBgn0038224 Dpse\CG3321 1 2 agEG10809 3 2L;26B
ATPK F chain CG4692 2 2R;60D8-9 FBgn0035032 Dpse\CG4692 2 3 agEG1544 1 ND
ATPN G chain l(2)06225 2 2L;32C1 FBgn0010612 Dpse\CG6105 2 ND agEG8590 2 3R;34B
ATPR Coupling factor 6 ATPsyn-Cf6 2 3R;94E13 FBgn0016119 Dpse\CG4412 2 2 agEG19097 2 2R;19D
AT91 Lipid-binding protein P1 CG1746 3 3R;100B7 FBgn0039830 Dpse\CG1746 3 2 agEG14837 3 X;2B
ATPO OSCP Oscp 3 3R;88E8-9 FBgn0016691 Dpse\CG4307 3 2 agEG9393 3 2R;15D
Others
ATPW ATP synthase coupling factor B CG10731 1 2R;52F FBgn0034081 Dpse\CG10731 1 3 agEG15185 1 2R;19B
CI30 Complex I intermediate- associate protein 30 CG7598 2 3R;99B9 FBgn0039689 Dpse\CG7598 2 2 agEG7818 2 X;5A
CYC Cytochrome C Cyt-c-p 1 2L;36A11 FBgn0000409 Dpse\CG17903 1 4 agEG17602 1 3R;34C
COXZ Complex IV assembly protein COX11 CG6922 1 2L;25E5 FBgn0031712 Dpse\CG6922 1 4 agEG19985 2 3L;38B
COXS Complex IV copper chaperone CG9065 2 X;13A9 FBgn0030610 Dpse\CG9065_1 2 XL agEG23169 1 3L;44C
OXA1 Biogenesis protein OXA1 CG6404 3 3L;67F1 FBgn0027615 Dpse\CG6404 3 XR agEG11581 3 2L;22C
ETFA Electron transfer flavoprotein alpha subunit wal 3 2R;48C1-2 FBgn0010516 Dpse\CG8996 3 3 agEG11798 2 2R;17B
ETFB Electron transfer flavoprotein beta subunit CG7834 2 3R;99C1 FBgn0039697 Dpse\CG7834 2 2 agEG13614 2 2R;19D
ETFD Electron transfer flavoprotein- ubiquinone oxidoreductase CG12140 5 2R;46C4 FBgn0033465 Dpse\CG12140 5 3 agEG10998 4 2L;23B
COXX Protoheme IX farnesyltransferase CG5037 4 2L;31D9 FBgn0032222 Dpse\CG5037 3 ND agEG11452 4 3R;32B
SCO1 Sco1 protein homolog CG8885 2 2L;25B5 FBgn0031656 Dpse\CG8885 2 4 agEG10475 1 3R;31C
SUR1 Surfeit locus protein 1 Surf1 4 3L65D4 FBgn0029117 Dpse\CG9943 4 XR agEG8998 4 2L;25C
*IDs in this column are taken from Swiss-Prot [20]. †Only coding exons were considered. ND, map position not determined. D. melanogaster, D. pseudoobscura and A. gambiae sequences used to determine intron-exon gene structures are available as supplementary material at the MitoComp website [22]
Table 2 OXPHOS gene duplications in the genomes of D. melanogaster, D. pseudoobscura and A. gambiae
Protein name D. melanogaster gene name Number of exons Number of ESTs* Map position D. pseudoobscura gene name Number of exons Map position A. gambiae gene name Number of exons Number of ESTs Map position
Complex I: NADH:ubiquinone oxidoreductase
18 kDa subunit CG12203 3 X;18C7 Dpse\CG12203.1 3 XL agEG18985 4 2L;27A
Dpse\CG12203.2 3 2
20 kDa subunit CG9172 1 36 (1) X;14A5 Dpse\CG9172 1 ND agEG16939 1 47 X;4A
CG2014 1 0 3R;99B2 Dpse\CG2014 1 2 agEG12298 1 2 2R;14D
24 kDa subunit CG5703 3 33 (2) X;16B10 Dpse\CG5703 3 XL agEG16953 5 2R;11A
CG6485 1 4 (4) 3L;74A4 Dpse\CG6485 1 XR
49 kDa subunit CG1970 6 47 (0) 4;102C2 Dpse\CG1970 6 ND agEG18856 1 38 X;1B
CG11913 2 0 3R;96D2 Dpse\CG11913 2 2 agEG19332 1 0 2L;26B
51 kDa subunit CG9140 4 135 (2) 2L;26B6-7 Dpse\CG9140 4 4 agEG9927 4 3R;36D
CG11423 1 4 (4) 2R;54C12 Dpse\CG11423 1 3
CG8102 2 3 (3) 2R;51F3-4 Dpse\CG8102 2 3
B14.5A subunit CG3621 2 16 (1) X;2D6-E1 Dpse\CG3621 2 XL agEG14707 4 2R;17A
CG6914 1 3 (3) 3L;79F2 Dpse\CG6914 1 XR
Complex II: Succinate dehydrogenase
Flavoprotein subunit
Scs-fp
4 54 (0) 2R; 56D3 Dpse\CG17246 4 3 agEG7754 3 3L:38B
CG5718 1 5 (14) 3L;68E3 Dpse\CG5718 1 XR
Iron-sulfur protein
SdhB
3 83 (0) 2R;42D3-4 Dpse\CG3283 3 3 agEG13539 4 2L;27D
CG7349 3 14 (12) X;17F3 Dpse\CG7349 1 XL
Complex III: Ubiquinol-cytochrome c reductase
Cytochrome C1, heme protein CG4769 6 246 (3) 3L;64C13 Dpse\CG4769 6 XR agEG19223 4 2L;26C
CG14508 1 7 (7) 3R;99A1 Dpse\CG14508 1 2
11 kDa protein
Ucrh
2 16 (0) 3R Dpse\Ucrh 2 2 agEG19398 2 2R;11B
CG30354 1 1 (1) 2R;44E2
14 kDa protein CG3560 3 8 (0) X;14B10 Dpse\CG3560 3 XL agEG11611 2 3L;46A
CG17856 1 0 3R;98C3
Core protein 1 CG3731 6 3R;88D6 Dpse\CG3731 6 2 agEG21302 3 56 X;5C
agEG10358 1 2 2R;9A
gEG15332 1 46 2L;22D
Core protein 2 CG4169 4 3L;73A10 Dpse\CG4169.1 4 XR agEG17930 4 2L;24A
Dpse\CG4169.2 1 XR
Complex IV: Cytochrome c oxidase
Subunit IV CG10664 2 138 (3) 2L;38A8 Dpse\CG10664 2 4 agEG13327 2 3R;31C
CG10396 1 9 (7) 2R;41F3 Dpse\CG10396.1 1 2
Dpse\CG10396.2 1 XL
Polypeptide VB CG11015 3 41 (0) 2L;26E3 Dpse\CG11015 3 4 agEG8633 4 3R;31C
CG11043 2 4 (4) 2L;26E3 Dpse\CG11043 2 4
Polypeptide VIA CG17280 2 90 (2) 2R;59E3 Dpse\CG17280 2 3 agEG7821 2 63 X;5A
CG30093 1 1 (1) 2R;52D3 agEG4851 1 0 3R;32A
Polypeptide VIIA CG9603 2 30 (0) 3R;84F13 Dpse\CG9603 2 XR agEG17423 3 X;4B
CG18193 2 4 (4) 3R;84F13
Complex V: ATP synthase
Beta chain ATPsyn-beta 3 484 (6) 4;102D1 Dpse\CG11154 3 ND agEG14379 1 3L;45C
CG5389 3 3 (3) 3L;72D5-6 Dpse\CG5389 3 XR
Epsilon chain
sun
4 11 (0) X;13F12 Dpse\CG9032 4 ND agEG10095 4 15 X;3D
CG12810 1 0 3R;85F11 agEG20782 4 6 3R,34C
agEG8173 1 0 2L;21D
G chain
l(2)06225
2 90 (6) 2L;32C1 Dpse\CG6105 2 ND agEG8590 2 3R;34B
CG7211 2 1 (1) 2L;28C2 Dpse\CG7211 2 4
Coupling factor 6 ATPsyn-Cf6 2 55 (0) 3R;94E13 Dpse\CG4412 2 2 agEG19097 2 2R;19D
CG12027 2 2 (2) 3L;64C4 Dpse\CG12027 1 XR
Lipid-binding protein P1 CG1746 3 3R;100B7 Dpse\CG1746 3 2 agEG14837 3 408 X;2B
agEG12441 3 4 3L;42A
Others
Complex IV, copper chaperone CG9065 2 X;13A9 Dpse\CG9065.1 2 XL agEG23169 1 3L;44C
Dpse\CG9065.2 1 4
Cytochrome c
Cyt-c-p
1 134 (0) 2L;36A11 Dpse\CG17903 1 4 agEG17602 1 3R;34C
Cyt-c-d 1 25 (25) 2L;36A11 Dpse\CG13263 1 4
*The number of ESTs in testis-derived libraries is in parentheses. Because insufficient information on D. pseudoobscura ESTs is available in the public EST databases, only D. melanogaster and A. gambiae ESTs were considered. Bold type is used to identify the putative orthologous genes in the three species (see text). Only coding exons were considered. ND, location not determined. D. melanogaster, D. pseudoobscura and A. gambiae OXPHOS sequences used are available at the MitoComp website [22]
Table 3 Chromosomal location and interarm homology of the orthologous D. melanogaster, D. pseudoobscura and A. gambiae OXPHOS genes
D. pseudoobscura chromosomal arm A. gambiae chromosomal arm
2 3 XL XR 4 ND 2L 2R 3L 3R X ND
D. mel. 2L 18→ 16 2 1 1 16
D. mel. 2R 15→ 15 6 3 4 1 1
D. mel. 3L 12→ 12 7 2 3
D. mel. 3R 16→ 15 1 1 5 4 1 5
D. mel. X 14→ 11 3 3 4 2 2 3
D. mel. 4 2→ 2 1 1
D. mel. ND 1→ 1 1
The first column shows the distributions of the OXPHOS genes on D. melanogaster chromosomal arms (D. mel). Arrows show the direction of counting; D. melanogaster → D. pseudoobscura or D. melanogaster → A. gambiae. Bold type is used when inter arm homology is conserved betweeen two species. Note that Dm 2L, Ag 3R is the only correspondence between D. melanogaster and A. gambiae chromosomal arms. ND, location not determined.
==== Refs
Boffelli D McAuliffe J Ovcharenko D Lewis KD Ovcharenko I Pachter L Rubin EM Phylogenetic shadowing of primate sequences to find functional regions of the human genome. Science 2003 299 1391 1394 12610304 10.1126/science.1081331
Kellis M Patterson N Endrizzi M Birren B Lander ES Sequencing and comparison of yeast species to identify genes and regulatory elements. Nature 2003 423 241 254 12748633 10.1038/nature01644
Bergman CM Pfeiffer BD Rincon-Limas DE Hoskins RA Gnirke A Mungall CJ Wang AM Kronmiller B Pacleb J Park S Assessing the impact of comparative genomic sequence data on the functional annotation of the Drosophila genome. Genome Biol 2002 3 research0086.1 0086.20 12537575 10.1186/gb-2002-3-12-research0086
Zdobnov EM Von Mering C Letunic I Torrents D Suyama M Copley RR Christophides GK Thomasova D Holt RA Subramanian GM Comparative genome and proteome analysis of Anopheles gambiae and Drosophila melanogaster. Science 2002 298 149 159 12364792 10.1126/science.1077061
Saraste M Oxidative phosphorylation at the fin de siècle. Science 1999 283 1488 1493 10066163 10.1126/science.283.5407.1488
Skladal D Halliday J Thorburn DR Minimum birth prevalence of mitochondrial respiratory chain disorders in children. Brain 2003 126 1905 1912 12805096 10.1093/brain/awg170
Mootha VK Bunkenborg J Olsen JV Hjerrild M Wisniewski JR Stahl E Bolouri MS Ray HN Sihag S Kamal M Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria. Cell 2003 115 629 640 14651853 10.1016/S0092-8674(03)00926-7
Rubin GM Yandell MD Wortman JR Gabor Miklos GL Nelson CR Hariharan IK Fortini ME Li PW Apweiler R Fleischmann W Comparative genomics of the eukaryotes. Science 2000 287 2204 2215 10731134 10.1126/science.287.5461.2204
Reiter LT Potocki L Chien S Gribskov M Bier E A systematic analysis of human disease-associated gene sequences in Drosophila melanogaster. Genome Res 2001 11 1114 1125 11381037 10.1101/gr.169101
Green C Brown G Dafforn TR Reichhart JM Morley T Lomas DA Gubb D Drosophila necrotic mutations mirror disease-associated variants of human serpins. Development 2003 130 1473 1478 12588861 10.1242/dev.00350
Adams MD Celniker SE Holt RA Evans CA Gocayne JD Amanatides PG Scherer SE Li PW Hoskins RA Galle RF The genome sequence of Drosophila melanogaster. Science 2000 287 2185 2195 10731132 10.1126/science.287.5461.2185
Celniker SE Wheeler DA Kronmiller B Carlson JW Halpern A Patel S Adams M Champe M Dugan SP Frise E Finishing a whole-genome shotgun: release 3 of the Drosophila euchromatic genome sequence. Genome Biol 2002 3 research0079.1 0079.14 12537568 10.1186/gb-2002-3-12-research0079
Human Genome Sequencing Center at Baylor College of Medicine: Drosophila Genome Project
Holt RA Subramanian GM Halpern A Sutton GG Charlab R Nusskern DR Wincker P Clark AG Ribeiro JM Wides R The genome sequence of the malaria mosquito Anopheles gambiae. Science 2002 298 129 149 12364791 10.1126/science.1076181
Powell JR Progress and Prospects in Evolutionary Biology: The Drosophila Model 1997 Oxford: Oxford University Press
Gaunt MW Miles MA An insect molecular clock dates the origin of the insects and accords with palaeontological and biogeographic landmarks. Mol Biol Evol 2002 19 748 761 11961108
Bolshakov VN Topalis P Blass C Kokoza E della Torre A Kafatos FC Louis C A comparative genomic analysis of two distant Diptera, the fruit fly, Drosophila melanogaster, and the malaria mosquito, Anopheles gambiae. Genome Res 2002 12 57 66 11779831 10.1101/gr.196101
Thomasova D Ton LQ Copley RR Zdobnov EM Wang X Hong YS Sim C Bork P Kafatos FC Collins FH Comparative genomic analysis in the region of a major Plasmodium-refractoriness locus of Anopheles gambiae. Proc Natl Acad Sci USA 2002 99 8179 8184 12060762 10.1073/pnas.082235599
Sardiello M Licciulli F Catalano D Attimonelli M Caggese C MitoDrome: a database of Drosophila melanogaster nuclear genes encoding proteins targeted to the mitochondrion. Nucleic Acids Res 2003 31 322 324 12520013 10.1093/nar/gkg123
ExPASy - Swiss-Prot and TrEMBL
Blast server for Anopheles sequences
mitoloc_index: MITOCOMP
Schatz G Dobberstein B Common principles of protein translocation across membranes. Science 1996 271 1519 1526 8599107
Voos W Martin H Krimmer T Pfanner N Mechanisms of protein translocation into mitochondria. Biochim Biophys Acta 1999 1422 235 254 10548718
Roise D Schatz G Mitochondrial presequences. J Biol Chem 1988 263 4509 4511 9729103
von Heijne G Steppuhn J Herrmann RG Domain structure of mitochondrial and chloroplast targeting peptides. Eur J Biochem 1989 180 535 545 2653818
Ragone G Caizzi R Moschetti R Barsanti P De Pinto V Caggese C The Drosophila melanogaster gene for the NADH:ubiquinone oxidoreductase acyl carrier protein: developmental expression analysis and evidence for alternatively spliced forms. Mol Gen Genet 1999 261 690 697 10394906 10.1007/s004380050012
Copley RR Evolutionary convergence of alternative splicing in ion channels. Trends Genet 2004 20 171 176 15101391 10.1016/j.tig.2004.02.001
Graur D Li W-H Fundamentals of Molecular Evolution 2000 2 Sunderland, MA: Sinauer Associates Inc
Mount SM Burks C Hertz G Stormo GD White O Fields C Splicing signals in Drosophila: intron size, information content, and consensus sequences. Nucleic Acids Res 1992 20 4255 4262 1508718
Deutsch M Long M Intron-exon structures of eukaryotic model organisms. Nucleic Acids Res 1999 27 3219 3228 10454621 10.1093/nar/27.15.3219
Yu J Yang Z Kibukawa M Paddock M Passey DA Wong GK Minimal introns are not "junk". Genome Res 2002 12 1185 1189 12176926 10.1101/gr.224602
Ohno S Evolution by Gene Duplication 1970 Heidelberg, Germany: Springer-Verlag
Doolittle RF Of URFs and ORFs: A Primer on How to Analyze Derived Amino Acid Sequences 1986 Mill Valley, CA: University Science Books
Rost R Twilight zone of protein sequence alignments. Protein Eng 1999 12 85 94 10195279 10.1093/protein/12.2.85
Ensembl mosquito genome server
Lynch M Conery JS The evolutionary fate and consequences of duplicate genes. Science 2000 290 1151 1155 11073452 10.1126/science.290.5494.1151
Davis JC Petrov DA Preferential duplication of conserved proteins in eukaryotic genomes. PLoS Biol 2004 2 e55 15024414 10.1371/journal.pbio.0020055
Papp B Pàl C Hurst LD Dosage sensitivity and the evolution of gene families in yeast. Nature 2003 424 194 197 12853957 10.1038/nature01771
Parisi M Nuttall R Edwards P Minor J Naiman D Lu J Doctolero M Vainer M Chan C Malley J A survey of ovary-, testis-, and soma-biased gene expression in Drosophila melanogaster adults. Genome Biol 2004 5 R40 15186491 10.1186/gb-2004-5-6-r40
Lynch M Force A The probability of duplicate gene preservation by subfunctionalization. Genetics 2000 154 459 473 10629003
FlyBase
BDGP: Berkeley Drosophila Genome Project
Shields DC Sharp PM Higgins DG Wright F 'Silent' sites in Drosophila genes are not neutral: evidence of selection among synonymous codons. Mol Biol Evol 1988 5 704 716 3146682
Laird CD DNA of Drosophila chromosomes. Annu Rev Genet 1973 7 177 204 4593302 10.1146/annurev.ge.07.120173.001141
Cohen BA Mitra RD Hughes JD Church GM A computational analysis of whole-genome expression data reveals chromosomal domains of gene expression. Nat Genet 2000 26 183 186 11017073 10.1038/79896
Blumenthal T Evans D Link CD Guffanti A Lawson D Thierry-Mieg J Thierry-Mieg D Chiu WL Duke K Kiraly M Kim SK A global analysis of Caenorhabditis elegans operons. Nature 2002 417 851 854 12075352 10.1038/nature00831
Boutanaev AM Kalmykova AI Shevelyov YY Nurminsky DI Large clusters of co-expressed genes in the Drosophila genome. Nature 2002 420 666 669 12478293 10.1038/nature01216
Spellman PT Rubin GM Evidence for large domains of similarly expressed genes in the Drosophila genome. J Biol 2002 1 5 12144710 10.1186/1475-4924-1-5
Caron H van Schaik B van der Mee M Baas F Riggins G van Sluis P Hermus MC van Asperen R Boon K Voute PA The human transcriptome map: clustering of highly expressed genes in chromosomal domains. Science 2001 291 1289 1292 11181992 10.1126/science.1056794
Parisi M Nuttall R Naiman D Bouffard G Malley J Andrews J Eastman S Oliver B Paucity of genes on the Drosophila X chromosome showing male-biased expression. Science 2003 299 697 700 12511656 10.1126/science.1079190
Ranz JM Castillo-Davis CI Meiklejohn CD Hartl DL Sex-dependent gene expression and evolution of the Drosophila transcriptome. Science 2003 300 1742 1745 12805547 10.1126/science.1085881
Betran E Thornton K Long M Retroposed new genes out of the X in Drosophila. Genome Res 2002 12 1854 1859 12466289 10.1101/gr.6049
Emerson JJ Kaessmann Betran E Long M Extensive gene traffic on the mammalian X chromosome. Science 2004 303 537 540 14739461 10.1126/science.1090042
Lifschytz E Lindsley DL The role of X-chromosome inactivation during spermatogenesis (Drosophila-allocycly-chromosome evolution-male sterility-dosage compensation). Proc Natl Acad Sci USA 1972 69 182 186 4621547
Richler C Soreq H Wahrman J X inactivation in mammalian testis is correlated with inactive X-specific transcription. Nat Genet 1992 2 192 195 1345167 10.1038/ng1192-192
Sardiello M Tripoli G Romito A Minervini C Viggiano L Caggese C Pesole G Energy biogenesis: one key for coordinating two genomes. Trends Genet 2005 21 12 16 15680507 10.1016/j.tig.2004.11.009
Thornton K Long M Rapid divergence of gene duplicates on the Drosophila melanogaster X chromosome. Mol Biol Evol 2002 19 918 925 12032248
Zhang J Evolution by gene duplication: an update. Trends Ecol Evol 2003 18 292 298 10.1016/S0169-5347(03)00033-8
Pairwise alignments algorithm: Emboss-Align
Corpet F Multiple sequence alignment with hierarchical clustering. Nucleic Acids Res 1988 16 10881 10890 2849754
MultAlin
Andrews J Bouffard GG Cheadle C Lu J Becker KG Oliver B Gene discovery using computational and microarray analysis of transcription in the Drosophila melanogaster testis. Genome Res 2000 10 2030 2043 11116097 10.1101/gr.10.12.2030
Misra S Crosby MA Mungall CJ Matthews BB Campbell KS Hradecky P Huang Y Kaminker JS Millburn GH Prochnik SE Annotation of the Drosophila euchromatic genome: a systematic review. Genome Biol 2002 3 research0083.1 0083.22 12537572 10.1186/gb-2002-3-12-research0083
| 15693940 | PMC551531 | CC BY | 2021-01-04 16:05:37 | no | Genome Biol. 2005 Jan 31; 6(2):R11 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r11 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r121569394110.1186/gb-2005-6-2-r12ResearchIdentifying genetic networks underlying myometrial transition to labor Salomonis Nathan 12Cotte Nathalie 13Zambon Alexander C 13Pollard Katherine S 4Vranizan Karen 15Doniger Scott W 1Dolganov Gregory 3Conklin Bruce R [email protected] Gladstone Institute of Cardiovascular Disease, 1650 Owens Street, San Francisco, CA 94158, USA2 Pharmaceutical Sciences and Pharmacogenomics Graduate Program, University of California, 513 Parnassus Avenue, San Francisco, CA 94143, USA3 Department of Medicine, Cardiovascular Research Institute, University of California, 505 Parnassus Avenue, San Francisco, CA 94143, USA4 Center for Biomolecular Science and Engineering, University of California, 1156 High Street, Santa Cruz, CA 95064, USA5 Functional Genomics Laboratory, University of California, Berkeley, CA 94720-3860, USA6 Cellular and Molecular Pharmacology, University of California, 600 16th Street, San Francisco, CA 94143-2140, USA2005 28 1 2005 6 2 R12 R12 25 10 2004 3 12 2004 29 12 2004 Copyright © 2005 Salomonis 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.
A time course of gene expression at the onset of labor reveals transcriptional networks associated with activation of the uterine muscle and identifies targets for drugs to prevent premature labor.
Background
Early transition to labor remains a major cause of infant mortality, yet the causes are largely unknown. Although several marker genes have been identified, little is known about the underlying global gene expression patterns and pathways that orchestrate these striking changes.
Results
We performed a detailed time-course study of over 9,000 genes in mouse myometrium at defined physiological states: non-pregnant, mid-gestation, late gestation, and postpartum. This dataset allowed us to identify distinct patterns of gene expression that correspond to phases of myometrial 'quiescence', 'term activation', and 'postpartum involution'. Using recently developed functional mapping tools (HOPACH (hierarchical ordered partitioning and collapsing hybrid) and GenMAPP 2.0), we have identified new potential transcriptional regulatory gene networks mediating the transition from quiescence to term activation.
Conclusions
These results implicate the myometrium as an essential regulator of endocrine hormone (cortisol and progesterone synthesis) and signaling pathways (cyclic AMP and cyclic GMP stimulation) that direct quiescence via the transcripitional upregulation of both novel and previously associated regulators. With term activation, we observe the upregulation of cytoskeletal remodeling mediators (intermediate filaments), cell junctions, transcriptional regulators, and the coordinate downregulation of negative control checkpoints of smooth muscle contractile signaling. This analysis provides new evidence of multiple parallel mechanisms of uterine contractile regulation and presents new putative targets for regulating myometrial transformation and contraction.
==== Body
Background
The initiation of mammalian labor is a complex physiological process that requires the expression and secretion of many factors, both maternal and fetal [1,2]. The majority of these factors exert their effect on the myometrium, the smooth muscle responsible for expelling the fetus from the uterus. While species differences in labor regulation have been observed, several common signaling pathways and factors have been implicated as key regulators across species. During mid to late gestation, myometrial quiescence is maintained by several contractile inhibitors, such as relaxin, adrenomedullin, nitric oxide, prostacyclin and progesterone [1,2]. A number of these regulators stimulate cyclic AMP (cAMP)- and cGMP-mediated signaling pathways. Smooth muscle contraction is inhibited by the phosphorylation of myosin light-chain kinase by the cAMP-dependent protein kinase. This inhibition is believed to promote quiescence. In addition, the myometrium undergoes major structural changes throughout pregnancy that are required to generate the necessary contractile force for labor, including hypertrophy and hyperplasia of smooth muscle, connective tissue, focal adhesion, and cytoskeletal remodeling [3].
The transition to labor results in synchronous contractions of high amplitude and high frequency by the myometrium. Factors previously associated with the regulation of myometrial activation include the oxytocin receptor, gap junction protein connexin-43, voltage-gated calcium channels, prostaglandin receptor subtypes, estrogen, cortisol and transcription factors c-Jun and c-Fos. Most of these proteins participate in pathways that stimulate calcium release (for example, calcium-calmodulin G protein signaling) and the formation of intracellular junctions, leading to stimulation of contractions. Although several important components that regulate the initiation of labor have been identified, the mechanisms that guide this transition are poorly understood.
A difficult challenge in identifying the regulatory events that control the switch from myometrial quiescence to activation has been developing tools for examining whole-genome expression profiles in the context of known biology. Recent efforts to identify transcriptional changes from laboring and non-laboring human myometrium have proved valuable in identifying putative physiological regulators [4-8]; however, the lack of gestational time points examined has limited these approaches to interrogating only those genes with large fold-changes at term activation without exploring the global patterns of gene expression over the time-course of myometrial transformation. While gene profiling of the rodent uterus during gestation has proved fruitful in revealing some of the large-scale patterns of gene expression throughput pregnancy [5,9], there is still a critical need to improve the global view of myometrial gene expression with greater temporal resolution using newly developed bioinformatic tools.
To identify molecular mechanisms involved in the transition from myometrial quiescence to labor, we analyzed gene-expression changes in mouse myometrium at mid-gestation, throughout late gestation, and during the postpartum period. Our results reveal several novel patterns of expression occurring along the phases of myometrial quiescence to term activation and postpartum involution. Analysis of putative quiescence and term activation regulators in the context of well defined biological pathways revealed new putative functional roles for several previously unassociated genes in the suppression of contraction throughout gestation and activation of phase-dependent contractions at labor. This analysis further implicates the regulation of several novel pathways, including smooth muscle-extracellular matrix interactions throughout late gestation and cell junction-cytoskeletal interactions immediately before the onset of labor.
Results
Clustering of expression changes in gestational myometrium
Messenger RNA transcript levels were measured from isolated myometrium of 35 time-mated mice at four time-points of late gestation (14.5-18.5 days), at postpartum (6 and 24 hours after labor), and from a non-pregnant control group. In all, approximately 13,000 probe sets corresponding to around 9,000 unique cDNAs and expressed sequence tags (ESTs) were probed with oligonucleotide microarrays. About 35% of these transcripts were regulated throughout gestation and postpartum (14.5 days through 24 hours postpartum) using the criteria of p < 0.05 and a change in level of expression of more than 20% (fold-change 0.2).
Analysis of these probe sets with HOPACH [10-12] revealed eight primary cluster groups and 133 subclusters. The majority of these clusters showed a clear association with known physiological phases of uterine gestation: quiescence (clusters 2, 3, 7 and 8), term activation (cluster 6), and postpartum involution (clusters 3, 4 and 7). In addition to these clusters, we observed two cluster groups with genes downregulated or upregulated throughout the analyzed time-course (clusters 1 and 5) (Figure 1).
MAPPFinder analysis
To characterize the major biological processes, molecular functions, and cellular components associated with the HOPACH pattern groups, we used MAPPFinder (a component of GenMAPP version 2.0) [13-16]. MAPPFinder produced a statistically ranked list (based on p-value) of Gene Ontology (GO) biological categories associated with each cluster, from which the most significant nonsynonymous groups are listed (Figure 1, GO categories). In each cluster, several highly significant biological associations were identified (adjusted permutation p < 0.05).
Association of expression clusters with previously associated uterine quiescence and activation genes
Gene expression groups associated with the maintenance of pregnancy (quiescence) or induction of labor (activation) were confirmed by mapping lists of previously associated regulators of uterine quiescence and activation onto our HOPACH cluster map. Extensive literature searches for such regulators identified 66 genes, of which 23 were regulated in our dataset (Figure 1, previously associated regulators). Genes hypothesized to regulate quiescence by transcriptional upregulation or secretion were largely associated with clusters 7 and 8 ('increased quiescence'), while putative activators of uterine activation were largely associated with cluster 6 ('increased term activation'). Although only three downregulated quiescence regulators were associated with HOPACH clusters, two of them mapped to cluster 2 ('decreased quiescence'), as predicted.
Functional analysis of quiescence and term activation pattern groups
To further elucidate specific genes and pathways linked to the regulation of uterine quiescence and the initiation of labor, we examined pattern groups linked to quiescence and term activation, in the context of GO categories, GenMAPP pathway maps and literature associations. While low-magnitude fold-changes have been included within these functional analyses to broaden our survey of biological groups, we have largely restricted our discussion to transcripts with fold-changes greater than two.
Upregulation of pathways of relaxation and remodeling during quiescence
Analysis of genes upregulated throughout gestation (increased quiescence) revealed a number of biological categories associated with uterine quiescence. These categories contain a large number of highly regulated genes coupled to the inhibition of prostaglandin and cortisol synthesis, stimulation of cAMP and cGMP signaling pathways, extracellular matrix remodeling, cytolysis and regulation of cell growth (Figure 2, Table 1). To explore the potential relationships between the products of these transcriptionally regulated genes, we mapped the data onto respective metabolic and signaling pathways (Figure 3a,b).
Besides well established quiescence regulators (Adm, Cgrp, Hsd11b2, Gnas, Cnn1 and Utg; see Tables 1, 2, 3 for full gene names), several genes previously unassociated with the maintenance of quiescence were identified along the same or related biological pathways. The most highly regulated of these genes were those implicated in the induction of cGMP and cAMP signaling pathways (Guca2b and Cmkor1), genes for calcium-dependent phospholipid binding proteins (Anxa1, Anxa2, Anxa3 and Anxa8), and for the Anxa2 dimerization partner S100A10 (Figure 3a). Other changes in expression from this pattern group were observed among cytolysis-inducing proteases (granzymes B-G), regulators of cell growth (Igfbp2 and Il1r2), and transcriptional regulation (Sfrp4 and Klf4). Several of these and other genes were found to have highly reproducible patterns of expression using quantitative real-time PCR (TaqMan), with typically larger fold-changes produced by TaqMan than by GeneChip (consistent with the more conservative fold-changes typically produced after robust multi-array average (RMA) normalization) (see Additional data file 1).
Several genes for cAMP-response element transcription factors were also found within the increased quiescence group (Atf4, Crebl1, and Creb3, see Figure 3b). These are all members of a larger group of basic leucine zipper (bZip) transcription factors not previously associated with quiescence, which also includes the CCAAT/enhancer binding protein Cebpd, the Maf protein Mafk, the nuclear factor, interleukin-3, regulated Nfil3, and the X-box binding protein Xbp1, also upregulated with quiescence.
Downregulation of mRNA processing and contraction-associated signaling during quiescence
MAPPFinder analysis of genes in the decreased quiescence group identified a wide variety of cell maintenance, transcription, and cell-signaling biological processes. Many of these GO categories were associated with the onset of labor (calcium-ion transport and protein tyrosine phosphatase activity) or myometrial postpartum involution (programmed cell death, collagen catabolism and ubiquitin-conjugating enzyme activity). These results are in accordance with the inhibition of contraction and suppression of cell death in late gestation. Unlike term-related biological processes, categories shared between the decreased quiescence and 'increased postpartum involution' group appear to be largely the result of a common transcript expression profile (Figure 1, cluster 3; Figure 2).
Although similar numbers of genes were downregulated or upregulated with quiescence (approximately 480-520 genes), very few genes were downregulated more than twofold at 14.5 days of gestation (Table 2). One of the most downregulated transcripts was the myosin light-chain gene Myl4; the Myl4 protein is the primary target for oxytocin-induced phosphorylation leading to uterine contraction at term. Several additional putative components of the oxytocin contractile signaling pathway (calcium-calmodulin signaling pathway) were also present in this expression group (Iptr1, Ryr3, Plcg1, and Atp2a2) (Figure 3b). Another large set of coordinately downregulated genes includes factors involved in RNA processing. Alternative splicing of putative quiescence and term activation regulators has been proposed to be a critical mechanism of the physiological switch to labor [17,18].
Transition from remodeling and relaxation to cell-cell signaling and transcriptional regulation with activation of the myometrium at term
A large percentage of genes regulated with quiescence continued to be highly regulated at term. This result emphasizes the importance of expression changes immediately before labor to counteract the effects of quiescence. Consistent with the number of upregulated genes, MAPPFinder analysis of the increased term activation group identified a smaller set of GO terms and pathways. Prominent among these were genes associated with the formation of cell junctions, kinesin complexes and endopeptidase inhibitors. In addition, functionally related transcription factors (members of the basic helix-loop-helix (bHLH) family), ion transport proteins and ion transport regulators were coordinately upregulated at term.
Within these biological categories, several contractile regulators, both associated and unassociated with parturition, were highly upregulated. These genes include those for cell junction proteins (Cx43, Cx26, Ocln, and Dsp), the pulmonary smooth muscle contractile regulator and complement component C3, the estrogen signaling regulator Hsp70, the chloride conductance regulator Fxyd3 and the ryanodine receptor regulator Gsto1 (Table 3). These changes occurred in concert with the upregulation of signaling molecules, such as growth factors (Inhba, Inhbb), G-protein signaling components (Edg2, Gng12) (Figure 3b) and collagen catabolism proteins (Pep4, Mmp7). On the whole, however, this pattern group was dominated by the upregulation of genes encoding proteins that are largely epithelial-cell specific. Most prominent among these are the genes for the cytokeratin intermediate filament proteins, Krt2-7, Krt2-8, Krt1-18, and Krt1-19, and for the cytokeratin transcriptional regulator Elf3, which are among the most highly upregulated genes at term.
Downregulation of pathways of calcium mobilization and G-protein signaling in term myometrium
HOPACH analysis with a metric that disregarded the direction of fold-change (see Additional data file 2) revealed a small number of downregulated genes at term that mirror the increased term activation group. Among these, we observed two highly downregulated genes: regulator of G-protein signaling 2 (Rgs2), a potent inactivator of Gαq-GTP bound activity, and inhibitor of DNA binding 2 (Idb2), a bHLH factor that heterodimerizes with other HLH proteins to inhibit their function. Rgs2 is one of the most downregulated genes throughout the gestation-postpartum time-course, in addition to being highly expressed in non-pregnant myometrium and throughout gestation. Additional term-downregulated G-protein signaling proteins that act to antagonize calcium-calmodulin signaling are illustrated in Figure 3b.
Global mechanisms of transcriptional regulation
One of the most prominent observations in this dataset is the highly significant correlation in the expression and genomic position of genes for eight serine-type endopeptidases (Gzmb through Gzmg, Mcpt8 and Ctsg) during the phase of quiescence. Genes within this multigene cluster undergo tight coordinate regulation in response to cell stimulation [19,20]. Examination of this expression cluster group in the context of genomic position reveals a novel pattern of positional gene regulation, where relative fold-change in expression increases from the peripheral members in the cluster to the center of the gene cluster (Figure 4a).
To determine whether other gene clusters exhibit a similar form of positional co-regulation, we developed a program to identify genomic intervals containing several coexpressed genes. Searching for regions with three or more members in a broad genomic interval (500 kilobases (kb)), we identified 11 clusters of genes that are co-localized and co-regulated (the same HOPACH cluster) [21]. Among these, we were able to identify at least one other gene cluster that possessed a genomic pattern of gene expression similar to that of the granzyme cluster, with genes maximally upregulated postpartum (Figure 4b). These genes, which encode several of the collagen catabolism matrix metalloproteinases, Mmp3, Mmp10, Mmp12 and Mmp13, are among the most highly upregulated genes postpartum. Because we do not have data from full genome arrays, it is difficult to determine if these co-regulated clusters of genes occur more frequently. However, these co-regulated gene clusters suggest coordinated gene regulation by an unknown mechanism.
Discussion
This time-course analysis provides the first global view of gene-expression changes in mouse myometrium from uterine quiescence through the activation of the myometrium before labor and to its postpartum involution. Examination of multiple time points, the use of replicates, robust array normalization and powerful clustering tools enabled us to delineate and characterize unique patterns of gene expression throughout this physiological process. In addition to partitioning clusters of genes, analysis with the program HOPACH also provides us with a continuum of expression changes that reveals an overall transition in the expression of genes from one cluster group to another (Figure 1). Annotation of these clusters with GO terms provides a bird's eye view of the major processes regulating each of these pattern groups. These results support the hypothesis that mid-to-late gestation is dominated by changes in the expression of genes related to cell growth and extracellular-matrix remodeling (cluster 7), term gestation by changes in the content of cell junctions (cluster 6), and postpartum by targeted protein degradation, collagen digestion and apoptosis (clusters 3 and 4). Furthermore, results from genes upregulated throughout gestation and through postpartum suggest a continual local uterine immune response throughout this process (cluster 5). To help visualize the large-scale gene-expression changes in the context of myometrial physiology, we have depicted the data in an animation (see Additional data file 3) that summarizes our major findings.
A number of studies emphasize the importance of fetal regulation of the switch from quiescence to term activation, particularly increased cortisol and estrogen output from the fetal adrenal gland [1,2]. Interestingly, our studies provide evidence of a dynamic interplay between the myometrium and the fetus, particularly at the level of cortisol and progesterone synthesis (Figure 3a). Genes highly upregulated with quiescence include Hsd11b, which encodes an enzyme that converts cortisol to the inactive cortisone, and Cyp11a1, encoding an enzyme that promotes the synthesis of progesterone. Conversely, Hsd11a, coding for an enzyme that catalyzes the synthesis of cortisol, increased expression from 11- to 18-fold throughout gestation, suggesting that local regulation of cortisol levels are important for myometrial activation. While we observed the upregulation of the estrogen signaling regulator Hsp70, with term activation, downstream markers of estrogen action are among the most highly upregulated genes with term activation, supporting the role of the fetus in myometrial activation.
Examination of highly upregulated putative quiescence and term activation genes revealed several novel changes within important associated pathways for quiescence and activation (cAMP and cGMP signaling, calcium and calmodulin signaling and prostaglandin synthesis). Proteins encoded by these genes include Guca2b (uroguanylin), Anxa3, and Anxa8 with quiescence, and C3, Edg2, Gsto1 and Fxyd3 during activation (see Figure 3). These factors may represent novel targets for controlling the length of gestation. This is evidenced by the parallel observed upregulation of Guca2b from a recent microarray analysis of rat uterine gestation, where this factor has also been proposed to be a crucial regulator of cGMP-mediated smooth muscle relaxation throughout late pregnancy [9,22]. We have validated the expression patterns of a number of these genes using quantitative real-time PCR (see Additional data file 1). In addition to the candidates mentioned here, a number of other highly upregulated genes, whose functions have not been elucidated are also found in these two expression groups (see Additional data file 6).
Although a number of genes upregulated with quiescence or with term activation can be clearly implicated in the regulation of contractile pathways or uterine growth, several more groups of genes with little known functional connection to these processes were coordinately expressed. Highlighted among these groups are serine endopeptidases (granzymes) and bZip transcription factors, upregulated during quiescence, and endopeptidase inhibitors and bHLH factors, upregulated with term activation. In addition to its role in cytolysis, granzyme expression and secretion by T lymphocytes has been associated with the breakdown of extracellular matrix proteins in the uterus during pregnancy [18,23,24]. Interestingly, the upregulation of serine endopeptidases appears to be antagonized before the onset of labor by the upregulation of several serine endopeptidase inhibitors with term activation. A similar antagonistic relationship may also exist for bHLH factors upregulated at term with inhibitors of HLH function that are upregulated with quiescence and become downregulated at term.
Although the myometrium is considered to be relatively homogeneous, many of the largest changes in gene expression at term occurred in genes that are not normally associated with muscle, such as the keratins, tight junction and desmosome junction proteins. Indeed, altered gene expression due to changes in cell-type distribution or the invasion of the myometrium by the decidua and endometrium would not be distinguished if those changes occur consistently between gestational myometrium preparations. Further inspection of the literature reveals that the cytokeratins, which compose the bulk of this group, are expressed within smooth muscle and probably function as components of intermediate filaments of the cytoskeleton [25-28]. Furthermore, several components of desmosome spot junctions and hemidesmosomes, which interact with keratin intermediate filaments and the extracellular matrix to impart tensile strength between cells, are also upregulated with term activation (see Additional data file 3). These data suggest that an increase in rigidity-imparting cell junctions and remodeling of the cytoskeleton immediately before labor may promote coordinate contractions. However, further studies are needed to determine if cytokeratin expression at term occurs within resident or infiltrating cells.
In addition to the capability to group and annotate clusters of genes, pattern analysis with HOPACH can be used to interrogate gene clusters in the context of genomic location. For this analysis, we developed a program to isolate gene clusters that are likely to be co-regulated on the basis of genomic location, similar to other reported methods [29-32]. Using this program, we identified genomic regions that undergo correlated changes in gene expression associated with specific phases of the myometrial time-course. These groups highlight novel forms of gene regulation during quiescence and postpartum to coordinate cell responses (serine-protease activation and collagen catabolism). The prominent co-regulation among members of these two gene clusters further suggests that immune-cell trafficking and activation also play important roles in the progression towards labor and recovery from pregnancy.
Conclusions
We have identified several highly regulated genes not previously associated with myometrial quiescence or activation, in addition to families of genes co-regulated at different phases of the myometrial time-course. In addition to providing new hypotheses about how the switch from quiescence to term activation may be facilitated (Figure 5), these data highlight several proteins that may serve as new candidate pharmacological targets for regulating myometrial contraction and thus the onset of labor. Such analyses will also be useful in predicting and correlating gene-expression changes in human pregnancy, where several time-points are often difficult to obtain [4-8]. Similar studies in other species using complementary methods of transcript measurement will also be necessary to validate these changes and understand the species-specific and regional myometrium transcriptional differences that probably occur. A detailed examination of the precise physiological roles of these regulators and mechanisms of regulation will be essential for developing a more detailed view of the regulation of labor.
Materials and methods
Tissue harvesting
FVB/N mice (Jackson Laboratory) were sacrificed in the morning (10 to noon) at 14.5 (n = 3), 16.5 (n = 4), 17.5 (n = 5), or 18.5 days (n = 7) after timed mating, and 6 (n = 4) or 24 h (n = 4) after delivery. Control myometrium was harvested from non-pregnant littermate females (n = 8) 1 day after timed mating with a vasectomized male. After dissection of both uterine horns, the tissue closest to the cervix was removed. Each horn was washed with PBS and opened longitudinally. Pups and placenta were discarded, and the decidua was removed by blunt dissection. The myometrium from each horn was then immediately frozen in liquid nitrogen and stored at -80°C.
Sample preparation and microarray data normalization
For each sample, labeled cRNA was prepared from 20 μg purified total RNA and hybridized to Affymetrix Mu11k A and B arrays according to the manufacturer's instructions. Tissue from each mouse was hybridized individually to one array set. Microarrays were scanned at a photomultiplier tube (PMT) setting of 100%. Resulting .cel files were generated with Affymetrix Microarray Suite 5.0 and analyzed with RMA [33].
Statistical analysis
To identify transcripts differing in mean expression across the seven experimental groups, p-values were calculated from a permutation test with the F-statistic function from the multtest package of Bioconductor [12,34]. Fold-changes in transcript levels were calculated from the mean log2 expression values of each time-point group versus the mean of non-pregnant controls. For cluster analysis, the dataset was filtered for probe sets with a p < 0.05 across the full expression time-course and a greater than 20% change in level of expression (positive or negative) for at least one time-point group versus non-pregnant controls. Additional filters were used downstream of clustering for genes related to uterine quiescence and term activation. For clusters related to quiescence and term activation, a change of more than 20% was required for the midgestation (14.5 days) and term (18.5 days) time points, respectively, versus non-pregnant controls.
Clustering and pattern analysis
Gene expression clustering for 4,510 significant probe sets was performed using the program HOPACH (hierarchical ordered partitioning and collapsing hybrid), with uncentered correlation distance [10-12]. HOPACH produced a tree with six levels of clusters (eight primary level clusters and 133 main clusters). To examine expression patterns independently of the direction of the fold change, HOPACH was re-run with absolute uncentered correlation distance. Associations with GO biological process, molecular function, cellular component groups, and GenMAPP biological pathways were obtained with MAPPFinder 2.0, a part of the GenMAPP 2.0 application package [13-16]. A permuted p-value was calculated by MAPPFinder 2.0 to adjust for multiple hypothesis testing (see Additional data file 7). Because of the highly redundant nature of the oligonucleotide arrays used, redundant probe sets corresponding to a single gene were identified from the Affymetrix NetAFFX website [35].
Real-time PCR validation of microarray data
Real-time reverse transcription PCR (RT-PCR) was used to validate the expression patterns of several highly regulated genes associated with specific phases of myometrium gestation. Gene-specific primers for multiplex real-time RT-PCR were designed for each gene of interest (n = 18) using Primer Express software (Perkin Elmer) and based on sequencing data from the National Center for Biotechnology Information (NCBI) databases and purchased from Biosearch Technologies. Sequence data for all oligos are available online [36]. Total RNA concentration and quality was assessed using the Agilent Bioanalyzer 2001. First-strand cDNA synthesis was performed using total cellular RNA (BD Biosciences Clontech), Powerscript reverse transcriptase (BD Biosciences Clontech), and random hexamer primers. Finally, an equivalent of 10 ng of total RNA from the first-strand cDNA synthesis reaction was used in 10 μl of each TaqMan gene quantification in 384-well format. Universal Master Mix for real-time PCR was purchased from Invitrogen Life Technologies. Raw data from an ABI Prism 7900 (Applied Biosystems) were processed into Excel spreadsheets and conversion of raw Ct values to relative gene copy numbers (GCN) was done as described previously [37]. Gene-expression analysis requires proper internal control genes for normalization. By using an endogenous control as an active reference, quantification of an mRNA target can be normalized for differences in the amount of total RNA added to each reaction. For this purpose, we used four mouse housekeeping genes - PPIA, GAPDH, PGK1 and S9. Moreover, using GeNorm [38], we selected PGK1 and GAPDH as the two most stable housekeeping genes across all 12 specimens and used their geometric means for normalization. Normalized data were graphed and compared to the data generated on similar specimens via microarrays. Genes could be broken down into the following groups: 13 genes with concordant microarray-TaqMan patterns; one false-negative result by microarray (Acta2); three genes with high TaqMan variability (Mmp9, Krt19, Id1); and one gene with evidence of alternative splicing (Csb) (see Additional data file 1). It should be noted that Acta2 baseline expression was relatively high for both microarray and TaqMan results. As both of these techniques probed different regions of the Acta2 gene, we cannot exclude the possibility of alternative splicing.
Chromosomal localization analysis
We constructed a program to link HOPACH expression data to chromosome transcription start-site location and strand orientation, obtained from the Ensembl database [39]. Co-localized clusters of genes were identified as those genes clustered within a 500-kb genomic interval, belonging to the same HOPACH cluster, with a z-score >1.96, and an average pairwise Pearson correlation among cluster members of r >0.65 (see Additional data file 7 for calculation details and [21] for the full supplemental chromosome cluster lists).
Additional data files
The following additional data are available with the online version of this article. Additional data file 1 is a figure showing the TaqMan vs GeneChip gene expression patterns. Relative fold changes (log base 2) are shown for 18 genes identified by these GeneChip studies to be differentially regulated throughout the myometrium gestation time-course. Combined standard errors are shown for each gestational time-point as compared to the non-pregnant control group. Additional data file 2 is a figure showing the HOPACH Absolute Value Pearson Correlation of Myometrial Expression Data. Gene expression data used for Pearson correlation HOPACH was used to generate a new set of clusters with a metric that disregards the direction of fold-change. Genes downregulated with term are identified based on association with genes upregulated at term from the non-absolute HOPACH analysis. Additional data file 3 is an animation of the summary and results, with a cartoon representation of myometrial transformation, general experimental design, results and conclusions. Additional data files 4 and 5 are Excel tables listing the MAPPFinder results. Nonsynonymous MAPPFinder GO categories for each expression pattern group are provided. Reanalysis with GenMAPP version 2.0 is required to visualize the genes that associate with each GO term. To download GenMAPP version 2.0, go to [16]. Additional data file 6 is a set of tables of cluster groups with annotations. Expression data, statistics, and biological groupings based on Gene Ontology annotations (via MAPPFinder analysis) and the literature are provided for 'Quiescence', 'Activation', and Postpartum 'Involution' gene lists. Additional data file 7 contains additional details of methods. Additional data file 8 contains the full expression dataset as an Excel file and Additional data file 9 is a GenMAPP format GEX file for use with GenMAPP format pathway maps (MAPP files). MAPP files can be downloaded from [16].
Supplementary Material
Additional data file 1
A figure showing the TaqMan vs GeneChip gene expression patterns
Click here for additional data file
Additional data file 2
A figure showing the HOPACH Absolute Value Pearson Correlation of Myometrial Expression Data
Click here for additional data file
Additional data file 3
An animation of the summary and results, with a cartoon representation of myometrial transformation, general experimental design, results and conclusions
Click here for additional data file
Additional data file 4
Excel tables listing the MAPPFinder results: HOPACH unique GO
Click here for additional data file
Additional data file 5
Excel tables listing the MAPPFinder results: MAPPFinder quiescence, activation and postpartum
Click here for additional data file
Additional data file 6
A set of tables of cluster groups with annotations
Click here for additional data file
Additional data file 7
Additional details of methods
Click here for additional data file
Additional data file 8
The full expression dataset as an Excel file
Click here for additional data file
Additional data file 9
A GenMAPP format GEX file for use with GenMAPP format pathway maps (MAPP files)
Click here for additional data file
Acknowledgements
We thank Chris Barker, Kristina Hanspers, Yanxia Hao, and Anita Chow from the Gladstone Genomics Core, and Michael McMaster for his assistance with uterine dissections. We thank Susan Fisher, Janet A. Warrington, Gary Howard, Bethany Taylor and members of the Conklin lab for helpful discussions and editorial assistance. This work is supported by the J. David Gladstone Institutes and grants from the National Institutes of Health: NHLBI, R01-HL61689 (B.R.C.) NHGRI R01-HG002766 (B.R.C) and SBIR 1R44DK53325-01 (Janet A. Warrington and B.R.C.), and T32 GM07175 (N.S.).
Figures and Tables
Figure 1 Clustering of myometrial expression profiles with HOPACH. Gene-expression profiles for 27 microarrays (vertical axis) and 4,510 probe sets (horizontal axis) are shown in the context of the HOPACH cluster map (non-pregnant data excluded). The array groups correspond to mid to late gestation (14.5, 16.5, 17.5 and 18.5 days) and postpartum (6 and 24 h). Eight clusters of genes are arranged vertically. Physiological phase groups are assigned on the basis of visual observation and association with previously associated regulators. MAPPFinder results are shown for the top-ranking distinct biological process, molecular function and cellular component groups based on a permuted p-value. Previously associated regulators of uterine quiescence and activation are indicated by a colored line next to the location of the corresponding gene probe set in the cluster map.
Figure 2 Association of quiescence and term activation pattern groups with biological pathways. Significant associations to GO classification groups and GenMAPP pathways were determined for each of the four expression pattern groups examined: Displayed are representative gene expression patterns for increased and decreased quiescence and term activation. (a) increased quiescence (yellow curve), increased activation (red curve); (b) decreased quiescence (green curve) and decreased activation (blue curve). GO terms and GenMAPP pathways highlighted by analysis with the program MAPPFinder are indicated by italicized blue text. Biological processes identified by literature association are indicated in black text. Parent biological categories are designated by bold text.
Figure 3 Analysis of pathways of uterine smooth muscle contraction. (a) Prostaglandin synthesis and (b) G-protein signaling pathways in the myometrium are overlaid with gene-expression color criterion and fold-changes from the program GenMAPP. Interactions suggested by results of this microarray analysis are included in these figures. Detailed gene-expression data, statistics and full gene annotations are available on the GenMAPP interactive version of these pathways online [40].
Figure 4 Association of genomic localization with co-regulation of expression. (a,b) Chromosomal gene clusters contain highly correlated expression changes among multiple members. Global patterns of gene expression within these genomic intervals are visualized by representing mean log expression for four of the myometrium time-point groups (non-pregnant, 14.5 and 18.5 days gestation, and 24 h postpartum), versus relative gene position on the chromosome. Gene strand orientation and position is designated by the orientation of arrows. Gene symbols above and below arrows are shown, where italicized black text indicates co-regulated genes (same HOPACH cluster) and italicized gray genes not co-regulated for (a) increased quiescence and (b) increased postpartum involution. Non-italicized gray text indicates genes not probed by the arrays.
Figure 5 Proposed maternal model of uterine-directed contractile regulation. Theoretical model based on the major gene-expression pattern groups for quiescence, term activation and postpartum involution (light gray box outline). Arrows next to gene processes and functional groups indicate the predominant direction of fold-change as indicated by HOPACH analysis. This model proposes new roles for transcriptional regulators, regulators of mRNA processing, local hormone regulation, protease activity and cell junction formation in the control of both contractile signaling and contraction propagation in the myometrium during pregnancy. A model of postpartum involution is also presented, based on additional data (see Additional data files 4-6).
Table 1 Genes upregulated with quiescence
Increased gestation pattern group Gene symbol Fold-change at 14 days
Prostaglandin and cortisol synthesis
Hydroxysteroid 11-beta dehydrogenase 1 Hsd11b1 10.6
Decidual/trophoblast prolactin-related protein Dtprp 6.2
Hydroxysteroid 11-beta dehydrogenase 2 Hsd11b2 3.6
Cytochrome P450, 11a Cyp11a1 2.3
Prostaglandin-endoperoxide synthase 1 Ptgs1 2.0
Phospholipase inhibition
Annexin A8 Anxa8 4.4
Annexin A3 Anxa3 3.1
Uteroglobin Utg 2.7
Calpactin S100a10 2.5
Annexin A1 Anxa1 2.4
Annexin A2 Anxa2 2.1
Proteolysis and peptidolysis
Kidney-derived aspartic protease-like protein Kdap 8.2
CTLA-2-beta Ctla2b 8.1
Cathepsin Z Ctsz 3.1
Dipeptidase 1 Dpep1 3.1
Procollagen C-proteinase enhancer protein Pcolce 2.6
Lipocalin 7 Lcn7 2.6
Serine-type endopeptidases
Granzyme G Gzmg 71.4
Granzyme D Gzmd 45.7
Granzyme F Gzmf 40.2
Granzyme E Gzme 19.8
Granzyme C Gzmc 10.7
RIKEN cDNA 2210021K23 gene 2210021K23Rik 2.9
Cathepsin G Ctsg 2.2
Protease, serine, 11 (Igf binding) Prss11 2.2
Granzyme B Gzmb 2.1
Protease inhibitors
Tissue factor pathway inhibitor 2 Tfpi2 4.2
Serine protease inhibitor 14 Serpinb9e 3.3
Plasma protease C1 inhibitor Serping1 2.7
Extracellular matrix remodeling and cell growth
Regulation of cell growth
Insulin-like growth factor binding protein 2 Igfbp2 12.4
Interleukin 1 receptor, type II Il1r2 5.0
Glucocorticoid-induced leucine zipper Gilz 3.8
Tumor necrosis factor, alpha-induced protein 2 Tnfaip2 3.3
c-Fos induced growth factor Figf 3.2
Related RAS viral (r-ras) oncogene homolog 2 Rras2 3.0
Cysteine rich protein 2 Crip2 2.9
MORF-related gene X Morf4l2 2.6
Epithelial membrane protein 1 Emp1 2.5
Four and a half LIM domains 1 Fhl1 2.3
S100 calcium binding protein A6 (calcyclin) S100a6 2.3
Insulin-like growth factor binding protein 6 Igfbp6 2.1
Transforming growth factor, beta 2 Tgfb2 2.0
Integrin-mediated signaling pathway
Secreted phosphoprotein 1 Spp1 17.3
Connective tissue growth factor Ctgf 2.8
Caveolin, caveolae protein Cav 2.5
Ras homolog gene family, member A2 Arha 2.4
Structural constituent of cytoskeleton
Gelsolin Gsn 2.4
Tropomyosin 4 Tpm4 3.1
Tubulin, beta 2 Tubb2 2.2
Extracellular matrix structural constituent
Microfibrillar associated protein 5 Mfap5-pending 6.9
Elastin Eln 3.1
Procollagen, type XI, alpha 1 Col11a1 3.0
Fibromodulin Fmod 2.4
Fibrillin 1 Fbn1 2.3
Procollagen, type V, alpha 2 Col5a2 2.2
Laminin, gamma 1 Lamc1 2.2
Procollagen, type I, alpha 2 Col1a2 2.2
G-protein signaling
Guanylate cyclase activator 2b Guca2b 15.2
Chemokine orphan receptor 1 Cmkor1 5.0
Adrenomedullin Adm 2.0
Guanine nucleotide binding protein, gamma 11 Gng11 2.0
Transcriptional regulation
Secreted frizzled-related sequence protein 4 Sfrp4 4.2
Kruppel-like factor 4 Klf4 3.0
C/EBP delta Cebpd 2.3
Inhibitor of DNA binding 1 Idb1 2.1
X-box binding protein 1 Xbp1 2.0
Kruppel-like factor 2 Klf2 2.0
Only upregulated genes with a relative fold-change of 2 or more versus non-pregnant mice at 14.5 days gestation and linked to biological categories highlighted by the expression analysis are shown. Full gene lists can be obtained online (see Additional data file 6).
Table 2 Genes downregulated with quiescence
Decreased gestation pattern group Gene symbol Fold-change at 14 days
Regulation of cell growth
Myosin light chain, alkali, cardiac atria Myl4 -2.8
N-myc downstream regulated 2 Ndr2 -2.7
Actin, beta, cytoplasmic Actb -2.2
Calmodulin signaling
MARCKS-like protein Mlp -2.2
Proteolysis
Matrix metalloproteinase 3 Mmp3 -2.2
Ion channels
Expressed sequence AW538430 Kctd12 -2.9
Transcriptional regulation
SRY-box containing gene 4 Sox4 -2.9
Homeobox protein Meis2 Mrg1 -2.5
Special AT-rich sequence binding protein 1 Satb1 -2.1
D site albumin promoter binding protein Dbp -2.1
RIKEN cDNA 1110033A15 gene 1110033A15Rik -2.1
Myeloid ecotropic viral integration site 1 Meis1 -2.0
Regulation of alternative splicing
CDC-like kinase Clk -2.1
Only downregulated genes with a relative fold-change of 2 or more versus non-pregnant mice at 14.5 days gestation and linked to biological categories highlighted by the expression analysis are shown. Full gene lists can be obtained online (see Additional data file 6).
Table 3 Genes upregulated with term activation
Gene symbol Fold-change at 18 days
Regulation of cell growth
Inhibin beta-B Inhbb 3.1
Inhibin beta-A Inhba 2.2
Cell death
Growth arrest and DNA-damage-inducible 45 γ Gadd45g 3.2
Baculoviral IAP repeat-containing 1a Birc1a 2.1
Clusterin Clu 2.0
Cell junctions
Occludin Ocln 2.8
Gap junction membrane channel protein α1 Cx43 2.8
Desmoplakin Dsp 2.8
G-protein signaling
Lysophosphatidic acid receptor Edg-2 Edg2 2.8
Guanine nucleotide binding protein, γ12 Gng12 2.1
Structural constituent of cytoskeleton
Villin 2 Vil2 3.1
Kinesin complex
Keratin complex 1, acidic, gene 19 Krt1-19 7.8
Keratin complex 2, basic, gene 7 Krt2-7 4.6
Keratin complex 2, basic, gene 8 Krt2-8 4.6
Keratin complex 1, acidic, gene 18 Krt1-18 4.5
Surfactant associated protein D Sftpd 3.4
Metabolism and biosynthetic reactions
Lipoprotein lipase Lpl 4.5
Aldehyde dehydrogenase family 1, subfamily A2 Aldh1a2 3.9
Glutathione S-transferase omega 1 Gsto1 3.7
Branched chain aminotransferase 1, cytosolic Bcat1 3.4
Protein phosphatase 1, regulatory subunit 3C Ppp1r3c 2.2
Carbonic anhydrase 2 Car2 2.1
Proteolysis and peptidolysis
Cytosolic nonspecific dipeptidase 0610010E05Rik 3.2
Transmembrane protease, serine 2 Tmprss2 2.1
Kallikrein 5 Klk5 2.1
Collagen catabolism
Peptidase 4 Pep4 2.3
Matrix metalloproteinase 7 Mmp7 2.2
Proteolysis inhibitors
Complement component 3 C3 4.3
RIKEN cDNA 1600023A02 gene 1600023A02Rik 2.9
Extracellular proteinase inhibitor Expi 2.8
Transcriptional regulation
Transcription factors
Myeloblastosis oncogene Myb 2.5
Hairy and enhancer of split 1 Hes1 2.3
E74-like factor 3 Elf3 2.1
Androgen regulation
Kidney androgen regulated protein Kap 33.9
Heat shock protein 4 Hspa4 3.1
Alpha fetoprotein Afp 3.1
Transport
FXYD domain-containing ion transport regulator 3 Fxyd3 2.8
Lipocalin 2 Lcn2 2.5
Lactotransferrin Ltf 2.2
Solute carrier family 16, member 1 Slc16a1 2.1
Fatty acid binding protein 5, epidermal Fabp5 2.0
Only upregulated genes with a relative fold-change of 2 or more versus non-pregnant mice at 18.5 days gestation and linked to biological categories highlighted by the expression analysis are shown. Full gene lists can be obtained online (see Additional data file 6).
==== Refs
Challis JRG Matthews SG Gibb W Lye SJ Endocrine and paracrine regulation of birth at term and preterm. Endocr Rev 2000 21 514 550 11041447 10.1210/er.21.5.514
Norwitz ER Robinson JN Challis JR The control of labor. N Engl J Med 1999 341 660 666 10460818 10.1056/NEJM199908263410906
Lopez BA Tamby-Raja RL Preterm labour. Baillieres Best Pract Res Clin Obstet Gynaecol 2000 14 133 153 10789265 10.1053/beog.1999.0068
Aguan K Carvajal JA Thompson LP Weiner CP Application of a functional genomics approach to identify differentially expressed genes in human myometrium during pregnancy and labour. Mol Hum Reprod 2000 6 1141 1145 11101697 10.1093/molehr/6.12.1141
Bethin KE Nagai Y Sladek R Asada M Sadovsky Y Hudson TJ Muglia LJ Microarray analysis of uterine gene expression in mouse and human pregnancy. Mol Endocrinol 2003 17 1454 1469 12775764 10.1210/me.2003-0007
Charpigny G Leroy MJ Breuiller-Fouche M Tanfin Z Mhaouty-Kodja S Robin P Leiber D Cohen-Tannoudji J Cabrol D Barberis C Germain G A functional genomic study to identify differential gene expression in the preterm and term human myometrium. Biol Reprod 2003 68 2289 2296 12606369 10.1095/biolreprod.102.013763
Havelock JC Keller P Muleba N Mayhew BA Casey BM Rainey WE Word RA Human myometrial gene expression before and during parturition. Biol Reprod 2004 DOI:10.1095/biolreprod.104.032979
Rehman KS Yin S Mayhew BA Word RA Rainey WE Human myometrial adaptation to pregnancy: cDNA microarray gene expression profiling of myometrium from non-pregnant and pregnant women. Mol Hum Reprod 2003 9 681 700 14561811 10.1093/molehr/gag078
Girotti M Zingg HH Gene expression profiling of rat uterus at different stages of parturition. Endocrinology 2003 144 2254 2265 12746283 10.1210/en.2002-0196
Pollard KS van der Laan MJ A method to identify significant clusters in gene expression data. Proc 6th World Multiconf Systemics, Cybernetics Informatics (SCI2002) 2002 II 318 325
van der Laan MJ Pollard KS A new algorithm for hybrid clustering with visualization and the bootstrap. J Stat Planning Infer 2003 117 275 303 10.1016/S0378-3758(02)00388-9
Bioconductor
Dahlquist KD Salomonis N Vranizan K Lawlor SC Conklin BR GenMAPP, a new tool for viewing and analyzing microarray data on biological pathways. Nat Genet 2002 31 19 20 11984561 10.1038/ng0502-19
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
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Gene Ontology: Tool for the unification of biology. Nat Genet 2000 25 25 29 10802651 10.1038/75556
GenMAPP
Pollard AJ Sparey C Robson SC Krainer AR Europe-Finner GN Spatio-temporal expression of the trans-acting splicing factors SF2/ASF and heterogeneous ribonuclear proteins A1/A1B in the myometrium of the pregnant human uterus: a molecular mechanism for regulating regional protein isoform expression in vivo. J Clin Endocrinol Metab 2000 85 1928 1936 10843177 10.1210/jc.85.5.1928
Benkusky NA Fergus DJ Zucchero TM England SK Regulation of the Ca2+-sensitive domains of the maxi-K channel in the mouse myometrium during gestation. J Biol Chem 2000 275 27712 27719 10871603
Pham CT MacIvor DM Hug BA Heusel JW Ley TJ Long-range disruption of gene expression by a selectable marker cassette. Proc Natl Acad Sci USA 1996 93 13090 13095 8917549 10.1073/pnas.93.23.13090
Allen MP Nilsen-Hamilton M Granzymes D, E, F, and G are regulated through pregnancy and by IL-2 and IL-15 in granulated metrial gland cells. J Immunol 1998 161 2772 2779 9743335
Interactive chromosomal cluster lists
Buxton IL Regulation of uterine function: a biochemical conundrum in the regulation of smooth muscle relaxation. Mol Pharmacol 2004 65 1051 1059 15102932 10.1124/mol.65.5.1051
Croy BA McBey BA Villeneuve LA Kusakabe K Kiso Y van den Heuvel M Characterization of the cells that migrate from metrial glands of the pregnant mouse uterus during explant culture. J Reprod Immunol 1997 32 241 263 9080386 10.1016/S0165-0378(96)01008-X
Garcia-Sanz JA MacDonald HR Jenne DE Tschopp J Nabholz M Cell specificity of granzyme gene expression. J Immunol 1990 145 3111 3118 2212674
Yu JT Lopez Bernal A The cytoskeleton of human myometrial cells. J Reprod Fertil 1998 112 185 198 9538344
Stiemer B Graf R Neudeck H Hildebrandt R Hopp H Weitzel HK Antibodies to cytokeratins bind to epitopes in human uterine smooth muscle cells in normal and pathological pregnancies. Histopathology 1995 27 407 414 8575730
Gown AM Boyd HC Chang Y Ferguson M Reichler B Tippens D Smooth muscle cells can express cytokeratins of 'simple' epithelium. Immunocytochemical and biochemical studies in vitro and in vivo. Am J Pathol 1988 132 223 232 2456700
Brown DC Theaker JM Banks PM Gatter KC Mason DY Cytokeratin expression in smooth muscle and smooth muscle tumours. Histopathology 1987 11 477 486 2440790
Megy K Audic S Claverie JM Positional clustering of differentially expressed genes on human chromosomes 20, 21 and 22. Genome Biol 2003 4 P1 12620117 10.1186/gb-2003-4-2-p1
Caron H van Schaik B van der Mee M Baas F Riggins G van Sluis P Hermus MC van Asperen R Boon K Voute PA The human transcriptome map: clustering of highly expressed genes in chromosomal domains. Science 2001 291 1289 1292 11181992 10.1126/science.1056794
Gabrielsson BL Carlsson B Carlsson LM Partial genome scale analysis of gene expression in human adipose tissue using DNA array. Obes Res 2000 8 374 384 10968729
Trinklein ND Aldred SF Hartman SJ Schroeder DI Otillar RP Myers RM An abundance of bidirectional promoters in the human genome. Genome Res 2004 14 62 66 14707170 10.1101/gr.1982804
Irizarry RA Bolstad BM Collin F Cope LM Hobbs B Speed TP Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003 31 e15 12582260 10.1093/nar/gng015
Dudoit S Gentleman RC Quackenbush J Open source software for the analysis of microarray data. Biotechniques 2003 Suppl 45 51 12664684
Liu G Loraine AE Shigeta R Cline M Cheng J Valmeekam V Sun S Kulp D Siani-Rose MA NetAffx: Affymetrix probe sets and annotations. Nucleic Acids Res 2003 31 82 86 12519953 10.1093/nar/gkg121
Real-time PCR oligonucleotide sequences
Dolganov GM Woodruff PG Novikov AA Zhang Y Ferrando RE Szubin R Fahy JV A novel method of gene transcript profiling in airway biopsy homogenates reveals increased expression of a Na+-K+-Cl- cotransporter (NKCC1) in asthmatic subjects. Genome Res 2001 11 1473 1483 11544191 10.1101/gr.191301
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 research0034.1 0034.11 12184808 10.1186/gb-2002-3-7-research0034
Ensembl EnsMart Genome Browser (MartView)
Interactive myometrium GenMAPP HTML pathways
| 15693941 | PMC551532 | CC BY | 2021-01-04 16:05:37 | no | Genome Biol. 2005 Jan 28; 6(2):R12 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r12 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r131569394210.1186/gb-2005-6-2-r13ResearchVariation in tissue-specific gene expression among natural populations Whitehead Andrew [email protected] Douglas L [email protected] Rosenstiel School of Marine and Atmospheric Science, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149, USA2005 26 1 2005 6 2 R13 R13 28 6 2004 2 9 2004 6 12 2004 Copyright © 2005 Whitehead and Crawford; 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 expression of a selected suite of 192 metabolic genes in brain, heart and liver in three populations of the teleost fish Fundulus heteroclitus was examined. Only a small subset (31%) of tissue-specific differences was consistent in all three populations, indicating that many tissue-specific differences in gene expression are unique to one population and thus are unlikely to contribute to fundamental differences between tissue types.
Background
Variation in gene expression is extensive among tissues, individuals, strains, populations and species. The interactions among these sources of variation are relevant for physiological studies such as disease or toxic stress; for example, it is common for pathologies such as cancer, heart failure and metabolic disease to be associated with changes in tissue-specific gene expression or changes in metabolic gene expression. But how conserved these differences are among outbred individuals and among populations has not been well documented. To address this we examined the expression of a selected suite of 192 metabolic genes in brain, heart and liver in three populations of the teleost fish Fundulus heteroclitus using a highly replicated experimental design.
Results
Half of the genes (48%) were differentially expressed among individuals within a population-tissue group and 76% were differentially expressed among tissues. Differences among tissues reflected well established tissue-specific metabolic requirements, suggesting that these measures of gene expression accurately reflect changes in proteins and their phenotypic effects. Remarkably, only a small subset (31%) of tissue-specific differences was consistent in all three populations.
Conclusions
These data indicate that many tissue-specific differences in gene expression are unique to one population and thus are unlikely to contribute to fundamental differences between tissue types. We suggest that those subsets of treatment-specific gene expression patterns that are conserved between taxa are most likely to be functionally related to the physiological state in question.
==== Body
Background
The regulation of gene expression varies extensively among tissues, individuals, strains, populations and species [1-6] and variation in gene expression has a genetic basis [7,8]. Despite such biological variance, differences in gene expression are used to describe cancers [9-12], heart failure [13,14] and metabolic diseases [15]. It is common for these pathologies to be associated with changes in tissue-specific gene expression or changes in metabolic gene expression. For example, many different cancers have unique tissue-specific patterns of gene expression [16], and thyroid cancers are associated with increases in aerobic metabolic gene expression [17].
Although tissue-specific gene expression patterns are often used as a method to identify functionally relevant genes, how conserved these differences are among outbred individuals and among populations has not been well documented. It is possible that many of these changes represent polymorphism among individuals or populations and are not specifically associated with disease. To address this we used a well established system (tissue-specific gene expression) and genes with well defined function and tissue-specific distributions (metabolic genes).
Given the high variance in gene expression among individuals and populations, our goal was to examine the conservation of tissue-specific gene expression among populations of the same species. Specifically, we assessed the among-population variance of tissue-specific patterns of gene expression (in brain, heart and liver) in the teleost fish Fundulus heteroclitus. A cDNA microarray was used to measure levels of expression in normal healthy male fish for 192 genes involved in central metabolic pathways. We used this compact array in order to impose a high degree of technical and biological replication (24 replicates for each of three tissues from nine individuals with two samples per array). Also, this array was used because metabolic genes are essential, are known to have tissue-specific expression, especially in fish, and are often misused as controls with little characterization of variation in expression among individuals or tissues. Analysis of variance (ANOVA) was used as a statistical test to determine which genes were differentially expressed among tissues and populations. Tissue-specific patterns of gene expression were compared among populations. As expected, we detected extensive variation in gene expression among tissues. Unexpectedly, only a fraction (31%) of tissue-specific differences was conserved between all populations.
Results
Variation among
Variation among individuals within groups was high (groups included the nine tissue-by-population groupings; Figure 1). Nearly half of genes (92 genes, 48%) were differentially expressed (p < 0.05) among individuals within populations and tissues (Figure 1), and inter-individual differences ranged over fivefold.
Variation among tissues
Although variation among individuals was high, added variation due to tissues was significant. Considering 192 genes and a p-value of 5%, one would expect less than 10 false-positive differences among tissues under the null hypothesis. We detected 76% of genes (146 of 192 genes) differentially expressed among brains, hearts and livers (ANOVA, p < 0.05). Selecting the α level at which differences between treatments are considered significant is problematic because of the large number of comparisons performed. As such, we present a volcano plot to illustrate the range of expression differences between tissues and associated p-values (Figure 2). When α is set at 0.01, 0.001 or at the Bonferroni-corrected value (2.6 × 10-4), the proportion of significant genes is 67% (129 genes), 50% (96) and 39% (75), respectively. Significant differences in expression ranged from less than 1.2-fold to nearly 16-fold (Figure 2). The predominant pattern of tissue-specific expression can be described by expression significantly different in the liver compared to the other two tissues (Figure 3).
Many expected tissue-specific patterns emerged. For example, the brain-specific fatty-acid-binding protein was typically more highly expressed in the brain than in other tissues (p = 0.005), hepatocyte nuclear factor 4-alpha (a transcription factor) was more highly expressed in liver than in other tissues (p < 0.001), and two genes involved in glycerolipid metabolism -lipoprotein lipase and phopholipase XIII A2 - were more highly expressed in liver than other tissues (p < 0.001 for both genes).
Liver-specific expression accounted for 61% of the expression differences among tissues (Figure 4). Heart-specific and brain-specific expression accounted for 24% and 15% of differences among tissues, respectively. Regardless of population, expression patterns were typically most similar between heart and brain, and least similar between liver and heart (Figure 5). There were 67 genes printed on the array that code for proteins involved in oxidative phosphorylation, and 88% (59 genes) were differentially expressed between tissues (genes highlighted in green, Figure 3). Of differentially expressed oxidative phosphorylation genes, only 10% (six genes) were expressed more highly in the liver than in other tissues, whereas the remaining 90% (53 genes) had lower expression in the liver compared to brain or heart.
Variation among taxa
A small proportion of genes (six genes, 3%) differed in expression among populations (p < 0.05). However, it should be noted that although the split-plot design is powerful for detecting differences between split-plot factors (tissues), it is considered to have low power for detecting differences between blocks (populations) [18]. As such, it is likely that 3% is an underestimate of true among-population differences in gene expression. Indeed, two-way ANOVA (data not shown), which has higher power for detecting population differences but is less valid than the split-plot model for testing individual and tissue differences, detected among-population differences in expression for 18% of genes at p < 0.05, or 6.3% of genes at p < 0.01. Each tissue contributed a similar number of genes differentially expressed among populations.
Surprisingly, differences among tissues in gene expression were not consistent across all three populations. More than one-third (37%) of the genes differentially expressed between tissues were significant in only one of the three populations (Figure 6). Population-specific differences were distributed among the three populations; Georgia had 40% of the population-specific genes, and New Jersey and Maine had 34% and 26%, respectively. A proportion of these inconsistencies could be due to false-positive or false-negative differences between tissues in individual populations. However, statistically significant interaction between tissue and population was detected for many (30%) of these inconsistencies (see Additional data file 1).
A relatively small proportion of tissue-specific genes (31%) have consistent expression patterns in all three populations (Figure 6; also see Additional data file 1 for details). This subset of genes also reflects the different metabolic status of brain, heart and liver; most of the genes involved in oxidative phosphorylation were more highly expressed in brain and heart than in liver (Figure 7a, Table 1), and most of the genes involved in fatty-acid metabolism, glycerolipid metabolism, steroid metabolism and detoxification were more highly expressed in liver. The majority of the tissue-specific genes were not consistent among populations (a subset of these genes are illustrated in Figure 7b, Table 1).
Quality control
Variation among technical replicates was low, and permutation tests indicated that the ANOVA model was robust. Sample coefficients of variation (CVs (standard deviation/mean) × 100), which estimate technical variance due to replicate spots (six spots per hybridization), repeated measures (two hybridizations per dye), and dye (two dyes per sample), were calculated for each gene of each of the 27 samples. CVs less than 5% accounted for 95% of sample/genes, respectively. Of the many comparisons performed (differences among tissues, populations, interaction), permutation tests results agreed with ANOVA results (the same comparisons identified as significant or not significant) for 99.1% of comparisons, suggesting that our ANOVA model was robust.
Discussion
Considerable variation occurs among the 27 samples (three tissues from each of three individuals from three populations) used to measure inter-individual and tissue-specific variation in gene expression. We are able to precisely describe the patterns of gene expression for 192 metabolic genes because of the low experimental variation; for 95% of the replicate measures of gene expression the standard deviation is less than 5% of the mean. Notably, gene expression is statistically different for many genes among individuals within a population for a tissue (48%), between tissues (76%), and between populations (3%). For genes with tissue-specific expression, only a fraction (31%) had expression patterns consistent across all three populations. These data do not specifically identify tissue-specific differences that are inconsistent across populations, but rather emphasize that tissue-specific differences detected can vary from one population to another. When measured from a single population, highly significant differences in tissue-specific expression do not necessarily represent genes relevant to general functional or morphological differences between tissues.
Variation among individuals
Variation in gene expression among healthy male individuals raised under controlled laboratory conditions was high. Nearly half of the metabolic genes (48%) were differentially expressed among individuals within a population for any one tissue (Figure 1), with fold differences ranging from 1.2- to 5-fold and p-values ranging down to 10-7. Differences in gene expression among individuals are unlikely to be due to common reversible environmental factors that affect physiological performance (acclimation effects) since all individuals used in this study were housed in a common environment and fed the same food for at least two months. However, the differences could be due to irreversible developmental effects or genetic variations that affect gene expression. Regardless of this, if these differences are heritable or due to developmental plasticity, they represent variation one would expect to find among outbred organisms, including humans.
Other studies that have measured inter-individual differences in gene expression have also detected high levels of variation in a variety of taxa. Among crosses of different yeast strains a large number of differences in expression (6% of genes varying more than twofold) were detected between morphotypes [1]. A previous study of the same Maine and Georgia Fundulus populations assayed here detected 18% of genes differentially expressed among healthy individuals [3]. Although inter-individual variance in gene expression seems prevalent, our observation that 48% of genes are differentially expressed among individuals is high. This may reflect the greater precision of these measurements as a result of extensive technical replication (24 replicate measures per sample) as coefficients of variation for technical replicates was less than 5% for 95% of the genes. Indeed, using similar methods and tools, a concurrent study assessing variation in Fundulus also detected a very high proportion of genes (94%) differentially expressed among individuals [19]. Alternatively, since our array is heavily biased toward metabolic genes, detected variance may also reflect a greater variation in metabolic gene expression. We could speculate that the high variation in metabolic genes reflects a greater allowable variation. That is, there may be less selective pressure to constrain metabolic variation either because varying the amount of an enzyme does not affect metabolism or variation in metabolism is phenotypically acceptable. One could test this by using an array with more comprehensive representation of the genome and comparing variances of different gene classes defined by function.
Considering the high inter-individual variation detected, the data presented here underscore the importance of including biological replicates within treatment groups in order to ascribe differences in expression to treatment rather than to inter-individual variation. Statistically, an analysis of variance can be used to examine the effects of technical and biological variation, and these tests have proved powerful for detecting significant differences in gene expression [3,4], even differences as small as 1.2-fold. The cost of resources in microarray experiments should no longer excuse lack of biological and technical replication. Often, microarray experiments pool individual samples within treatment groups to capture biological variation. However, this approach only estimates an average level of expression and fails to estimate biological variation. When only small quantities of RNA can be extracted from samples, one can estimate biological variation by pooling multiple independent samples [20].
A variety of factors can contribute to differences in gene expression among individuals. Pritchard et al. [21] proposed that differences in immune status may explain the 3.3% difference in gene expression among genetically identical mice. Sex explained a large portion of among-individual variation in gene expression in Drosophila, whereas genotype was less of an influence, and the influence of age was weak [4]. Furthermore, this type of variation can be biologically relevant. For example recent work in Fundulus indicates that most inter-individual variation in metabolism can be accounted for by differences in metabolic gene expression [19].
Variation among tissues
Another important source of biological variation in gene expression is differences in expression among different tissues; 76% of genes were differentially expressed between brain, heart and liver, and expression in the liver was the most distinct compared to heart and brain. In this study, genes printed on our array are primarily enzymes functional in central metabolic pathways such as fatty-acid metabolism, glycolysis and oxidative phosphorylation. Of the oxidative phosphorylation genes differentially expressed between tissues, 92% were more highly expressed in heart or brain than in liver (Figure 3). The primary purpose of the heart is to act as a pump, and contraction is highly dependent on oxidative metabolism [22]. The metabolic rate in the brain is 7.5 times the average rate in the rest of the body [23]. High metabolic demand in the brain supports pumping of ions across neuronal membranes during action potentials and metabolism is primarily oxidative. Mitochondria are the principal sites for oxidative phosphorylation, and are most numerous in heart, brain and skeletal muscle cells. The liver, in contrast, is much more functionally diverse, as it is involved in carbohydrate storage, synthesis of proteins, glucose, fatty acids, cholesterol and lipids, and metabolism of xenobiotics and endogenous compounds, and has a relatively low respiration rate. Accordingly, transcripts of genes functional in oxidative phorphorylation appear to represent a much smaller portion of the cell's RNA transcripts in liver tissues than in the heart or brain. In addition, genes involved in fatty acid and phospholipid synthesis were more highly expressed in liver than the other tissues. Differences in expression among tissues detected using our array appear to reflect differences in the metabolic status of brain, heart, and liver. Because data presented here support well established patterns of metabolism, they suggest that measuring mRNA expression using microarrays accurately reflects changes in proteins and their phenotypic effect.
Many microarray studies have used expression levels of 'housekeeping' genes as an internal control for comparisons among arrays, individuals and treatments. Housekeeping genes may be defined as those that are involved in routine cellular metabolism and always expressed in all cells. Accordingly, many, if not most, of the genes studied here could be considered housekeeping genes. Nearly half of these genes were expressed at different levels between individuals, with fold differences ranging from 1.2- to 5-fold and p-values ranging down to 10-7. Lee et al. [24] applied ANOVA to screen four previously published datasets for housekeeping genes across a variety of biological contexts. They found that all genes that are commonly used as controls had fold changes ranging from greater than 2.0 to more than 300 within at least one dataset, and coefficients of variation were concordantly high, reflecting high variance in expression of these genes. It appears that upon application of ANOVA, statistically significant differences in expression of housekeeping genes can be detected among individuals and across different biological contexts, and scaling for differences among arrays using expression levels of these genes ought to be approached with caution.
Although genes differentially expressed among tissues reflect their different metabolic requirements, it should be noted that the purpose of the current study was not to comprehensively identify suites of genes responsible for functional differences between tissues. The relatively small number of printed probes was useful for a high degree of technical replication, and obviously represents a small portion of the expressed genes. However, this approach shows that highly significant differences in gene expression among tissues may be apparent but not consistent among closely related taxa. Therefore, highly significant differences in gene expression found only within a single population may not necessarily represent genes relevant to general functional or morphological differences between tissues.
Variation among taxa
Although the pattern of metabolic gene expression among tissues reflects established patterns of tissue-specific metabolism, there is additional variation due to population. It should be noted that the split-plot statistical design is not as powerful for detecting among-block differences (among populations) as for detecting differences among split-plot factors [18]. We detected 3% of genes (6 of 192) differentially expressed among populations. This proportion is similar to that detected in a previous study [3] in which 2.6% of genes were differentially expressed between Maine and Georgia Fundulus hearts. Similarly, approximately 1% of genes were differentially expressed in brain tissue among inbred strains of mice [2]. Differences in gene expression are to be expected among taxa (phylogenetically distinct groups of organisms which may include strains, populations or species), with the majority of differences most likely to be attributable to random genetic drift. For more distantly related groups, one would expect expression patterns to be more divergent than for closely related groups. Indeed, expression patterns between humans and chimpanzees are more similar than those between humans and orangutans, and similar results were obtained from comparisons among three mouse species [5,6].
An unexpected finding is that the tissue-specific differences depend on which population was assayed. Differences in gene expression are expected between tissues because of functional divergence and between populations because of neutral genetic divergence. In addition, one might expect that the number of genes significantly different between populations would depend on the tissue. One might also expect tissue-specific differences to be consistent in all taxa. Yet our data indicate that tissue-specific expression patterns are not fixed within a species. The genes for which expression is significantly different between tissues are not all the same in all three populations. Of the 128 genes that have tissue-specific patterns of expression in any population, 37% are tissue-specific in only one of the three populations and 32% are found in only two of the three populations. Overall, it would appear that only 31% of tissue-specific differences in gene expression are consistent among all populations of F. heteroclitus. One needs to be careful about this interpretation, however. Our emphasis was not to specifically identify genes that have significant interaction between tissue and population. Rather, we emphasize that genes detected as tissue specific will vary from one population to another, and most microarray studies measure treatment-specific expression patterns in only one population of test organism. Because inter-individual variation is high, it is probable that inclusion of more replicate individuals in each group would increase the sensitivity of ANOVA, and the number of genes that distinguish tissues consistently in all populations may change.
The consistent tissue-specific differences still support expectations based on the metabolic requirements of each tissue (for example, genes involved in oxidative phosphorylation were more highly expressed in heart and brain, and those involved in fatty-acid and lipid metabolism were more highly expressed in the liver; Figure 7a). Accordingly, those differences in expression that are consistent across several groups of organisms are most likely to account for functional and morphological differences among tissues, emphasizing that this type of comparative approach may be powerful for testing the biological relevance of other functional traits. For example, expression differences between diseased and non-diseased tissues may vary among mouse strains, so that the subset of differences that are consistent across strains are more likely to be functionally related to the diseased state.
Our data suggest that many of the differences in gene expression detected between experimental groups may be of little functional importance because they vary among taxa. We suggest that patterns of expression that are consistent in different populations are more likely to be functionally important. Elucidation of adaptively important variation, such as variation related to antibiotics, pesticides or temperature adaptation, may also benefit from such a comparative approach that screens for conserved patterns. However, there is the possibility that partitioning of genetic polymorphisms among populations may allow distinct groups of organisms to reach different physiological or biochemical solutions to the same biological challenges. For example, patterns of polymorphism in a gene that regulates coat color in mammals indicated recent directional selection and was associated with coat color in one pocket mouse population, but not in a second population [25]. Other loci were probably responsible for adaptive variation in coat color in the second population.
Conclusions
These data indicate high variation in metabolic gene expression among individuals and thus expression of these housekeeping genes is unreliable as an internal control or as a method of normalization across samples. Second, concordance between tissue-specific expression patterns and established metabolic functions of brain, heart and liver indicate that measuring mRNA levels accurately reflects physiological status. Furthermore, since many metabolic genes differ in expression among brain, heart and liver, those studies using whole organisms need to rule out whether changes in expression reflect differences in the proportions of various tissues among samples. Finally, studies seeking to identify patterns of gene expression related to physiological states, such as disease or toxic stress, must consider both variation between individuals and differences between populations. Because of this biological variation, not all differences between treatments in any one population of test organism are likely to be generally relevant. We suggest that conserved patterns of treatment-specific gene expression among taxa are most likely to be functionally related to the physiological state in question.
Methods and materials
Animals and maintenance
Teleost fish Fundulus heteroclitus were collected from the field by seine and minnow trap in June 2003, transported to the University of Miami RSMAS laboratory under controlled temperature and aeration conditions, and acclimated to common conditions (20°C, 15 parts per thousand salinity) in recirculating 100-gallon tanks for at least two months before experiments. Fish were sacrificed by cervical dislocation and tissues were excised and stored in RNAlater (Ambion) at -20°C. Fish were collected at Wiscasset, Maine; Stone Harbor, New Jersey, and Sapelo Island, Georgia. Only healthy male fish were used for the following experiments.
Microarrays
Microarrays were printed using 192 cDNAs from a F. heteroclitus cardiac library encoding essential proteins for cellular metabolism [26]. These cDNAs were a subset of over 40,000 expressed sequences in our online database Funnybase [27]. These 192 cDNAs were amplified with amine-linked primers and printed on 3-D Link Activated slides (Surmodics) using a SpotArray Enterprise piezoelectric microarray printer (PerkinElmer Life Sciences) at Louisiana State University. Slides were blocked following slide manufacturer protocols. The suite of 192 amplified cDNAs was printed as a group in six spatially separated replicates. Four hybridization zones of these six replicate arrays were printed per slide, with each zone set separated by a hydrophobic barrier.
Hybridization experimental design
Microarray analyses were applied to three tissues (brain, heart and liver) from three individuals collected from three populations of F. heteroclitus. Each of these 27 samples was measured four times, twice with Cy3 and twice with Cy5 (Figure 8). In addition, since a hybridization zone covered six replicate printed arrays, total experimental replication per sample per gene was 24-fold. A total of 108 hybridizations were performed (27 × 4), and Cy3-Cy5 hybridizations were balanced (although incompletely) among tissues and populations in a sheet-loop design (Figure 8).
Sample preparation
RNA was extracted from tissue homogenate in a chaotropic buffer using phenol/cholorform/isoamyl alcohol. All reagents were from Sigma unless otherwise noted. Tissues were removed from RNAlater, blotted dry, and homogenized using an electric homogenizer in 400 μl chaotropic buffer (4.5 M guanidinium thiocyanate, 2% N-lauroylsarcosine, 50 mM EDTA pH 8.0, 25 mM Tris-HCl pH 7.5, 0.1 M β-mercaptoethanol, 2% antifoam A). An equal volume of 2 M sodium acetate (pH 4.0) was added to the homogenate, followed by 400 μl acidic phenol (pH 4.4), and 120 μl chloroform/isoamyl alcohol (23:1). The mixture was kept at 4°C for 10 min then centrifuged at 4°C at 16,000g for 20 min. Supernatant was removed and combined with 400 μl isopropanol, stored at -20°C for 30 min, then centrifuged at 4°C at 16,000g for 30 min. The remaining RNA pellet was rinsed twice with 400 μl of 70% ethanol, then further purified using the Qiagen RNeasy Mini kit (Qiagen) following the manufacturer's protocols. Purified RNA was quantified spectrophotometrically, and RNA quality was assessed using the Agilent 2100 Bioanalyzer. RNA was stored in 1/10 volumes 3 M sodium acetate and 2.5 volumes 100% ethanol at -20°C.
RNA for hybridization was prepared by amplification using a modified Eberwine protocol [28]. The Ambion Amino Allyl MessageAmp aRNA Kit was used (according to manufacturer's protocols) to copy template RNA by T7 amplification following incorporation of a T7 promoter, resulting in amplified template in the form of antisense RNA. Amino-allyl UTP was incorporated into targets during T7 transcription, and resulting amino-allyl antisenseRNA was coupled to Cy3 and Cy5 dyes (Amersham Biosciences).
Hybridization
Labeled aRNA aliquots of the two individual samples for each hybridization (18 pmol each of Cy3 and Cy5) were vacuum dried together and resuspended in 12 μl hybridization buffer (final concentration of each labeled sample = 1.5 pmol/μl). Hybridization buffer consisted of 5 × SSPE, 1% SDS, 50% formamide, 1 mg/ml poly(A), 1 mg/ml sheared herring sperm carrier DNA, and 1 mg/ml BSA. Slides were washed in sodium borohydride solution according to Raghavachari et al. [29] to reduce autofluorescence. Following rinsing, slides were boiled for 2 min and spin-dried in a centrifuge at 800 rpm for 3 min. Samples (12 μl) were heated to 90°C for 2 min, quick cooled to 42°C, applied to slide (hybridization zone area was 350 mm2), and covered with a coverslip. Slides were placed in an airtight chamber humidified with paper soaked in 1 × SSC buffer and incubated 12-18 h at 42°C. Following hybridization, slides were scanned using the Packard Bioscience ScanArray Express microarray scanner (PerkinElmer Life Sciences). Resulting .tiff images were imported into spot grids built in ImaGene (Biodiscovery) for each array, and spot signals were collected as fluorescence intensities for each dye channel.
Data processing and statistical analysis
Raw data were first sum normalized [30], which involves summing the total signal from each replicate array to the same value. Then spatial bias on each array was smoothed using a lowess transformation in MAANOVA Version 0.93-2 for R [31]. Other methods of normalization have also been proposed [32-34]. Log2 values of lowess-transformed sum-normalized data were used for all subsequent statistical analyses. MIAME-compliant data [35] have been submitted to the Gene Expression Omnibus as accession number GLP1224. Data were analyzed in a split-plot ANOVA design with population as blocks and tissues as split-plot factors using scripts written in MatLab Version 6 (The MathWorks). MatLab code is available upon request from the authors. Nested within tissue-by-population samples were technical replicates. Replicate spots within hybridization (six), replicate hybridizations per labeling (two) and replicate labelings per sample (two; Cy3 and Cy5) represent the three levels of technical variance nested within the tissue-by-population sample. The ANOVA structure is presented in Figure 9 and Table 2, and the model can be written as:
y = grand mean + population + tissue + population-tissue interaction + individual in population + tissue-by-individual within population + dye within individual + hybridization within dye + spot within hybridization
where y is the normalized log2 expression and individual in population and tissue-by-individual within population are random effects. To test for differences among multiple means (for example, among population and tissue groups), and to correct for multiple comparisons, the T-method [36] was applied. The T-method calculates the minimum significant range defined as
MSR = Qα[kv] × SE
where the critical value Qα[kv] is the studentized range [37], k = number of groups in the comparison (for example, if comparisons are among tissues then k = 3), v = degrees of freedom of MStissue-by-individual within population, and SE is the standard error among tissue-by-individual samples within populations. The T-method following ANOVA was used to identify genes differentially expressed among tissues in each population. These data were then used to contrast tissue-specific and population-specific expression patterns. Robustness of ANOVA data was tested using a permutation test; means for the 27 biological samples were randomly permuted 1,000 times between population and tissue and test statistics were recalculated for differences among populations, tissues and tissue-by-population interaction. Agreement between ANOVA and permutation test results would indicate the robustness of the ANOVA model. Finally, in order to graphically illustrate expression similarity among tissues, expression distance between samples was calculated as the sum of differences of log2 expression values over all genes, and neighbor-joining trees of global similarity of expression patterns among tissues (L, liver; H, heart; B, brain) were constructed [38] for each population.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 lists the results from statistical analyses for all genes. Listed for each gene are p-values associated with statistical tests for differences in expression between populations, tissues, tissue-by-population interaction, and among individuals within populations. Also listed are mean expression for each sample, and columns comparing differences in expression between tissues within each population. Final columns tabulate whether a tissue difference was detected for each comparison, whether this difference was consistent between populations, and whether significant interaction was detected for that gene.
Supplementary Material
Additional data file 1
The results from statistical analyses for all genes
Click here for additional data file
Acknowledgements
Much credit is due to Marjorie Oleksiak for construction of the expressed sequence tag (EST) library, array printing, and constructive criticisms. We also thank Steve Hand at Louisiana State University for the use of their facilities and assistance in printing the microarray. We thank Gary Churchill for statistical advice and Justin Paschall for EST database management and bioinformatics. Valuable assistance was provided by Jen Roach and Jeff VanWye, and Jen Roach provided helpful comments on the manuscript. We especially thank two reviewers for insightful criticisms and comments on the manuscript. This project was supported by a National Science Foundation OCE grant 0221879 to D.L.C. and a National Institutes of Health grant NHLBI R01 HL65470 to D.L.C.
Figures and Tables
Figure 1 Variation within individuals (technical variance) and among individuals within populations and tissues (biological variance) for each of 192 genes indicated by the mean square error (MS) of measurements. Points above the dashed line indicate genes with greater variance among individuals than within. F-crit is the critical value of the F-statistic (F = MSamong/MSwithin, with 12 and 27 degrees of freedom and α = 0.05) for testing significant differences in gene expression between individuals. For 48% of genes, MSamong/MSwithin > F-crit (solid red line). These genes are therefore differentially expressed among individuals within treatments.
Figure 2 Volcano plot of differences between tissues and corresponding p-values. Differences in expression for each gene is the log2 ratio of tissue mean expression minus grand mean; a twofold difference in expression between tissues is indicated by one unit separation along the x-axis. p-values for differences in gene expression among tissues were calculated using ANOVA, and illustrated as -log(p). A p-value of 10-4 is expressed as 4 on the y-axis, and the α = 0.05 threshold is indicated by the red dashed line (1 - log(0.05) = 1.3).
Figure 3 Dendrogram of gene expression patterns across samples for genes significantly different between tissues (ANOVA, p < 0.05). Clustering indicates similar expression patterns among samples (top axis) and among genes (left axis). Samples cluster as livers (yellow), hearts (pink) and brains (blue). Genes involved in oxidative phosphorylation are highlighted in green, and expression patterns that are consistent across all three populations are highlighted with a blue triangle.
Figure 4 Number of genes differentially expressed among tissue groups for each population. Tissue-specific genes are those that are expressed more highly in a tissue than in the other tissues (for example, L > H, B) or lower in a tissue than in the other tissues (for example, L < H, B).
Figure 5 Similarity of expression patterns among tissues. (a) Proportion of 192 genes that are similarly expressed between heart and brain (black bar), brain and liver (gray bar) and liver and heart (white bar), for each population including Maine (ME), New Jersey (NJ) and Georgia (GA). (b) Neighbor-joining trees of global similarity of expression patterns among samples (L, liver; H, heart; B, brain) for each population. Distance between samples is the sum of differences of log2 expression values over all genes.
Figure 6 Shared expression patterns among populations.
Figure 7 Gene expression in liver, brain and heart (three symbols for each line) for the three different populations (three lines per gene). Each letter represents a gene, expression values are log2 transformed and are indicated for liver, brain and heart (left to right) in each of Maine (circles), New Jersey (triangles) and Georgia (squares) populations. (a) Genes consistently different among tissues in all three populations are grouped as those involved in oxidative phosphorylation (upper panel) and those involved in other metabolic pathways (lower panel). (b) A representative subset of genes not consistently different among tissues in all populations. Gene names associated with letters are provided in Table 1 and Additional data file 1.
Figure 8 Experimental design for hybridizations. Each arrow represents an array hybridization, with the samples at arrow base and head labeled with Cy3 and Cy5, respectively. Liver, heart and brain samples are indicated as purple, red and blue circles, respectively. Three individuals were assayed per tissue and from each of three populations. ME, Maine; NJ, New Jersey; GA, Georgia.
Figure 9 Split-plot ANOVA statistical design. Populations (ME, Maine; NJ, New Jersey; GA, Georgia) are treated as blocks, replicate individuals within each population (1, 2 and 3) as plots, and tissue (L, liver; H, heart; B, brain) within an individual as the split-plot factor. Nested within each tissue-by-individual sample are technical replicates including two dyes (Cy3 and Cy5) within each sample, two replicate hybridizations (A and B) per dye, and six replicate spots per hybridization. GM, grand mean.
Table 1 Identity of tissue-specific genes with expression patterns consistent in all three populations, and those inconsistent in all three populations
Gene (see Figure 7) Grid Short name Swiss-Prot name
Consistent - oxidative phosphorylation
a E8 Aldo keto reductase 1 A1 Aldo-keto reductase family 1 member A1 (aldehyde reductase)
b E7 Aldo keto reductase 1 D1 Aldo-keto reductase family 1 member D1; steroid-5-beta-reductase beta polypeptide 1 (3-oxo-5 beta-steroid delta 4-dehydrogenase beta 1); steroid 5-beta-reductase
c F1 G3PDH Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
d D10 Glucose 6 phosphatase Glucose-6-phosphatase (G6PASE)
e H3 Pyruvate kinase muscle Pyruvate kinase (muscle isozyme)
f G10 Pyruvate kinase R Pyruvate kinase isoform R (erythroid)
g I6 NADH dehydrb 6 (17 kD) NADH dehydrogenase (ubiquinone) 1 beta subcomplex 6 (17 kD B17)
h G4 NADH Ubiq Oxi ASHI NADH-ubiquinone oxidoreductase ASHI subunit precursor (complex I-ASHI) (CI-ASHI)
i M4 NADH Ubiq Oxi MNLL NADH-ubiquinone oxidoreductase MNLL subunit (complex I-MNLL) (CI-MNLL)
j L1 ATP syn H+ FO c 9 2 ATP synthase H+ transporting mitochondrial F0 complex subunit c (subunit 9) isoform 2
k P3 ATP syn H+ FO F6 ATP synthase H+ transporting mitochondrial F0 complex subunit F6; coupling factor 6
l I9 Cyto C oxi III Cytochrome c oxidase subunit III
m J10 Cyto C oxi VA Cytochrome C oxidase polypeptide VA
n N8 Cyto C oxi VIa Cytochrome c oxidase subunit VIa precursor polypeptide 2
o J5 Cyto C oxi VIIC Cytochrome C oxidase polypeptide VIIC precursor (VIIIA)
p K12 Cyto C oxi VIIIb Cytochrome c oxidase subunit VIIIb
Consistent - other metabolism
q H12 Isocitrate dehyd 2 Isocitrate dehydrogenase 2 (mitochondrial IDH2)
r A9 PEP carboxykinase PEP carboxykinase phosphoenolpyruvate carboxykinase
s D8 Fatty acid binding liver basic Liver-basic fatty acid binding protein (LB-FABP)
t B12 Delta 6 fatty acid desaturase Delta-6 fatty acid desaturase
u H1 Triglyceride lipase triacylglycerol Triglyceride lipase triacylglycerol
v I5 Glycerol kinase Glycerol kinase
w M10 Lipoprotein lipase Lipoprotein lipase
x P9 Phospholipase XIII A2 Group XIII secreted phospholipase A2
y F4 Cystathionine beta synthase Cystathionine-beta-synthase
z K11 Cold inducible RNA binding Cold inducible RNA-binding protein; (CIRBP) glycine-rich RNA binding protein;
aa F2 Hepatocyte nuclear F 4 A Hepatocyte nuclear factor 4-alpha (HNF-4-alpha) (transcription factor HNF-4)
bb M1 p450 2P1 (CYP2P1) Cytochrome P450 2P1 (CYP2P1)
cc D6 Glutathione peroxidase 4 Glutathione peroxidase 4 (phospholipid hydroperoxidase)
dd O11 Methylmalonate semialdehyde dehyd Methylmalonate-semialdehyde dehydrogenase (acylating)
ee N7 Phosphatidylcholine sterol acyltrans Phosphatidylcholine-sterol acyltransferase
ff B1 Prostaglandin D syn Prostaglandin D synthase
Inconsistent - oxidative phosphorylation
gg A6 ADH class II mito Aldehyde dehydrogenase, mitochondrial precursor (ALDH class 2)
hh E12 Aldolase 1 A Aldolase 1 A. muscle
ii A2 Enolase beta muscle enolase (beta muscle specific)
jj F7 LDHB lactate dehydrogenase B (LDHB)
kk O6 PFK 6-phosphofructokinase
ll K4 NADH dehyd MLRQ NADH dehydrogenase (ubiquinone) MLRQ subunit (complex I-MLRQ)
mm L9 NADH dehyd I NADH dehydrogenase subunit 1
nn C6 NADH dehydr a 1 (7.5 kD MWFE) NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 1 (7.5 kD MWFE)
oo E6 NADH dehydr a 9 (39 kD) NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 9 (39 kD)
pp M6 ATP syn B ATP synthase subunit B
Inconsistent - other metabolism
qq C7 Transketolase Transketolase
rr H8 Fatty acid binding 7 brain Fatty acid binding protein 7 brain (B-FABP)
ss A3 Fatty acid binding H6 Fatty acid binding protein H6-isoform
tt O10 Fatty acid binding heart Heart-type fatty acid-binding protein (H-FABP)
uu D9 Fatty acid syn Fatty acid synthase
vv F9 Glutamate decarboxylase Glutamate decarboxylase
Letters in the first column refer to genes illustrated in Figure 7; the grid column identifies genes as reported in our data entry to the NCBI Gene Expression Omnibus (GLP1224). Gene identities are listed as those identified by Swiss-Prot and as shortened names, and grouped as genes involved in oxidative phosphorylation or in other biochemical pathways.
Table 2 Sources of variance and calculation of variables for the split-plot ANOVA statistical design [18]
Source of variance df Sum of squares Fs
Among populations P - 1 2 SS_P I × T × Σ(popmeans - GM)2 MS_P/MS_I(P)*
Among tissues T - 1 2 SS_T I × P × Σ(tissuemeans - GM)2 MS_T/MS_TI(P)†
Interaction (P - 1) × (T - 1) 4 SS_PT I × Σ(tissue(population)means - popmeans - tissuemeans + GM)2 MS_PT/MS_TI(P)‡
Among inviduals in population P(I - 1) 6 SS_I(P) T × Σ(ind(population)means - popmeans)2
Tissue-by-individual in population P × (T - 1) × (I-1) 12 SS_TI(P) Σ(samplemean - tissue(population)mean - ind(population)mean + popmean)2 MS_TI(P)/MS_dye§
Dye within individuals P × T × I × (D - 1) 27 SS D(I)
Replicate hybridization in dye P × T × I × D × (H - 1) 54
Spot in replicate hybridization P × T × I × D × H × (S - 1) 540
Total P × T × I × D × H × S - 1 647
==== Refs
Cavalieri D Townsend JP Hartl DL Manifold anomalies in gene expression in a vineyard isolate of Saccharomyces cerevisiae revealed by DNA microarray analysis. Proc Natl Acad Sci USA 2000 97 12369 12374 11035792 10.1073/pnas.210395297
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
Oleksiak MF Churchill GA Crawford DL Variation in gene expression within and among natural populations. Nat Genet 2002 32 261 266 12219088 10.1038/ng983
Jin W Riley RM Wolfinger RD White KP Passador-Gurgel G Gibson G The contributions of sex, genotype and age to transcriptional variance in Drosophila melanogaster. Nat Genet 2001 29 389 395 11726925 10.1038/ng766
Enard W Khaitovich P Klose J Zoellner S Heissig F Giavalisco P Nieselt-Struwe K Muchmore E Varki A Ravid R Intra- and interspecific variation in primate gene expression patterns. Science 2002 296 340 343 11951044 10.1126/science.1068996
Hsieh WP Chu TM Wolfinger RD Gibson G Mixed-model reanalysis of primate data suggests tissue and species biases in oligonucleotide-based gene expression profiles. Genetics 2003 165 747 757 14573485
Cheung VG Conlin LK Weber TM Arcaro M Jen KY Morley M Spielman RS Natural variation in human gene expression assessed in lymphoblastoid cells. Nat Genet 2003 33 422 425 12567189 10.1038/ng1094
Brem RB Yvert G Clinton R Kruglyak L Genetic dissection of transcriptional regulation in budding yeast. Science 2002 296 752 755 11923494 10.1126/science.1069516
Zhang L Zhou W Velculescu VE Kern SE Hruban RH Hamilton SR Vogelstein B Kinzler KW Gene expression profiles in normal and cancer cells. Science 1997 276 1268 1272 9157888 10.1126/science.276.5316.1268
Elek J Park KH Narayanan R Microarray-based expression profiling in prostate tumors. In Vivo 2000 14 173 182 10757075
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JG Sabet H Tran T Yu X Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000 403 503 511 10676951 10.1038/35000501
Paez JG Janne PA Lee JC Tracy S Greulich H Gabriel S Herman P Kaye FJ Lindeman N Boggon TJ EGFR mutations in lung cancer: correlation with clinical response to Gefitinib therapy. Science 2004 304 1497 500 15118125 10.1126/science.1099314
Archacki SR Angheloiu G Tian X-L Tan FL DiPaola N Shen G-Q Moravec C Ellis S Topol EJ Wang Q Identification of new genes differentially expressed in coronary artery disease by expression profiling. Physiol Genomics 2003 15 65 74 12902549
Iemitsu M Miyauchi T Maeda S Sakai S Fujii N Miyazaki H Kakinuma Y Matsuda M Yamaguchi I Cardiac hypertrophy by hypertension and exercise training exhibits different gene expression of enzymes in energy metabolism. Hypertens Res 2003 26 829 837 14621187 10.1291/hypres.26.829
Kunz WS Different metabolic properties of mitochondrial oxidative phosphorylation in different cell types - important implications for mitochondrial cytopathies. Exp Physiol 2003 88 149 154 12525863 10.1113/eph8802512
Ramaswamy S Tamayo P Rifkin R Mukherjee S Yeang CH Angelo M Ladd C Reich M Latulippe E Mesirov JP Multiclass cancer diagnosis using tumor gene expression signatures. Proc Natl Acad Sci USA 2001 98 15149 15154 11742071 10.1073/pnas.211566398
Baris O Savagner F Nasser V Loriod B Granjeaud S Guyetant S Franc B Rodien P Rohmer V Bertucci F Transcriptional profiling reveals coordinated up-regulation of oxidative metabolism genes in thyroid oncocytic tumors. J Clin Endocrinol Metab 2004 89 994 1005 14764826 10.1210/jc.2003-031238
Steel RGD Torrie JH Principles and Procedures of Statistics 1980 2 New York, NY: McGraw-Hill
Oleksiak MF Roach JL Crawford DL Natural variation in cardiac metabolism and gene expression in Fundulus heteroclitus. Nat Genet 2005 37 67 72 15568023
Kendziorski CM Zhang Y Lan H Attie AD The efficiency of pooling mRNA in microarray experiments. Biostatistics 2003 4 465 477 12925512 10.1093/biostatistics/4.3.465
Pritchard CC Hsu L Delrow J Nelson PS Project normal: defining normal variance in mouse gene expression. Proc Natl Acad Sci USA 2001 98 13266 13271 11698685 10.1073/pnas.221465998
Weiss L (Ed) Cell and Tissue Biology: A Textbook of Histology 1983 6 Baltimore, MD: Urban and Schwarzenberg
Guyton AC Textbook of Medical Physiology 1991 8 Philadelphia: W.B. Saunders Company
Lee PD Sladek R Greenwood CMT Hudson TJ Control genes and variability: absence of ubiquitous reference transcripts in diverse mammalian expression studies. Genome Res 2002 12 292 297 11827948 10.1101/gr.217802
Nachman MW Hoekstra HE D'Agostino SL The genetic basis of adaptive melanism in pocket mice. Proc Natl Acad Sci USA 2003 100 5268 5273 12704245 10.1073/pnas.0431157100
Oleksiak MF Kolell KJ Crawford DL Utility of natural populations for microarray analyses: isolation of genes necessary for functional genomic studies. Mar Biotechnol (NY) 2001 3 (Supplement 1) S203 S211 14961317 10.1007/s10126-001-0043-0
FunnyBase gene expression database
Van Gelder RN Von Zastrow ME Yool A Dement WC Barchas JD Eberwine JH Amplified RNA synthesized from limited quantities of heterogeneous complementary DNA. Proc Natl Acad Sci USA 1990 87 1663 1667 1689846
Raghavachari N Bao YP Li G Xie X Muller UR Reduction of autofluorescence on DNA microarrays and slide surfaces by treatment with sodium borohydride. Anal Biochem 2003 312 101 105 12531193 10.1016/S0003-2697(02)00440-2
Quackenbush J Microarray data normalization and transformation. Nat Genet 2002 32 Suppl 496 501 12454644 10.1038/ng1032
Wu H Kerr K Cui X Churchill GA 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
Chu TM Weir B Wolfinger R A systematic statistical linear modeling approach to oligonucleotide array experiments. Math Biosci 2002 176 35 51 11867082 10.1016/S0025-5564(01)00107-9
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
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 10.1093/nar/30.4.e15
Brazma A Hingamp P Quackenbush J Sherlock G Spellman P Stoeckert C Aach J Ansorge W Ball CA Causton HC Minimum information about a microarray experiment (MIAME): toward standards for microarray data. Nat Genet 2001 29 365 371 11726920 10.1038/ng1201-365
Sokal RR Rohlf FJ Biometry 2001 3 New York: W.H. Freeman
Rohlf FJ Sokal RR Statistical Tables 2002 3 New York: W.H. Freeman
Kumar S Tamura K Jakobsen IB Nei M MEGA: Molecular Evolutionary Genetics Analysis, version 2.1
| 15693942 | PMC551533 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 26; 6(2):R13 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r13 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r131569394210.1186/gb-2005-6-2-r13ResearchVariation in tissue-specific gene expression among natural populations Whitehead Andrew [email protected] Douglas L [email protected] Rosenstiel School of Marine and Atmospheric Science, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149, USA2005 26 1 2005 6 2 R13 R13 28 6 2004 2 9 2004 6 12 2004 Copyright © 2005 Whitehead and Crawford; 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 expression of a selected suite of 192 metabolic genes in brain, heart and liver in three populations of the teleost fish Fundulus heteroclitus was examined. Only a small subset (31%) of tissue-specific differences was consistent in all three populations, indicating that many tissue-specific differences in gene expression are unique to one population and thus are unlikely to contribute to fundamental differences between tissue types.
Background
Variation in gene expression is extensive among tissues, individuals, strains, populations and species. The interactions among these sources of variation are relevant for physiological studies such as disease or toxic stress; for example, it is common for pathologies such as cancer, heart failure and metabolic disease to be associated with changes in tissue-specific gene expression or changes in metabolic gene expression. But how conserved these differences are among outbred individuals and among populations has not been well documented. To address this we examined the expression of a selected suite of 192 metabolic genes in brain, heart and liver in three populations of the teleost fish Fundulus heteroclitus using a highly replicated experimental design.
Results
Half of the genes (48%) were differentially expressed among individuals within a population-tissue group and 76% were differentially expressed among tissues. Differences among tissues reflected well established tissue-specific metabolic requirements, suggesting that these measures of gene expression accurately reflect changes in proteins and their phenotypic effects. Remarkably, only a small subset (31%) of tissue-specific differences was consistent in all three populations.
Conclusions
These data indicate that many tissue-specific differences in gene expression are unique to one population and thus are unlikely to contribute to fundamental differences between tissue types. We suggest that those subsets of treatment-specific gene expression patterns that are conserved between taxa are most likely to be functionally related to the physiological state in question.
==== Body
Background
The regulation of gene expression varies extensively among tissues, individuals, strains, populations and species [1-6] and variation in gene expression has a genetic basis [7,8]. Despite such biological variance, differences in gene expression are used to describe cancers [9-12], heart failure [13,14] and metabolic diseases [15]. It is common for these pathologies to be associated with changes in tissue-specific gene expression or changes in metabolic gene expression. For example, many different cancers have unique tissue-specific patterns of gene expression [16], and thyroid cancers are associated with increases in aerobic metabolic gene expression [17].
Although tissue-specific gene expression patterns are often used as a method to identify functionally relevant genes, how conserved these differences are among outbred individuals and among populations has not been well documented. It is possible that many of these changes represent polymorphism among individuals or populations and are not specifically associated with disease. To address this we used a well established system (tissue-specific gene expression) and genes with well defined function and tissue-specific distributions (metabolic genes).
Given the high variance in gene expression among individuals and populations, our goal was to examine the conservation of tissue-specific gene expression among populations of the same species. Specifically, we assessed the among-population variance of tissue-specific patterns of gene expression (in brain, heart and liver) in the teleost fish Fundulus heteroclitus. A cDNA microarray was used to measure levels of expression in normal healthy male fish for 192 genes involved in central metabolic pathways. We used this compact array in order to impose a high degree of technical and biological replication (24 replicates for each of three tissues from nine individuals with two samples per array). Also, this array was used because metabolic genes are essential, are known to have tissue-specific expression, especially in fish, and are often misused as controls with little characterization of variation in expression among individuals or tissues. Analysis of variance (ANOVA) was used as a statistical test to determine which genes were differentially expressed among tissues and populations. Tissue-specific patterns of gene expression were compared among populations. As expected, we detected extensive variation in gene expression among tissues. Unexpectedly, only a fraction (31%) of tissue-specific differences was conserved between all populations.
Results
Variation among
Variation among individuals within groups was high (groups included the nine tissue-by-population groupings; Figure 1). Nearly half of genes (92 genes, 48%) were differentially expressed (p < 0.05) among individuals within populations and tissues (Figure 1), and inter-individual differences ranged over fivefold.
Variation among tissues
Although variation among individuals was high, added variation due to tissues was significant. Considering 192 genes and a p-value of 5%, one would expect less than 10 false-positive differences among tissues under the null hypothesis. We detected 76% of genes (146 of 192 genes) differentially expressed among brains, hearts and livers (ANOVA, p < 0.05). Selecting the α level at which differences between treatments are considered significant is problematic because of the large number of comparisons performed. As such, we present a volcano plot to illustrate the range of expression differences between tissues and associated p-values (Figure 2). When α is set at 0.01, 0.001 or at the Bonferroni-corrected value (2.6 × 10-4), the proportion of significant genes is 67% (129 genes), 50% (96) and 39% (75), respectively. Significant differences in expression ranged from less than 1.2-fold to nearly 16-fold (Figure 2). The predominant pattern of tissue-specific expression can be described by expression significantly different in the liver compared to the other two tissues (Figure 3).
Many expected tissue-specific patterns emerged. For example, the brain-specific fatty-acid-binding protein was typically more highly expressed in the brain than in other tissues (p = 0.005), hepatocyte nuclear factor 4-alpha (a transcription factor) was more highly expressed in liver than in other tissues (p < 0.001), and two genes involved in glycerolipid metabolism -lipoprotein lipase and phopholipase XIII A2 - were more highly expressed in liver than other tissues (p < 0.001 for both genes).
Liver-specific expression accounted for 61% of the expression differences among tissues (Figure 4). Heart-specific and brain-specific expression accounted for 24% and 15% of differences among tissues, respectively. Regardless of population, expression patterns were typically most similar between heart and brain, and least similar between liver and heart (Figure 5). There were 67 genes printed on the array that code for proteins involved in oxidative phosphorylation, and 88% (59 genes) were differentially expressed between tissues (genes highlighted in green, Figure 3). Of differentially expressed oxidative phosphorylation genes, only 10% (six genes) were expressed more highly in the liver than in other tissues, whereas the remaining 90% (53 genes) had lower expression in the liver compared to brain or heart.
Variation among taxa
A small proportion of genes (six genes, 3%) differed in expression among populations (p < 0.05). However, it should be noted that although the split-plot design is powerful for detecting differences between split-plot factors (tissues), it is considered to have low power for detecting differences between blocks (populations) [18]. As such, it is likely that 3% is an underestimate of true among-population differences in gene expression. Indeed, two-way ANOVA (data not shown), which has higher power for detecting population differences but is less valid than the split-plot model for testing individual and tissue differences, detected among-population differences in expression for 18% of genes at p < 0.05, or 6.3% of genes at p < 0.01. Each tissue contributed a similar number of genes differentially expressed among populations.
Surprisingly, differences among tissues in gene expression were not consistent across all three populations. More than one-third (37%) of the genes differentially expressed between tissues were significant in only one of the three populations (Figure 6). Population-specific differences were distributed among the three populations; Georgia had 40% of the population-specific genes, and New Jersey and Maine had 34% and 26%, respectively. A proportion of these inconsistencies could be due to false-positive or false-negative differences between tissues in individual populations. However, statistically significant interaction between tissue and population was detected for many (30%) of these inconsistencies (see Additional data file 1).
A relatively small proportion of tissue-specific genes (31%) have consistent expression patterns in all three populations (Figure 6; also see Additional data file 1 for details). This subset of genes also reflects the different metabolic status of brain, heart and liver; most of the genes involved in oxidative phosphorylation were more highly expressed in brain and heart than in liver (Figure 7a, Table 1), and most of the genes involved in fatty-acid metabolism, glycerolipid metabolism, steroid metabolism and detoxification were more highly expressed in liver. The majority of the tissue-specific genes were not consistent among populations (a subset of these genes are illustrated in Figure 7b, Table 1).
Quality control
Variation among technical replicates was low, and permutation tests indicated that the ANOVA model was robust. Sample coefficients of variation (CVs (standard deviation/mean) × 100), which estimate technical variance due to replicate spots (six spots per hybridization), repeated measures (two hybridizations per dye), and dye (two dyes per sample), were calculated for each gene of each of the 27 samples. CVs less than 5% accounted for 95% of sample/genes, respectively. Of the many comparisons performed (differences among tissues, populations, interaction), permutation tests results agreed with ANOVA results (the same comparisons identified as significant or not significant) for 99.1% of comparisons, suggesting that our ANOVA model was robust.
Discussion
Considerable variation occurs among the 27 samples (three tissues from each of three individuals from three populations) used to measure inter-individual and tissue-specific variation in gene expression. We are able to precisely describe the patterns of gene expression for 192 metabolic genes because of the low experimental variation; for 95% of the replicate measures of gene expression the standard deviation is less than 5% of the mean. Notably, gene expression is statistically different for many genes among individuals within a population for a tissue (48%), between tissues (76%), and between populations (3%). For genes with tissue-specific expression, only a fraction (31%) had expression patterns consistent across all three populations. These data do not specifically identify tissue-specific differences that are inconsistent across populations, but rather emphasize that tissue-specific differences detected can vary from one population to another. When measured from a single population, highly significant differences in tissue-specific expression do not necessarily represent genes relevant to general functional or morphological differences between tissues.
Variation among individuals
Variation in gene expression among healthy male individuals raised under controlled laboratory conditions was high. Nearly half of the metabolic genes (48%) were differentially expressed among individuals within a population for any one tissue (Figure 1), with fold differences ranging from 1.2- to 5-fold and p-values ranging down to 10-7. Differences in gene expression among individuals are unlikely to be due to common reversible environmental factors that affect physiological performance (acclimation effects) since all individuals used in this study were housed in a common environment and fed the same food for at least two months. However, the differences could be due to irreversible developmental effects or genetic variations that affect gene expression. Regardless of this, if these differences are heritable or due to developmental plasticity, they represent variation one would expect to find among outbred organisms, including humans.
Other studies that have measured inter-individual differences in gene expression have also detected high levels of variation in a variety of taxa. Among crosses of different yeast strains a large number of differences in expression (6% of genes varying more than twofold) were detected between morphotypes [1]. A previous study of the same Maine and Georgia Fundulus populations assayed here detected 18% of genes differentially expressed among healthy individuals [3]. Although inter-individual variance in gene expression seems prevalent, our observation that 48% of genes are differentially expressed among individuals is high. This may reflect the greater precision of these measurements as a result of extensive technical replication (24 replicate measures per sample) as coefficients of variation for technical replicates was less than 5% for 95% of the genes. Indeed, using similar methods and tools, a concurrent study assessing variation in Fundulus also detected a very high proportion of genes (94%) differentially expressed among individuals [19]. Alternatively, since our array is heavily biased toward metabolic genes, detected variance may also reflect a greater variation in metabolic gene expression. We could speculate that the high variation in metabolic genes reflects a greater allowable variation. That is, there may be less selective pressure to constrain metabolic variation either because varying the amount of an enzyme does not affect metabolism or variation in metabolism is phenotypically acceptable. One could test this by using an array with more comprehensive representation of the genome and comparing variances of different gene classes defined by function.
Considering the high inter-individual variation detected, the data presented here underscore the importance of including biological replicates within treatment groups in order to ascribe differences in expression to treatment rather than to inter-individual variation. Statistically, an analysis of variance can be used to examine the effects of technical and biological variation, and these tests have proved powerful for detecting significant differences in gene expression [3,4], even differences as small as 1.2-fold. The cost of resources in microarray experiments should no longer excuse lack of biological and technical replication. Often, microarray experiments pool individual samples within treatment groups to capture biological variation. However, this approach only estimates an average level of expression and fails to estimate biological variation. When only small quantities of RNA can be extracted from samples, one can estimate biological variation by pooling multiple independent samples [20].
A variety of factors can contribute to differences in gene expression among individuals. Pritchard et al. [21] proposed that differences in immune status may explain the 3.3% difference in gene expression among genetically identical mice. Sex explained a large portion of among-individual variation in gene expression in Drosophila, whereas genotype was less of an influence, and the influence of age was weak [4]. Furthermore, this type of variation can be biologically relevant. For example recent work in Fundulus indicates that most inter-individual variation in metabolism can be accounted for by differences in metabolic gene expression [19].
Variation among tissues
Another important source of biological variation in gene expression is differences in expression among different tissues; 76% of genes were differentially expressed between brain, heart and liver, and expression in the liver was the most distinct compared to heart and brain. In this study, genes printed on our array are primarily enzymes functional in central metabolic pathways such as fatty-acid metabolism, glycolysis and oxidative phosphorylation. Of the oxidative phosphorylation genes differentially expressed between tissues, 92% were more highly expressed in heart or brain than in liver (Figure 3). The primary purpose of the heart is to act as a pump, and contraction is highly dependent on oxidative metabolism [22]. The metabolic rate in the brain is 7.5 times the average rate in the rest of the body [23]. High metabolic demand in the brain supports pumping of ions across neuronal membranes during action potentials and metabolism is primarily oxidative. Mitochondria are the principal sites for oxidative phosphorylation, and are most numerous in heart, brain and skeletal muscle cells. The liver, in contrast, is much more functionally diverse, as it is involved in carbohydrate storage, synthesis of proteins, glucose, fatty acids, cholesterol and lipids, and metabolism of xenobiotics and endogenous compounds, and has a relatively low respiration rate. Accordingly, transcripts of genes functional in oxidative phorphorylation appear to represent a much smaller portion of the cell's RNA transcripts in liver tissues than in the heart or brain. In addition, genes involved in fatty acid and phospholipid synthesis were more highly expressed in liver than the other tissues. Differences in expression among tissues detected using our array appear to reflect differences in the metabolic status of brain, heart, and liver. Because data presented here support well established patterns of metabolism, they suggest that measuring mRNA expression using microarrays accurately reflects changes in proteins and their phenotypic effect.
Many microarray studies have used expression levels of 'housekeeping' genes as an internal control for comparisons among arrays, individuals and treatments. Housekeeping genes may be defined as those that are involved in routine cellular metabolism and always expressed in all cells. Accordingly, many, if not most, of the genes studied here could be considered housekeeping genes. Nearly half of these genes were expressed at different levels between individuals, with fold differences ranging from 1.2- to 5-fold and p-values ranging down to 10-7. Lee et al. [24] applied ANOVA to screen four previously published datasets for housekeeping genes across a variety of biological contexts. They found that all genes that are commonly used as controls had fold changes ranging from greater than 2.0 to more than 300 within at least one dataset, and coefficients of variation were concordantly high, reflecting high variance in expression of these genes. It appears that upon application of ANOVA, statistically significant differences in expression of housekeeping genes can be detected among individuals and across different biological contexts, and scaling for differences among arrays using expression levels of these genes ought to be approached with caution.
Although genes differentially expressed among tissues reflect their different metabolic requirements, it should be noted that the purpose of the current study was not to comprehensively identify suites of genes responsible for functional differences between tissues. The relatively small number of printed probes was useful for a high degree of technical replication, and obviously represents a small portion of the expressed genes. However, this approach shows that highly significant differences in gene expression among tissues may be apparent but not consistent among closely related taxa. Therefore, highly significant differences in gene expression found only within a single population may not necessarily represent genes relevant to general functional or morphological differences between tissues.
Variation among taxa
Although the pattern of metabolic gene expression among tissues reflects established patterns of tissue-specific metabolism, there is additional variation due to population. It should be noted that the split-plot statistical design is not as powerful for detecting among-block differences (among populations) as for detecting differences among split-plot factors [18]. We detected 3% of genes (6 of 192) differentially expressed among populations. This proportion is similar to that detected in a previous study [3] in which 2.6% of genes were differentially expressed between Maine and Georgia Fundulus hearts. Similarly, approximately 1% of genes were differentially expressed in brain tissue among inbred strains of mice [2]. Differences in gene expression are to be expected among taxa (phylogenetically distinct groups of organisms which may include strains, populations or species), with the majority of differences most likely to be attributable to random genetic drift. For more distantly related groups, one would expect expression patterns to be more divergent than for closely related groups. Indeed, expression patterns between humans and chimpanzees are more similar than those between humans and orangutans, and similar results were obtained from comparisons among three mouse species [5,6].
An unexpected finding is that the tissue-specific differences depend on which population was assayed. Differences in gene expression are expected between tissues because of functional divergence and between populations because of neutral genetic divergence. In addition, one might expect that the number of genes significantly different between populations would depend on the tissue. One might also expect tissue-specific differences to be consistent in all taxa. Yet our data indicate that tissue-specific expression patterns are not fixed within a species. The genes for which expression is significantly different between tissues are not all the same in all three populations. Of the 128 genes that have tissue-specific patterns of expression in any population, 37% are tissue-specific in only one of the three populations and 32% are found in only two of the three populations. Overall, it would appear that only 31% of tissue-specific differences in gene expression are consistent among all populations of F. heteroclitus. One needs to be careful about this interpretation, however. Our emphasis was not to specifically identify genes that have significant interaction between tissue and population. Rather, we emphasize that genes detected as tissue specific will vary from one population to another, and most microarray studies measure treatment-specific expression patterns in only one population of test organism. Because inter-individual variation is high, it is probable that inclusion of more replicate individuals in each group would increase the sensitivity of ANOVA, and the number of genes that distinguish tissues consistently in all populations may change.
The consistent tissue-specific differences still support expectations based on the metabolic requirements of each tissue (for example, genes involved in oxidative phosphorylation were more highly expressed in heart and brain, and those involved in fatty-acid and lipid metabolism were more highly expressed in the liver; Figure 7a). Accordingly, those differences in expression that are consistent across several groups of organisms are most likely to account for functional and morphological differences among tissues, emphasizing that this type of comparative approach may be powerful for testing the biological relevance of other functional traits. For example, expression differences between diseased and non-diseased tissues may vary among mouse strains, so that the subset of differences that are consistent across strains are more likely to be functionally related to the diseased state.
Our data suggest that many of the differences in gene expression detected between experimental groups may be of little functional importance because they vary among taxa. We suggest that patterns of expression that are consistent in different populations are more likely to be functionally important. Elucidation of adaptively important variation, such as variation related to antibiotics, pesticides or temperature adaptation, may also benefit from such a comparative approach that screens for conserved patterns. However, there is the possibility that partitioning of genetic polymorphisms among populations may allow distinct groups of organisms to reach different physiological or biochemical solutions to the same biological challenges. For example, patterns of polymorphism in a gene that regulates coat color in mammals indicated recent directional selection and was associated with coat color in one pocket mouse population, but not in a second population [25]. Other loci were probably responsible for adaptive variation in coat color in the second population.
Conclusions
These data indicate high variation in metabolic gene expression among individuals and thus expression of these housekeeping genes is unreliable as an internal control or as a method of normalization across samples. Second, concordance between tissue-specific expression patterns and established metabolic functions of brain, heart and liver indicate that measuring mRNA levels accurately reflects physiological status. Furthermore, since many metabolic genes differ in expression among brain, heart and liver, those studies using whole organisms need to rule out whether changes in expression reflect differences in the proportions of various tissues among samples. Finally, studies seeking to identify patterns of gene expression related to physiological states, such as disease or toxic stress, must consider both variation between individuals and differences between populations. Because of this biological variation, not all differences between treatments in any one population of test organism are likely to be generally relevant. We suggest that conserved patterns of treatment-specific gene expression among taxa are most likely to be functionally related to the physiological state in question.
Methods and materials
Animals and maintenance
Teleost fish Fundulus heteroclitus were collected from the field by seine and minnow trap in June 2003, transported to the University of Miami RSMAS laboratory under controlled temperature and aeration conditions, and acclimated to common conditions (20°C, 15 parts per thousand salinity) in recirculating 100-gallon tanks for at least two months before experiments. Fish were sacrificed by cervical dislocation and tissues were excised and stored in RNAlater (Ambion) at -20°C. Fish were collected at Wiscasset, Maine; Stone Harbor, New Jersey, and Sapelo Island, Georgia. Only healthy male fish were used for the following experiments.
Microarrays
Microarrays were printed using 192 cDNAs from a F. heteroclitus cardiac library encoding essential proteins for cellular metabolism [26]. These cDNAs were a subset of over 40,000 expressed sequences in our online database Funnybase [27]. These 192 cDNAs were amplified with amine-linked primers and printed on 3-D Link Activated slides (Surmodics) using a SpotArray Enterprise piezoelectric microarray printer (PerkinElmer Life Sciences) at Louisiana State University. Slides were blocked following slide manufacturer protocols. The suite of 192 amplified cDNAs was printed as a group in six spatially separated replicates. Four hybridization zones of these six replicate arrays were printed per slide, with each zone set separated by a hydrophobic barrier.
Hybridization experimental design
Microarray analyses were applied to three tissues (brain, heart and liver) from three individuals collected from three populations of F. heteroclitus. Each of these 27 samples was measured four times, twice with Cy3 and twice with Cy5 (Figure 8). In addition, since a hybridization zone covered six replicate printed arrays, total experimental replication per sample per gene was 24-fold. A total of 108 hybridizations were performed (27 × 4), and Cy3-Cy5 hybridizations were balanced (although incompletely) among tissues and populations in a sheet-loop design (Figure 8).
Sample preparation
RNA was extracted from tissue homogenate in a chaotropic buffer using phenol/cholorform/isoamyl alcohol. All reagents were from Sigma unless otherwise noted. Tissues were removed from RNAlater, blotted dry, and homogenized using an electric homogenizer in 400 μl chaotropic buffer (4.5 M guanidinium thiocyanate, 2% N-lauroylsarcosine, 50 mM EDTA pH 8.0, 25 mM Tris-HCl pH 7.5, 0.1 M β-mercaptoethanol, 2% antifoam A). An equal volume of 2 M sodium acetate (pH 4.0) was added to the homogenate, followed by 400 μl acidic phenol (pH 4.4), and 120 μl chloroform/isoamyl alcohol (23:1). The mixture was kept at 4°C for 10 min then centrifuged at 4°C at 16,000g for 20 min. Supernatant was removed and combined with 400 μl isopropanol, stored at -20°C for 30 min, then centrifuged at 4°C at 16,000g for 30 min. The remaining RNA pellet was rinsed twice with 400 μl of 70% ethanol, then further purified using the Qiagen RNeasy Mini kit (Qiagen) following the manufacturer's protocols. Purified RNA was quantified spectrophotometrically, and RNA quality was assessed using the Agilent 2100 Bioanalyzer. RNA was stored in 1/10 volumes 3 M sodium acetate and 2.5 volumes 100% ethanol at -20°C.
RNA for hybridization was prepared by amplification using a modified Eberwine protocol [28]. The Ambion Amino Allyl MessageAmp aRNA Kit was used (according to manufacturer's protocols) to copy template RNA by T7 amplification following incorporation of a T7 promoter, resulting in amplified template in the form of antisense RNA. Amino-allyl UTP was incorporated into targets during T7 transcription, and resulting amino-allyl antisenseRNA was coupled to Cy3 and Cy5 dyes (Amersham Biosciences).
Hybridization
Labeled aRNA aliquots of the two individual samples for each hybridization (18 pmol each of Cy3 and Cy5) were vacuum dried together and resuspended in 12 μl hybridization buffer (final concentration of each labeled sample = 1.5 pmol/μl). Hybridization buffer consisted of 5 × SSPE, 1% SDS, 50% formamide, 1 mg/ml poly(A), 1 mg/ml sheared herring sperm carrier DNA, and 1 mg/ml BSA. Slides were washed in sodium borohydride solution according to Raghavachari et al. [29] to reduce autofluorescence. Following rinsing, slides were boiled for 2 min and spin-dried in a centrifuge at 800 rpm for 3 min. Samples (12 μl) were heated to 90°C for 2 min, quick cooled to 42°C, applied to slide (hybridization zone area was 350 mm2), and covered with a coverslip. Slides were placed in an airtight chamber humidified with paper soaked in 1 × SSC buffer and incubated 12-18 h at 42°C. Following hybridization, slides were scanned using the Packard Bioscience ScanArray Express microarray scanner (PerkinElmer Life Sciences). Resulting .tiff images were imported into spot grids built in ImaGene (Biodiscovery) for each array, and spot signals were collected as fluorescence intensities for each dye channel.
Data processing and statistical analysis
Raw data were first sum normalized [30], which involves summing the total signal from each replicate array to the same value. Then spatial bias on each array was smoothed using a lowess transformation in MAANOVA Version 0.93-2 for R [31]. Other methods of normalization have also been proposed [32-34]. Log2 values of lowess-transformed sum-normalized data were used for all subsequent statistical analyses. MIAME-compliant data [35] have been submitted to the Gene Expression Omnibus as accession number GLP1224. Data were analyzed in a split-plot ANOVA design with population as blocks and tissues as split-plot factors using scripts written in MatLab Version 6 (The MathWorks). MatLab code is available upon request from the authors. Nested within tissue-by-population samples were technical replicates. Replicate spots within hybridization (six), replicate hybridizations per labeling (two) and replicate labelings per sample (two; Cy3 and Cy5) represent the three levels of technical variance nested within the tissue-by-population sample. The ANOVA structure is presented in Figure 9 and Table 2, and the model can be written as:
y = grand mean + population + tissue + population-tissue interaction + individual in population + tissue-by-individual within population + dye within individual + hybridization within dye + spot within hybridization
where y is the normalized log2 expression and individual in population and tissue-by-individual within population are random effects. To test for differences among multiple means (for example, among population and tissue groups), and to correct for multiple comparisons, the T-method [36] was applied. The T-method calculates the minimum significant range defined as
MSR = Qα[kv] × SE
where the critical value Qα[kv] is the studentized range [37], k = number of groups in the comparison (for example, if comparisons are among tissues then k = 3), v = degrees of freedom of MStissue-by-individual within population, and SE is the standard error among tissue-by-individual samples within populations. The T-method following ANOVA was used to identify genes differentially expressed among tissues in each population. These data were then used to contrast tissue-specific and population-specific expression patterns. Robustness of ANOVA data was tested using a permutation test; means for the 27 biological samples were randomly permuted 1,000 times between population and tissue and test statistics were recalculated for differences among populations, tissues and tissue-by-population interaction. Agreement between ANOVA and permutation test results would indicate the robustness of the ANOVA model. Finally, in order to graphically illustrate expression similarity among tissues, expression distance between samples was calculated as the sum of differences of log2 expression values over all genes, and neighbor-joining trees of global similarity of expression patterns among tissues (L, liver; H, heart; B, brain) were constructed [38] for each population.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 lists the results from statistical analyses for all genes. Listed for each gene are p-values associated with statistical tests for differences in expression between populations, tissues, tissue-by-population interaction, and among individuals within populations. Also listed are mean expression for each sample, and columns comparing differences in expression between tissues within each population. Final columns tabulate whether a tissue difference was detected for each comparison, whether this difference was consistent between populations, and whether significant interaction was detected for that gene.
Supplementary Material
Additional data file 1
The results from statistical analyses for all genes
Click here for additional data file
Acknowledgements
Much credit is due to Marjorie Oleksiak for construction of the expressed sequence tag (EST) library, array printing, and constructive criticisms. We also thank Steve Hand at Louisiana State University for the use of their facilities and assistance in printing the microarray. We thank Gary Churchill for statistical advice and Justin Paschall for EST database management and bioinformatics. Valuable assistance was provided by Jen Roach and Jeff VanWye, and Jen Roach provided helpful comments on the manuscript. We especially thank two reviewers for insightful criticisms and comments on the manuscript. This project was supported by a National Science Foundation OCE grant 0221879 to D.L.C. and a National Institutes of Health grant NHLBI R01 HL65470 to D.L.C.
Figures and Tables
Figure 1 Variation within individuals (technical variance) and among individuals within populations and tissues (biological variance) for each of 192 genes indicated by the mean square error (MS) of measurements. Points above the dashed line indicate genes with greater variance among individuals than within. F-crit is the critical value of the F-statistic (F = MSamong/MSwithin, with 12 and 27 degrees of freedom and α = 0.05) for testing significant differences in gene expression between individuals. For 48% of genes, MSamong/MSwithin > F-crit (solid red line). These genes are therefore differentially expressed among individuals within treatments.
Figure 2 Volcano plot of differences between tissues and corresponding p-values. Differences in expression for each gene is the log2 ratio of tissue mean expression minus grand mean; a twofold difference in expression between tissues is indicated by one unit separation along the x-axis. p-values for differences in gene expression among tissues were calculated using ANOVA, and illustrated as -log(p). A p-value of 10-4 is expressed as 4 on the y-axis, and the α = 0.05 threshold is indicated by the red dashed line (1 - log(0.05) = 1.3).
Figure 3 Dendrogram of gene expression patterns across samples for genes significantly different between tissues (ANOVA, p < 0.05). Clustering indicates similar expression patterns among samples (top axis) and among genes (left axis). Samples cluster as livers (yellow), hearts (pink) and brains (blue). Genes involved in oxidative phosphorylation are highlighted in green, and expression patterns that are consistent across all three populations are highlighted with a blue triangle.
Figure 4 Number of genes differentially expressed among tissue groups for each population. Tissue-specific genes are those that are expressed more highly in a tissue than in the other tissues (for example, L > H, B) or lower in a tissue than in the other tissues (for example, L < H, B).
Figure 5 Similarity of expression patterns among tissues. (a) Proportion of 192 genes that are similarly expressed between heart and brain (black bar), brain and liver (gray bar) and liver and heart (white bar), for each population including Maine (ME), New Jersey (NJ) and Georgia (GA). (b) Neighbor-joining trees of global similarity of expression patterns among samples (L, liver; H, heart; B, brain) for each population. Distance between samples is the sum of differences of log2 expression values over all genes.
Figure 6 Shared expression patterns among populations.
Figure 7 Gene expression in liver, brain and heart (three symbols for each line) for the three different populations (three lines per gene). Each letter represents a gene, expression values are log2 transformed and are indicated for liver, brain and heart (left to right) in each of Maine (circles), New Jersey (triangles) and Georgia (squares) populations. (a) Genes consistently different among tissues in all three populations are grouped as those involved in oxidative phosphorylation (upper panel) and those involved in other metabolic pathways (lower panel). (b) A representative subset of genes not consistently different among tissues in all populations. Gene names associated with letters are provided in Table 1 and Additional data file 1.
Figure 8 Experimental design for hybridizations. Each arrow represents an array hybridization, with the samples at arrow base and head labeled with Cy3 and Cy5, respectively. Liver, heart and brain samples are indicated as purple, red and blue circles, respectively. Three individuals were assayed per tissue and from each of three populations. ME, Maine; NJ, New Jersey; GA, Georgia.
Figure 9 Split-plot ANOVA statistical design. Populations (ME, Maine; NJ, New Jersey; GA, Georgia) are treated as blocks, replicate individuals within each population (1, 2 and 3) as plots, and tissue (L, liver; H, heart; B, brain) within an individual as the split-plot factor. Nested within each tissue-by-individual sample are technical replicates including two dyes (Cy3 and Cy5) within each sample, two replicate hybridizations (A and B) per dye, and six replicate spots per hybridization. GM, grand mean.
Table 1 Identity of tissue-specific genes with expression patterns consistent in all three populations, and those inconsistent in all three populations
Gene (see Figure 7) Grid Short name Swiss-Prot name
Consistent - oxidative phosphorylation
a E8 Aldo keto reductase 1 A1 Aldo-keto reductase family 1 member A1 (aldehyde reductase)
b E7 Aldo keto reductase 1 D1 Aldo-keto reductase family 1 member D1; steroid-5-beta-reductase beta polypeptide 1 (3-oxo-5 beta-steroid delta 4-dehydrogenase beta 1); steroid 5-beta-reductase
c F1 G3PDH Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
d D10 Glucose 6 phosphatase Glucose-6-phosphatase (G6PASE)
e H3 Pyruvate kinase muscle Pyruvate kinase (muscle isozyme)
f G10 Pyruvate kinase R Pyruvate kinase isoform R (erythroid)
g I6 NADH dehydrb 6 (17 kD) NADH dehydrogenase (ubiquinone) 1 beta subcomplex 6 (17 kD B17)
h G4 NADH Ubiq Oxi ASHI NADH-ubiquinone oxidoreductase ASHI subunit precursor (complex I-ASHI) (CI-ASHI)
i M4 NADH Ubiq Oxi MNLL NADH-ubiquinone oxidoreductase MNLL subunit (complex I-MNLL) (CI-MNLL)
j L1 ATP syn H+ FO c 9 2 ATP synthase H+ transporting mitochondrial F0 complex subunit c (subunit 9) isoform 2
k P3 ATP syn H+ FO F6 ATP synthase H+ transporting mitochondrial F0 complex subunit F6; coupling factor 6
l I9 Cyto C oxi III Cytochrome c oxidase subunit III
m J10 Cyto C oxi VA Cytochrome C oxidase polypeptide VA
n N8 Cyto C oxi VIa Cytochrome c oxidase subunit VIa precursor polypeptide 2
o J5 Cyto C oxi VIIC Cytochrome C oxidase polypeptide VIIC precursor (VIIIA)
p K12 Cyto C oxi VIIIb Cytochrome c oxidase subunit VIIIb
Consistent - other metabolism
q H12 Isocitrate dehyd 2 Isocitrate dehydrogenase 2 (mitochondrial IDH2)
r A9 PEP carboxykinase PEP carboxykinase phosphoenolpyruvate carboxykinase
s D8 Fatty acid binding liver basic Liver-basic fatty acid binding protein (LB-FABP)
t B12 Delta 6 fatty acid desaturase Delta-6 fatty acid desaturase
u H1 Triglyceride lipase triacylglycerol Triglyceride lipase triacylglycerol
v I5 Glycerol kinase Glycerol kinase
w M10 Lipoprotein lipase Lipoprotein lipase
x P9 Phospholipase XIII A2 Group XIII secreted phospholipase A2
y F4 Cystathionine beta synthase Cystathionine-beta-synthase
z K11 Cold inducible RNA binding Cold inducible RNA-binding protein; (CIRBP) glycine-rich RNA binding protein;
aa F2 Hepatocyte nuclear F 4 A Hepatocyte nuclear factor 4-alpha (HNF-4-alpha) (transcription factor HNF-4)
bb M1 p450 2P1 (CYP2P1) Cytochrome P450 2P1 (CYP2P1)
cc D6 Glutathione peroxidase 4 Glutathione peroxidase 4 (phospholipid hydroperoxidase)
dd O11 Methylmalonate semialdehyde dehyd Methylmalonate-semialdehyde dehydrogenase (acylating)
ee N7 Phosphatidylcholine sterol acyltrans Phosphatidylcholine-sterol acyltransferase
ff B1 Prostaglandin D syn Prostaglandin D synthase
Inconsistent - oxidative phosphorylation
gg A6 ADH class II mito Aldehyde dehydrogenase, mitochondrial precursor (ALDH class 2)
hh E12 Aldolase 1 A Aldolase 1 A. muscle
ii A2 Enolase beta muscle enolase (beta muscle specific)
jj F7 LDHB lactate dehydrogenase B (LDHB)
kk O6 PFK 6-phosphofructokinase
ll K4 NADH dehyd MLRQ NADH dehydrogenase (ubiquinone) MLRQ subunit (complex I-MLRQ)
mm L9 NADH dehyd I NADH dehydrogenase subunit 1
nn C6 NADH dehydr a 1 (7.5 kD MWFE) NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 1 (7.5 kD MWFE)
oo E6 NADH dehydr a 9 (39 kD) NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 9 (39 kD)
pp M6 ATP syn B ATP synthase subunit B
Inconsistent - other metabolism
qq C7 Transketolase Transketolase
rr H8 Fatty acid binding 7 brain Fatty acid binding protein 7 brain (B-FABP)
ss A3 Fatty acid binding H6 Fatty acid binding protein H6-isoform
tt O10 Fatty acid binding heart Heart-type fatty acid-binding protein (H-FABP)
uu D9 Fatty acid syn Fatty acid synthase
vv F9 Glutamate decarboxylase Glutamate decarboxylase
Letters in the first column refer to genes illustrated in Figure 7; the grid column identifies genes as reported in our data entry to the NCBI Gene Expression Omnibus (GLP1224). Gene identities are listed as those identified by Swiss-Prot and as shortened names, and grouped as genes involved in oxidative phosphorylation or in other biochemical pathways.
Table 2 Sources of variance and calculation of variables for the split-plot ANOVA statistical design [18]
Source of variance df Sum of squares Fs
Among populations P - 1 2 SS_P I × T × Σ(popmeans - GM)2 MS_P/MS_I(P)*
Among tissues T - 1 2 SS_T I × P × Σ(tissuemeans - GM)2 MS_T/MS_TI(P)†
Interaction (P - 1) × (T - 1) 4 SS_PT I × Σ(tissue(population)means - popmeans - tissuemeans + GM)2 MS_PT/MS_TI(P)‡
Among inviduals in population P(I - 1) 6 SS_I(P) T × Σ(ind(population)means - popmeans)2
Tissue-by-individual in population P × (T - 1) × (I-1) 12 SS_TI(P) Σ(samplemean - tissue(population)mean - ind(population)mean + popmean)2 MS_TI(P)/MS_dye§
Dye within individuals P × T × I × (D - 1) 27 SS D(I)
Replicate hybridization in dye P × T × I × D × (H - 1) 54
Spot in replicate hybridization P × T × I × D × H × (S - 1) 540
Total P × T × I × D × H × S - 1 647
==== Refs
Cavalieri D Townsend JP Hartl DL Manifold anomalies in gene expression in a vineyard isolate of Saccharomyces cerevisiae revealed by DNA microarray analysis. Proc Natl Acad Sci USA 2000 97 12369 12374 11035792 10.1073/pnas.210395297
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
Oleksiak MF Churchill GA Crawford DL Variation in gene expression within and among natural populations. Nat Genet 2002 32 261 266 12219088 10.1038/ng983
Jin W Riley RM Wolfinger RD White KP Passador-Gurgel G Gibson G The contributions of sex, genotype and age to transcriptional variance in Drosophila melanogaster. Nat Genet 2001 29 389 395 11726925 10.1038/ng766
Enard W Khaitovich P Klose J Zoellner S Heissig F Giavalisco P Nieselt-Struwe K Muchmore E Varki A Ravid R Intra- and interspecific variation in primate gene expression patterns. Science 2002 296 340 343 11951044 10.1126/science.1068996
Hsieh WP Chu TM Wolfinger RD Gibson G Mixed-model reanalysis of primate data suggests tissue and species biases in oligonucleotide-based gene expression profiles. Genetics 2003 165 747 757 14573485
Cheung VG Conlin LK Weber TM Arcaro M Jen KY Morley M Spielman RS Natural variation in human gene expression assessed in lymphoblastoid cells. Nat Genet 2003 33 422 425 12567189 10.1038/ng1094
Brem RB Yvert G Clinton R Kruglyak L Genetic dissection of transcriptional regulation in budding yeast. Science 2002 296 752 755 11923494 10.1126/science.1069516
Zhang L Zhou W Velculescu VE Kern SE Hruban RH Hamilton SR Vogelstein B Kinzler KW Gene expression profiles in normal and cancer cells. Science 1997 276 1268 1272 9157888 10.1126/science.276.5316.1268
Elek J Park KH Narayanan R Microarray-based expression profiling in prostate tumors. In Vivo 2000 14 173 182 10757075
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JG Sabet H Tran T Yu X Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000 403 503 511 10676951 10.1038/35000501
Paez JG Janne PA Lee JC Tracy S Greulich H Gabriel S Herman P Kaye FJ Lindeman N Boggon TJ EGFR mutations in lung cancer: correlation with clinical response to Gefitinib therapy. Science 2004 304 1497 500 15118125 10.1126/science.1099314
Archacki SR Angheloiu G Tian X-L Tan FL DiPaola N Shen G-Q Moravec C Ellis S Topol EJ Wang Q Identification of new genes differentially expressed in coronary artery disease by expression profiling. Physiol Genomics 2003 15 65 74 12902549
Iemitsu M Miyauchi T Maeda S Sakai S Fujii N Miyazaki H Kakinuma Y Matsuda M Yamaguchi I Cardiac hypertrophy by hypertension and exercise training exhibits different gene expression of enzymes in energy metabolism. Hypertens Res 2003 26 829 837 14621187 10.1291/hypres.26.829
Kunz WS Different metabolic properties of mitochondrial oxidative phosphorylation in different cell types - important implications for mitochondrial cytopathies. Exp Physiol 2003 88 149 154 12525863 10.1113/eph8802512
Ramaswamy S Tamayo P Rifkin R Mukherjee S Yeang CH Angelo M Ladd C Reich M Latulippe E Mesirov JP Multiclass cancer diagnosis using tumor gene expression signatures. Proc Natl Acad Sci USA 2001 98 15149 15154 11742071 10.1073/pnas.211566398
Baris O Savagner F Nasser V Loriod B Granjeaud S Guyetant S Franc B Rodien P Rohmer V Bertucci F Transcriptional profiling reveals coordinated up-regulation of oxidative metabolism genes in thyroid oncocytic tumors. J Clin Endocrinol Metab 2004 89 994 1005 14764826 10.1210/jc.2003-031238
Steel RGD Torrie JH Principles and Procedures of Statistics 1980 2 New York, NY: McGraw-Hill
Oleksiak MF Roach JL Crawford DL Natural variation in cardiac metabolism and gene expression in Fundulus heteroclitus. Nat Genet 2005 37 67 72 15568023
Kendziorski CM Zhang Y Lan H Attie AD The efficiency of pooling mRNA in microarray experiments. Biostatistics 2003 4 465 477 12925512 10.1093/biostatistics/4.3.465
Pritchard CC Hsu L Delrow J Nelson PS Project normal: defining normal variance in mouse gene expression. Proc Natl Acad Sci USA 2001 98 13266 13271 11698685 10.1073/pnas.221465998
Weiss L (Ed) Cell and Tissue Biology: A Textbook of Histology 1983 6 Baltimore, MD: Urban and Schwarzenberg
Guyton AC Textbook of Medical Physiology 1991 8 Philadelphia: W.B. Saunders Company
Lee PD Sladek R Greenwood CMT Hudson TJ Control genes and variability: absence of ubiquitous reference transcripts in diverse mammalian expression studies. Genome Res 2002 12 292 297 11827948 10.1101/gr.217802
Nachman MW Hoekstra HE D'Agostino SL The genetic basis of adaptive melanism in pocket mice. Proc Natl Acad Sci USA 2003 100 5268 5273 12704245 10.1073/pnas.0431157100
Oleksiak MF Kolell KJ Crawford DL Utility of natural populations for microarray analyses: isolation of genes necessary for functional genomic studies. Mar Biotechnol (NY) 2001 3 (Supplement 1) S203 S211 14961317 10.1007/s10126-001-0043-0
FunnyBase gene expression database
Van Gelder RN Von Zastrow ME Yool A Dement WC Barchas JD Eberwine JH Amplified RNA synthesized from limited quantities of heterogeneous complementary DNA. Proc Natl Acad Sci USA 1990 87 1663 1667 1689846
Raghavachari N Bao YP Li G Xie X Muller UR Reduction of autofluorescence on DNA microarrays and slide surfaces by treatment with sodium borohydride. Anal Biochem 2003 312 101 105 12531193 10.1016/S0003-2697(02)00440-2
Quackenbush J Microarray data normalization and transformation. Nat Genet 2002 32 Suppl 496 501 12454644 10.1038/ng1032
Wu H Kerr K Cui X Churchill GA 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
Chu TM Weir B Wolfinger R A systematic statistical linear modeling approach to oligonucleotide array experiments. Math Biosci 2002 176 35 51 11867082 10.1016/S0025-5564(01)00107-9
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
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 10.1093/nar/30.4.e15
Brazma A Hingamp P Quackenbush J Sherlock G Spellman P Stoeckert C Aach J Ansorge W Ball CA Causton HC Minimum information about a microarray experiment (MIAME): toward standards for microarray data. Nat Genet 2001 29 365 371 11726920 10.1038/ng1201-365
Sokal RR Rohlf FJ Biometry 2001 3 New York: W.H. Freeman
Rohlf FJ Sokal RR Statistical Tables 2002 3 New York: W.H. Freeman
Kumar S Tamura K Jakobsen IB Nei M MEGA: Molecular Evolutionary Genetics Analysis, version 2.1
| 15693943 | PMC551534 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 14; 6(2):R14 | latin-1 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r14 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r151569394410.1186/gb-2005-6-2-r15ResearchPolarized monocyte response to cytokine stimulation Nagorsen Dirk [email protected] Sara [email protected] Kina [email protected] Ena [email protected] Vladia [email protected] Paola [email protected] Francesco M [email protected] Monica C [email protected] Immunogenetics Section, Department of Transfusion Medicine, Clinical Center, National Institutes of Health, Bethesda, MD 20892-1502, USA2 Department of Oncology and Surgical Sciences, Oncology Section, University of Padova, Padova 35100, Italy3 Current address: Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Medizinische Klinik III, Hindenburgdamm 30, 12200 Berlin, Germany2005 21 1 2005 6 2 R15 R15 5 10 2004 1 12 2004 22 12 2004 Copyright © 2005 Nagorsen 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.
A comprehensive study of the transcriptional response of mononuclear phagocytes to cytokines reveals distinct classes of cytokines that elicit either the classical or alternative pathway of monocyte activation.
Background
Mononuclear phagocytes (MPs) stand at the crossroads between the induction of acute inflammation to recruit and activate immune effector cells and the downmodulation of the inflammatory process to contain collateral damage. This decision is extensively modulated by the cytokine microenvironment, which includes a broad array of cytokines whose direct effect on MPs remains largely unexplored. Therefore, we tested whether polarized responses of MPs to pathogens are related to the influence of selected cytokines or represent a mandatory molecular switch through which most cytokines operate.
Results
Circulating CD14+ MPs were exposed to bacterial lipopolysaccharide (LPS) followed by exposure to an array of cytokines, chemokines and soluble factors involved in the immune response. Gene expression was studied by global transcript analysis. Two main classes of cytokines were identified that induced a classical or an alternative pathway of MP activation. Expression of genes affected by NFκB activation was most predictive of the two main classes, suggesting that this pathway is a fundamental target of cytokine regulation. As LPS itself induces a classical type of activation, the most dramatic modulation was observed toward the alternative pathway, suggesting that a broad array of cytokines may counteract the pro-inflammatory effects of bacterial components.
Conclusions
This analysis is directly informative of the primary effect of individual cytokines on the early stages of LPS stimulation and, therefore, may be most informative of the way MP maturation may be polarized at the early stages of the immune response.
==== Body
Background
Resident and recruited mononuclear phagocytes (MPs) display a versatile phenotype that reflects the plasticity of these cells in response to microenvironmental signals. This heterogeneity spans a continuous spectrum that can be polarized into two extremes recently described by Mantovani et al. [1]. Pathogen stimulation exemplified, for instance, by lipopolysaccharide (LPS) stimulation in the presence of interferon (IFN)-γ induces M1 MPs through engagement of Toll-like receptors (TLRs). M1 MPs are true antigen-presenting cells capable not only of killing invading organisms but of concomitantly recruiting and activating immune effector cells [2,3]. Treatment of MPs with type II cytokines such as interleukin (IL)-4, IL-13 and, partly, IL-10 [4], polarizes their function towards tissue repair, angiogenesis and containment of collateral damage through reduction of inflammation (the M2 macrophage phenotype). This alternative mode of macrophage activation accounts for a distinct phenotype with a key role in humoral immunity and tissue repair [5].
It has been suggested that the extreme dichotomy between a classical M1 and an alternative M2 polarization of macrophage function may not take into account intermediate regulation by cytokines such as IL-10, transforming growth factors (TGF-α and TGF-β), macrophage colony-stimulating factor (M-CSF), IFN-α and IFN-β and tumor necrosis factor (TNF) [5]. Most important, a rigid dichotomy in MP function may not directly apply to physiological and/or pathological conditions in which these cells are exposed to an array of cytokines produced by innate or adaptive immune mechanisms during infection, tissue damage or in other conditions. Indeed, a comprehensive overview of the modulatory properties of cytokines on the MP reaction to a pathogen is missing. In addition, little is known about the transcriptional changes occurring in MPs on exposure to pathogen components such as LPS.
A recent study analyzed the transcriptional profile induced by the exposure of circulating MP conditioned in vitro for 7 days with IL-4 and GM-CSF (immature dendritic cell, (DC)) to pathogen components [6]. Bacterial, viral and fungal components elicited distinctive pathways that were, however, largely overlapping. The predominant response of these DCs to most pathogen components encompassed a rapid upregulation of genes associated with the innate arm of the immune response followed by induction of adaptive immune response genes.
The response of circulating MP-derived DCs is short-lived as these cells can exhaust their production of effector molecules (cytokines and chemokines) within a few hours of LPS stimulation [7]. The transience of mRNA and protein expression can cause DCs to redirect the immune response in different ways at different time points. For instance, soon after stimulation, DCs elicit T-cell responses of the Th-1 type, whereas at later stage of activation they prime T-cell responses of the Th-2 type, suggesting that their function is strongly dependent on the timing and duration of exposure to individual and/or combined stimulatory conditions in the surrounding microenvironment. At the transcriptional level, the dual function of DCs shifted from an early pre-inflammatory phase occurring within 3 hours to a later regulatory phase occurring approximately 8 hours following LPS exposure. During these evolving stages of DC activation, cytokines play a dominant role in shaping the function of DCs and other immune cells, providing a malleable link between the innate and adaptive immune responses [8]. In natural conditions, circulating or resident MPs may encounter a pathogen before the surrounding microenvironment has a chance to influence their maturation. Therefore, it is unknown whether non-conditioned circulating CD14+ MPs would react similarly to DCs on engagement with infectious agents. Thus, a preliminary aim of this study was to evaluate the kinetics of the response of non-conditioned circulating CD14+ MP to LPS. The results suggested that these cells respond to LPS similarly to immature CD14- DCs, with a surge in transcriptional activity that peaks around 3 hours after stimulation and in which the activation of genes associated with a classical activation of innate immune mechanisms predominates [6].
The stringent dichotomy describing a classical activation of MPs into mature antigen-presenting cells caused by IFN-γ and an alternative induction into macrophages induced by IL-4 and IL-13 may not apply to physiological conditions in which the microenvironment responds to pathogen exposure with a broad array of cytokine secretion. We therefore investigated whether polarized responses of MPs to pathogens are extreme behaviors that can be observed in vitro by studying a few illustrative cytokines or whether they represent a mandatory molecular switch through which most cytokines operate. Thus, we stimulated non-conditioned CD14+ MPs with LPS for 1 hour. The MPs were then exposed to an array of different cytokines that may be expressed in distinct pathologic conditions by different immune-cell subsets. The 1-hour interval was empirically selected to induce a biphasic model in which the presumed modulation by cytokines occurred during an ongoing reaction to LPS. This allowed mapping of cytokines into conditional subclasses based on their effects on the global transcriptional changes responsible for MP activation and differentiation.
Two main classes of cytokines were identified that induced a classical or alternative pathway of MP activation, respectively. An intermediate class (including IL-10) was also identified, while TNF-α, TNF-β and GM-CSF displayed a quite distinct behavior from the other cytokines. Expression of genes affected by NFκB activation was most predictive of the two main classes, suggesting that in most cases the NFκB pathway is a central target of cytokine regulation that modulates the cascade of events following LPS stimulation. Overall, it seems that MP maturation/differentiation goes through a molecular switch that is partly independent of the fine differences in the stimulatory properties of the various cytokines and, with few exceptions, is pre-programmed towards a classical or an alternative route. As LPS itself induces a classical type of activation, the most dramatic modulation in this model was observed toward the alternative pathway, suggesting that a broad array of cytokines may counteract the pro-inflammatory effects of bacterial components.
Results
Effect of LPS stimulation on circulating CD14+ MPs
Enriched MP preparations (about 90% CD14+) were exposed to LPS. Total RNA extracts were obtained 4 and 9 hours after. These time points were selected to catch salient stages of the biphasic response of MP to LPS described by others [6,7]. Amplified antisense RNA (aRNA) [9] was hybridized to a custom-made 17,000 (17K)-clone cDNA microarray chip enriched with genes relevant to immune function. The transcriptional profile of LPS-induced MPs was similar to that described by others in DCs [6]. In particular, genes associated with the innate response of CD14-, immature DCs to pathogen components [6,10] were similarly upregulated in CD14+ MPs (data not shown). This finding suggests that the differential expression of the LPS co-receptor CD14 between the two cell populations has a relatively minor impact on the transcriptional regulation of the innate immune response [11].
Kinetics of the response of CD14+ MPs to LPS and its modulation by cytokines
Aliquots of CD14+ MPs were stimulated in parallel with LPS and exposed 1 hour later to individual cytokines selected from a library of recombinant proteins possibly relevant to MP regulation. MPs were kept in culture for 4 and 9 hours, at which times aRNA was prepared for transcriptional analysis. Unsupervised Eisen's clustering [12] was applied to the complete dataset (Figure 1). The kinetics of the response to LPS had the greatest influence on the global transcriptional profile of MP induction; samples preferentially clustered according to time of stimulation rather than type of treatment. This was underlined by the observation that MPs stimulated with LPS alone clustered with the cytokine-stimulated MPs according to the time elapsed after stimulation. In addition, a cluster containing most of the samples obtained 9 hours after stimulation (9') included three control samples consisting of CD14+ MPs not exposed to LPS or cytokines (no stimulation). These three samples were prepared at times 0, 4 and 9 hours to parallel the culture conditions used for stimulation. This finding suggested that the transcriptional profile of CD14+ MPs 9 hours after LPS stimulation and 8 hours after treatment with most cytokines converges toward a less reactive metabolic state closer to that of unstimulated MPs. Several cytokines, however, maintained a more active metabolic profile and after 9 hours retained a transcriptional footprint relatively close to that of samples treated for 4 hours (9"). This group included the genes for most IFN-α isoforms, IFN-β, vascular endothelial growth factor (VEGF), FLT-3 ligand, TGF-α, the chemokine RANTES (CCL5), IL-2, IL-4, IL-15 and the chemokines MIP1α (CCL3) and MIP1β (CCL4), suggesting that these cytokines may have relatively prolonged kinetics of MP activation.
The average number of genes whose expression was increased compared to unstimulated MPs was higher (420 genes) after 4 hours than after 9 hours (265 genes). About half of the genes upregulated after 4 hours remained upregulated at 9 hours (223 genes). This is in accordance with Huang's observation [6] that transcriptional changes in DCs occur predominantly in the early phase of the response to pathogen components, with about half the genes displaying transitory expression and the other half sustained expression. For this reason, subsequent analyses were limited to the 4-hour time point.
Transcriptional modulation by cytokines 4 hours after LPS stimulation of CD14+ MPs
Although LPS alone strongly affected the transcriptional program of MPs, it was possible to discern the contribution of individual cytokines. This analysis was limited to samples treated for 4 hours, when the most dramatic effects on gene modulation were noted. Genes differentially expressed in samples treated with cytokines compared to those treated with LPS alone were identified. Stringent criteria were applied to select genes expressed in at least 80% of samples with a threefold or greater increase or reduction in expression over LPS in at least one of the cytokine-treated samples. This gave 2,057 genes that were deemed most relevant to the analysis. Using these genes, all cytokine-treated samples were subjected to unsupervised clustering to evaluate their relatedness (Figure 2a).
Two main classes of cytokines were identified. One class (Figure 2a, blue horizontal line) included IL-4, IL-13 TGF-α, TGF-β and VEGF, which are unquestionably associated with the alternative pathway of MP activation [1,5]. The second class (Figure 2a, red line) included IFN-α2, IFN-β, IFN-γ, CD40 ligand (CD40L), and FLT-3 ligand, which are generally associated with the classical pathway of MP activation [2,3]. Therefore, we considered the first cluster representative of the alternative and the second of the classical pathway of MP activation in response to LPS. As predicted by Gordon [5], few cytokines (Figure 2a, green line) did not properly belong to either class. These included IL-10, IL-1β, IL-15 and two IFN-α isoforms. Because several of these cytokines have previously been associated with the alternative pathway, we referred to this class as alternative II. Not surprisingly [5], MPs stimulated with TNF-α (Figure 2a, purple line) and, most dramatically, TNF-β and GM-CSF (Figure 2a, black line) had a totally independent effect on the transcriptional regulation of LPS-induced CD14+ MPs [5].
Cytokines classified according to the previous groups were tested for class prediction by applying unsupervised principal component analysis (PCA) to the global, unfiltered 17K gene dataset (Figure 2b). This analysis independently classified cytokines in two groups corresponding to the alternative (Figure 2b, blue circles) and classical (Figure 2b, red circles) cytokine classes. Most of the cytokines belonging to the alternative II class (Figure 2b, green circles) grouped with the alternative group, whereas TNF-α (Figure 2b, purple circle and arrow), TNF-β and GM-CSF (Figure 2b, gray circles) remained separate. The sample treated with LPS alone (Figure 2b, yellow circle and arrow) grouped with the classical cytokines, confirming the predominant pro-inflammatory effects of this bacterial product and its alignment with the classical pathway of MP activation [1,11]. This finding based on the complete dataset indicates the intrinsic bias of this study aimed at exploring the alternative modulation of the MP response to LPS.
Particular mention should be made of the erratic behavior of various IFN-α subtypes, which clustered indiscriminately between the two main cytokine classes. Interestingly, however, alignment of the IFN-α protein sequences through the EMBL-EBI Clustal W database identified, with the exception of IFN-αG, a close relationship among the IFN-α subtypes that clustered with the alternative type of cytokines (data not shown). This subclassification was also supported by the phylogenetic relationship among interferons described by Henco [13]. This information suggests that specific domains of the IFN-α molecules may have dramatically different effects in the modulation of the MP response to pathogen [14,15]. Interestingly, IFN-α2, which is the one most commonly used in clinical trials as a pro-inflammatory cytokine, clustered with the classical cytokines adjacent to IFN-γ.
Cytokine-mediated modulation of LPS-stimulated CD14+ MPs predominantly affects pathways downstream of NFκB
Signatures associated with several pathways of immune-cell activation were constructed by selecting genes from the global pool of 17K clones according to literature information without pre-existing information about the association of their expression to either class of cytokines. Signature genes were then subjected to supervised clustering according to the cytokine classification shown in Figure 2a. This independent process identified virtual signatures, in some cases portraying opposite transcriptional regulation by the two classes. The signature that most strongly discriminated the two classes comprised 121 genes whose expression is closely dependent on NFκB modulation [11] (Figure 3a). This is not surprising as LPS acts through engagement of Toll receptor 4 (TLR4) and CD14, with resulting activation of NFκB [3]. It would, therefore, seem intuitive that the strongest modulation in the present experimental conditions would target this pathway. In particular, several TNF- and IL-1-related genes classically modulated by NFκB during the acute phases of the innate immune response [11] were strongly and inversely modulated by the two cytokine classes.
The same 121 genes were used for unsupervised class prediction by reclustering cytokine-treated samples (Figure 3b). This independent analysis segregated cytokines into two classes that with the exception of one (IFN-α2b) matched the respective original classical and alternative classification (Figure 3a, red and blue horizontal bars, respectively). Interestingly, the cytokines that belonged to the alternative II class clustered with the alternative cytokines (Figure 3a, green horizontal bars) while TNF-α (Figure 3a, purple horizontal bar), TNF-β, and GM-CSF (Figure 3a, dark gray horizontal bar) clustered separately but in proximity of the classical group. Analysis of early signaling events occurring 1 hour and 30 minutes after LPS stimulation and, therefore, 30 minutes after the additional cytokine exposure, demonstrated significantly increased levels of the free p50 subunit of NFκB in MP whole-cell extracts treated with alternative class cytokines (IL-4 and IL-13). In addition, IL-1α and TNF-α significantly upregulated p50, whereas no significant changes were caused by classical cytokines (IFN-γ, IL-6 and IL-3). Extracts from MPs stimulated only with LPS also failed to demonstrate changes in NFκB subunit release; this is probably related to the 90-minute period from stimulation that allowed a return of signaling molecules to baseline conditions (Figure 3c). The similarity of the IL-1α and TNF-α effects on p50 to those of alternative cytokines contrasts with the dramatic differences observed on the respective transcriptional profiles, suggesting that other pathways induced by these cytokines may prevail in the conditions tested here.
Among the additional pathways tested, those mediated through STATs, Janus kinases (JAKs), and interferon regulatory factor (IRF) did not appear consequential to the experimental conditions tested in this study (data not shown).
Cytokine-mediated modulation of metalloproteinase expression in LPS-stimulated CD14+ MPs
Matrix metalloproteinases (MMP) are tightly connected to MP activation. MMP released by MPs contribute to normal and pathological tissue remodeling and MP migration. In addition MMP function as regulatory proteins by promoting the activation or degradation of cytokines. Finally, MMP are susceptible to cytokine stimulation. Thirty-six MMP and MMP-related genes (filtered from a larger group of 184 MMPs, disintegrins, α-defensins, TGF-β, TNF-α, insulin-like growth factor 1 (IGF-1), epidermal growth factor (EGF), fibroblast growth factor (FGF), IL-1 and monocyte chemotactic protein-3 (CCL7/MCP-3) genes) [16,17] were clustered according to the alternative and classical group denomination and the 11 most representative are shown in Figure 4. The alternative group of cytokines induced the transcription of MMPs (MMP 7, 9, 10, 19), enzymes related to MMP function (disintegrin ADAM 9, pro-collagen proline dioxygenase, MEK1 kinase, serine protease inhibitor, cathepsin L) and structural proteins (gap junction connexin 26, laminin A/C). This confirms the role of alternative MP activation in promoting tissue remodeling, cell-cell interactions and local control of the inflammatory process through activation (via MMP-7, 9) [18-21] or degradation (via MMP-19) of cytokine activity [17,22]. In addition, these observations suggest a role for the cytokines in the alternative group (other than the well documented IL-4 or IL-13) in polarizing MPs toward an M2 regulatory phenotype.
Characterization of the alternative and classical groups of cytokines
Comparison of gene-expression patterns induced by the two cytokine classes (classical and alternative) identified 2,007 genes that were differentially expressed at a less than 0.001 significance level (t-test, p2-value). Genes associated with immune function were proportionally over-represented. A selection of genes relevant to MP function is shown in Figure 5. In most cases, the pattern of expression echoed the class allocation suggested by the literature [23]. Classical cytokines induced genes responsible for the cytotoxic and migratory properties of MPs such as those for CD95, TRAIL, granzymes, perforin, CD16 (stimulatory Fcγ receptor) and CD62L. In addition, antigen presentation was enhanced as suggested by the coordinate expression of several HLA class II genes. IFN-γ and the other classical cytokines inhibited the expression of the macrophage-derived chemokine CCL22/MDC, while upregulating the expression of CXCR3, as previously reported [23].
Alternative cytokines induced the expression of several cytokines and their respective receptors involved in the chemotaxis and activation of neutrophils, MPs, natural killer (NK) cells, DCs, helper T lymphocytes (Th2) and B lymphocytes. Several genes known to be associated with alternative MP activation [5,23] were consistently upregulated. These included the mannose receptor, the inhibitory Fc-IIb receptor CD32, and the cell-surface molecule CD44 [24], which is associated with the disposal of inflammatory cell corpses without expansion of the inflammatory process. Several inducible chemokines were expressed in response to alternative stimulation such as CCL22/MDC, supporting the emerging role of this cytokine as an enhancer of polarized Th2 responses [25]. Other chemokines known to be induced by master type II cytokines and associated with the induction of Th2 responses were also induced by alternative activation; these included CCL11 (eotaxin), CCL1 (I-309), CCL2 (MCP-1) and CCL7 (MCP-3) [23]. Of interest was the relatively higher expression of IL-24, a cytokine belonging to the IL-10 family constitutively expressed by MPs [26]. Although the true role of this cytokine in inflammatory processes is not known, it is likely that its pro-apoptotic and angioregulatory properties have important roles in tissue repair and remodeling during inflammation. Finally, various chemokine receptors responsible for MP trafficking and localization were differentially regulated by the two cytokine groups, including in particular CCR1 and CCR5, which were induced by alternative stimulation, and CCR2, induced by classical stimulation.
This early transcriptional profile underlines the primary role of MPs during the acute phases of the response to pathogen as effector cells that can kill pathogens, take up antigen and migrate to local regional lymph nodes to recruit adaptive immune responses (classical activation). In contrast, alternative cytokines may be produced in the microenvironment to maintain a resident MP phenotype rich in chemokine production, which can attract Th2-type immune responses while continuing pathogen clearance through retention of phagocyte properties and promoting tissue remodeling.
Surprisingly, genes of the IL-1 family and its receptors, TNF and IL-6 were consistently upregulated by the alternative class of cytokines, suggesting that LPS alters MP polarization with regard to these cytokines [5,10]. Particularly interesting is the alternative induction of IL-6 and its receptor, which may play a central role in mediating the transition from neutrophils to MP recruitment during progression from acute to chronic inflammation [27].
Validation of microarray analysis by TaqMan real-time PCR
To define the validity and accuracy of our global microarray analysis, quantitative TaqMan real-time PCR was performed on amplified RNA material isolated after stimulation of monocytes obtained from five additional normal donors with representative cytokines/soluble factors selected from the alternative and from the classical groups. Comparison of monocyte stimulation with a candidate cytokine from the alternative/M2 (IL-4) and the classical/M1 (IFN-γ) group, respectively, is shown in Figure 6. Ten genes whose expression was upregulated by alternative cytokines were tested, as the thrust of the analysis was the evaluation of the effect of alternative cytokines on LPS-stimulated MPs. The relative expression of five genes out of 10 (IL-1α, IL-1 receptor, mannose receptor, NFKb-p105/50 and TRAF) was significantly higher after treatment with IL-4, as suggested by the array data. Also, the expression profiles of the other genes tested reproduced the pattern observed in the array experiments even if they did not reach statistical significance for each individual gene. This is not surprising because array experiments summarize in signature fine differences in gene expression, that often describe patterns rather than absolute significance for individual genes.
Reproducibility of the estimates of mannose receptor, NFKb-p105/50 and TRAF gene expression in MPs obtained from the same five donors was evaluated after stimulation with four cytokines/soluble factors selected from the alternative (IL-13, IL-1α, IL-4, TGF-β) and four from the classical groups (FLT-3 ligand, IFN-γ, IL-3, CD40L) (Figure 7). In all cases gene expression significantly reflected the pattern of gene expression detected by microarray analysis (Figures 3, 5).
Discussion
It has been suggested that MP activation and maturation progresses through a polarized mechanism whereby two extreme products result that promote inflammation on one side and tissue repair on the other [1,5]. The first mechanism has been called the classical pathway of MP activation. It induces M1 monocytes specialized for pathogen killing and activation of innate and adaptive immune effector cells. A second, alternative, pathway induces M2-type monocytes committed to clearing pathogen through internal metabolism while reducing inflammation. This process limits the collateral damage induced by an excessive immune response, and, upon cessation of the pathogenic stimulus, promotes tissue repair. This dichotomy is based on the study of a few cytokines deemed representative of the classical mode (LPS, IFN-γ) or the alternative mode (IL-4, IL-13) of MP activation. Other cytokines such as IL-10, TGF, M-CSF, IFN-α/β and TNF, although partially overlapping both pathways, display functional effects that diverge significantly enough that they would be inaccurately grouped in either class [5].
In this study we evaluated whether MP commitment is dependent on an early bipolar switch through which most cytokines operate. This was done by testing in parallel a library of 42 stimulatory molecules possibly present in the tissue microenvironment following a pathogenic insult. This modulation was tested on MPs triggered by a pathogenic stimulus, exemplified in this case by LPS. This was done on the assumption that in most circumstances resident or migratory MPs reaching an infected area are exposed concomitantly to a pathogen and to the cytokine milieu resulting from the infection. The transcriptional profile identified by this study cannot distinguish between the direct effect of each soluble factor analyzed and the downstream activation of transcriptional pathways by secondary paracrine or autocrine secretion of biological modifiers by MPs. However, the main goal of this study was to identify the overall effect on MPs of the exposure to individual cytokines. Future analyses, focused on specific cytokine patterns described here, should possibly include the addition of blocking antibodies to segregate secondary from primary MP responses.
The results suggest that MP activation is in most cases a bipolar process regulated by an internal switch through which cytokines modulate the yin and yang of the MP transcriptional program. In fact, most cytokines preferentially induced one or other pattern of transcriptional activation. We, therefore, mapped most of the cytokines within a classical or an alternative classification according to their effects on CD14+, LPS-induced MPs. In particular, it appeared that transcriptional programs down-stream of NFκB activation [11] were mostly associated with either class, suggesting that the NFκB system is at the center of the switch regulating MP activation/differentiation in the conditions tested in the present study. This is not surprising as LPS signaling is mediated through the TLR-4 and CD14, which in turn directly regulate the IκB kinase (IKK)-NFκB pathway [3,11,28,29]. It appears that cytokine regulation modulates the release of the p50 subunit of NFκB, whch is in turn responsible for the downstream effects on the transcriptional program. Interestingly, reclustering of cytokines based on NFκB-dependent genes consolidated the two classes of cytokines adding TNF-α and TNF-β, IFN-α2b and GM-CSF to the classical group and the alternative type II cytokines with the alternative group, suggesting that NFκB may be central to MP polarization.
This information cannot of course be generalized to other conditions as the model tested is strongly biased by NFκB induction by LPS. This is underlined by the unexpected alternative upregulation of genes associated with IL-1, TNF and IL-6 [11]. This observation suggests that in different conditions, cytokines may differently modulate MP activation, possibly through various modulatory feedback mechanisms [5]. In addition, genes associated with the interrelated arginine and tryptophan pathways that modulate nitric oxide induction and are indirectly associated with NFκB function were, at least in part, differentially regulated by the two classes of cytokines [30-32]. It has been suggested that inducible nitric oxide is produced rapidly after LPS stimulation of MPs and inhibits NFκB through the stabilization of IκB [16]. It is possible that cytokines may counteract this effect through modulation of this metabolic junction.
Cytokine and chemokine effects on MP function are tightly intertwined with the enzymatic activities of MMP. Membrane-bound cytokine receptors and adhesion molecules can be released from the cell surface by MMPs acting as 'sheddases' or 'convertases'. This, in turn, can downregulate cell-surface signaling by removal of receptors, or induce paracrine activity by release of soluble proteins. Not surprisingly, IL-13 overexpression results in production of several MMPs [33]. For instance, MMP-9 activates latent TNF-α on the surface of MPs or soluble VEGF [19,34]. Furthermore, Yu and Stamenkovic [21] observed that gelatinase B/MMP9 bound to CD44 activates latent TGF-β stored in the pericellular matrix. MMP-7 can enhance tissue repair by facilitating migration of epithelial cells [16]. In agreement with these reports we observed that the transcription of MMP-7, MMP-9 and CD44 are coordinately induced (Figures 4, 6) by the alternative activation of MPs. Conversely, the downregulation of MMP-7 and MMP-9 by the classical cytokines confirms their inhibitory effects on MMP expression. Inhibition of MMP-9 production by IFN-β and IFN-γ has been recently reported by Sanceau et al. [35], who noted that interferons regulate MMP expression through IRF/NFκB interaction. Binding of NFκB p50/p65 to the MMP-9 promoter is competitively inhibited by IFN-β and IFN-γ-induced IRF-1. Possibly, NFκB regulation of MMP promoters through release of p50 (Figure 3) was responsible for the transcriptional activation of MMP expression by alternative cytokines observed in this study (Figure 4). These observations confirm the tight specificity of the relationship between cytokine and MMP regulation, which is finely toned at several check points and strongly polarized, in these experimental conditions, toward an M2 phenotype.
Enhanced transcription of gap junction/connexin 26 by the alternative cytokines is also of particular interest in view of the hypothesized junctional communications among MPs or between MPs and endothelial cells [36,37]. The finding that MPs activated by alternative cytokines induce the transcription of genes for gap junction components is of physiological importance and may be the missing link in the identification of factors that regulate the expression of gap junction connexins in MPs. In addition, it opens up the possibility that in a milieu dominated by alternative cytokines, where the ultimate goal is to return to homeostasis, the induction of gap junctions increases the ability to transmit or receive regulatory signals [38] that could facilitate the return to normal housekeeping functions.
Conclusions
The early-phase transcriptional profile presented in this study may not comprehensively parallel the plethora of biological effects that a given cytokine can induce under in vitro or, most importantly, in vivo conditions. Secondary, autocrine and paracrine modulation through the cytokine network following a primary stimulation may introduce novel on and off switches that could override the original signal. Nevertheless, this analysis is directly informative on the primary effect of individual cytokines on the early stages of LPS stimulation and, therefore, may be most informative on the way MP maturation may be polarized at the early stages of the immune response. The clustering of most cytokines into two main groups suggests that their control of central switches (NFκB), or regulatory molecules (cytokines, MMPs, gap junctions, cytotoxic molecules, migratory markers) is essentially bimodal. This polarization program turns MPs to a 'cytotoxic' or a 'symbiotic' phenotype [18]. In physiologic conditions, this dualism is probably modulated by a multiplicity of factors: the extent and duration of the environmental insult and the conditions of the resulting microenvironment. Possibly, predominant and persistent stimulation by pathogen components (such as LPS) may polarize MP towards the cytotoxic phenotype. A predominantly regulatory response is then mounted by the host, mediated by alternative cytokines that would take over at a later stage to induce a symbiotic phenotype aimed at resuming homeostasis upon pathogen clearance.
Materials and methods
MP separation and FACS staining
Peripheral blood mononuclear cells (PBMCs) from an HLA-A*0201-positive healthy caucasian male donor age 35 were collected at the Department of Transfusion Medicine, NIH. PBMCs were isolated by Ficoll gradient separation and frozen until analysis. After thawing, PBMCs were kept overnight in 175-cm2 tissue-culture flasks (Costar) in complete medium (CM) consisting of Iscove's medium (Biofluids) supplemented with 10% heat-inactivated human AB-serum (Gemini Bioproducts), 10 mM HEPES buffer (Cellgro; Mediatech), 0.03% l-glutamine (Biofluids), 100 U/ml penicillin/streptomycin (Biofluids), 10 μg/ml ciprofloxacin (Bayer), and 0.5 mg/ml amphotericin B (Biofluids). Adherent and non-adherent cells were gently removed from the flask and centrifugated. MPs were separated by negative selection using the MP isolation kit and an autoMACS system (Miltenyi). Before and after separation cells were stained with anti-CD14-FITC (Becton Dickinson), and analyzed using a FACScalibur flow cytometer and CellQuest software (Becton Dickinson).
Stimulation of MPs and RNA isolation
Negatively selected CD14+ cells were washed twice with serum-free OPTI-MEM (OM) medium (Gibco-BRL) prepared similarly to CM. CD14+ cells were then seeded at a concentration of 1 × 106/ml in 10 ml OM in 25 cm2 flasks (Falcon) and stimulated with 5 μg/ml LPS (Sigma) for 1 h. LPS was used at 5 μg/ml to simulate maximal pathogen exposure as in [6]. No LPS was added to the non-stimulation control flask. After 1 h, 42 cytokines, chemokines and soluble factors were added individually to the MP suspensions (Table 1). Then, 4 and 9 h after LPS stimulation, MPs were harvested, washed twice in PBS and lysed for RNA isolation using 700 μl RNeasy lysis buffer (Qiagen) per 25 cm2 flask, according to the manufacturer's protocol.
Probe preparation, amplification and hybridization to microarrays
Total RNA was isolated using RNeasy minikits (Qiagen). Amplified antisense RNA (aRNA) was prepared from total RNA (0.5-3 μg) according the protocol previously described by us [9,39]. Test samples were labeled with Cy5-dUTP (Amersham) while the reference sample (pooled normal donor PBMCs) was labeled with Cy3-dUTP. Test-reference sample pairs were mixed and co-hybridized to 17K cDNA microarrays.
Microarrays and statistical analyses
Hybridized arrays were scanned at 10-μm resolution on a GenePix 4000 scanner (Axon Instruments) at variable PMT voltage to obtain maximal signal intensities with less than 1% probe saturation. Resulting jpeg and data files were analyzed via mAdb Gateway Analysis tool [40]. Data were further analyzed using Cluster and TreeView software [12] and Partek Pro software (Partek). The global gene-expression profiling of 4- and 9-h treated and untreated MP consisted of 98 experimental samples. Subsequent low-stringency filtering (80% gene presence across all experiments and removal of genes that did not have a log2 ≥ 1.2: 2.3 ratio in at least one of the samples) selected 10,370 genes for further analysis. Clustering of experimental samples according to Eisen et al. [12] was based on these genes. Gene ratios were average corrected across experimental samples and displayed according to the central method for display using a normalization factor as recommended by Ross [41].
NFκB protein activation analysis
MPs separated from peripheral blood by adherence were stimulated for 1 h with LPS and for an additional 30 min with cytokines selected from the alternative group (IL-4, IL-13, IL-1α) or the classical group (IFN-γ, IL-6, IL-3). In addition, TNF-α was tested. After 90 min stimulation, cytoplasmic cell extracts were isolated using a cytoplasmic and nuclear extract kit (Active Motif), and the TransAM NFκB transcription factor kit (Active Motif) was used to detect activation of NFκB subunits p50, p52, p65, c-Rel and RelB, according to the manufacturer's protocol.
Real-time quantitative RT-PCR
MPs obtained from PBMC of five normal caucasian donors (three males, two females, age range: 35-55 years old) were stimulated with four cytokines/soluble factors selected from the alternative group (IL-13, IL-1α, IL-4, TGF-β) and four from the classical groups (FLT-3L, IFN-γ, IL-3, CD40L). TaqMan real-time PCR was performed on amplified RNA material isolated after stimulation for 4 h in conditions identical to those applied for the cDNA array study to validate the expression of the following 10 genes: TRAF binding protein, NFκB-p105/50, MMP9, MMP19, MCP-1, mannose receptor, IL-24, IL-1R, IL-1A and FADD-MORT. An ABI Prism 7900 HT sequence detection system with 384-well capability (Applied Biosystems) was used for detection. Primers and TaqMan probes (Biosource) were designed to span exon-intron junctions and to generate amplicons of less than 150 bp. TaqMan probes were labeled at the 5' end with the reporter dye molecule FAM (6-carboxyfluorescein; emission λmax = 518 nm) and at the 3' end with the quencher dye molecule TAMRA (6-carboxytetramethylrhodamine; emission λmax = 582 nm). The following are the sequences for forward (f) and reverse (r) primer and probe (p) pairs:
IL-1α f: TGTATGTGACTGCCCAAGATGAA
IL-1α r: ACTACCTGTGATGGTTTTGGGTATC
IL-1α p: FAM-AGTGCTGCTGAAGGAGATGCCTG-TAMRA
IL-1 rec. f: TGTCACCGGCCAGTTGAGT
IL-1 rec. r: GCACTGGGTCATCTTCATCAATT
IL1 rec p: FAM-ACATTGCTTACTGGAAGTGGAATGGGTCAG-TAMRA
TRAF bp f: TTGCTTACAG AGGTGTCTCAACAAG
TRAF bp r: CTCCGGATTTGTTCTGTCAGTTC
TRAF bp p: FAM-AGCAAAGTGTATTCCAGCAATGGTGTGTCC-TAMRA
MMP9 f: TGGATCCAAAACTACTCGGAAGA
MMP9 r: GAAGGCGCGGGCAAA
MMP9 p: FAM-CGCGGGCGGTGATTGACGAC-TAMRA
MMP19 f: GACGAGCTAGCCCGAACTGA
MMP19 r: TTTGGCACTCCCGTAAACAAA
MMP19 p: FAM-TCAGCAGCTACCCCAAACCAATCAAGG-TAMRA
Mannose receptor f: CTAAACCTACTCATGAATTACTTACAACAAAAG
Mannose receptor. r: CTCCGGCCACGTTGGA
Mannose receptor p: FAM-ACACAAGGAAGATGGACCCTTCTAAACCGTC-TAMRA
FADD-MORT f: GGTGGCTGACCTGGTACAAGA
FADD-MORT r: ACATGGCCCCACTCCTGTT
FADD-MORT p: FAM-TTCAGCAGGCCCGTGACCTCCA-TAMRA
NFκB p105/50 f: CTACACCGAAGCAATTGAAGTGA
NFκB p105/50 r: CAGCGAGTGGGCCTGAGA
NFκB p105/50 p: FAM-CAGGCAGCCTCCAGCCCAGTGA-TAMRA
IL-24 f: AAGAAAATGAGATGTTTTCCATCAGA
IL-24 r: CTGTTTGAATGCTCTCCGGAAT
IL-24 p: FAM-ACAGTGCACACAGGCGGTTTCTGC-TAMRA
MCP-1 f: CATGGTACTAGTGTTTTTTAGATACAGAGACTT
MCP-1 r: TAATGATTCTTGCAAAGACCCTCAA
MCP-1 p: FAM-AACCACAGTTCTACCCCTGGGATG-TAMRA
Standards for the selected genes were amplified by reverse transcriptase primer-specific amplification of 6 μg antisense RNA obtained from PBMCs stimulated in vitro with IL-2 (300 IU/ml and Flu M1 peptide) and reverse transcribed using random dN6 primers (Boehringer Mannheim). Amplified cDNA standards were quantified by spectrometry and the number of copies was calculated using the Oligo Calculator software [42]. Six micrograms of test antisense RNA samples were converted to cDNA using random primers and were immediately used for quantitative real-time PCR (RT-PCR). RT-PCR reactions of cDNA samples were conducted in a total volume of 20 μl, including 1 μl cDNA, 1x TaqMan Master Mix (Applied Biosystems), 2 μl of 20 μM primers and 1 μl of 12.5 μM probe. Thermal cycler parameters included 2 min at 50°C, 10 min 95°C and 40 cycles involving denaturation at 95°C for 15 sec, annealing-extension at 60°C for 1 min.
Linear regression analyses of all standard curves were 0.98 or greater. Standard curve extrapolation of copy number and quantity means were performed using the ABI Prism SDS 2.1 software (Applied Biosystems). Normalization of samples was performed by dividing the quantity mean of the gene of interest run in duplicate by the quantity mean of reference actin filament associated protein (AFAP) gene × 105 [43].
Additional data files
The following additional data are available with the online version of this article. Additional data file 1 is a spread sheet containing microarray raw data (intensity ratios of test versus reference samples) subsequently subjected to cluster analysis.
Supplementary Material
Additional data file 1
A spread sheet containing microarray raw data (intensity ratios of test versus reference samples) subsequently subjected to cluster analysis
Click here for additional data file
Acknowledgements
D.N. was supported by a grant from Deutsche Krebshilfe.
Figures and Tables
Figure 1 Unsupervised clustering of LPS-stimulated CD14+ MPs exposed to distinct cytokine treatments. CD14+ MPs were stimulated in parallel with LPS and exposed after 1 h to 42 individual cytokines (see Table 1 for cytokines used). Antisense RNA obtained 4 and 9 h following LPS stimulation was hybridized to custom-made 17K cDNA arrays. Unsupervised Eisen clustering [12] was applied to the complete, unfiltered dataset of 98 experiments. Arrowheads represent control samples that did not receive any stimulation and were obtained at different time points to parallel culture conditions during stimulation (0, 4 and 9 h). Light blue and gold represent samples obtained 4 and 9 h after stimulation, respectively and black represents no stimulation. A375 is a melanoma cell line that was used for quality control alternating conventional (Cy5, red) or reciprocal (Cy3, green) labeling every 25 experiments as previously described [9]. Experiments cluster closer together according to time rather than type of stimulation, with samples obtained after 4 h clustering together (4') with the exception of few cytokines (4"). With few exceptions (9"), cytokine treatments at 9 h clustered together with non-stimulated MPs (9').
Figure 2 Definition of cytokine classes based on their modulatory effect on the response of CD14+ circulating MP to LPS. (a) Definition of cytokine classes based on their modulatory effect on the response of CD14+ circulating MPs to LPS. CD14+ MPs were stimulated with LPS and exposed after 1 h to 42 individual cytokine stimulations. The clusterogram represents 2,057 genes obtained by Eisen hierarchical clustering of the complete 17K dataset filtered for genes that are expressed in a minimum of 80% of the samples 4 h after LPS stimulation, and that, at least in one experiment, displayed a greater than threefold change in expression over stimulation with LPS alone. Two main classes of cytokines are distinguished: the blue bar indicates type II cytokines (for example, IL-4 and IL-13) and the red bar indicates type I cytokines (such as IFN-γ, CD40L and FLT-3L). A smaller third class including IL-10 is also shown (green bar). TNF-α (purple bar), TNF-β and GM-CSF (dark-gray bars) clustered separately from all other cytokines. (b) Unsupervised principal component analysis (PCA) of the unfiltered 17K gene dataset. Cytokine treatments are color-coded according to the groups classified in (a): blue circles, alternative type I cytokines; green circles, alternative type II cytokines; red circles, classical cytokines; purple circle (arrowed), TNF-α; dark-gray circles, TNF-β and GM-CSF; yellow circle (arrowed), LPS alone.
Figure 3 Effect of cytokines on the expression of genes dependent on NFκB activation. (a) One hundred and twenty-one genes associated with the downstream effects of NFκB activation were selected from the unfiltered 17K gene dataset and reclustered without changing the cytokine treatment grouping as per Figure 2. The genes most significantly different in expression between the two classes are listed on the right. (b) All samples were reclustered (dendrogram) using the 121 genes described in (a). Color coding is identical to that in Figure 2. (c) The histogram depicts detection of the free NFκB subunit p50 in four consecutive experiments. p = p2 value, NS, nonsignificant. Blue bars, MPs stimulated with LPS (1 h) + alternative cytokines (additional 30 min stimulation); red bars, classical cytokines; yellow bar, MPs stimulated with LPS alone; gray bar, unstimulated MP control; purple bar, TNF-α alternative II class of cytokines.
Figure 4 Effect of cytokines on the expression of genes for matrix metalloproteinases (MMPs) and MMP-related genes. MMP genes and MMP-related genes (such as disintegrins, IFN-γ and related genes, IFN-α, α-defensins, TGF-β, TNF-α, IGF-1, EGF, FGF, IL-1, MCP-3 - complete list available from the authors on request) were clustered according to the alternative and classical groups and filtered for 70% presence and at least one value equal to or greater than a 1.5-fold change in expression (a total of 36 genes). The clusterogram displays 11 of the most representative genes. Color coding is as Figure 2.
Figure 5 Selection of genes differentially expressed between classically and alternatively activated MPs based on previous annotations linking their function to MP activation. Eighty-one representative genes with known MP-associated function are shown among 2,007 genes differentially expressed (p2 < 0.001, Student's t-test). The nomenclature for chemokines and chemokine receptors follows the recommendations of the IUIS/WHO subcommittee on chemokine nomenclature [44]. Color coding as in Figure 2.
Figure 6 Induction of gene expression by stimulation of LPS-activated MPs with IFN-γ and IL-4. MPs obtained from PBMC from five normal donors were stimulated for 4 h with one cytokine representative of the alternative (IL-4) and one of the classical group (IFN-γ) and gene transcription measured by TaqMan real-time PCR. The relative quantification of 10 genes was calculated by normalizing the ratio of the mean copy number for each gene with the mean copy number of the reference AFAP gene in MPs from five donors. Statistically significant differences (p-value < 0.05) between the two cytokine treatments as assessed by Student's t-test are represented by an asterisk.
Figure 7 Induction of gene expression by cytokines of the alternative and classical groups. MPs derived from PBMC from five normal donors were stimulated for 4 h with four cytokines/soluble factors selected from the alternative group (IL-13, IL-1α, IL-4, TGF-β) and four from the classical group (FLT-3 L, IFN-γ, IL-3, CD40L). Gene expression was assessed by TaqMan real-time PCR. Relative estimates of (a) mannose receptor, (b) NFκB and (c) TRAF binding protein genes were calculated by averaging the ratio of the quantity mean of each gene normalized to the reference AFAP gene in five donors. Statistical significance is expressed as Student's t-test p-value.
Table 1 Concentration and doses of cytokines/soluble factors used for stimulation of monocytes
Cytokine Concentration in vitro Source
Aldara 3 μM 3M Pharmaceuticals
BCA-1 10 ng/ml Peprotech
CD40 L 500 ng/ml Peprotech
FLT-3 ligand 100 ng/ml Peprotech
GM-CSF 1,000 IU/ml Peprotech
IFN-γ 1,000 U/ml Biogen
IFN-α2 1,000 U/ml Peprotech
IFN-α2b (α2, α2b,) 1,000 U/ml PBL Biomedical Laboratories
IFN-αA (2a) 1,000 U/ml PBL Biomedical Laboratories
IFN-αI 1,000 U/ml PBL Biomedical Laboratories
IFN-αB2 1,000 U/ml PBL Biomedical Laboratorios
IFN-α4b 1,000 U/ml PBL Biomedical Laboratories
IFN-αC 1,000 U/ml PBL Biomedical Laboratories
IFN-αF 1,000 U/ml PBL Biomedical Laboratories
IFN-αG 1,000 U/ml PBL Biomedical Laboratories
IFN-αH2 1,000 U/ml PBL Biomedical Laboratories
IFN-αJ1 1,000 U/ml PBL Biomedical Laboratories
IFN-αK 1,000 U/ml PBL Biomedical Laboratories
IFN-αWA 1,000 U/ml PBL Biomedical Laboratories
IFN-β 1,000 U/ml PBL Biomedical Laboratories
IL-1α 10 ng/ml National Cancer Institute (NCI) Biological Research Branch
IL-1β 10 ng/ml NCI Biological Research Branch
IL-2 6,000 IU/ml Chiron
IL-3 15 ng/ml NCI Biological Research Branch
IL-4 1,000 IU/ml Peprotech
IL-5 10 ng/ml Peprotech
IL-6 100 ng/ml Peprotech
IL-8 100 ng/ml Peprotech
IL-10 10 ng/ml Peprotech
IL-12 10 ng/ml Peprotech
IL-13 20 ng/ml Peprotech
IL-15 20 ng/ml Peprotech
LPS only 5 μg/ml Sigma Aldrich
MIP-1α (CCL3) 100 ng/ml Peprotech
MIP-4 (PARC, CCl18) 100 ng/ml Peprotech
MIP-1β (CCL4) 100 ng/ml Peprotech
RANTES (CCL5) 100 ng/ml Peprotech
TARC (CCL17) 100 ng/ml Peprotech
TGF-α 10 ng/ml Peprotech
TNF-β 20 ng/ml Peprotech
TGF-β 5 ng/ml Peprotech
TNF-α 100 ng/ml Peprotech
VEGF 10 ng/ml R&D Systems
==== Refs
Mantovani A Sozzani S Locati M Allavena P Sica A Macrophage polarization: tumor-associated macrophages as a paradigm for polarized M2 mononuclear phagocytes. Trends Immunol 2002 23 549 555 12401408 10.1016/S1471-4906(02)02302-5
Takeda K Kaisho T Akira S Toll-like receptors. Annu Rev Immunol 2003 21 335 376 12524386 10.1146/annurev.immunol.21.120601.141126
Anderson KV Toll signaling pathways in the innate immune response. Curr Opin Immunol 2000 12 13 19 10679407 10.1016/S0952-7915(99)00045-X
Mocellin S Panelli MC Wang E Nagorsen D Marincola FM The dual role of IL-10. Trends Immunol 2003 24 36 43 12495723 10.1016/S1471-4906(02)00009-1
Gordon S Alternative activation of macrophages. Nat Rev Immunol 2003 3 23 35 12511873 10.1038/nri978
Huang Q Liu D Majewski P Schulte LC Korn JM Young RA Lander ES Hacohen N The plasticity of dendritic cell responses to pathogens and their components. Science 2001 294 870 875 11679675 10.1126/science.294.5543.870
Langenkamp A Messi M Lanzavecchia A Sallusto F Kinetics of dendritic cell activation: impact on priming of Th1, Th2 and nonpolarized T cells. Nat Immunol 2000 1 311 316 11017102 10.1038/79758
Belardelli F Ferrantini M Cytokines as a link between innate and adaptive antitumor immunity. Trends Immunol 2002 23 201 208 11923115 10.1016/S1471-4906(02)02195-6
Wang E Miller L Ohnmacht GA Liu E Marincola FM High fidelity mRNA amplification for gene profiling using cDNA microarrays. Nat Biotechnol 2000 18 457 459 10748532 10.1038/74546
Locati M Deuschle U Massardi ML Martinez FO Sironi M Sozzani S Bartfai T Mantovani A Analysis of the gene expression profile activated by the CC chemokine ligand 5/Rantes and by lipopolysaccharide in human monocytes. J Immunol 2002 168 3557 3562 11907119
Hatada EN Krappmann D Scheidereit C NF-κB and the innate immune response. Curr Opin Immunol 2000 12 52 58 10679399 10.1016/S0952-7915(99)00050-3
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998 95 14863 14868 9843981 10.1073/pnas.95.25.14863
Henco K Brosius J Fujisawa A Fujisawa J-I Haynes JR Hochstadt J Kovacic T Pasek M Schambock A Schmid J Structural relationship of human interferon alpha genes and pseudogenes. J Mol Biol 1985 185 227 260 4057246
Mogensen KE Lewerenz M Reboul J Lutfalla G Uze G The type I interferon receptor: structure, function, and evolution of a family business. J Interferon Cytokine Res 1999 19 1069 1098 10547147 10.1089/107999099313019
Cook JR Cleary CM Mariano TM Izotova L Pestka S Differential responsiveness of a splice variant of the human Type I intereferon receptor to intereferons. J Biol Chem 1996 271 13448 13453 8662801 10.1074/jbc.271.23.13448
Parks WC Shapiro SD Matrix metalloproteinases in lung biology. Respir Res 2001 2 10 19 11686860 10.1186/rr33
de Fougerolles AR Chi-Rosso G Bajardi A Gotwals P Green CD Koteliansky VE Global expression of extracellular matrix-integrin interactions in monocytes. Immunity 2000 13 749 758 11163191 10.1016/S1074-7613(00)00073-X
Kreutz M Fritsche J Andreesen R Burke B, Lewis CE Macrophages in tumor biology. The Macrophage 2002 Oxford: Oxford University Press 457 489
Haro H Crawford HC Fingleton B Shinomiya K Spengler DM Matrisian LM Matrix metalloproteinase-7-dependent release of tumor necrosis factor-alpha in a model of herniated disc resorption. J Clin Invest 2000 105 143 150 10642592
Yu WH Woessner JF Jr Heparan sulfate proteoglycan as extracellular docking molecules for matrilysin (matrix metalloproteinase 7). J Biol Chem 2000 275 4183 4191 10660581 10.1074/jbc.275.6.4183
Yu Q Stamenkovic I Cell surface localized metalloproteinase 9 proteolytically activates TGF beta and promotes tumor invasion and angiogenesis. Genes Dev 2000 14 163 176 10652271
Sadowski T Dietrich S Koschinsky F Sedlacek R Matrix metalloproteinase 19 regulates IGF-mediated proliferation, migration and adhesion in human keratinocytes through proteolysis of insulin-like growth factor binding protein-3. Mol Biol Cell 2003 14 4569 4580 12937269 10.1091/mbc.E03-01-0009
Locati M Otero K Schioppa T Signorelli P Perrier P Baviera S Sozzani S Mantovani A The chemokine system: tuning and shaping by regulation of receptor expression and coupling in polarized responses. Allergy 2002 57 972 982 12358993 10.1034/j.1398-9995.2002.02166.x
Vivers S Dransfield I Hart SP Role of macrophage CD44 in the disposal of inflammatory cell corpses. Clin Sci (Lond) 2002 103 441 449 12401116
Mantovani A Gray PA Van Damme J Sozzani S Macrophage-derived chemokine (MDC). J Leukocyte Biol 2000 68 400 404 10985257
Conti P Kempuraj D Frydas S Kandere K Boucher W Letourneau R Madhappan B Sagimoto K Christodoulou S Theoharides TC IL-10 subfamily members: IL-19, IL-20, IL-22, IL-24 and IL-26. Immunol Lett 2003 88 171 174 12941475 10.1016/S0165-2478(03)00087-7
Kaplanski G Marin V Montero-Julian F Mantovani A Farnarier C IL-6: a regulator of the transition from neutrophil to monocyte recruitment during inflammation. Trends Immunol 2003 24 25 29 12495721 10.1016/S1471-4906(02)00013-3
Beutler B Tlr4: central component of the sole mammalian LPS sensor. Curr Opin Immunol 2000 12 20 26 10679411 10.1016/S0952-7915(99)00046-1
Guha M Mackman N LPS induction of gene expression in human monocytes. Cell Signal 2001 13 85 94 11257452 10.1016/S0898-6568(00)00149-2
Alberati-Giani D Malherbe P Ricciardi-Castagnoli P Kohler C Denis-Donini S Cesura AM Differential regulation of indoleamine 2,3-dioxygenase expression by nitric oxide and inflammatory mediators in IFN-γ-activated murine macrophages and microglial cells. J Immunol 1997 159 419 426 9200481
Bertazzo A Ragazzi E Biasiolo M Costa CVL Allegri G Enzyme activities involved in tryptophan metabolism along the kynurenine pathway in rabbits. Biochim Biophys Acta 2001 1527 167 175 11479034
Thomas SR Mohr D Stocker R Nitric oxide inhibits indoleamine 2,3-dioxygenase activity in interferon-γ primed mononuclear phagocytes. J Biol Chem 1994 269 14457 14464 7514170
Zheng T Zhu Z Wang Z Homer RJ Ma B Riese RJ JrChapman HA JrShapiro SD Elias JA Inducible targeting of IL-13 to the adult lung causes matrix metalloproteinase- and cathepsin-dependent emphysema. J Clin Invest 2000 106 1081 1093 11067861
Munaut C Noel A Hougrand O Foidart JM Boniver J Deprez M Vascular endothelial growth factor expression correlates with matrix metalloproteinases MT1-MMP, MMP-2 and MMP-9 in human glioblastomas. Int J Cancer 2003 106 848 855 12918061 10.1002/ijc.11313
Sanceau J Boyd DD Seiki M Bauvois B Interferons inhibit tumor necrosis factor-alpha-mediated matrix metalloproteinases-9 activation via interferon regulatory factor-1 binding competition with NF-kappa B. J Biol Chem 2002 277 35766 35775 12105194 10.1074/jbc.M202959200
Lee SW Tomasetto C Keyomarsi K Sager R Transcriptional downregulation of gap-junction proteins block junctional communication in human mammary tumor cell lines. J Cell Biol 1992 118 1213 1221 1324944 10.1083/jcb.118.5.1213
Eugenin EA Branes MC Berman JW Saez JC TNF-alpha plus IFN-gamma induce connexin43 expression and formation of gap junctions between human monocyte/macrophages that enhance physiological responses. J Immunol 2003 170 1320 1328 12538692
Alves LA Coutinho-Silva R Persechini PM Spray DC Savino W Campos de Carvalho AC Are there functional gap junctions or junctional hemichannels in macrophages? Blood 1996 88 328 334 8704191
Wang E Marincola FM Bowtell D, Sambrook J Amplification of small quantities of mRNA for transcript analysis. DNA Arrays: A Molecular Cloning Manual 2002 Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press 204 213
Bacon K Baggiolini M Broxmeyer H Horuk R Lindley I Mantovani A Maysushima K Murphy P Nomiyama H Oppenheim J Chemokine/chemokine receptor nomenclature. J Interferon Cytokine Res 2002 22 1067 1068 12433287 10.1089/107999002760624305
mAdb Gateway Analysis tool
Ross DT Scherf U Eisen MB Perou CM Rees C Spellman P Iyer V Jeffrey SS Van de Rijn M Waltham M Systematic variation in gene expression patterns in human cancer cell lines. Nat Genet 2000 24 227 235 10700174 10.1038/73432
Oligo Calculator software
Jin P Zhao Y Ngalame Y Panelli MC Nagorsen D Monsurro V Smith K Hu N Su H Taylor PR Selection and validation of endogenous reference genes using a high throughput approach. BMC Genomics 2004 5 55 15310404 10.1186/1471-2164-5-55
| 15693944 | PMC551535 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 21; 6(2):R15 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r15 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r161569394510.1186/gb-2005-6-2-r16ResearchPreferred analysis methods for Affymetrix GeneChips revealed by a wholly defined control dataset Choe Sung E [email protected] Michael [email protected] Alan M [email protected] George M 1Halfon Marc S [email protected] Department of Genetics, Harvard Medical School, New Research Building, 77 Avenue Louis Pasteur, Boston, MA 02115, USA2 Division of Genetics, Department of Medicine, Brigham and Women's Hospital, New Research Building, 77 Avenue Louis Pasteur, Boston, MA 02115, USA3 Howard Hughes Medical Institute, Brigham and Women's Hospital, 20 Shattuck Street, Boston, MA 02115, USA4 Department of Biochemistry, 140 Farber Hall, 3435 Main St., SUNY at Buffalo, Buffalo, NY 14214, USA5 Center of Excellence in Bioinformatics, 140 Farber Hall, 3435 Main St., SUNY at Buffalo, Buffalo, NY 14214, USA6 German Cancer Research Center (DKFZ/B110), Im Neuenheimer Feld 580, 69120 Heidelberg, Germany2005 28 1 2005 6 2 R16 R16 3 8 2004 20 10 2004 2 12 2004 Copyright © 2005 Choe 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.
A 'spike-in' experiment for Affymetrix GeneChips is described that provides a defined dataset of 3,860 RNA species. A 'best route' combination of analysis methods is presented which allows detection of approximately 70% of true positives before reaching a 10% false discovery rate.
Background
As more methods are developed to analyze RNA-profiling data, assessing their performance using control datasets becomes increasingly important.
Results
We present a 'spike-in' experiment for Affymetrix GeneChips that provides a defined dataset of 3,860 RNA species, which we use to evaluate analysis options for identifying differentially expressed genes. The experimental design incorporates two novel features. First, to obtain accurate estimates of false-positive and false-negative rates, 100-200 RNAs are spiked in at each fold-change level of interest, ranging from 1.2 to 4-fold. Second, instead of using an uncharacterized background RNA sample, a set of 2,551 RNA species is used as the constant (1x) set, allowing us to know whether any given probe set is truly present or absent. Application of a large number of analysis methods to this dataset reveals clear variation in their ability to identify differentially expressed genes. False-negative and false-positive rates are minimized when the following options are chosen: subtracting nonspecific signal from the PM probe intensities; performing an intensity-dependent normalization at the probe set level; and incorporating a signal intensity-dependent standard deviation in the test statistic.
Conclusions
A best-route combination of analysis methods is presented that allows detection of approximately 70% of true positives before reaching a 10% false-discovery rate. We highlight areas in need of improvement, including better estimate of false-discovery rates and decreased false-negative rates.
==== Body
Background
Since their introduction in the mid 1990s [1,2], expression-profiling methods have become a widespread tool in numerous areas of biological and biomedical research. However, choosing a method for analyzing microarray data is a daunting task. Dozens of methods have been proposed for the analysis of both high-density oligonucleotide (for example, Affymetrix GeneChip) and spotted cDNA or long oligonucleotide arrays, with more being put forward on a regular basis [3]. Moreover, it is clear that different methods can produce substantially different results. For example, two lists of differentially expressed genes generated from the same dataset can display as little as 60-70% overlap when analyzed using different methods ([4] and see Additional data file 1). Despite the large number of proposed algorithms, there are relatively few studies that assess their relative performance [5-9]. A significant challenge to undertaking such studies is the scarcity of control datasets that contain a sufficiently large number of known differentially expressed genes to obtain adequate statistics. The comparative studies that have been performed have used a small number of positive controls, and have included a background RNA sample in which the concentrations of the various genes are unknown, preventing an accurate assessment of false-positive rates and nonspecific hybridization.
The most useful control datasets to date for evaluating the effectiveness of analysis methods for Affymetrix arrays are cRNA spike-in datasets from Affymetrix and Gene Logic. The Affymetrix Latin square dataset [10] is a series of transcriptional profiles of the same biological RNA sample, into which 42 cRNAs have been spiked at various known concentrations. The dataset is designed so that, when comparing any two hybridizations in the series, all known fold changes are powers of two. The Gene Logic dataset [11] has a similar experimental design, but with 11 cRNAs spiked in at varying fold changes, ranging from 1.3-fold upwards.
Here we present a new control dataset for the purpose of evaluating methods for identifying differentially expressed genes (DEGs) between two sets of replicated hybridizations to Affymetrix GeneChips. This dataset has several features to facilitate the relative assessment of different analysis options. First, rather than containing a limited number of spiked-in cRNAs, the current dataset has 1309 individual cRNAs that differ by known relative concentrations between the spike-in and control samples. This large number of defined RNAs enables us to generate accurate estimates of false-negative and false-positive rates at each fold-change level. Second, the dataset includes low fold changes, beginning at only a 1.2-fold concentration difference. This is important, as small fold changes can be biologically relevant, yet are frequently overlooked in microarray datasets because of a lack of knowledge as to how reliably such small changes can be detected. Third, our dataset uses a defined background sample of 2,551 RNA species present at identical concentrations in both sets of microarrays, rather than a biological RNA sample of unknown composition. This background RNA population is sufficiently large for normalization purposes, yet also enables us to observe the distribution of truly nonspecific signal from probe sets which correspond to RNAs not present in the sample.
We have used this dataset to compare several algorithms commonly used for microarray analysis. To perform a direct comparison of the selected methods at each stage of analysis, we applied all possible combinations of options to the data. Thus, it was possible to assess whether some steps are more critical than others in maximizing the detection of true DEGs. Our results show that at several steps of analysis, large differences exist in the effectiveness of the various options that we considered. These key steps are: first, adjusting the perfect match probe signal with an estimate of nonspecific signal (the method from MAS 5.0 [12] performs best); second, checking that the log fold changes are roughly distributed around 0 (by observing the so-called M versus A plot [13], the plot of log fold change (M) versus average log signal intensity (A)), and if necessary, performing a normalization at the probe-set level to center this plot around M = 0; and third, choosing the best test statistic (the regularized t-statistic from CyberT [14] is most accurate). Overall, we find a significant limit to the sensitivity of microarray experiments to detect small changes: in the best-case scenario we could detect approximately 95% of true DEGs with changes greater than twofold, but less than 30% with changes below 1.7-fold before exceeding a 10% false-discovery rate. We propose a 'best-route' combination of existing methods to achieve the most accurate assessment of DEGs in Affymetrix experiments.
Results and discussion
Experimental design
A common use of microarrays is to compare two samples, for example, a treatment and a control, to identify genes that are differentially expressed. We constructed a control dataset to mimic this scenario using 3,860 individual cRNAs of known sequence in a concentration range similar to what would be used in an actual experimental situation (see Materials and methods). The cRNAs were divided into two samples - 'constant' (C) and 'spike' (S) - and each sample was hybridized in triplicate to Affymetrix GeneChips (six chips total). The S sample contains the same cRNAs as the C sample, except that selected groups of approximately 180 cRNAs each are present at a defined increased concentration compared to the C sample (Figure 1, Table 1). Out of the 3,860 cRNAs, 1,309 were spiked in with differing concentrations between the S and C samples. The rest (2,551) are present at identical relative concentration in each sample, to serve as a guide for normalization between the two sets of microarrays. For the sake of consistency with typical discussions of microarray experiments, we sometimes refer to the cRNAs with positive log fold changes as DEGs, despite their not representing true gene-expression data.
Assignment of Affymetrix probe sets to DGC clones
In the Affymetrix GeneChip design, the expression level of each RNA species is reported by a probe set, which in the DrosGenome1 chip [15] comprises 14 oligonucleotide probe pairs. Each probe pair contains two 25-mer DNA oligonucleotide probes; the perfect match (or PM) probe matches perfectly to the target RNA, and the mismatch (or MM) probe is identical to its PM partner probe except for a single homomeric mismatch at the central base-pair position, and thus serves to estimate nonspecific signal.
The DrosGenome1 chip used in this experiment is based on release version 1.0 of the Drosophila genome sequence and thus does not represent the most up-to-date annotated version of the genome. To ensure that probe-target assignments are made correctly, we assigned the 14,010 probe sets on the DrosGenome1 GeneChip to the spiked-in RNAs by BLAST of the individual PM probe sequences against the Drosophila Gene Collection release 1.0 (DGC [16]) clone sequences that served as the template for the cRNA samples (Materials and methods). Of the 3,860 DGC clones used in this study, 3,762 (97%) have full-length cDNA sequence available at the DGC web site, 90 have 3' and 5'-end sequence only, and eight have no available sequence. For each probe set, all clone sequences with BLAST matches to PM probe sequences in that probe set are collected, allowing at most two (out of 25 base-pair (bp)) mismatches, and only allowing matches on the correct strand. If at least three PM sequences match to a given clone, then the probe set is assigned to that clone. Matches of one probe set to more than one clone are allowed. In this manner, 3,866 probe sets are assigned to at least one DGC clone each. Among these probe sets, 1,331 have an increased concentration between the S and C chips, whereas 2,535 represent RNAs with equal concentration between the two samples. Among those probe sets which do not have any assignment using this criterion, if fewer than three PM probes within the probe set have a BLAST match to any clone, the probe set is then called 'empty' (that is, its signal should correspond to nonspecific hybridization). There are 10,131 empty probe sets; combined with the 2,535 1x probe sets, about 90% of the probe sets on the chip represent RNAs with constant expression level between the C and S samples. The rest of the probe sets are then called 'mixed', meaning that they match to more than one clone, but each with only a few PM probe matches. There are only 13 mixed probe sets. The numbers of probe sets assigned to each fold-change class are depicted in Table 1.
Assessment of absent/present call metrics
Our dataset design provides the rare knowledge of virtually all of the RNA sequences within a complex sample (excepting the small number (3%) of clones for which only partial sequence was available, and the possible rare mistakenly assigned or contaminated clone). We can therefore evaluate various absent/present call metrics on the basis of their ability to distinguish between the known present and absent RNAs. We investigate this issue at both the probe pair level and probe set level. For the probe pair level assessment, we first identify the probe pairs which we expect to show signal, and those which should not. We thus define two classes of probe pairs: first, perfect probe pairs, whose PM probe matches perfectly to a target RNA sequence, and neither PM nor MM probe matches to any other RNA in the sample with a BLAST E-value cutoff of 1 and word size of 7, and second, empty probe pairs, whose PM and MM probes do not match to any RNA sequence when using the same criteria.
On the chip, which contains 195,994 probe pairs, there are 50,859 perfect probe pairs and 117,904 empty ones. Observation of the signal for these probe pairs (Figure 2a,b) clearly shows that there is considerable signal intensity for the empty probe pairs. Figure 2c shows the ability of several metrics - log2(PM/MM), PM-MM, , and log2(PM) - to distinguish between perfect and empty probe pairs, by calculating receiver-operator characteristics (ROC) curves using the perfect probe pairs as true positives and the empty ones as true negatives. Each point on a curve depicts the specificity and sensitivity for RNA detection, when using a specific value of the corresponding metric as a cutoff for classifying probe sets as present or absent. Instead of depicting the false-positive rate (the fraction of true negatives that are detected as present) on the x-axis, which is customary for these types of graphs, we show the false-discovery rate (the fraction of detected probe sets which are true negatives), which distinguishes between the metrics more effectively for the top-scoring probe sets. Figure 2 clearly shows that metrics that compare the PM signal with the MM signal, such as log2(PM/MM) and PM-MM, are the most successful at distinguishing perfect from empty probe pairs. This indicates that the PM signal alone is a less effective indicator of RNA presence, probably because the probe hybridization affinity is highly sequence-dependent. However, even with the more successful metrics, only about 60% of the perfect probe sets are detected before reaching a 10% false-discovery rate, indicating that there is still a high level of variability in probe pair sensitivity, even when using the MM signal to estimate the probe hybridization affinity.
When signals from the 14 probe pairs in each probe set are combined to create a composite absence/presence call, a much larger fraction of the spiked-in RNA species can be detected reliably. To obtain absent/present calls at the probe-set level, we perform the Wilcoxon signed rank test using each of the metrics listed above [17]. The p-values from this test are used to generate the ROC curves in Figure 2d. Again, the best results are obtained when the metric compares PM with MM signals, as opposed to monitoring signal alone. The metric used in MAS 5.0 ((PM-MM)/(PM+MM)), which is equivalent to log2(PM/MM), performs best. Therefore, the MM signals are important in generating accurate presence/absence calls. In our dataset, about 85% of the true positives could be detected before having a 10% false-discovery rate. The detection of perfect probe pairs is not improved when we include additional information from replicates. The 15% of probe sets which are called absent may represent truly absent RNAs, owing to failed transcription or labeling (see Additional data file 5). However, as we do not have an independent measure of failed transcription for the individual cRNA sequences in the target sample, we cannot completely rule out the possibility that they are the result of non-responsive probes or a suboptimal absent/present metric that fails to score low-abundance cRNAs. Regardless, as non-responsive probes or missing target cRNAs should affect both the C and S chips identically, these factors should not limit the value of this dataset in making relative assessments of different analysis methods.
Generating expression summary values
The first task in analyzing Affymetrix microarrays is to combine the 14 PM and 14 MM probe intensities into a single number ('expression summary') which reflects the concentration of the probe set's target RNA species. Generating this value involves several discrete steps designed to subtract background levels, normalize signal intensities between arrays and correct for nonspecific hybridization. To compare the effectiveness of different analysis packages at each of these steps, we created multiple expression summary datasets using every possible (that is, compatible) combination of the options described below. Algorithms were chosen for their popularity with microarray researchers and their open-source availability, and were generated using the implementations found in the Bioconductor 'affy' package [18]. Figure 3 summarizes the options that we chose within Bioconductor. We also used the dChip [19] and MAS 5.0 [12] executables made available by the respective authors in order to cross-check with the open-source implementations within Bioconductor. In addition, we applied two analysis methods that incorporate probe sequence-dependent models of nonspecific signal (Perfect Match [20] and gcrma [21]). The combinations of options that were used to generate the 152 expression summary datasets are detailed in Additional data file 2.
Background correction
An estimate of the background signal, which is the signal due to nonspecific binding of fluorescent molecules or the autofluorescence of the chip surface, was generated using two possible metrics. The MAS background [17] is calculated on the basis of the 2nd percentile signal in each of 16 subsections of the chip, and is thus a spatially varying metric. The Robust Multi-chip Average (RMA) algorithm [22] subtracts a background value which is based on modeling the PM signal intensities as a convolution of an exponential distribution of signal and a normal distribution of nonspecific signal.
Normalization at the probe level
The signal intensities are normalized between chips to allow comparisons between them. Because in our dataset, a large number of RNAs are increased in S versus C (and none are decreased), commonly used methods often result in apparent downregulation for spiked-in probe sets in the 1x change category. We thus added a set of modified normalization methods which used our knowledge of the 1x probe sets. The following different methods were applied. Constant is a global adjustment by a constant value to equalize the chip-wide mean (or median) signal intensity between chips. Constantsubset is the same global adjustment but equalizing the mean intensity for only the probe sets with fold change equal to 1. Invariantset [23] is a nonlinear, intensity-dependent normalization based on a subset of probes which have similar ranks (the rank-invariant set) between two chips. Invariantsetsubset is the same as invariantset but the rank-invariant set is selected as a subset of the probe sets with fold change equal to 1. Loess normalization [24] is a nonlinear intensity-dependent normalization which uses a local regression to make the median fold change equal to zero, at all average intensity levels. Loesssubset normalization is the same as loess but using only the probe sets with fold change equal to 1. Quantile normalization [24] enforces all the chips in a dataset to have the same distribution of signal intensity. Quantilesubset normalization is the same as quantile but normalizes the spiked-in and non-spiked-in probe sets separately.
PM correction
We chose three ways to adjust the PM signal intensities to account for nonspecific signal. The first is to subtract the corresponding MM probe signal (subtractmm). The second is the method used in MAS 5.0, in which negative values are avoided by estimating the nonspecific signal when the MM value exceeds its corresponding PM intensity [17]. The third is PM only (no correction). The subtractmm and MAS methods are compatible only with the MAS background correction method; that is, it does not make sense to combine these with RMA background correction.
Expression summary
The 14 probe intensity values were combined using one of the following robust estimators: Tukey-biweight (MAS 5.0); median polish (RMA); or the model-based Li-Wong expression index (dChip). Analyses including the subtractmm PM correction method require dealing with negative values when PM is less than MM, which occurs in about a third of the cases. Within Bioconductor, the Li-Wong estimator can handle negative values, but the other two metrics mostly output 'not applicable' (NA) for the probe set when any of the constituent probe pairs has negative PM - MM. The result for MAS and median polish is NA for about 85% of the probe sets on the chip. To study the consequence of losing so many probe sets, we modified one of these two metrics (median polish) to accept negative (PM - MM) (medianpolishna), and added this metric whenever subtractmm was used.
Normalization at the probe set level
Many of the expression summary datasets that were produced still show a dependence of fold change on the signal intensity (Figure 4a). To correct this, a second set of expression summary datasets was created, in which a loess normalization at the probe set level was used to center the log-fold changes around zero (Figure 4b).
Comparison of the observed fold changes with known fold changes
For each of the 150 expression summary datasets that we generated, fold changes between the S and C samples were calculated and then compared with the actual fold changes. Most expression summary datasets show good correlation between the observed and actual fold changes (Figure 5). The greatest sources of variability are probe sets with low signal intensity; as Figure 5b shows, the correlation improves dramatically when we filter out the probe sets with low signal. For all the expression summary datasets, the agreement between observed and actual fold changes is good (R2 = 86 ± 3%) when the probe sets in the lowest quartile of signal intensity are filtered out. The expression summary datasets which involve correcting the PM signal by subtracting the MM signal (subtractmm) have the highest correlation coefficient, because low-intensity probe sets have been filtered out during processing, as described above. We therefore suggest that an important feature of a successful microarray analysis is to account for probe sets with low signal intensity, either by filtering them out or by using a signal-dependent metric for significance. Several ways of accomplishing such filtering are described below.
We also observed that the fold changes resulting from the chips are consistently lower than the actual fold changes. Apparently, the decrease in fold change is only partly the result of signal saturation (Figure 5b-c), and is not a byproduct of the robust estimators used to calculate expression summaries (because the low fold changes are also observed at the probe pair level; see Additional data file 3). In other experiments we have also observed that our Affymetrix fold-change levels are smaller than those obtained by quantitative reverse transcription (RT)-PCR (data not shown). One likely explanation is that we do not have an adequate estimate for nonspecific signal. For example, if we choose the MM signal as the nonspecific signal (thus calculating PM - MM, or PM - CT from MAS 5.0), we are probably overestimating the nonspecific signal, as the MM intensity value responds to increasing target RNA concentrations, and therefore contains some real signal. On the other hand, if we choose not to use a probe sequence-dependent nonspecific signal (such as in RMA), we are likely to underestimate the nonspecific signal for a large number of probes. In either case, the result is decreased fold change magnitudes. Artificially low fold-change values have been noted by others, including those investigating the Affymetrix Latin square [6], GeneLogic [22] and other [25] datasets, although some of the differences they report are smaller than are observed here.
Test statistics and ROC curves
Because a typical microarray experiment contains a large number of hypotheses (here 14,010) and a limited number of replicates (in this case three), high false-positive rates are a common problem in identifying DEGs. An important factor in minimizing false positives is to incorporate an appropriate error model into the signal/noise metric. We compared three t-statistic variants, which differ in their calculations of noise. The first is significance analysis for microarrays (SAM) [26], in which the t-statistic has a constant value added to the standard deviation. This constant 'fudge factor' is chosen to minimize the dependence of the t-statistic variance on standard deviation levels. The second is CyberT [14], in which the standard deviation is modeled as a function of signal intensity. The third is the basic (Student's) t-statistic. For CyberT and the basic t-test, we performed the tests on the expression summaries after log transformation, as well as on the raw data. As shown in the example ROC curve, the CyberT statistic outperforms the other statistics for the vast majority of expression summary datasets (Figure 6a). Inspection of the false positives and false negatives shows the reason for the different performance. Because CyberT uses a signal intensity-dependent standard deviation, probe sets at low signal intensities have reduced significance even when their observed fold change is high (Figure 6b). As shown in Figure 6c, the SAM algorithm (using the authors' Excel Add-in) does not effectively filter out these same false-positive probe sets (with low signal intensity and high fold change). Upon further inspection, we observed that the SAM algorithm favors using large values for the constant fudge factor, so that the t-statistic depends more on the fold change value, than on the noise level. The basic t-statistic is prone to false positives resulting from artificially low standard deviations, owing to the limited number of replicates in a typical microarray experiment (scattered magenta spots in Figure 6d). This comparison agrees with the result of Broberg [9], who also found that the CyberT approach (there called 'samroc') outperforms several other methods. Because the CyberT statistic clearly performs the best, we use only this statistic to compare the options for the other steps in microarray analysis, below.
Comparison of options at each of the other analysis steps
Performance of the various options that were investigated varied significantly, as seen by the ROC curves shown in Figure 7. First, we find that a second loess normalization at the probe set level generally yields a superior result (Figure 7a,f), as could be expected by observing the strong intensity-dependence of the fold-change values in Figure 4. This intensity-dependence is most likely the result of the unequal concentrations of labeled cRNA for the C and S chips. However, this artifact is not unique to this dataset. We routinely observe similar intensity-dependent fold changes in comparisons of biological samples, especially when there are small differences in starting RNA amounts between the two samples (see Additional data file 4 for an example). Therefore, in the absence of a biological reason to suppose that the fold change should depend on signal intensity, it is important to view the plot of log fold change versus signal and recenter it around y = 0 when necessary. Owing to the significant improvement seen when the second normalization is used, the subsequent figures (Figure 7b-f) only show the comparison of the remaining options in conjunction with this step (blue curves in Figure 7a).
Among the background correction methods, the MAS 5.0 method generally performs better than the RMA method (Figure 7b). No clearly superior normalization method was found at the probe level (Figure 7c), even when using the subset normalization variants, although quantile normalization tended to underperform in the absence of the second normalization step.
With respect to adjusting the PM probe intensity with an estimate of nonspecific signal, Figure 7d clearly shows that either subtracting the MM signal (subtractmm), or using the MAS 5.0 correction method, is better than using uncorrected or RMA-corrected PM values (PM-only). The MAS 5.0 method performs the best because it does not create any negative values. This result is in apparent conflict with the conclusions of Irizarry et al. [5], who show drastically reduced noise at low signal intensity levels when the PM signal is not adjusted with MM values, and therefore better detection of spiked-in probe sets when using the fold change as the cutoff criterion. However, when Irizarry et al. use a test statistic that takes the variance into account, PM-only and MM-corrected methods (MAS) have similar sensitivity/specificity (Figure 3d,e from [5]). In the dataset presented here, the MAS PM-correction method yields a high variance at low signal-intensity levels, which effectively reduces the false-positive calls at this intensity range when using CyberT, thus resulting in better performance than when using PM-only. We can reconcile the Irizarry et al. result with our observations by considering a major difference between the datasets used by the two studies. Both the Affymetrix and GeneLogic Latin square datasets used in [5] involve a small number (10-20) of spiked-in cRNAs in a common biological RNA sample, and therefore comparisons are made between two samples that are almost exactly the same. As a result, the nonspecific component of any given probe's signal is expected to be almost identical in the two samples, and should not contribute to false-positive differential expression calls. In contrast, a large fraction of our dataset is differentially expressed; in addition, the C sample contains a high concentration of (unlabeled) poly(C) RNA. Because nonspecific hybridization depends both on a probe's affinity and on the concentrations of RNAs that can hybridize to it in a nonspecific fashion, we expect that each probe's signal can have different contributions of nonspecific hybridization between the C and S chips. Figure 2a shows that nonspecific hybridization can be a large component of a probe's signal. We hypothesize that, for our dataset, PM-only performs worse than MM-corrected methods (subtractmm or MAS) because PM-only does not try to correct for nonspecific hybridization in a probe-specific fashion. In contrast, for the Latin square datasets used in [5], PM-only works just as well as MM-corrected methods because the contribution of nonspecific hybridization is constant. Therefore, datasets which compare substantially different RNA samples (such as two different tissue types) should probably be processed using the MAS 5.0 method for PM correction.
Figure 7e compares the different robust estimators that were used to create expression summaries. Of these, median polish (RMA) and the Tukey Biweight methods (MAS 5.0) perform the best. Figure 7f highlights the 10 best summary method option sets, which are also depicted in Figure 3, as well as straight applications of some popular software, with or without an additional normalization step at the probe-set level. The result from the MAS 5.0 software, when adjusted with the second loess normalization step, ranks among the top 10. However, the other methods (dChip, RMA and MAS 5.0 without probe-set normalization) are not as sensitive or specific at detecting DEGs.
We were concerned that some of our analyses might be confounded by a possible correlation between low fold change and low expression summary levels, which could affect the interpretations of Figure 7 (comparing different methods) and the detection of small fold changes (see below). We therefore examined the distribution of expression levels within each spiked-in fold change group, and compared the methods with respect to their ability to detect a subset of probe sets with low expression summary levels (Additional data file 5). We found that the distribution of expression levels for the known DEGs was comparable among all the fold-change groups, and that all the conclusions reported here are similarly applicable to the low expression subset. However, the sensitivity of all methods was reduced, suggesting that they perform less well on weakly expressed than on highly expressed genes. As the number of low signal spike-ins was relatively small (265 probe sets), resulting in reduced accuracy for the ROC curves, the development of additional control datasets specifically focusing on DEG detection at low cRNA concentrations will be an important extension of this study.
Models dependent on probe sequence provide a promising route to improving the accuracy of nonspecific signal measures. Here, we applied two different models (perfect match and gcrma) to the control dataset. With respect to detecting the true DEGs, these two models perform reasonably well, although slightly less well than the MAS 5.0 PM correction method. When we consider only the low signal DEGs (Additional data file 5), gcrma outperforms perfect match, and is similar in effectiveness to the top analysis option combinations.
Estimating false discovery rates
We have identified a set of analysis choices that optimally ranks genes according to significance of differential expression. To decide how many of the top genes to investigate further in follow-up experiments, it would be useful to have accurate estimates of the false-discovery rate (FDR or q-value), which is the fraction of false positives within a list of genes exceeding a given statistical cutoff. We used our control dataset to compare the actual q-values for the 10 optimal expression summary datasets with q-value estimates from the permutation method implemented in SAM. As shown in Figure 8b, permutation-based q-value calculations using each of the top ten datasets underestimate the actual q-value for a given cutoff. We attempted to reduce the contribution of biases inherent in any given data-processing step by combining the results from the top 10 expression summary datasets. The goal is to pinpoint those genes that are called significant regardless of small changes in the analysis protocol (changes that only marginally affect the DEG detection sensitivity and specificity according to our control dataset). To identify these 'robustly significant' genes, we created a combined statistic from the top 10 datasets depicted in Figure 7f, taking into account the significance of each individual test, as well as the variation in fold change between datasets (see Materials and methods). This combined statistic distinguishes between true and false DEGs equally as well as the best of the 10 input datasets (Figure 8a). To make false-discovery rate estimates using this combined statistic, each of the 10 datasets was permuted (using the same permutation) and the combined statistic was recalculated. Figure 8b shows that this combined statistic gives a more accurate q-value estimate than any of the individual datasets. However, there is still considerable difference between the estimated and actual q-values. For example, if we estimate q = 0.05, the corresponding CyberT statistic has an actual q = 0.18, and if we estimate q = 0.1, then the actual q = 0.3. Therefore, until more accurate methods for estimating the false-discovery rate are developed, we recommend that a conservative choice of false-discovery rate cutoff be used (for example < 1%) to prevent actual numbers of false-positive DEG calls (that is, the true, rather than estimated, FDR) from being too high.
Assessment of sensitivity and specificity
As the identities and relative concentrations of each of the RNAs in the experiment were known, we were able to assess directly the sensitivity and specificity obtained by the best-performing methods. Examination of the ROC curves in Figure 7 reveals that sensitivity begins to plateau as the false discovery rate (q) increases from 10% to 30%. Taking an upper acceptable bound for q as 10%, the maximum sensitivity obtained is about 71%. Thus, under the best-performing analysis scheme, roughly 380 (29%) of the 1,309 DEGs are not detected as being differentially expressed, with the number of false positives equaling about 105. At q = 2%, sensitivity reduces to around 60%, meaning that more than 520 DEGs are missed, albeit with fewer than 20 false positives.
We next looked at the dependence of sensitivity and specificity on the magnitude of the spiked-in fold-change value. We find that at q = 10%, sensitivity is increased to 93% when only cRNAs that differ by twofold or more are considered as DEGs (Figure 9a). This sensitivity decreases only slightly (to 90%) when q is lowered to 5%. However, sensitivity drops off sharply as differences in expression below twofold are considered. At q = 10%, only 82% of DEGs with 1.5-fold or greater changes in expression are identified, dropping to 71% for all DEGs at 1.2-fold change or above (77% and 67% at q = 5%, respectively). The reduction in sensitivity is almost wholly due to the low-fold-change genes: less than 50% of DEGs with fold change 1.5, and none of the DEGs with fold change 1.2, are detected at q = 10% (Figure 9b).
It is tempting to conclude from this that we are achieving adequate sensitivity in our experiments and merely need not bother with DEGs below the twofold change level. However, we would argue that obtaining greater sensitivity should be an important goal. There is ample demonstration in the biological and medical literature that small changes in gene expression can have serious phenotypic consequences, as seen both from haploinsufficiencies and from mutations that reduce levels of gene expression through transcriptional regulation or effects on mRNA stability. Furthermore, effective fold changes seen in a microarray experiment might be considerably smaller than actual fold changes within a cell, if the sample contains additional cell populations that dilute the fold-change signal. As it is often not possible to obtain completely homogeneous samples (for example, when profiling an organ composed of several specialized cell types), this is likely to prove a very real limitation to detecting DEGs. In cases where pure cell populations can be obtained, for example by laser capture microdissection, the numbers of cells are often small and RNA needs to undergo amplification in order to have enough for hybridization. Here, non-linearities in RNA amplification might also lead to observed fold changes that fall below the twofold level. We used three microarray replicates for this study, as this is frequently the number chosen by experimentalists because of cost and limiting amounts of RNA. One possible extension of this work would be to examine how many replicates are necessary for reliable detection of DEGs at a given fold change level.
Conclusions
We have compared a number of popular analysis options for the purpose of identifying differentially expressed genes using an Affymetrix GeneChip control dataset. Clear differences in sensitivity and specificity were observed among the analysis method choices. By trying all possible combinations of options, we could see that choices at some steps of analysis are more critical than at others; for example, the normalization methods that we considered perform similarly, whereas the choice of the PM adjustment method can strongly influence the accuracy of the results. On the basis of our observations, we have chosen a best route for finding DEGs (Figure 3). As any single choice of analysis methods can introduce bias, we have proposed a way to combine the results from several expression summary datasets in order to obtain more accurate FDR estimates. However, these estimates remain substantially lower than actual false-discovery rates, demonstrating the need for continued development of ways to assess the false-discovery rate in experimental datasets. Our analysis further revealed the existence of a high false-negative rate (low sensitivity), especially for those DEGs with a small fold change, and thus suggests the need for improved analysis methods for Affymetrix microarrays. In order to be feasible, this study investigated only a fraction of the current options. The raw data from our hybridizations are available in Additional data files 6-7 and on our websites [27,28], and we encourage the use of this dataset for benchmarking existing and future algorithms. Also important will be the construction of additional control datasets to explore issues not well covered by the present study, such as performance of the analysis methods for specifically detecting low-abundance RNAs and the effects of including larger numbers of replicate arrays. We hope that these experiments will help researchers to choose the most effective analysis routines among those available, as well as guide the design of new methods that maximize the information that can be obtained from expression-profiling data.
Materials and methods
cRNA and hybridization
PCR products from Drosophila Gene Collection release 1.0 cDNA clones [16] were generated in 96-well format, essentially as described [29]. Each PCR product includes T7 and SP6 promoters located 5' and 3' to the coding region of the cDNA, respectively. Each PCR reaction was checked by gel electrophoresis for a band of detectable intensity and the correct approximate size. Those clones which did not yield PCR product were labeled as 'failed' and eliminated from subsequent analysis. From sequence verification of randomly selected clones, we estimate the number of mislabeled clones to be < 3%. The contents of the plates were collected into 19 pools, such that each pool contained the PCR product from one to four plates (approximately 96-384 clones). Biotinylated cRNA was generated from each pool using SP6 polymerase (detailed protocol available upon request) and the reactions were purified using RNeasy columns (Qiagen). Concentration and purity for each pool was determined both by spectrophotometry and with an Agilent Bioanalyzer. The labeled products were then divided into each of two samples - constant (C) and spike (S) - at specific relative concentrations (Table 1, Figure 1). Because the C sample contains less total RNA than the S sample, 20 μg of (unlabeled) poly(C) RNA was added to the C sample to equalize the nucleic acid concentrations. By mixing the labeled pools just before hybridization, we ensured that the fold change between C and S is uniform for all RNAs within a single pool, while still allowing the absolute concentrations of individual RNAs to vary. The two samples were then hybridized in triplicate to Affymetrix Drosophila arrays (DrosGenome1) using standard Affymetrix protocols. We chose to hybridize each replicate chip from an aliquot of a single C (or S) sample, resulting in technical replication; thus this dataset does not address the noise introduced by the labeling and mixing steps. The clones comprising each pool can be found in Additional data file 8, and the resulting Affymetrix chip intensity files (.CEL) files are available in Additional data files 6-7.
Estimate of RNA concentrations
The total amount of labeled cRNA that was added to each chip (approximately 18 μg) was comparable to a typical Affymetrix experiment (20 μg). Although we do not know the individual RNA concentrations, we estimate that these span the average RNA concentration in a biological GeneChip experiment. Our biological RNA samples typically result in about 40% of the probe sets on the DrosGenome1 chip called present, so the mean amount of individual RNA is 20 μg/(14,010 × 0.40) = 0.003 μg/RNA. In the C chips, the average concentration of individual RNAs in the different pools range from 0.0008 to 0.007 μg/RNA, so the concentrations are roughly similar to those in a typical experiment. We note, however, that there is no way to ensure that the concentration distribution is truly reflective of a real RNA distribution. This is especially true with respect to the low end of the range, as it is usually unknown how many of the absent genes on an array are truly absent versus weakly expressed and thus poorly detected by the analysis algorithms used. Therefore, our analysis possibly favors methods that perform best when applied to highly expressed genes.
Software
All of the analysis was performed using the statistical program R [30], including the affy and gcrma packages from Bioconductor [18], and scripts adapted from the hdarray library by Baldi et al. [31,32]. In addition, we used the dChip [19], MAS 5.0 [12], Perfect Match [20,21] and SAM [27] executables made available by the respective authors. Note that the false-discovery rate calculations were slightly different depending on the t-statistic variant: for the SAM statistic, false discovery rates from the authors' Excel Add-in software was used, whereas for the CyberT and basic t-statistics, the Bioconductor false-discovery rate implementation was applied, which includes an extra step to enforce monotonicity of the ROC curve. In our experience, this extra step does not qualitatively alter the results. All scripts generated in this study are available for use [27,28].
Calculation of the statistic that combines the results of multiple expression summary datasets
Say we have n datasets and Cij, Sij are the logged signals for a given probe set in the jth C and S chips, respectively, in dataset i. The mean signal (for this probe set) for the C chips in dataset i is:
where is the number of C chips in dataset i; similarly, the mean signal for the S chips in dataset i is:
The mean fold change over all datasets is:
The modified standard deviation for the C chips in dataset i is based on the CyberT estimate:
where const is the weight for the contribution of the average standard deviation for probe sets with the same average signal intensity as Cij. The modified standard deviation for the S chips in dataset i (sd.Si) is defined analogously. The pooled variance over all 10 datasets is defined as:
The variance between the 10 datasets is defined as:
Then the combined statistic was chosen to be:
Additional data files
Additional data is available with the online version of this article. Additional data file contains a figure and explanatory legend showing the degree of overlap between two lists of differentially expressed genes. Additional data file 2 lists all analysis option combinations used to generate the expression summary datasets in this study. Additional data file 3 is a plot of observed vs actual spiked-in fold changes at the probe level. Additional data file 4 shows an example of asymmetric M (log2 fold change) vs A (average log2 signal) plot for the comparison of two biological samples. Additional data file 5 contains a comparison of the analysis methods with respect to the detection of DEGs with low signal. Additional data file 6 is a Zip archive containing plain text files (in Affymetrix CEL format), Affymetrix *.CEL files for the C chips in this dataset. Additional data file 7 is a Zip archive containing plain text files (in Affymetrix CEL format), Affymetrix *.CEL files for the S chips in this dataset. Additional data file 8 contains detailed information for the individual DGC clones used in this study.
Supplementary Material
Additional data file 1
A figure and explanatory legend showing the degree of overlap between two lists of differentially expressed genes
Click here for additional data file
Additional data file 2
All analysis option combinations used to generate the expression summary datasets in this study
Click here for additional data file
Additional data file 3
A plot of observed vs actual spiked-in fold changes at the probe level
Click here for additional data file
Additional data file 4
An example of asymmetric M (log2 fold change) vs A (average log2 signal) plot for the comparison of two biological samples
Click here for additional data file
Additional data file 5
A comparison of the analysis methods with respect to the detection of DEGs with low signal
Click here for additional data file
Additional data file 6
A Zip archive containing plain text files (in Affymetrix CEL format), Affymetrix *.CEL files for the C chips in this dataset
Click here for additional data file
Additional data file 7
A Zip archive containing plain text files (in Affymetrix CEL format), Affymetrix *.CEL files for the S chips in this dataset
Click here for additional data file
Additional data file 8
Detailed information for the individual DGC clones used in this study
Click here for additional data file
Acknowledgements
We thank M. Ramoni and M. Morrissey for helpful comments on the manuscript, K. Kerr and A. Wohlheuter for assistance with, and N. Perrimon for resources for, the PCR, the HMS Biopolymers facility for assistance with robotics and GeneChip hybridization, and B. Estrada and L. Raj for sharing the data depicted in Additional Data Files 1 and 4, respectively. S.E.C. was supported by a PhRMA Foundation CEIGI grant, a Brigham and Women's Research Council bioinformatics grant, and NIH fellowship F32 GM67483-01A1. A.M.M. is an Associate Investigator of the Howard Hughes Medical Institute. G.M.C. is supported by a PhRMA Foundation CEIGI grant. M.S.H. is supported by NIH grant K22-HG002489.
Figures and Tables
Figure 1 Schematic depiction of the experimental protocol.
Figure 2 Signal of individual probes and dependence on present versus absent RNA molecules. (a, b) Plot of probe-pair signals for the three C chips, highlighting (a) the empty probe pairs or (b) the present probe pairs in green. (c) Receiver-operator characteristic (ROC) curves at the probe-pair level for several absent/present metrics. The metric (PM - MM)/(PM + MM) gives the same result as the green curve. (d)Receiver-operator characteristic curves at the probe-set level for several absent/present metrics combined using the Wilcoxon rank sum test.
Figure 3 The set of options that were investigated using Bioconductor's affy package. The choices that optimize the detection of DEGs are circled in red. Broken circles indicate choices that are slightly suboptimal but still rank within the top 10 datasets.
Figure 4 The dependence of log fold change on signal intensity (M versus A plots). (a)M versus A plot before the second normalization step and (b) after a loess fit at the probe set level. FC in the key denotes the spiked-in fold change value.
Figure 5 Correlation of observed with actual fold changes for a representative expression summary dataset (Additional data file 2, using dataset 9e.b). (a) The fold change for each probe set with spiked-in target RNA is depicted as a cross. Empty probe sets are not shown. For each actual fold-change level (on the x axis), a boxplot shows the distribution of the corresponding observed fold changes. A linear fit of the data is shown in cyan. Fit parameters: R2 = 0.508; slope = 0.505; y-intercept = -0.061. (b-d) Increasingly more of the low-intensity probe sets are filtered out of the plot. All probe sets are ranked according to average signal level, and those in the lowest 25th (b), 50th (c), or 75th (d) percentile of signal level are eliminated from (a). Fit parameters: (b) R2 = 0.870; slope = 0.546; y-intercept = -0.008; (c) R2 = 0.895; slope = 0.517; y-intercept = -0.015; (d) R2 = 0.906; slope = 0.457; y-intercept = -0.017.
Figure 6 Comparison of three t-statistic variants. (a)ROC curves for a particular expression summary dataset, using the different t-statistics. Location of false positives and false negatives are shown for the (b) CyberT, (c) SAM, and (d) basic t-statistic when considering the top 1,000 probe sets as positive DEG calls.
Figure 7 ROC curves for all expression summary datasets. The curves are color-coded to highlight how the ability to detect differential expression is dependent on the different options at each step of analysis, using the CyberT regularized t-statistic metric. (a) All 152 expression summary datasets are represented here, with the different colors depicting whether the second loess normalization step at the probe set level was performed. In general, the second loess normalization (blue) improves the detection of true DEGs. (b-f)To decrease clutter, only the 76 expression summary datasets involving the second normalization step are shown. (b) When comparing the two background correction methods, the MAS algorithm is superior to the RMA algorithm. (c) The various probe-level normalization methods do not show great differences between each other. (d) Among the different PM-correction options, using the method in MAS 5.0 clearly is the most successful. (e) Various robust estimators were examined, revealing that the median polish method is the most sensitive (with MAS 5.0's Tukey Biweight a close second). (f) Depiction (in blue and orange) of the 10 datasets which maximize detection of truly differentially expressed genes, while minimizing false positives. These datasets are generated using the options circled in Figure 3. MAS 5.0, with the inclusion of the second loess normalization step, falls within these top 10.
Figure 8 The accuracy of false discovery rate estimates (q-values). The top 10 expression summary datasets (named 9a-9e, 10a-10e in Additional data file 2) were combined to generate a composite statistic, which was used to rank genes based on the robustness of their significance over the 10 datasets. (a) The composite statistic performs as well as the best summary dataset in terms of sensitivity and specificity. (b) In addition, permutation tests carried out using this composite statistic yield q-value estimates which are more accurate than any of the 10 component datasets, although still lower than the true false-discovery rate.
Figure 9 DEG detection sensitivity and specificity as a function of spiked-in fold change level. (a, b)ROC curves using the composite statistic, and different definitions of the true-positive probe sets (criteria given in the legends; FC, spiked-in fold change). The true negatives remain the same for all curves (the probe sets which were not spiked in, or were spiked in at 1x).
Table 1 The number of clones and assigned fold change for each pool of PCR products
Pool number Number of clones Number of assigned Affymetrix probe sets Assigned fold change (S vs C) Amount of RNA added to each C chip (μg) Amount of RNA added to S chip (μg)
1 87 84 1.2 0.47 0.56
2 141 143 2 0.43 0.85
3 85 83 1.5 0.35 0.52
4 180 185 2.5 0.73 1.82
5 90 89 1.2 0.29 0.35
6 88 96 3 0.65 1.94
7 186 188 3.5 0.76 2.67
8 90 95 1.5 0.44 0.67
9 180 190 4 0.78 3.11
10 183 191 1.7 0.48 0.81
13 391 385 1 0.37 0.37
14 369 355 1 1.23 1.23
15 394 404 1 0.40 0.40
16 452 453 1 0.57 0.57
17 419 434 1 0.44 0.44
18 372 407 1 0.31 0.31
19 163 191 1 0.27 0.27
Also depicted is the total amount of cRNA for each pool that was placed on each chip, and the number of Affymetrix probe sets that are assigned to each pool. There were 10,131 probe sets not assigned to any spiked-in clone (called empty). Pools 11 and 12 were not included in this dataset.
==== Refs
Lockhart DJ Dong H Byrne MC Follettie MT Gallo MV Chee MS Mittmann M Wang C Kobayashi M Horton H Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 1996 14 1675 1680 9634850 10.1038/nbt1296-1675
Schena M Shalon D Davis RW Brown PO Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 1995 270 467 470 7569999
Parmigiani G Garrett ES Irizarry RA Zeger SL The analysis of gene expression data. 2003 New York: Springer Verlag
Barash Y Dehan E Krupsky M Franklin W Geraci M Friedman N Kaminski N Comparative analysis of algorithms for signal quantitation from oligonucleotide microarrays. Bioinformatics Adv Access 2004 1 1
Irizarry RA Bolstad BM Collin F Cope LM Hobbs B Speed TP Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003 31 e15 12582260 10.1093/nar/gng015
Rajagopalan D A comparison of statistical methods for analysis of high density oligonucleotide array data. Bioinformatics 2003 19 1469 1476 12912826 10.1093/bioinformatics/btg202
Lemon WJ Liyanarachchi S You M A high-performance test of differential gene expression for oligonucleotide arrays. Genome Biol 2003 4 R67 14519202 10.1186/gb-2003-4-10-r67
He YD Dai H Schadt EE Cavet G Edwards SW Stepaniants SB Duenwald S Kleinhanz R Jones AR Shoemaker DD Microarray standard dataset and figures of merit for comparing data processing methods and experiment designs. Bioinformatics 2003 19 956 965 12761058 10.1093/bioinformatics/btg126
Broberg P Statistical methods for ranking differentially expressed genes. Genome Biol 2003 4 R41 12801415 10.1186/gb-2003-4-6-r41
Affymetrix - Latin square data
Scientific studies
Affymetrix: technical support documentation
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 10.1093/nar/30.4.e15
Baldi P Long AD A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes. Bioinformatics 2001 17 509 519 11395427 10.1093/bioinformatics/17.6.509
Affymetrix - Drosophila genome array
BDGP: Drosophila gene collection
Affymetrix - Statistical Algorithms Description Document
Bioconductor
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
Zhang L Miles MF Aldape KD A model of molecular interactions on short oligonucleotide microarrays. Nat Biotechnol 2003 21 818 821 Corrigendum: Nat Biotechnol 2003, 21:941. 12794640 10.1038/nbt836
Wu Z Irizarry RA Stochastic models inspired by hybridization theory for short oligonucleotide arrays. Proc 8th Conf Res Comput Mol Biol 2004 New York: ACM Press 98 106
Irizarry RA Hobbs B Collin F Beazer-Barclay YD Antonellis KJ Scherf U Speed TP Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003 4 249 264 12925520 10.1093/biostatistics/4.2.249
Schadt EE Li C Ellis B Wong WH Feature extraction and normalization algorithms for high-density oligonucleotide gene expression array data. J Cell Biochem Suppl 2001 Suppl 37 120 125 11842437 10.1002/jcb.10073
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 10.1093/bioinformatics/19.2.185
Chudin E Walker R Kosaka A Wu SX Rabert D Chang TK Kreder DE Assessment of the relationship between signal intensities and transcript concentration for Affymetrix GeneChip arrays. Genome Biol 2002 3 research0005.1 0005.10 11806828 10.1186/gb-2001-3-1-research0005
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001 98 5116 5121 11309499 10.1073/pnas.091062498
The Golden Spike Experiment
Assessment of microarray analysis methods
BDGP Resources: PCR amplification of cDNAs from bacterial cultures: DGC/pOT2
The R Project for statistical computing
Welcome to Cyber-T
Downloading and installing Cyber-T / hdarray (R code)
| 15693945 | PMC551536 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 28; 6(2):R16 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r16 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r171569394610.1186/gb-2005-6-2-r17MethodIdentification of ciliated sensory neuron-expressed genes in Caenorhabditis elegans using targeted pull-down of poly(A) tails Kunitomo Hirofumi [email protected] Hiroko [email protected] Yuji [email protected] Yuichi [email protected] Molecular Genetics Research Laboratory, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan2 Genome Biology Laboratory, National Institute of Genetics, Mishima 411-8540, Japan2005 31 1 2005 6 2 R17 R17 17 9 2004 29 11 2004 21 12 2004 Copyright © 2005 Kunitomo 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.
An mRNA-tagging method was used to selectively isolate mRNA from a small number of cells for subsequent cDNA microarray analysis. The approach was used to identify genes specifically expressed in ciliated sensory neurons of Caenorhabditis elegans.
It is not always easy to apply microarray technology to small numbers of cells because of the difficulty in selectively isolating mRNA from such cells. We report here the preparation of mRNA from ciliated sensory neurons of Caenorhabditis elegans using the mRNA-tagging method, in which poly(A) RNA was co-immunoprecipitated with an epitope-tagged poly(A)-binding protein specifically expressed in sensory neurons. Subsequent cDNA microarray analyses led to the identification of a panel of sensory neuron-expressed genes.
==== Body
Background
Recent advances in technologies for analyzing whole-genome gene-expression patterns have provided a wealth of information on the complex transcriptional regulatory networks and changes in gene-expression patterns that are related to phenotypic changes caused by environmental stimuli or genetic alterations. Changes in gene expression are also fundamental during development and cellular differentiation, and differences in gene expression lead to different cell fates and eventually determine the structural and functional characteristics of each cell type. Comparative analyses of gene-expression patterns in various cell types will therefore provide a framework for understanding the molecular architecture of these cells as cellular systems.
Caenorhabditis elegans is an ideal model organism for investigating development and differentiation at high resolution, because adult hermaphrodites only have 959 somatic nuclei, whose cell lineages are all known. About 19,000 genes were identified by determination of the C. elegans genome sequence [1]. Functional genomic approaches, including systematic inhibition of gene functions by RNA interference [2-5], large-scale identification of interacting proteins [6], systematic generation of deletion mutants [7-9], and determination of the time and place of transcription [10-12], are currently in progress to accumulate information on all genes in the genome.
Genome-wide gene-expression profiling using DNA or oligonucleotide microarray technology has also been applied to this organism. Microarrays containing more than 90% of C. elegans genes have been constructed and used in global gene-expression analyses under a wide variety of developmental, environmental and genetic conditions [13-15]. Genome-wide gene expression analyses of the germline have also been carried out [16,17]. Mutants lacking functional gonads and those with masculinized or feminized gonads were used in these studies to identify germline-expressed genes and genes correlated with the germline sexes.
To analyze gene-expression patterns in various cells, particularly those forming small tissues, selective isolation of mRNA from these cells is necessary. As an example of this approach, mRNA was prepared from mechanosensory neurons after cell culture of their embryonic precursors followed by selection of the cells by flow cytometry [18]. Although embryonic cell cultures allow the collection of cells at early stages of development, methods for the separation, culture and collection of fully developed tissues have not been established and might be technically difficult.
C. elegans modifies its behavior by sensing environmental cues such as food, chemicals, temperature or pheromones. These cues are recognized by approximately 50 sensory neurons positioned in the head and tail. Although the overall functions of the chemosensory or thermosensory neurons have been examined by laser-killing experiments, the molecular mechanisms that underlie the functions of each sensory neuron have not yet been fully explored. Profiling of genes that are expressed in sensory neurons might therefore provide insights into the genes required for the specific functions of neurons.
To identify sensory neuron-expressed genes, we adopted the mRNA-tagging method [19]. In this method, poly(A)-binding protein (PABP), which binds the poly(A) tails of mRNA, is utilized to specifically pull-down poly(A) RNA from the target tissues. By employing this method, we successfully identified novel genes that are expressed in the ciliated sensory neurons of C. elegans.
Results
Preparation of mRNA from particular types of neurons using mRNA tagging
To isolate sensory neuron-expressed transcripts, we devised a method that utilizes PABP. This approach involves the generation of transgenic animals that express an epitope-tagged PABP using cell-specific promoters. Since PABP binds the poly(A) tails of mRNA [20], in situ crosslinking of RNA and proteins, followed by affinity purification of the tagged PABP from lysates of these animals, is expected to co-precipitate all the poly(A)+ RNA from cells expressing the tagged PABP (Figure 1). This method was independently devised by Roy et al. and used to identify muscle-expressed genes [19], but whether the procedure was applicable to smaller tissues, such as neurons, was unknown. We applied this technology, mRNA tagging [19], to the ciliated sensory neurons of C. elegans; these comprise approximately 50 cells whose cell bodies are typically 2 μm in diameter compared to the approximate animal body length of 1 mm.
PABP is encoded by the pab-1 gene in C. elegans. Nematode strains expressing FLAG-tagged PAB-1 from transgenes were generated using tissue-specific promoters. To prepare mRNA from sensory neurons, we generated the JN501 strain (hereafter called che-2::PABP) in which the transgene was expressed in most of the ciliated sensory neurons using a che-2 gene promoter [21]. A second strain, JN502 (acr-5::PABP), was generated to prepare mRNA from another subset of neurons using an acr-5 promoter, which is active in B-type motor neurons, as well as unidentified head and tail neurons [22]. A third strain, JN503 (myo-3::PABP), which expressed the transgene in non-pharyngeal muscles using the myo-3 promoter [23], was generated to serve as a non-neuronal control. Expression of FLAG-PAB-1 was confirmed by western blotting analyses, and immunohistochemistry using an anti-FLAG antibody (data not shown). Expression patterns were essentially the same as those reported for the promoters used, but we note that expression of FLAG-PAB-1 in ventral cord motor neurons was weak in the acr-5::PABP strain compared to that in sensory neurons in the che-2::PABP strain. As a measure of the functional integrity of FLAG-PAB-1-expressing cells, responses of the che-2::PABP strain to the volatile repellent 1-octanol, which is sensed by ASH amphid sensory neurons was tested. The sensitivity of the che-2::PABP animals was indistinguishable from the wild type (data not shown). The ability of the exposed sensory neurons to absorb the lipophilic dye diQ was also tested. Amphid sensory neurons in the head stained normally, whereas phasmid neurons, PHA and PHB, in the tail showed weak defects in dye-filling (90% staining of PHA and 91% staining of PHB, compared to 100% in wild type for both neurons). The acr-5::PABP and myo-3::PABP strains appeared to move normally, suggesting overall functional integrity of motor neurons and body-wall muscles, respectively.
Poly(A) RNA/FLAG-PAB-1 complexes were pulled-down from whole lysates of these transgenic worms using anti-FLAG monoclonal antibodies. Poly(A) RNA was then extracted and concentrated. The amounts of known tissue-specific transcripts were examined by reverse transcription PCR (RT-PCR) (Figure 2). The mRNA for tax-2, which is expressed in a subset of sensory neurons [24], was enriched in RNA from che-2::PABP. The mRNA for odr-10, which is expressed in only one pair of sensory neurons [25], was also highly enriched in che-2::PABP. On the other hand, mRNA for acr-5 and del-1, both of which are expressed in B-type motor neurons [22], was enriched in RNA from acr-5::PABP. The mRNA for unc-8, which is expressed in motor neurons and ASH and FLP sensory neurons in the head [26], was contained in RNA from both che-2::PABP and acr-5::PABP. The mRNA for unc-54, which is expressed in muscles [23], was enriched in RNA from myo-3::PABP. Representatives of housekeeping genes, eft-3 [27] and lmn-1 [28], were detected in RNA from all transgenic strains. Quantitative RT-PCR was performed to estimate the relative amounts of neuron type-specific transcripts. The amount of the odr-10 transcript in RNA from che-2::PABP was 39-fold higher than that from acr-5::PABP, and mRNA for gcy-6, which is expressed in only a single sensory neuron [29], was enriched 10-fold. On the other hand, the mRNA for acr-5 was enriched eightfold in RNA from acr-5::PABP compared with that from che-2::PABP. mRNA for the pan-neuronally expressed gene snt-1 [30] was equally represented in RNA from both acr-5::PABP and che-2::PABP. Therefore, selective enrichment of sensory neuron-, motor neuron- and muscle-expressed genes in RNA from che-2::PABP, acr-5::PABP and myo-3::PABP strains, respectively, have been achieved as intended. Of these, the enrichment of motor neuron-expressed genes appeared less efficient, because weak bands were sometimes seen for these genes in RT-PCR from che-2::PABP or myo-3::PABP RNA.
cDNA microarray experiments
We used a cDNA microarray to compare the properties of mRNA prepared from che-2-expressing ciliated sensory neurons with that from acr-5-expressing cells. RNA purified from che-2::PABP was labeled with Cy5 and that from acr-5::PABP was labeled with Cy3. The two types of labeled RNA were mixed and hybridized to the cDNA microarray and the che-2::PABP/acr-5::PABP (Cy5/Cy3) ratio was calculated for each cDNA spot. The cDNA microarray contained 8,348 cDNA spots corresponding to 7,088 C. elegans genes. Two sets of independently prepared RNA samples were hybridized to two separate arrays. The logarithm of the hybridization intensity ratio for each spot, log2(che-2::PABP/acr-5::PABP), was calculated and values from the two experiments were averaged. This calculation allowed us to order the genes represented on the microarrays according to the log2(che-2::PABP/acr-5::PABP) value (see Additional data file 1). Genes specifically expressed in che-2-expressing cells should have higher rank orders in this list, whereas those expressed in acr-5-expressing cells should have lower rank orders.
To evaluate the results of the microarray experiments, we searched for genes that are known to be expressed in amphid sensory neurons, but not in ventral cord motor neurons, or vice versa, using the WormBase database (WS94). Of these, 20 sensory neuron-specific genes and five motor neuron-specific genes were present on the arrays (see Additional data files 1 and 2). These genes showed a highly uneven distribution, with sensory neuron-specific genes concentrated in the highest rank orders and motor neuron-specific genes distributed in lower rank orders (Figure 3a). Muscle-expressed genes (also found using WormBase) were almost evenly distributed. However, intestine-expressed genes were concentrated in the lower rank orders. These results demonstrate that our mRNA isolation procedure specifically enriched ciliated sensory neuron- and motor neuron-expressed genes as intended. The unexpected distribution of the intestine-expressed genes will be discussed later.
daf-19 encodes a transcription factor similar to mammalian RFX2. Several genes expressed in ciliated sensory neurons and essential for ciliary morphogenesis, such as che-2 and osm-6, are under the control of daf-19 and have one or more copies of the cis-regulatory element X-box in their promoter regions [31]. We therefore examined the distribution of genes that harbor X-boxes in their promoter regions. Again, the distribution of X-box-containing genes was highly uneven (Figure 3b, see also Additional data files 1 and 2), further demonstrating the successful enrichment of ciliated neuron-expressed genes.
Expression analysis of candidate sensory neuron-expressed genes by reporter fusions
The above analyses showed that sensory neuron-expressed genes were enriched in the mRNA population purified from che-2::PABP. However, only a few genes were previously known to be expressed in these tissues. In fact, the expression patterns for most top-ranked genes in our list were not known. To determine which of these genes were actually expressed in sensory neurons, we examined the expression patterns of 17 genes with the highest rank orders using translational green fluorescent protein (GFP) fusions. The expression patterns for these genes had not been reported previously.
We did not observe any GFP fluorescence for two clones, K07B1.8 and C13B9.1, probably because the promoter region we selected did not contain all the functional units or expression was below the level of detection. GFP-expressing cells were identified for all the remaining 15 genes (Figure 4, Table 1). For 13 of these GFP fusions, expression was observed in ciliated sensory neurons, namely amphid, labial and/or phasmid sensory neurons. Of these, expression in the intestine, in addition to the sensory neurons, was observed for Y55D5A.1a and T07C5.1c, whereas expression of K10D6.2a was also observed in seam cells and the main body hypodermis (hyp7). Expression of K10G6.4 was observed in many other neurons in addition to sensory neurons. Expression in the intestine and coelomocytes, but not in sensory neurons, was observed for two other clones, C35E7.11 and F10G2.1, respectively. In summary, of the 15 genes whose expression patterns could be determined, 13 (87%) were expressed in sensory neurons. These results showed that most of the genes with the highest rank orders were expressed in ciliated sensory neurons.
We also examined the expression patterns of two genes with the lowest rank orders (Y44A6D.2 and T08A9.9/spp-5). Expression in the ventral nerve cord was observed for Y44A6D.2, while only weak expression in the intestine was observed for T08A9.9 (data not shown). These results also suggested that our procedure was somewhat less effective in enriching motor neuron-expressed genes than sensory neuron-expressed genes (Figure 3a).
Categorization of che-2::PABP-enriched genes reveals specific features
In an attempt to characterize ciliated sensory neuron-expressed genes as a set, we first referred to functional annotations of each gene generated by the WormBase. It was noted that the fraction of genes with functional annotations was smaller for the highest ranked genes (Figure 5a). BLASTP searches of the nonredundant (nr) protein sequence database and proteome datasets for several representative animal and yeast species showed that nematode-specific genes were enriched, while those with homologs in yeast and other animals tended to be under-represented in the top-ranked genes (Figure 5b,c).
Among the genes with Gene Ontology (GO) annotations, top-ranked genes showed a significantly larger fraction with a 'nucleic acid binding' functional capacity (P = 0.004, Figure 6). Protein motifs found to be enriched among the che-2::PABP-enriched genes included 'cuticle collagen', 'chromo domain', 'linker histone' and 'laminin G domain'.
Another prominent characteristic of the che-2::PABP-derived mRNA fraction was enrichment of genes homologous to nephrocystins. Nephrocystins are responsible for a hereditary cystic kidney disease, nephronophthisis, and to date, nephrocystin 1 (NPHP1) through nephrocystin 4 (NPHP4) have been identified [32-35]. C. elegans homologs of NPHP1 and NPHP4 were ranked at positions 15 and 25 in our list, suggesting a link between these disease genes and the functions of worm sensory neurons.
Discussion
Preparation of mRNA from a subset of neurons in C. elegans
We prepared poly(A) RNA from a subset of neurons using the mRNA-tagging technique. The genome-wide identification of muscle-expressed genes demonstrated that mRNA tagging is a powerful technique for collecting tissue-specific transcripts in C. elegans [19]. The method is especially useful in this organism because dissection and separation of the tissues are difficult because of the worm's small size and the presence of cuticles. However, it was not known whether this method was applicable to smaller tissues, such as subsets of neurons. In this study, we attempted to isolate mRNA from ciliated sensory neurons using mRNA tagging. Although the volume of target neurons was much smaller than that of muscles, transcripts of various sensory neuron-expressed genes, ranging from those expressed in many sensory neurons to those expressed in only one or two sensory neurons, were successfully enriched.
The procedure of mRNA tagging is based on immunoprecipitation of poly(A)-RNA/FLAG-PAB-1 complexes. A potential problem with this technique is that once the cells are broken, poly(A) RNA released from non-target cells might bind unoccupied FLAG-PAB-1. To reduce this possibility, we adopted stringent washing conditions in addition to in situ formaldehyde crosslinking. Although this procedure reduced the recovery of immunocomplexes, it ensured minimal contamination by mRNA from non-target cells. As there are many characterized promoters that can deliver FLAG-PAB-1 to small numbers of neurons in C. elegans, profiling of the gene-expression pattern of each type of neuron should be possible with this technique.
Another potential problem with this method is that PABP might have different binding affinities for different transcript species, rendering some tissue-specific transcripts difficult to recover. Although PABP binds tightly to the poly(A) tails of most mRNA [36], RNA species co-immunoprecipitated with PABP from cultured cells do not represent the total RNA of the cells [37]. This might also cause another problem in that transcripts with strong PABP affinity might be undesirably enriched in the precipitates and cause unexpected biases.
Analysis of purified mRNA using a cDNA microarray
Preparation and characterization of EST clones led to the identification of more than 10,000 cDNA groups corresponding to different genes of C. elegans ([12,38] and Y.K., unpublished results). We used a cDNA microarray on which such cDNA clones were spotted to identify the genes expressed in ciliated sensory neurons. Using a cDNA microarray rather than a genome DNA microarray has the advantage that genes on the array have guaranteed expression, and hybridization to the corresponding mRNA species is efficient. The microarray we used contained 7,088 genes of C. elegans, representing 40% of the predicted genes on the genome [1]. On the other hand, there are also genes that were not represented in our cDNA collection, including characterized sensory neuron-specific genes such as osm-6 [39] and most seven-transmembrane receptor genes including odr-10; this might be a disadvantage of using a cDNA microarray. Acquisition of cDNA clones for rare mRNA species and use of whole-genome microarrays are complementary approaches for improving the applicability of the method described here.
Evaluation of microarray experiments
Previously known sensory neuron- and motor neuron-expressed genes were used to evaluate the results of our microarray analyses. Most genes were enriched in our che-2::PABP-derived mRNA preparations or in the acr-5::PABP-derived mRNA preparations depending on their expression patterns. However, several genes were not enriched as expected. Furthermore, enrichment of motor neuron-expressed genes in the acr-5::PABP-derived mRNA preparations appeared less efficient. The reasons for these occurrences are unknown, but the expression of FLAG-PAB-1 in motor neurons were low in the acr-5::PABP strain, which could account for the low efficiency of enrichment for this tissue. Another potential problem is that the expression pattern of the acr-5 promoter has not been fully characterized [22], and both the che-2 and acr-5 promoters are active in labial neurons, where expression of the acr-5 promoter was relatively strong compared to motor neurons (data not shown). Genes expressed in the intestine were enriched in the acr-5::PABP-derived mRNA preparations. FLAG-PAB-1 was weakly expressed in both the che-2::PABP and acr-5::PABP strains in intestine, with the latter showing higher level of expression (data not shown). Low-level expression of artificially manufactured genes in the intestine seems to be quite common, either due to readthrough transcription from the vector or the 3' regulatory sequences. Our results may suggest that in future applications one must be very careful about this type of low-level expression of FLAG-PAB-1.
We determined the expression patterns of genes highly enriched in the che-2::PABP-derived mRNA preparations. Thirteen of 15 genes that showed clear expression patterns of GFP reporters were expressed in multiple sensory neurons. None of these genes has previously been characterized. In addition, quantitative PCR analysis shows that genes expressed in only one or two neurons, gcy-6 and odr-10, respectively, can be enriched. Therefore, our procedure is effective for identifying genes that are preferentially expressed in a particular subset of cells. On the other hand, the presence of small fractions of genes that are predominantly expressed in tissues other than sensory neurons was also evident. Therefore, mRNA-tagging technology should be regarded as enrichment of candidate cell-specific genes and the real expression pattern of each gene should be verified independently.
Characterization of the sensory neuron-expressed gene set
Since most of the genes enriched in the che-2::PABP-derived mRNA preparations proved to be sensory neuron-expressed, characterization of the enriched genes as a set should lead to molecular characterization of the sensory neurons of C. elegans. A prominent feature of the genes enriched in the che-2::PABP-derived mRNA preparations is that they include nematode-specific genes more often than the rest of the genes, as judged from inter-species BLASTP comparisons, suggesting that many of the genes identified have functions unique to nematode sensory neurons. The existence of many nematode-specific gene families has previously been noted, and was proposed to be related to the nematode-specific body plan [40]. Since we were obviously counter-selecting for ubiquitously expressed genes that serve common cellular functions, a lower representation of highly conserved genes is expected. In addition, these observations indicate that our approach is effective for identifying hitherto uncharacterized genes that might be important for specific functions of differentiated cell types.
Identification of panels of genes expressed in particular cells will also be useful for understanding the regulatory network of gene expression. In this context, it is of interest to examine whether we can identify cis-acting elements commonly found in the promoter regions of the sensory neuron-expressed genes. The enrichment of X-boxes in the che-2::PABP fraction suggests that this might be plausible. In addition, other reports have identified cis-acting elements in genes expressed during particular developmental stages or in particular neurons (see, for example [41,42]). However, searches for common sequences using the MEME program did not reveal any motifs that were enriched in the che-2::PABP fraction. This is likely to be due to the heterogeneity of our sensory neuron-expressed gene collection (see the expression patterns in Table 1). Further refinement of our gene sets by expression analysis of each gene will be required to identify cis-acting elements that regulate cell-specific gene expression.
By surveying GO annotations and protein motifs, genes whose predicted functions are related to nucleic acids and/or chromatin were found to be enriched in the che-2::PABP gene set. This might indicate that C. elegans sensory neurons have specialized regulatory mechanisms for gene expression, although it remains to be seen which of these 'chromatin' genes are actually expressed in a sensory neuron-specific manner. It was also apparent from visual inspection or computer searches that two homologs of nephrocystins are included in the highest rank orders. It has recently been shown that nephrocystin 1 and nephrocystin 4 interact with each other and are both components of cilia. These studies have led to the hypothesis that the kidney disease nephronophthisis is caused by malfunctions of cilia on the tubular epithelium [33-35,43]. C. elegans ciliated sensory neurons also have prominent ciliary structures [44], but none of the other cell types in this organism has any cilia. It has also been found that all C. elegans homologs (bbs-1, 2, 7 and 8) of the human genes responsible for Bardet-Biedl syndrome, which is also thought to be a ciliary disease, are specifically expressed in ciliated sensory neurons [45]. It is therefore likely that the gene set revealed by our analysis includes C. elegans homologs of as yet unidentified ciliary disease genes.
Conclusions
The present study demonstrates that a combination of mRNA tagging and microarray analysis is an effective strategy for identifying genes expressed in subsets of neurons. Systematic reporter expression analyses following this approach will facilitate the accumulation of information regarding gene expression patterns. In particular, profiling of the gene expression patterns of subsets of neurons, in combination with analyses of neural functions, might provide insights into understanding the distinct roles of cells within the neural network.
Materials and methods
Generation of strains expressing FLAG-PAB-1 in a tissue-specific manner
The initiation codon of a cDNA for pab-1, yk28d10, was replaced with a linker composed of two complementary oligonucleotides, 5'-AATTGCTAGCATGGATTACAAGGATGATGACGATAAGT-3' and 5'-CTAGACTTATCGTCATCATCCTTGTAATCCATGCTAGC-3', in which the underlined sequence encodes an initiation codon followed by a FLAG peptide. The resulting epitope-tagged gene was cloned into the pPD49.26 vector (donated by Andy Fire, Stanford University). The promoter of che-2 [21], acr-5 [22] or myo-3 [23] was inserted 5' upstream to the fusion gene to generate the FLAG-PAB-1 expression plasmids pche2-FLAG-PABP(FL), pacr5-FLAG-PABP(FL) and pmyo3-FLAB-PABP(FL), respectively. Wild-type animals were transformed with each expression construct, along with the pRF4 plasmid, which carries a dominant rol-6 allele, as a marker [46]. Stable integrated transgenic strains were generated from unstable transgenic lines as described [47]. Each integrated strain was outcrossed twice with wild-type N2. The genotypes of these strains were: JN501: Is [che-2p::flag-pab-1 pRF4]; JN502: Is [acr-5p::flag-pab-1 pRF4]; and JN503: Is [myo-3p::flag-pab-1 pRF4].
mRNA tagging
To purify poly(A)-RNA/FLAG-PAB-1 complexes from subsets of neurons, we modified a protocol for chromosome immunoprecipitation [48]. Transgenic animals were grown in liquid as described previously [49]. The worms were then harvested and washed twice with M9 [50]. To crosslink poly(A) RNA with FLAG-PAB-1 in vivo, worms were treated with 1% formaldehyde in M9 for 15 min at 20°C with gentle agitation. The formaldehyde was then inactivated by 125 mM glycine for 5 min at 20°C and washed out by replacing the buffer with four changes of TBS (20 mM Tris-HCl pH 7.5, 150 mM NaCl). At this point, worms were dispensed into 0.4 g aliquots, placed in 2-ml microtubes and stored frozen until lysate preparation.
Worms were resuspended in 0.45 ml lysis buffer (50 mM HEPES-KOH pH 7.3, 1 mM EDTA, 140 mM KCl, 10% glycerol, 0.5% Igepal CA-630 (Sigma), 1 mM DTT, 0.2 mM PMSF, protease inhibitor cocktail (Complete-EDTA, Roche) at the recommended concentration) supplemented with 20 mM ribonucleoside vanadyl complexes (RVC, Sigma) and 1000 U/ml of human placental ribonuclease inhibitor (Takara). Animals were disrupted by vigorous shaking with 2 g acid-washed glass beads (Sigma), and worm debris was removed by centrifugation at 18,000 g for 20 min. Five hundred microliters of supernatant, with the protein concentration roughly adjusted to 20 mg/ml, was incubated with 50 μl of anti-FLAG M2 affinity gel beads (Sigma) for 2 h. The affinity beads were sequentially washed three times with lysis buffer supplemented with PMSF, twice with wash buffer (50 mM HEPES-KOH pH 7.3, 1 mM EDTA, 1 M KCl, 10% glycerol, 0.5% Igepal CA-630, 1 mM DTT) and once with TE (10 mM Tris-HCl pH 7.5, 0.5 mM EDTA). Lysate preparation and purification of RNA-protein complexes were performed at 4°C. Precipitated materials were eluted with 100 μl elution buffer (50 mM Tris-HCl pH 7.5, 10 mM EDTA, 1% SDS, 20 mM RVC) by incubation for 5 min at 65°C. Elution was repeated and the two supernatant fractions were combined. The eluted RNA/FLAG-PAB-1 complexes were incubated for 6 h at 65°C to reverse the formaldehyde crosslinks. Proteins were digested with proteinase K and removed by phenol-chloroform extraction. Nucleic acid was recovered by ethanol precipitation. Typically, 100 ng nucleic acid was obtained from 0.5 ml cleared lysate of myo-3::PABP. Under the above washing conditions, binding of free poly(A) RNA to PABP was severely impaired (data not shown).
Examination of the functional integrity of FLAG-PAB-1-expressing cells in the che-2::PABP strain
For staining of living animals with lipophilic dye, we followed the procedure described before [51] except that diQ (Molecular Probes) was used instead of FITC. Forty-six wild type and 56 che-2::PABP worms at L4 to young adult were observed. Cells were identified by their positions and the percentage of stained cells was scored. Responses of the che-2::PABP strain to 1-octanol was assessed as described [52] except that Eppendorf Microloader (Eppendorf) was used to deliver 1-octanol to animals' noses.
RT-PCR
Fifty nanograms of RNA was converted to cDNA using an RNA PCR Kit (AMV) Ver. 2.1 (Takara) according to the manufacturer's protocol. One-tenth of the cDNA from each sample was subjected to a gene-specific PCR reaction in a total volume of 20 μl. Quantification of the PCR products was performed using a FastStart DNA Master SYBR Green I Kit (Roche) with the Light Cycler system (Roche). Serial dilutions of cDNA prepared from poly(A) RNA of wild-type worms were used to generate a standard curve. The ratio of expression levels for each gene was calculated using the amount of eft-3 as a reference, and the results of three independent experiments were averaged. The primers used for the amplification of each gene were: lmn1-52: 5'-CGTTCACCACCCACCAGAA-3' and lmn1-32: 5'-CAAGACGAGCTGATGGGTTATCT-3' for lmn-1; eft3-52: 5'-ATTGCCACACCGCTCACA-3' and eft3-32: 5'-CCGGTACGACGGTCAACCT-3' for eft-3; tax2-54: 5'-GATTAATCCAAGACAAGTTCCTAAATTGAT-3' and tax2-34: 5'-TTCAATTCTTGAACTCCTTTGTTTTC-3' for tax-2; unc8-52: 5'-TCTCAGATTTTGGAGGTAATATTGGA-3', and unc8-32: 5'-GATCTCGCAGAAAAGTTCTGCAA-3' for unc-8; unc54-52: 5'-AACAGAAGTTGAAGACCCAGAAGAA-3', and unc54-32: 5'-TGGTGGGTGAGTTGCTTGTACT-3' for unc-54; snt1-51: 5'-GAGCTGAGGCATTGGATGGA-3' and snt1-31: 5'-CCAAGTGTATGCCATTGAGCAA-3' for snt-1; acr5-52: 5'-AATCGATTTATGGACAGAATTTGGA-3' and acr5-32: 5'-ATGTTGCAAAAGAAGTGGGTCTAGA-3' for acr-5; odr10-51: 5'-TCATTGTGTTTTGCTCATTTCTGTAC-3' and odr10-31: 5'-ATATTGTTCTTCGGAAATCACGAAT-3' for odr-10; del1-51: 5'-TAAACTGCCTCACGACAGAAG-3' and del1-31: 5'-GCCATCAAGTTGAACCAAGAAT-3' for del-1. All primers were designed to include one intron in the PCR product amplified from the genomic DNA for each gene, such that the length and melting point were different from the product amplified from the cDNA. In Figure 2, eft-3 was amplified for 25 cycles, lmn-1, snt-1 and unc-54 for 30 cycles and tax-2, unc-8, odr-10, del-1 and acr-5 for 35 cycles. Amplified DNA was visualized by electrophoresis followed by staining with ethidium bromide.
cDNA microarray analysis
Microarrays were prepared using a 16-pin arrayer constructed according to the format of Patrick Brown (Stanford University [53]) on CMT-GAPS-coated glass slides. Two micrograms of RNA prepared from JN501 was reverse-transcribed using oligo(dT) primers and SuperScript II reverse transcriptase (Lifetech) with the addition of Cy5-dCTP to generate Cy5-labeled probes. RNA prepared from JN502 was similarly used for the generation of Cy3-labeled probes. Equal amounts of the two probes were mixed and hybridized to a single array overnight at 42°C in Gene TAC Hyb Buffer (Genomic Solutions). Each array was then washed in 1× SSC/0.03% SDS at 42°C, followed by successive washes in 0.2× SSC and 0.05× SSC at room temperature. The fluorescence intensity of each spot was scanned using a ScanArray Lite (Perkin Elmer) and analyzed by QuantArray (GSI Lumonics).
Reporter constructs for determination of expression patterns
A genomic DNA fragment for each gene was amplified by PCR such that it contained an upstream promoter region followed by a partial or full-length predicted coding region. The 3' PCR primers were designed to introduce a restriction site in-frame with GFP in the vectors. The 5' PCR primers were designed to anneal to a sequence 0.6-5.0 kilobases (kb) upstream of the predicted coding region of each gene. The upstream-predicted gene was essentially not included in the promoter fragment. The amplified PCR fragments were cloned into pPD95.70, pPD95.75 (donated by A. Fire) or a Gateway vector (Invitrogen) to create GFP fusions (see Additional data file 3 for details). The resulting reporter plasmid for each gene was introduced into wild-type animals. Transgenic worms at all developmental stages were observed under a differential interference contrast (DIC)-fluorescence microscope. Cells were identified according to their positions [54] by comparing the fluorescence images of GFP and DiQ staining with Nomarski images of the same animal. At least two independent transgenic lines were observed to confirm the expression patterns. Images were obtained as described previously [17].
Bioinformatics
cDNA clones were mapped to the C. elegans genome using BLAT [55] and BLASTN programs, corresponding gene models were identified in the WormBase annotations [56] and protein sequences were obtained from WormPep122. The gene ontology annotation dataset for C. elegans was obtained from the Gene Ontology Consortium [57].
For Figure 3, genes known to be expressed in specific tissues were searched for using the expression pattern search interface of WormBase [58] or using the AcePerl AceDB server [59]. The gene set 'Sensory neurons' was defined as genes expressed in all or some ciliated sensory neurons (including amphid neurons), but not in motor neurons or the ventral nerve cord, according to WormBase. The gene set 'Motor neurons' was genes expressed in VB or DB ventral cord motor neurons and no more than one type of ciliated sensory neuron, or those expressed in cholinergic neurons. The gene set 'Muscles' was genes expressed in some muscles, but not in neurons or the intestine, while 'Intestine' was genes expressed in the intestine, but not in neurons or muscles. X-boxes were searched for using MEME and MAST [60] based on the definition matrix deduced from Swoboda et al. [31]. Only between 60 and 160 base-pairs (bp) upstream of the initiation codon of each gene were considered, since Swoboda et al. observed that X -boxes were present about 100 bp upstream of the initiation codons [31].
The proteome data set for Caenorhabditis briggsae, whose draft genome sequence has recently been released, was downloaded from WormBase [61]. Proteome data sets for humans, mice, Drosophila melanogaster, Schizosaccharomyces pombe and Saccharomyces cerevisiae were obtained from NCBI [62], and BLASTP searches were performed using the WormPep protein sequence as a query for each C. elegans gene.
Protein motifs that preferentially appeared in genes at the higher rank orders were searched for as follows. For all genes represented in our microarrays, the protein motifs contained in each gene product were obtained from the AcePerl server. For each motif, deviation of the average log2(che-2::PABP/acr-5::PABP) value for all genes that carried the motif was calculated. To avoid artifactual results due to gene families with close sequence similarities (such as major sperm proteins), groups of genes whose mRNA are expected to cross-hybridize were treated as a single imaginary gene with the average log2(che-2::PABP/acr-5::PABP) value.
Additional data files
The following additional data are available with the online version of this article. Additional data file 1 is a table listing the results of the microarray experiments. Additional data file 2 lists genes expressed in sensory neurons, motor neurons, muscles and the intestine, and those with X-boxes shown in Figure 3. Additional data file 3 lists the primers and vectors used for reporter constructions. Additional data file 4 contains the legends to the above three tables.
Supplementary Material
Additional data file 1
a table listing the results of the microarray experiments
Click here for additional data file
Additional data file 2
Genes expressed in sensory neurons, motor neurons, muscles and the intestine, and those with X-boxes shown in Figure 3
Click here for additional data file
Additional data file 3
The primers and vectors used for reporter constructions
Click here for additional data file
Additional data file 4
The legends to the above three tables
Click here for additional data file
Acknowledgements
We thank Andrew Fire (Stanford University) for providing vectors, Jim Kent for the BLAT program, and R.C. Chan (UC Berkeley) for sharing the chromatin IP protocol with us before publication. We thank Toshiko Tanaka for technical assistance. We also thank Takayo Hamanaka for information on microarray experiments.
Figures and Tables
Figure 1 Principle of the mRNA-tagging method. Step 1, FLAG-tagged poly(A)-binding protein (PABP) is expressed from a transgene using a cell-specific promoter. Step 2, PABP and poly(A)+ RNA are crosslinked in situ by formaldehyde. Step 3, poly(A)-RNA/FLAG-PABP complexes are purified by anti-FLAG affinity purification. Step 4, RNA-PABP crosslinks are reversed and RNA is isolated. Step 5, purified RNA is used for microarray analysis.
Figure 2 Quantification of tissue-specific transcripts in RNA prepared by mRNA tagging. The transcript indicated on the left of each row was amplified by RT-PCR using gene-specific primers. Poly(A)+RNA from wild-type (WT) animals was used as a template in lane 1. RNA prepared by mRNA tagging from che-2::PABP (JN501), acr-5::PABP (JN502) and myo-3::PABP (JN503) was used in lanes 2, 3 and 4, respectively.
Figure 3 Rank orders of che-2::PABP/acr-5::PABP values for specific genes in the microarray analyses. (a) Distribution of genes with known expression patterns. Genes known to be specifically expressed in sensory neurons, motor neurons, muscles or the intestine, respectively, were collected from WormBase (see Materials and methods) and the rank orders of their che-2::PABP/acr-5::PABP signal ratios were plotted. Vertical bars indicate the medians. Genes expressed in sensory neurons are specifically enriched in the che-2::PABP RNA preparations, while motor neuron- and intestine-expressed genes are enriched in the acr-5::PABP RNA preparations. Note that although only five genes were found as motor neuron-expressed genes, nine data points were plotted in (a), because multiple cDNA clones were present on the microarray for three of the genes (see Additional data file 2). (b) Distribution of genes with X-boxes in their promoter regions. Genes that carry one or more X-boxes in their promoter regions were collected from the genome database (see Materials and methods) and their rank orders of che-2::PABP/acr-5::PABP signal ratios were plotted. These genes, which are expected to be expressed in ciliated sensory neurons under the control of the DAF-19 transcription factor, are also enriched in the che-2::PABP RNA preparations.
Figure 4 Expression patterns of newly identified sensory neuron-expressed genes. The genes indicated were each fused to GFP in-frame, and the reporters introduced into wild-type animals. Overlaid images of the Nomarski and GFP fluorescence images of transgenic worms between larval stages 1 and 3 are shown. Gene expression is indicated by the green fluorescence. Scale bar, 50 μm. See Table 1 for the identity of the expressing cells.
Figure 5 Sensory neuron-specific genes are less likely to be classified into Gene Ontology categories and more likely to be worm-specific. (a) All genes on the microarray were ordered by descending che-2::PABP/acr-5::PABP value and the fraction of GO-annotated genes in each bin is indicated for a bin width of 50 rank orders. Only the top 1,500 genes are shown in (a)-(c). (b) The fraction of genes with homologs in C. briggsae, and not in humans, mice, flies, fission yeast or budding yeast (cutoff BLASTP score E = 1 × 10-20) in each bin is indicated as in (a). (c) The fraction of genes with homologs in both animals and yeasts, namely in humans, mice or flies and in fission yeast or budding yeast (cutoff BLASTP score E = 1 × 10-20) in each bin is indicated as in (a). In all panels, the red dotted line indicates the average of all the genes, and the blue dotted lines indicate the 95% confidence limits assuming a random binominal distribution.
Figure 6 Categories of genes enriched in the sensory neuron fraction. Genes were categorized according to the GO molecular function categories. (a) Categorization of all the genes on the microarray; (b) categorization of genes within the top 500 che-2::PABP/acr-5::PABP ranks. In both panels, the fraction of genes in each category in respect of all annotated genes is shown.*P < 0.05; **P < 0.01 (binominal distribution).
Table 1 Expression patterns of the top-ranked genes
Rank Clone Gene Locus Expression pattern
1 yk380a6 R102.2 ADF, ADL, ASH, ASI, ASJ, ASK, PHA, PHB
2 yk305a7 C33A12.4 ADF, ADL, ASE, ASH, ASI, ASJ, ASK, AVJ, AWA, AWB, PHA, PHB, labial neurons
3 yk139b4 C34D4.1 ADL, ASH, ASI, ASJ, ASK, PHA
4 yk534e12
5 yk91d12 C02H7.1 ADF, ADL, AFD, ASG, ASH, ASI, ASJ, ASK, AWB, PHA, PHB, URX
6 yk261h1 Y43F8C.4
7 yk538c3 K07C11.10 ADF, ADL, ASE, ASG, ASH, ASI, ASJ, ASK, AWA, AWB, AWC, PHA, PHB
8 yk561g1 F40H3.6
9 yk267a7 ZK938.2 ADL, ASE, ASG, ASH, ASI, ASJ, ASK, AWB, AWC, PHB, URX
10 yk509b4 Y55D5A.1a ADF, ADL, AFD, ASE, ASG, ASH, ASI, ASJ, ASK, AWA, AWB, AWC, BAG, PHA, PHB, URX, intestine
11 yk609e11 T27E4.3 hsp-16.48
12 yk561g1 F40H3.6
13 yk341h9 F53A9.4 ADL, ASE, ASH, ASI, ASJ, ASK, AWC, PHA, PHB, labial neurons
14 yk604g4 C35E7.11 Intestine, RMF, RMH
15 yk467b4 M28.7 ADF, ADL, AFD, ASE, ASG, ASH, ASI, ASJ, ASK, AWA, AWB, AWC, PHA, PHB, URX, labial neurons
16 yk284g4 K10D6.2a ADL, ASE, ASI, ASJ, ASK, PHB, URX, labial neurons, seam cells, hypodermis
17 yk610e5 Y9D1A.1
18 yk295d7 K07B1.8 No GFP
19 yk252h2 C29H12.3a rgs-3
20 yk225f3 C27A7.4 che-11
21 yk488h9 C13B9.1 No GFP
22 yk373g4 T07C5.1c AFD, ASG, AUA, PVQ, intestine
23 yk305c8 F10G2.1 Coelomocytes
24 yk450c2 K10G6.4 ADF, ADL, AFD, ASE, ASG, ASH, ASI, ASJ, ASK, AVA, AVD, AWA, AWB, AWC, PHB, RMD, ventral nerve cord neurons, many other neurons
25 yk76f1 R13H4.1 ADF, ADL, ASE, ASG, ASH, ASI, ASJ, ASK, PHA, PHB, URX, labial neurons
The expression patterns of the genes indicated in bold were examined. Only the cells and tissues in which GFP expression was consistently observed are listed. It is therefore possible that the genes are weakly expressed in cells or tissues other than those listed here. Cells and cell groups in bold are ciliated sensory neurons.
==== Refs
The C. elegans Sequencing Consortium Genome sequence of the nematode C. elegans : a platform for investigating biology. Science 1998 282 2012 2018 9851916 10.1126/science.282.5396.2012
Fraser AG Kamath RS Zipperlen P Martinez-Campos M Sohrmann M Ahringer J Functional genomic analysis of C. elegans chromosome I by systematic RNA interference. Nature 2000 408 325 330 11099033 10.1038/35042517
Gonczy P Echeverri C Oegema K Coulson A Jones SJ Copley RR Duperon J Oegema J Brehm M Cassin E Functional genomic analysis of cell division in C. elegans using RNAi of genes on chromosome III. Nature 2000 408 331 336 11099034 10.1038/35042526
Kamath RS Fraser AG Dong Y Poulin G Durbin R Gotta M Kanapin A Le Bot N Moreno S Sohrmann M Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 2003 421 231 237 12529635 10.1038/nature01278
Maeda I Kohara Y Yamamoto M Sugimoto A Large-scale analysis of gene function in Caenorhabditis elegans by high-throughput RNAi. Curr Biol 2001 11 171 176 11231151 10.1016/S0960-9822(01)00052-5
Li S Armstrong CM Bertin N Ge H Milstein S Boxem M Vidalain PO Han JD Chesneau A Hao T A map of the interactome network of the metazoan C. elegans. Science 2004 303 540 543 14704431 10.1126/science.1091403
Jansen G Hazendonk E Thijssen KL Plasterk RH Reverse genetics by chemical mutagenesis in Caenorhabditis elegans. Nat Genet 1997 17 119 121 9288111 10.1038/ng0997-119
Edgley M D'Souza A Moulder G McKay S Shen B Gilchrist E Moerman D Barstead R Improved detection of small deletions in complex pools of DNA. Nucleic Acids Res 2002 30 e52 12060690 10.1093/nar/gnf051
Gengyo-Ando K Mitani S Characterization of mutations induced by ethyl methanesulfonate, UV, and trimethylpsoralen in the nematode Caenorhabditis elegans. Biochem Biophys Res Commun 2000 269 64 69 10694478 10.1006/bbrc.2000.2260
McKay SJ Johnsen R Khattra J Asano J Baillie DL Chan S Dube N Fang L Goszczynski B Ha E Gene expression profiling of cells, tissues, and developmental stages of the nematode C. elegans. Cold Spring Harb Symp Quant Biol 2003 68 159 69 15338614 10.1101/sqb.2003.68.159
Dupuy D Li QR Deplancke B Boxem M Hao T Lamesch P Sequerra R Bosak S Doucette-Stamm L Hope IA A first version of the Caenorhabditis elegans promoterome. Genome Res 2004 14 2169 75 15489340 10.1101/gr.2497604
NEXTDB (The Nematode Expression Pattern Database)
Jiang M Ryu J Kiraly M Duke K Reinke V Kim SK Genome-wide analysis of developmental and sex-regulated gene expression profiles in Caenorhabditis elegans. Proc Natl Acad Sci USA 2001 98 218 223 11134517 10.1073/pnas.011520898
Kim SK Lund J Kiraly M Duke K Jiang M Stuart JM Eizinger A Wylie BN Davidson GS A gene expression map for Caenorhabditis elegans. Science 2001 293 2087 2092 11557892 10.1126/science.1061603
Romagnolo B Jiang M Kiraly M Breton C Begley R Wang J Lund J Kim SK Downstream targets of let-60 Ras in Caenorhabditis elegans. Dev Biol 2002 247 127 136 12074557 10.1006/dbio.2002.0692
Reinke V Smith HE Nance J Wang J Van Doren C Begley R Jones SJ Davis EB Scherer S Ward S A global profile of germline gene expression in C. elegans. Mol Cell 2000 6 605 616 11030340 10.1016/S1097-2765(00)00059-9
Hanazawa M Mochii M Ueno N Kohara Y Iino Y Use of cDNA subtraction and RNA interference screens in combination reveals genes required for germ-line development in Caenorhabditis elegans. Proc Natl Acad Sci USA 2001 98 8686 8691 11447255 10.1073/pnas.141004698
Zhang Y Ma C Delohery T Nasipak B Foat BC Bounoutas A Bussemaker HJ Kim SK Chalfie M Identification of genes expressed in C. elegans touch receptor neurons. Nature 2002 418 331 335 12124626 10.1038/nature00891
Roy PJ Stuart JM Lund J Kim SK Chromosomal clustering of muscle-expressed genes in Caenorhabditis elegans. Nature 2002 418 975 979 12214599
Gallie DR A tale of two termini: a functional interaction between the termini of an mRNA is a prerequisite for efficient translation initiation. Gene 1998 216 1 11 9714706 10.1016/S0378-1119(98)00318-7
Fujiwara M Ishihara T Katsura I A novel WD40 protein, CHE-2, acts cell-autonomously in the formation of C. elegans sensory cilia. Development 1999 126 4839 4848 10518500
Winnier AR Meir JY Ross JM Tavernarakis N Driscoll M Ishihara T Katsura I Miller DM 3rd UNC-4/UNC-37-dependent repression of motor neuron-specific genes controls synaptic choice in Caenorhabditis elegans. Genes Dev 1999 13 2774 2786 10557206 10.1101/gad.13.21.2774
Okkema PG Harrison SW Plunger V Aryana A Fire A Sequence requirements for myosin gene expression and regulation in Caenorhabditis elegans. Genetics 1993 135 385 404 8244003
Coburn CM Bargmann CI A putative cyclic nucleotide-gated channel is required for sensory development and function in C. elegans. Neuron 1996 17 695 706 8893026 10.1016/S0896-6273(00)80201-9
Sengupta P Chou JH Bargmann CI odr-10 encodes a seven transmembrane domain olfactory receptor required for responses to the odorant diacetyl. Cell 1996 84 899 909 8601313 10.1016/S0092-8674(00)81068-5
Tavernarakis N Shreffler W Wang S Driscoll M unc-8, a DEG/ENaC family member, encodes a subunit of a candidate mechanically gated channel that modulates C. elegans locomotion. Neuron 1997 18 107 119 9010209 10.1016/S0896-6273(01)80050-7
Mitrovich QM Anderson P Unproductively spliced ribosomal protein mRNAs are natural targets of mRNA surveillance in C. elegans. Genes Dev 2000 14 2173 2184 10970881 10.1101/gad.819900
Liu J Ben-Shahar TR Riemer D Treinin M Spann P Weber K Fire A Gruenbaum Y Essential roles for Caenorhabditis elegans lamin gene in nuclear organization, cell cycle progression, and spatial organization of nuclear pore complexes. Mol Biol Cell 2000 11 3937 3947 11071918
Yu S Avery L Baude E Garbers DL Guanylyl cyclase expression in specific sensory neurons: a new family of chemosensory receptors. Proc Natl Acad Sci USA 1997 94 3384 3387 9096403 10.1073/pnas.94.7.3384
Nonet ML Grundahl K Meyer BJ Rand JB Synaptic function is impaired but not eliminated in C. elegans mutants lacking synaptotagmin. Cell 1993 73 1291 1305 8391930 10.1016/0092-8674(93)90357-V
Swoboda P Adler HT Thomas JH The RFX-type transcription factor DAF-19 regulates sensory neuron cilium formation in C. elegans. Mol Cell 2000 5 411 421 10882127 10.1016/S1097-2765(00)80436-0
Hildebrandt F Otto E Rensing C Nothwang HG Vollmer M Adolphs J Hanusch H Brandis M A novel gene encoding an SH3 domain protein is mutated in nephronophthisis type 1. Nat Genet 1997 17 149 153 9326933 10.1038/ng1097-149
Mollet G Salomon R Gribouval O Silbermann F Bacq D Landthaler G Milford D Nayir A Rizzoni G Antignac C The gene mutated in juvenile nephronophthisis type 4 encodes a novel protein that interacts with nephrocystin. Nat Genet 2002 32 300 305 12244321 10.1038/ng996
Olbrich H Fliegauf M Hoefele J Kispert A Otto E Volz A Wolf MT Sasmaz G Trauer U Reinhardt R Mutations in a novel gene, NPHP3, cause adolescent nephronophthisis, tapeto-retinal degeneration and hepatic fibrosis. Nat Genet 2003 34 455 459 12872122 10.1038/ng1216
Otto EA Schermer B Obara T O'Toole JF Hiller KS Mueller AM Ruf RG Hoefele J Beekmann F Landau D Mutations in INVS encoding inversin cause nephronophthisis type 2, linking renal cystic disease to the function of primary cilia and left-right axis determination. Nat Genet 2003 34 413 420 12872123 10.1038/ng1217
Gorlach M Burd CG Dreyfuss G The mRNA poly(A)-binding protein: localization, abundance, and RNA-binding specificity. Exp Cell Res 1994 211 400 407 7908267 10.1006/excr.1994.1104
Tenenbaum SA Carson CC Lager PJ Keene JD Identifying mRNA subsets in messenger ribonucleoprotein complexes by using cDNA arrays. Proc Natl Acad Sci USA 2000 97 14085 14090 11121017 10.1073/pnas.97.26.14085
Reboul J Vaglio P Tzellas N Thierry-Mieg N Moore T Jackson C Shin-i T Kohara Y Thierry-Mieg D Thierry-Mieg J Open-reading-frame sequence tags (OSTs) support the existence of at least 17,300 genes in C. elegans. Nat Genet 2001 27 332 336 11242119 10.1038/85913
Collet J Spike CA Lundquist EA Shaw JE Herman RK Analysis of osm-6, a gene that affects sensory cilium structure and sensory neuron function in Caenorhabditis elegans. Genetics 1998 148 187 200 9475731
Blaxter M Caenorhabditis elegans is a nematode. Science 1998 282 2041 2046 9851921 10.1126/science.282.5396.2041
Beer MA Tavazoie S Predicting gene expression from sequence. Cell 2004 117 185 198 15084257 10.1016/S0092-8674(04)00304-6
Wenick AS Hobert O Genomic cis-regulatory architecture and trans-acting regulators of a single interneuron-specific gene battery in C. elegans. Dev Cell 2004 6 757 770 15177025 10.1016/j.devcel.2004.05.004
Watnick T Germino G From cilia to cyst. Nat Genet 2003 34 355 356 12923538 10.1038/ng0803-355
Ward S Thomson N White JG Brenner S Electron microscopical reconstruction of the anterior sensory anatomy of the nematode Caenorhabditis elegans. J Comp Neurol 1975 160 3 13 37 1112920 10.1002/cne.901600305
Ansley SJ Badano JL Blacque OE Hill J Hoskins BE Leitch CC Kim JC Ross AJ Eichers ER Teslovich TM Basal body dysfunction is a likely cause of pleiotropic Bardet-Biedl syndrome. Nature 2003 425 628 633 14520415 10.1038/nature02030
Mello CC Kramer JM Stinchcomb D Ambros V Efficient gene transfer in C. elegans : extrachromosomal maintenance and integration of transforming sequences. EMBO J 1991 10 3959 3970 1935914
CPC: C. elegans protocols
Chu DS Dawes HE Lieb JD Chan RC Kuo AF Meyer BJ A molecular link between gene-specific and chromosome-wide transcriptional repression. Genes Dev 2002 16 796 805 11937488 10.1101/gad.972702
Lewis JA Fleming JT Basic culture methods. Methods Cell Biol 1995 48 3 29 8531730
Brenner S The genetics of Caenorhabditis elegans. Genetics 1974 77 71 94 4366476
Hedgecock EM Culotti JG Thomson JN Perkins LA Axonal guidance mutants of Caenorhabditis elegans identified by filling sensory neurons with fluorescein dyes. Dev Biol 1985 111 158 170 3928418 10.1016/0012-1606(85)90443-9
Chao MY Komatsu H Fukuto HS Dionne HM Hart AC Feeding status and serotonin rapidly and reversibly modulate a Caenorhabditis elegans chemosensory circuit. Proc Natl Acad Sci USA 2004 101 15512 15517 15492222 10.1073/pnas.0403369101
Pat Brown's Lab
White J Southgate E Thomson J Brenner S The structure of the nervous system of the nematode Caenorhabditis elegans. Phil Trans R Soc Lond Ser B 1986 314 1 340
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
WormBase - bulk downloads (C. elegans)
Gene Ontology Consortium Current Annotations - WormBase
WormBase: expression pattern search
WormBase
MEME -introduction
WormBase - bulk downloads (C. briggsae)
NCBI genome assembly/annotation projects
| 15693946 | PMC551537 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 31; 6(2):R17 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r17 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r181569394710.1186/gb-2005-6-2-r18MethodFast and systematic genome-wide discovery of conserved regulatory elements using a non-alignment based approach Elemento Olivier [email protected] Saeed [email protected] Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, NJ 08544, USA2005 26 1 2005 6 2 R18 R18 1 9 2004 29 10 2004 3 12 2004 Copyright © 2005 Elemento and Tavazoie; 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 authors describe a powerful approach for discovering globally conserved regulatory elements between two genomes that does not require alignments. Its application to pairs of yeasts, worm, flies and mammals, yields a large number of known and novel putative regulatory elements, many of which show surprising conservation across large phylogenetic distances.
We describe a powerful new approach for discovering globally conserved regulatory elements between two genomes. The method is fast, simple and comprehensive, without requiring alignments. Its application to pairs of yeasts, worms, flies and mammals yields a large number of known and novel putative regulatory elements. Many of these are validated by independent biological observations, have spatial and/or orientation biases, are co-conserved with other elements and show surprising conservation across large phylogenetic distances.
==== Body
Background
One of the major challenges facing biology is to reconstruct the entire network of protein-DNA interactions within living cells. A large fraction of protein-DNA interactions corresponds to transcriptional regulators binding DNA in the neighborhood of protein-coding and RNA genes. By interacting with RNA polymerase or recruiting chromatin-modifying machinery, transcriptional regulators increase or decrease the transcription rate of these genes. Transcriptional regulators bind specific DNA sequences upstream, within or downstream of the genes they regulate, and a large number of experimental and computational studies are aimed at locating these sites and understanding their functions (for example [1,2]). The increasing availability of whole-genome sequences provides unprecedented opportunities for identifying binding sites and studying their evolution. The strong conservation of functional elements (binding sites, protein-coding genes, noncoding RNAs, and so on) across even distantly related species should make it possible to predict these functional elements and prioritize them for experimental validation. The few large-scale comparative genomics approaches for finding transcriptional regulatory elements have so far relied mostly on detecting locally conserved motifs within global alignments of orthologous upstream sequences [3,4]. Although very powerful and straightforward, these approaches cannot be used when upstream regions are very divergent or have undergone genomic rearrangements. For example, aligning the mouse and puffer fish orthologous upstream regions would be very difficult, because of the great reduction that the puffer fish intergenic regions have undergone [5]. Also, global alignments cannot be used when the positions of regulatory elements within functionally conserved promoter regions have been scrambled, for example through genomic rearrangements. Also, global alignment-based approaches often generate an overwhelming number of predictions because of the basal conservation between the genomes under study. To reduce the number of predictions, multiple global alignments of upstream sequences from several related species have been used, yielding many new candidate binding sites [3,4]. However, multiple (more than two) closely related genome sequences are not always available; moreover, by focusing only on regulatory elements that are conserved between several genomes, these approaches might miss elements that are conserved in more local areas of the phylogenetic tree.
Here we describe a simple and efficient comparative approach for finding short noncoding DNA sequences that are globally conserved between two genomes, independently of their specific location within their respective promoter regions. Our method, which we call FastCompare, is based on a principle that we have termed 'network-level conservation' [6], according to which the wiring of transcriptional regulatory networks should be largely conserved between two closely related genomes.
Our previous attempts at using network-level conservation relied on Gibbs sampling to find candidate regulatory elements [7]. However, Gibbs sampling and related algorithms are not fully appropriate in this context, because of the low density of actual binding sites in pairs of orthologous upstream regions. Moreover, these algorithms are non-deterministic, relatively slow, and rely on sequence sampling, which makes them likely to miss many regulatory elements. While our previous approach was successful at predicting a large fraction of functional regulatory elements in the relatively small yeast genome, analyzing larger and more complex metazoan genomes requires faster and more exhaustive algorithms. Here, we use a faster, simpler and more comprehensive approach for detecting conserved and probably functional regulatory elements using the network-level conservation principle. FastCompare allows comprehensive exploration of the conserved - but not aligned - motifs between two genomes, while retaining a linear time complexity. We apply our approach to a large number of species, including yeasts, worms, flies and mammals, and describe some of the most conserved known and unknown regulatory elements within these genomes. We also show how this approach may help reconstruct part of the transcriptional network and reveal some of its associated constraints. Finally, we show that a large number of predicted motifs are conserved within and across different phylogenetic groups.
Results
In the following sections, pairs of closely related species are termed phylogenetic groups. We applied FastCompare to the four following phylogenetic groups: yeasts (Saccharomyces cerevisiae and S. bayanus), worms (Caenorhabditis elegans and C. briggsae), flies (Drosophila melanogaster and D. pseudoobscura) and mammals (Homo sapiens and Mus musculus). For each phylogenetic group, we describe some of the most interesting, known and novel, predicted regulatory elements. For each of these regulatory elements, we perform independent validation using gene expression data, chromatin immunoprecipitation (IP) data, known motifs and data from several biological databases (Gene Ontology (GO)/MIPS, TRANSFAC), and show that the most globally conserved predicted regulatory elements are strongly supported by these independent sources.
Yeasts
The average nucleotide identity between S. cerevisiae and S. bayanus upstream regions is approximately 62% [4] (similar to the identity between human and mouse upstream regions) and divergence times are estimated between 5 and 20 million years [4]. The number of ortholog pairs between S. cerevisiae and S. bayanus is 4,358 (see Materials and methods). We chose to analyze 1 kb-long upstream regions, because most of the known transcription factor binding sites in S. cerevisiae are located within this range [8]. Using FastCompare, we calculated a conservation score for all possible 7-, 8- and 9-mers on the corresponding 8.6 megabase-pairs (Mbp) of sequences and sorted each list separately according to conservation score (see Figure 1; the raw sorted lists are available on our website [9]). On a typical desktop PC, this analysis took approximately 5 minutes (for example, the entire set (8,170) of 7-mers was processed in 35 seconds).
Distribution of conservation scores
As described in Materials and methods, conservation scores are calculated for all k-mers (with fixed k), and are relative measures of network-level conservation for these k-mers (the higher the conservation score, the more conserved the corresponding k-mer). We first describe the distribution of conservation scores for all 7-mers. As shown in Figure 2, the distribution of conservation scores has a very long tail and many 7-mers on the tail correspond to well known regulatory elements in S. cerevisiae (see below for a detailed description of these sites). To verify that such high conservation scores could not be obtained by chance, we generated randomized sequences as described in Materials and methods and re-ran FastCompare on these sequences. The corresponding distribution of conservation scores is shown on Figure 2 and clearly shows that the high conservation scores corresponding to known regulatory elements are extremely unlikely to arise by chance.
Validation using independent biological data
We used various independent sources of biological data to demonstrate that k-mers with the highest conservation scores are likely to be functional. For a given k-mer, we define the 'conserved set' as the set of ORFs corresponding to the overlap between the two sets of orthologous ORFs containing at least one exact match to the k-mer in their upstream regions (see Materials and methods). We found that conserved sets defined for the highest-scoring 7-mers are significantly enriched with genes whose upstream regions contain occurrences of known motifs in yeast (Figure 3a), significantly enriched with genes whose upstream regions were shown to be bound by known transcription factors in vivo (Figure 3b), and significantly enriched in at least one MIPS functional category (Figure 3c). We also show that the number of 7-mers found upstream of over- or underexpressed genes in at least one microarray condition increases with the conservation score (Figure 3d) and that the number of 7-mers matching at least one TRANSFAC consensus also increases with the conservation score (Figure 3e). Altogether, these data provide strong and independent evidence that our method identifies functional yeast regulatory elements by giving them a high conservation score.
Closer examination of Figure 3a-d shows that the 400 highest-scoring 7-mers are most strongly supported by independent data. Therefore we retain them for further analysis and, when possible, replace them by 8-mers and 9-mers with higher conservation scores and also add the high-scoring 8-mers and 9-mers without high-scoring substrings, as described in Materials and methods. This processing yields 398 k-mers (k = 7, 8 and 9).
Then, for each of these 398 k-mers, we determine the optimal window within the initial 1 kb which maximizes the conservation score (see Materials and methods); we then re-evaluate the functionality of each of the 398 k-mers with the independent biological information described above, using the new conserved sets. The full information for the 398 k-mers is available at [9].
Known regulatory elements
Using known transcription factor binding site motifs, genome-wide in vivo binding data, functional annotation and literature searches, we found at least 27 different known transcription factor binding sites among the 398 highest scoring k-mers. These regulatory elements, along with their support from independent biological data, are shown in Table 1. Some of the best-known binding sites are represented several times within the 398 top scoring k-mers, in the form of slightly distinct or overlapping sequences (see [9]). Note also that we use very stringent criteria for identifying known binding sites among our predictions. When we matched our predictions to the known motifs published in [4] (regular expressions), we predicted 42 out of 53 known motifs (Kellis et al. [4] predict exactly the same number of motifs, and essentially the same motifs, but using multiple alignments of four yeast genomes).
Among the 27 different known regulatory elements returned by FastCompare, several (Swi4, Mbp1, Sum1/Ndt80, Fkh1/2) are involved in regulating the yeast cell cycle. The other known sites are also involved in fundamental biological processes in yeast: amino-acid metabolism (Cbf1, Gcn4), meiosis (Ume6), rRNA transcription (PAC and RRPE), proteolytic degradation (Rpn4), stress response (Msn2/Msn4) and general activation/repression (Rap1, Reb1). As described in Materials and methods, our approach also handles gapped motifs. Thus, the binding sites for Abf1, a chromatin reorganizing transcription factor (CGTNNNNNNTGA), and Mcm1, a factor involved in cell-cycle regulation and pheromone response (CCCNNNNNGGA), were also identified as very high-scoring patterns and strongly supported by independent information (known motifs and chromatin immunoprecipitation).
When we used the same independent biological data to evaluate the 400 highest-scoring 7-mers obtained on randomized data, we found only three known binding sites (RRPE, FKH1 and BAS1).
Several known binding sites are not found among the 398 top-scoring k-mers, perhaps because their transcriptional network has undergone extensive rewiring since the speciation of the two yeasts, or because the corresponding transcription factors regulate few genes. In some cases, the presence of several known sites (clearly identified in terms of independent data) among the full set of 7-mers argues in favor of the rewiring hypothesis. For example, the binding site for the Rcs1 transcription factor, TGCACCC, only appears at the 1,883rd position within the list of ranked 7-mers. Despite its lack of conservation, this site is strongly backed by independent biological information: it is identified as a known motif, it is found in 33 microarray conditions, and its conserved set is significantly enriched in genes annotated with homeostasis of metal ions (p < 10-5), which is the known function for Rcs1 [10]. Similarly, the known binding sites for the Ace2/Swi5 and Hsf1 transcription factors were clearly identified (in terms of independent data) within the complete list of 7-mers, but not among the 398 highest scoring k-mers.
Positional constraints
It is now known that functional regulatory elements can be positionally constrained, relative to other regulatory elements or to the start of transcription [7,11,12]. To assess whether some of the predicted regulatory elements are positionally constrained in yeast, we calculated the median distance to ATG for the conserved sets of each of the 398 k-mers and independently built the distribution of median distances to ATG for all 7-mers as described in Materials and methods (the distribution is shown in Figure 4) and found d0.025 = 350 and d0.975 = 680. In other words, a median distance to ATG of less than 350 or higher than 680 should each arise by chance with only a 2.5% probability. Among the 398 most conserved k-mers, more than a fifth (86) have their median distance below 350 (p < 10-52), while only seven have a median distance greater than 680. A closer examination reveals that a few known sites are particularly constrained. For example, the binding sites for Reb1, PAC, TATA, Swi4, Rpn4, RRPE and Mbp1 are found to be situated relatively close to the start of translation, with a median distance to ATG between 150 and 300 bp. Some of these constraints were also found to be good predictors of gene expression in a recent study [11] (for RPN4, PAC and RRPE, for example). In contrast, binding sites for Met4, Ume6, Hap4, Rap1, Ino4 and Ste12 are found to be situated at a greater median distance, between 400 and 500 bp from ATG.
Novel predicted regulatory elements
We found many novel motifs among our highest-scoring predictions. For example, we found two strongly conserved motifs, AGGGTAA (rank 17) and TGTAAATA (rank 31), which are situated relatively close to ATG (with a median distance to ATG of 349 and 378.5 bp, respectively) and more often in upstream regions than in coding regions (with ratios of 1.95 and 1.83, respectively). Interestingly, TGTAAATA also has a statistically significant 5' to 3' orientation bias (binomial p-value < 10-7). However, neither of the two putative sites is supported by independent biological data. Additional expression data may help define their biological role. Other sites, such as CAGCCGC or GCGCCGC are found upstream of over- or underexpressed genes in many microarray conditions (15 and 6, respectively). While these two sites are similar to the canonical Ume6-binding site, the latter was not found in any microarray conditions (as none of the microarray experiments we used is related to meiosis, the biological process which Ume6 is known to be involved in), suggesting that the two sites are bound by other factors.
Comparing closer and more distant yeast species
We repeated the same analysis on distinct pairs of yeast species other than S. cerevisiae/S. bayanus. We first compared S. cerevisiae and S. paradoxus (a much closer relative of S. cerevisiae) and found 15 of the 27 known motifs we obtained when comparing S. cerevisiae and S. bayanus (results are available at [9]). We also compared S. cerevisiae with S. castellii, which is a more distant relative within the Saccharomyces phylogenetic group. S. castelli is interesting in that its upstream regions cannot be globally aligned with those of S. cerevisiae, because of extensive sequence divergence [3]. We also found 15 of the 27 known motifs found in the S. cerevisiae/S. bayanus comparison (results at [9]), although they were different from the S. cerevisiae/S. paradoxus conserved motifs. Interesting similarities and differences in conservation were revealed when comparing the known motifs discovered in each comparison. For example, the PAC, RRPE and Mbp1 motifs were found within the highest-scoring k-mers in all three comparisons, hinting at the conserved role of the corresponding proteins. However, the Reb1-binding site, which was found to be highly conserved between S. cerevisiae and S. bayanus (rank 1), is much less conserved between S. cerevisiae and S. castelli (rank 230). This argues for extensive rewiring in the Reb1 transcriptional network in the lineage that led to S. castelli.
Motif interactions
To discover interactions between regulatory elements, we searched for co-conservation of pairs of high-scoring predicted regulatory elements, as described in Materials and methods. Not surprisingly, the most conserved interaction is between RRPE (AAAAATTTT) and PAC (CTCATCGC), with a median distance D = 22 bp [11,13]. We also find that the Cbf1-binding site (CACGTGA) is strongly co-conserved with the Met4-binding site (CTGTGGC), and that these two sites are separated by a short distance (D = 44.5) in S. cerevisiae. Indeed, it has been shown that the binding of Cbf1 in the vicinity of a very similar sequence (AAACTGTG) enhances the DNA-binding affinity of a Met4-Met28-Met31 complex for this sequence [14], and that the median distance between the above Cbf1 and Met4 sites is small [15].
Many of the predicted interactions have not yet been experimentally studied. For example, we found that the highest scoring Reb1 motif (CGGGTAA) is significantly co-conserved with both the highest scoring RRPE motif (AAAAATTTT) and the highest scoring PAC motif (CTCATCGC), with a short median distance between the two sites in both cases (D = 38 and D = 63.5, respectively). The Reb1/RRPE interaction was also discovered independently as a good predictor of expression [11]. We also found that Reb1 interacts with the Cbf1 motif (CACGTGA), also at a short median distance (D = 30). An interesting interaction between RRPE and an unknown motif, TGAAGAA, displays a conserved set strongly enriched in translation (p < 10-11), while RRPE alone is more strongly enriched in rRNA transcription (p < 10-14). The full sorted list of interactions is available at [9].
Worms
In contrast to yeast, relatively little is known about cis-regulatory sequences in C. elegans. There is a dramatically greater complexity of transcriptional regulation in multicellular organisms. Indeed, transcription factors in multicellular organisms regulate cohorts of genes in different tissues and at different times during development [16]. C. elegans promoter regions often contain many domains of activation/repression and, as a result, are much larger than those in yeast.
We applied FastCompare to the genomes of C. elegans and C. briggsae, two worms that diverged about 50-120 million years ago [17]. The number of orthologous open reading frames (ORFs) between these two species is 13,046 and here we have only considered 2,000 bp upstream regions. It takes approximately 11 minutes for FastCompare to process the corresponding 50 Mbp of sequences and calculate a conservation score for all 7-, 8- and 9-mers on a typical desktop PC.
Validations
The distribution of conservation scores for all 7-mers shows that high conservation scores are unlikely to be obtained by chance (Figure 5a). As shown in Figure 5a, many known regulatory elements fall on the tail of the distribution. We then used functional categories, over- or underexpression, and TRANSFAC motifs to assess the ability of FastCompare to predict functional regulatory elements. Figure 5b-d shows that support for the highest-scoring k-mers by functional enrichment, expression and TRANSFAC strongly increases with conservation score. We have only retained the 400 highest-scoring 7-mers, which are particularly well supported by independent biological information as shown in Figure 5b,c. Starting from these 400 highest-scoring 7-mers, we obtain 437 k-mers (k = 7, 8 or 9) using the procedure described in Materials and methods.
Known regulatory elements
As shown in Table 2, at least 15 distinct known binding sites in C. elegans and other metazoan organisms were identified among the 437 predicted regulatory elements.
One of the most conserved is TGATAAG, the binding site for the GATA factors, a family of regulators controlling intestinal development (see [18] for review). Another motif returned by FastCompare, GTGTTTGC, corresponds to the binding site for the forkhead-related activator-4 (Freac-4) [19]. Note that this motif is also compatible with the PHA-4-binding site (published consensus: T[AG]TT[GT][AG][CT] [20]), present in the upstream regions of pharyngeal genes [20] (PHA-4 is also a member of the forkhead family of transcription factors). FastCompare also returned TGTCATCA, the known binding site for the SKN-1 transcription factor (published consensus [AT][AT]T[AG]TCAT). In C. elegans, SKN-1 is known to initiate mesendodermal development by inducing expression of the GATA factors MED-1 and MED-2 (required for mesendodermal differentiation in the EMS lineage) [21].
The GAGA-factor binding site (AGAGAGA) was also found as a highly conserved pattern. GAGA repeats in upstream regions have been shown to be functional in C. elegans in at least two separate studies [22,23]. At least one GAGA-binding protein has been identified in D. melanogaster, and is assumed to create nucleosome-free regions of DNA, thus allowing additional transcription factors to bind those regions [24]. However, the ortholog of this protein has not yet been identified in C. elegans [24].
We also found CAGCTGG, a site known to be bound by the myogenic basic helix-loop-helix (bHLH) family of transcription factors (in worms, flies and mammals) and AP-4 transcription factors (in mammals) [25,26] (published consensus CAGCTG [27-29]). The homolog of human AP-4 was found to be ubiquitously expressed in D. melanogaster and a C. elegans homolog has also been identified [25]. FastCompare returned GTAAACA, the known binding site for the DAF-16 transcription factor (published consensus GTAAACA [30,31]). DAF-16, a FOXO-family transcription factor, was shown to influence the rate of aging of C. elegans in response to insulin/insulin-like growth factor-1 signaling [31,32].
Searching for gapped motifs found few strongly conserved sites. However, when searching for 8-mers with a 5-bp gap, we found that TGGCNNNNNGCCA, the known binding site for nuclear factor I (NFI) [33], had a score comparable to those of the highest-scoring k-mers.
Several of the C. elegans sites returned by FastCompare and shown in Table 2 are known to be functional transcription factor binding sites in other species. For example, TGACTCAT, identical to the AP-1-binding site [34], is known to be bound in yeast (by Gcn4), Drosophila [35], mouse and human (see [36] for a review).
FastCompare also returns the CACGTGG motif, which is the binding site for the Myc/Max complex, a family of bHLH transcription factors [37]. Among the top-scoring motifs in Table 2, we also find AAGGTCA, the hormone response element (HRE), bound by several transcription factors in human, mouse, fruit fly and silkworm (published consensus [CT]CAAGG[CT]C[AG] [38,39]); TGACGTC, the cAMP response element (published consensus TGACGTCA [40]); CCCGCCC, the binding site for the mammalian Sp1 transcription factor (known consensus CCCCGCCCC); ATCAATCA, the known binding site for the human proto-oncogene Pbx-1 [41]. A similar site, ATCAATTA, has been shown to be bound in vitro by the Drosophila homolog of Pbx-1, the extradenticle (exd) protein [42]. Moreover, CEH-20C was identified as the C. elegans homolog of both Pbx-1 and exd. Other known sites discovered by FastCompare include CAGGTGA, similar to the known binding site for the Snail protein, a transcription factor involved in dorso-ventral pattern formation in Drosophila (published consensus [AG][AT][AG]ACAGGTG[CT]AC [43]), and TTCGCGC, the known binding site for the E2F proteins, a family of transcription factors involved in regulating the cell cycle in Drosophila and mammals (published consensus TTTCGCGC [44]). An E2F homolog has been identified in C. elegans and recently shown to be involved in cell-cycle regulation [45,46].
Position and orientation biases
As in yeast, several of the known binding sites in C. elegans appear to be constrained in terms of position. Using the distribution of median distances for all 7-mers (see Materials and methods), we found d0.025 = 690 and d0.975 = 1,135. Among the 437 highest-scoring k-mers, we found that 75 are located below the lower threshold, a proportion that is much higher than the expected 2.5% (p < 10-38). The binding sites for forkhead-related activator-4 (Freac-4), Sp1, E2F and AP-1 are particularly constrained (see Figure 6). We found only 21 k-mers to be located further away from the distant d0.975 threshold. Interestingly, the most conserved k-mer among these 21, CCACCAGGA (rank 96), is found in the upstream regions of over- or underexpressed genes in 57 microarray conditions.
Note that for a few predicted elements (for example, CAGGTGA, rank 111), the median distance falls outside of the optimal window; this is due to the fact that, for these elements, the median distance does not correspond to the peak of the distribution of distances to ATG. Hence, for these elements, the optimal window provides a better descriptor of the positional bias than the median distance. Additional analysis reveals that several of the known binding sites discovered in this study are constrained in term of orientation. For example, the binding site for the GATA-factor(s) (as shown in Table 2) is significantly more often found in the 3' to 5' orientation, relative to downstream genes. Probably the most interesting finding is that the GAGA repeats appear to be strongly oriented 3' to 5' relative to their downstream genes. Indeed, 2,375 out of 3,557 (67%) of the AGAGAGA sites are oriented 3' to 5', a proportion that is much larger than the expected 50% (p < 10-90). This bias is confirmed by the fact that TCTCTCT alone (not taking into account its reverse complement) has a much higher conservation score (129.2) than AGAGAGA (34.3). We also found that several related motifs display a similar, albeit weaker, orientation bias, for example, GAAGAAG (p < 10-16), GGAGGAG (p < 10-10). It is interesting that all the GAGA repeats found to be necessary for correct expression of the ceh-24 and unc-54 genes are in fact TCTC repeats [22,23]. The conserved sets for TCTCTCT or AGAGAGA were not found to be enriched in any GO category. Note that this orientation bias is not due to genes with the repeats in their upstream regions being predominantly located on one strand, as these genes are approximately identically distributed on each strand (1,065/1,122, p = 0.89). Interestingly, conserved GAGA repeats in D. melanogaster were also found to be constrained in terms of orientation, but at a much lower significance (p < 10-4, see below). Although it is possible that the TCTC repeats are bound at the 5' untranslated region (UTR) mRNA level, the positional distribution of the conserved AGAGAGA sites does not indicate a strong positional bias with respect to ATG (DATG = 893).
Novel predicted regulatory elements
FastCompare also returned many novel motifs; some of the most interesting ones are shown in Table 3. The top-scoring motif, CTGCGTCT, belongs to this category. A larger version of that motif, TCTGCGTCTCT, was found in a recent study to be necessary for the expression of several ethanol-response genes [47]. However, the very high conservation of this site suggests a broader role. It is interesting to note that this site was not significantly found upstream of under- or overexpressed genes in any microarray conditions (including the data from [47]). Interestingly, the most conserved k-mer found in yeast, the binding site for the Reb1 protein, had the same property. Moreover, this site displays a relatively strong orientation bias 5' to 3' (p < 10-10).
Several of the other novel predicted regulatory elements in Table 3 have interesting properties. For example, the fourth most-conserved k-mer, CGACACTCC, is one of the closest motifs to ATG, with a median distance of 234 bp, and its conserved set is strongly enriched in genes involved in positive regulation of growth (a biological process defined in GO as the increase in size or mass of all or part of the worm) (p < 10-7). Another predicted regulatory element, CGAGACC (rank 20), is found upstream of downregulated genes in 23 microarray conditions. Interestingly, it is found upstream of downregulated genes in a study measuring gene-expression changes at several time points during worm aging [48], in two distinct strains (fer-15 and spe-9;fer-15) and at similar time points (6, 9 and 10 days for fer-15, 9 and 11 for spe-9;fer-15). In addition, the functional enrichment of its conserved set points at a potential role in embryonic development (p < 10-7). Another strongly conserved and novel motif, CTCCGCCC (rank 14), was independently found upstream of almost all transcribed worm microRNA genes in a recent study [49].
Motif interactions
We found many interactions between the most conserved k-mers found at the previous stage. For example, the most conserved k-mer, TCTGCGTCT, is very often co-conserved with AGAGAGA. The high-scoring interaction between the DRE-like motif, AATCGAT and the putative E2F-binding site, TTTTCGC, also appears interesting. Indeed, the conserved sets for both k-mers are separately enriched significantly with genes involved in embryonic development, according to GO (p < 10-8 and p < 10-7, respectively). However, the conserved set of genes having both elements in their upstream regions is even more enriched in this GO category (p < 10-9). TTTTCGC also seems to interact with the novel site CGACACTCC, and the corresponding conserved set is enriched with genes involved in modification-dependent protein catabolism (p < 10-5). The full list of motif interactions is available at [9].
Flies
We applied FastCompare to the genomes of D. melanogaster and D. pseudoobscura, two species of Drosophila that diverged about 46 million years ago [50]. The number of orthologous ORFs between these two species is 11,306 and here we only consider 2,000-bp upstream regions. Using 5,000 bp instead produced similar results, but also produced additional putative binding sites (results are available at [9]). It takes approximately 10 minutes for FastCompare to process the corresponding 45 Mbp of sequences and calculate a conservation score for all 7-mers, 8-mers and 9-mers on a typical desktop PC.
Validations
The distribution of conservation scores shown in Figure 7a, for actual and randomized data, shows once again that the high conservation scores obtained with the real sequences are very unlikely to be achieved by chance. Also, as shown in Figure 7a, many known regulatory elements fall on the tail of the distribution.
As for the yeast and worm genomes, we used functional annotations (GO), expression data and known TRANSFAC sites to evaluate the FastCompare predictions. Unfortunately, expression data is often available for only a subset of genes and its analysis led to very few validations. However, Figure 7b,c clearly shows that functional enrichment of the conserved sets and TRANSFAC matches strongly correlate with conservation score. As with yeasts and worms, we focused on the 400 highest-scoring 7-mers, which are particularly well supported by the functional enrichment analysis (see Figure 7b). The simple processing described in Materials and methods yielded 469 k-mers (k = 7, 8 or 9), which we further analyze below.
Known regulatory elements
As shown in Table 4a, we found at least 16 distinct known regulatory elements among the 469 highest-scoring k-mers. The most conserved element, AACAGCTG, is similar to the site known to be bound by AP-4 (mammals) and MyoD (worms, flies and mammals). One of the most interesting predictions is TATCGATA (rank 12); this palindromic motif, known as the DNA replication-related element (DRE), has been experimentally proved to be necessary for proper expression of several cell proliferation-related genes in D. melanogaster [51] and, more recently, the genes encoding the TATA-binding protein (TBP) [52] and catalase [53] in the same organism. Interestingly, it is both the motif with the closest median distance to ATG (DATG = 168), and the most over-represented k-mer (among the 469 highest scoring ones) within D. melanogaster upstream regions compared to exons, with a ratio of 5.39.
Several of the other predicted sites are known to be bound by Drosophila transcription factors involved in development. For example, FastCompare predicts TTTATGGC (rank 14) and TAATTGA (rank 24), the binding sites for two homeodomain transcription factors. The first site matches the TRANSFAC consensus binding site for Abd-B ([CG]NTTTATGGC), while the second site is the known consensus binding site for the Antennapedia (Antp) class of homeodomain proteins [54] (TAATTGA matches the TRANSFAC consensus binding site for Ubx, a member of the Antp class). FastCompare also predicts ATTTATGC, a site matching the TRANSFAC consensus binding site for the chicken CdxA protein ([AC]TTTAT[AG]), the homolog of the Caudal protein in D. melanogaster. Also, FastCompare predicts CAGGTGC, the binding site for the Snail repressor/activator protein, a transcription factor required for proper mesodermal development [43].
FastCompare also predicts ATTTGCATA (rank 3) as one of the most conserved putative regulatory elements between the two flies. This site is the binding site for the POU-domain family of transcription factors, and it is probably bound by one or several of the three POU-domain transcription factors in Drosophila: DFR, PDM-1 and PDM-2. These three proteins are involved in different stages of Drosophila development: DFR is expressed in midline glia and in tracheal cells [55], whereas the redundant PDM-1 and PDM-2 are essential for proper neuronal development [56].
Many of the known motifs found when comparing the two Drosophila genomes were also found when analyzing the worm genomes. For example, GAGA repeats are found to be strongly conserved, slightly oriented 3' to 5' (p < 10-4), and very significantly found upstream of genes involved in morphogenesis (p < 10-23). GTAAACA (rank 147), the DAF16-binding site in C. elegans, is also one of the most conserved sites between the two Drosophila genomes. This site is probably bound by dFOXO, the unique homolog of the C. elegans DAF16 protein in D. melanogaster [57].
As for both previous phylogenetic groups (yeasts and worms), the median distances to ATG for the conserved elements show that some of the predicted regulatory elements are severely constrained in terms of position. Among the most constrained k-mers are the DRE site (TATCGATA, DATG = 168) and the known AP-4/MyoD binding site (AACAGCTG, DATG = 373). However, both the optimal windows and the median distances in Table 4a show that, compared to previously studied organisms, a smaller number of conserved regulatory element are constrained. Using the distribution of median distances for all 7-mers, we find that the d0.025 = 798 and d0.975 = 1,126. Among the 469 highest scoring k-mers, 45 fall below 798 (p < 10-13) and 36 above 1,126 (p < 10-8), once again suggesting weaker positional constraints than in yeasts and worms, at least when considering the first 2,000 bp of 5' upstream sequences.
Novel predicted regulatory elements
FastCompare predicts many putative regulatory elements in Drosophila that to the best of our knowledge are unknown (Table 4b). One of these novel sites, CAGGTAG (rank 143), was found upstream of several genes that are activated before widespread activation of zygotic transcription (which begins during the 14th nuclear cycle), in several Drosophila species [58]; it was also found to be necessary for the early expression of several of these genes (Sxl and sisterlessB) in a subsequent study (J.R. ten Bosch, J.A. Benavides and T.W. Cline, personal communication). It is interesting to see that this particular site is significantly conserved upstream of genes involved in cell fate commitment (p < 10-8).
Some of these sites, such as the palindromic TTAATTA (rank 31), are found much more often in upstream regions than in exons (with an over-representation ratio of 3.07). Others, such as ACACACAC, are found to be significantly enriched upstream of genes in known functional categories (embryonic development, p < 10-9). The same site appears to be strongly oriented 5' to 3' (p < 10-12). Others, such as GTGTGACC or AAATGGCG, appear to be located closer to ATG than most other sites (DATG = 296 and 592, respectively).
Motif interactions
We found many potential interactions between the most conserved sites discovered by FastCompare. For example, the POU-domain-binding site ATTTGCATA was found to be strongly co-conserved with TAATTGA, the Antp-binding site, and with many other potential homeodomain sites, such as AATAAAT and TAATTAA. The CACA repeats were also found to be co-conserved with several different sites, and in some cases, the set of genes having both sites simultaneously conserved in their upstream regions (conserved sets) was found to be enriched in certain functional categories, for example, ACACACAC and GAGAGAG, regulation of transcription (p < 10-12); ACACACAC and TAATTGC (an Antp variant site), embryonic development (p < 10-5). The full list of interactions is available at [9].
Mammals
The much larger noncoding regions of mammalian genomes present significant challenges for computational motif discovery. Also, many repeat elements (for example, Alu) have colonized mammalian genomes and are likely to be conserved between closely related genomes. The distance between enhancers and the transcriptional start of the genes they regulate can be extremely large, reaching tens of kilobases. Finally, gene predictions and gene boundaries are still largely unverified experimentally for a large number of genes.
We applied FastCompare to the genomes of H. sapiens and M. musculus,, which diverged about 75 million years ago [59]. The number of orthologous ORFs between these two species is 15,983 and again, we have only considered 2,000-bp upstream regions. As in flies, using 5,000-bp instead produced similar results. It takes approximately 15 minutes for FastCompare to process the corresponding 60 Mbp of sequences and calculate a conservation score for all 7-mers, 8-mers and 9-mers on a typical desktop PC.
Validations
Unlike the other genomes considered so far, the output of FastCompare from the mammalian genomes is dominated by GC-rich sequences, probably corresponding to CpG islands (GC-rich regions known to be associated with the promoters of many genes). However, analysis of the FastCompare output yielded the same validations as for other species. Indeed, the distribution of conservation scores obtained on actual and randomized sequences shows that high conservation scores are very unlikely to be obtained by chance (Figure 8a). As with other species, many known regulatory elements are on the tail of the distribution (Figure 8a). Also, as shown in Figure 8b-d, more k-mers are found upstream of over or underexpressed genes, more k-mers have their conserved set enriched with GO functional categories, and more k-mers match TRANSFAC consensus sites as the conservation score increases.
We found that masking Alu repeats did not influence the output of FastCompare (data not shown). To overcome the overabundance of GC-rich sequences in the FastCompare output, we use longer k-mers as starting points, namely 8-mers instead of 7-mers. We started with the 600 highest-scoring 8-mers, and replaced each of these 8-mers by one of its substrings (7-mer) or one of its superstrings (9-mer), when their conservation score is higher. We then removed duplicates in the list and added the high-scoring 9-mers that have no substrings within the list. This procedure yielded 284 k-mers (k = 7, 8, 9). Subsequent validation was limited to this small set of high-scoring predictions.
Known regulatory elements
As shown in Table 5a, we found 17 distinct known regulatory elements among the 284 highest-scoring k-mers. Among these are the well characterized sites for the Sp1, C/EBP, CREB and Myc/Max proteins or families of proteins. These four sites reside very close to ATG (their median distance to ATG is between 100 and 250 bp), suggesting that the four proteins (or families of proteins) may be involved in intimate interactions with the transcriptional complex. Sp1 is an ubiquitous transcription factor, involved in the basal expression of a large number of genes in mammals (see [60] for review). The CCAAT/enhancer binding protein (C/EBP) has been implicated in the regulation of cell-specific gene expression mainly in hepatocytes, adipocytes and hematopoietic cells (see [61] for review). Both Sp1 and C/EBP are constitutive transcription factors whose presence is necessary for significant induction of a large number of genes [62]. The CRE-binding protein (CREB or CBP) is a transcription factor that binds cyclic AMP (cAMP) response elements (CREs) in the promoters of specific genes, and functions as a co-activator for a large number of other transcription factors (see [63] for review). The Myc/Max heterodimer binds the CACGTG sequence, and also acts as a transcriptional activator (see [64] for review).
Interestingly, we found that some of the most conserved interactions between k-mers (see Materials and methods) involve Sp1-binding sites (CCCGCCC or CCGCCCC) with other known sites such as CACGTGAC (Myc/Max), TGACGTCA (CREB), CGCAGGCGC (unknown), GCCAATC (CCAAT-box) and ACTTCCG (Ets), and that the median distances between these sites are relatively small (138, 164, 200.5, 234 and 234, respectively).
Among the other predicted regulatory elements returned by FastCompare are CCGCCTC, a site known as the insulin response element [65]; CGGAAGTGA, a site known to be bound by the GA-binding protein in human [66]; CGCATGCG, a site known as the palindromic octamer sequence, which was found at 132 bp (relative to ATG) upstream of the inosine-5'-monophosphate dehydrogenase type II gene in human, and shown to be functional in resting and activated T cells using site-directed mutagenesis, in vivo footprinting and electrophoretic mobility shift assay (EMSA) [67]; TTTCGCGC, the E2F-binding site; TAATCCCAG, a site known to be bound in D. melanogaster by the anterior morphogen Bicoid, and also recently shown to be bound in human by Goosecoid-like (GSCL) [68]. Interestingly, this site has a relatively strong orientation bias 3' to 5' (p < 10-14). It is also the site with the strongest over-representation in upstream regions compared to exons that we observed, with a ratio of 7.06.
FastCompare also predicts ATTTGCAT, the binding site for the POU-domain Oct-1 and Oct-2 proteins, known to bind the promoter and intronic enhancer of immunoglobulin genes [69]; it also returns GGAAGTCCC, a site that was shown to bind NFκB [70,71], a transcription factor involved in a variety of pathways (including inflammation, response to infection and oxidative stress, and apoptosis).
The distribution of distances to ATG for all 7-mers (Figure 9) shows an interesting bimodal shape, indicating that a large number of short sequences are constrained to reside around 500 bp to ATG. We calculated d0.025 = 342 and d0.975 = 1,185 and found that 83 k-mers among the 284 highest-scoring ones have a shorter median distance than 342 (p < 10-63) and only 11 have a larger median distance than 1,185. Indeed, a majority of the known sites identified by FastCompare are preferentially located near the 5' start of genes, with some elements being very close to ATG (for example, the CREB site, whose median distance to ATG is 107, whereas the optimal window is [0;1,000]). Nonetheless, a few known motifs do not seem to show any positional constraints. For example, the Bicoid-like site TAATCCCAG has a median distance to ATG of 1,258.
Novel predicted regulatory elements
FastCompare identifies many putative regulatory elements which to the best of our knowledge are novel (Table 5b). Some of these predicted regulatory elements are found upstream of over- or underexpressed genes in many microarray conditions. One example is CCCCAGC, which is significantly found upstream of overexpressed genes in 21 conditions (out of 30) of the human cell-cycle experiment [72]. Other conserved elements are found much more often in upstream regions than in exons, for example, CCCCTCCC or TCTCGCGA, with ratios of 5.12 and 4.45, respectively. Others appear to be positionally constrained, for example, the palindromic CTGCGCA with an optimal window [0;300] and a median distance to ATG of 199, or constrained in terms of orientation, for example, GTGAGCCAC, which is significantly oriented 5' to 3' (p < 10-6).
Inter-groups comparisons
To gain a better understanding of the network-level conservation of regulatory elements between the different phylogenetic groups, we compared the results we obtained by applying FastCompare to yeasts, worms, flies and mammals in the previous sections. We calculated the overlap (and its significance) of the 400 highest-scoring 7-mers and 8-mers found for each phylogenetic group. As shown in Table 6a,b, the number of shared predicted sites correlates with phylogenetic distance (the number of high-scoring putative motifs that two phylogenetic groups have in common decreases as the phylogenetic distance between the groups increases). All of the overlaps were found to be statistically significant, except for the yeast-human comparison. For both 7-mers and 8-mers, the best overlap is the one obtained between the two invertebrate phylogenetic groups: worms and flies. Indeed, simple observation of the identified known regulatory elements in Tables 2 and 4a reveals that these two organisms have a large number of predicted binding sites in common.
However, when we looked at the overlap between conserved sets for identical high-scoring k-mers in different phylogenetic groups (after determination of reciprocal best BLAST hits between the considered species), we found little overlap. The only significant overlap we found (after Bonferroni correction) was between the GATA sites (GATAAGA) in worm and fly (p = 2.5 × 10-4). As a control, we performed the same analysis within the yeast phylogenetic group, using the S. cerevisiae/S. bayanus and S. paradoxus/S. mikatae 400 most conserved 7-mers. One hundred and ninety-five sites were found in both groups of 7-mers, and for all of them, the overlaps between the conserved sets obtained separately in the S. cerevisiae/S. bayanus and S. paradoxus/S. mikatae analyses were highly significant, with hypergeometric p-values < 10-40. Therefore, our results strongly suggest that, while transcription factors have largely retained their ability to recognize specific DNA sites, their targets have largely changed through appearance or disappearance of those binding sites in promoters. This hypothesis is supported by recent analysis of the fission yeast cell cycle using microarrays, which showed that the role and the binding sites for several of the main transcription factors involved in regulating the yeast cell cycle (Swi4/Mbp1, Fkh1/Fkh2, Swi5/Ace2) are conserved between budding and fission yeasts (which diverged about 1 billion years ago), but the sets of genes that they regulate overlap much less than expected (only about 50 orthologous genes are cell-cycle-regulated in both species) [73].
It is particularly interesting to consider the seven 8-mers that are top predictions for all three multicellular phylogenetic groups (note that many more 7-mers are conserved between these groups). These sites include the CRE (TGACGTCA, GACGTCAC and ATGACGTC), the POU-domain binding site (ATTTGCAT), and the HRE (CAAGGTCA). A fourth site is also shared (GCCACGCC, CCACGCCC), which to the best of our knowledge is a novel motif. Its strong over-representation in upstream regions compared to coding regions, and its closeness to ATG (median DATG = 230 for GCCACGCC) make it a promising candidate for experimental testing. Interestingly, the location constraints on these conserved sites can vary across phylogenetic groups. For example, the CRE appears weakly constrained in worms and flies in terms of distance to ATG (DATG = 708 and 825, respectively), but is very close to ATG in mammalian genomes (DATG = 107). However, the distances to ATG of the POU-domain-binding sites (862, 882 and 729, respectively) indicate that their positional constraints are shared among the phylogenetic groups. The same holds for the HRE binding site (845, 1,015.5 and 895, respectively).
Discussion and conclusions
We have presented a powerful approach for discovering transcriptional regulatory elements that are globally conserved between pairs of genomes. Our approach requires only two unaligned genomes, thus allowing the use of genomes of arbitrary divergence and those with extensive rearrangements of noncoding regions. Moreover, our motif-finding strategy does not use any parameters other than a conservation score threshold, used to separate presumptive functional from nonfunctional motifs. We have shown that such thresholds can be roughly estimated using independent biological data, when available. Our approach is also computationally efficient: whole eukaryotic genomes can be processed in minutes on a typical computer. In turn, this efficiency allows FastCompare to explore exhaustive pattern lists.
Our results show that FastCompare can recover most of the known functional binding sites in S. cerevisiae when its upstream regions are compared to those of a related species, S. bayanus. We comprehensively explored the globally conserved motif content between worms, flies and mammalian genomes, discovering large sets of known and novel motifs. The use of external information (expression data, functional categories, TRANSFAC, chromatin IP and known motifs) clearly shows that our method is able to detect conserved and functional motifs in all the phylogenetic groups that we studied. In all analyses, we have shown that some of the discovered known or novel motifs were severely constrained, either in terms of position relative to the start of translation or in orientation. We also observed that some of the known or novel motifs are co-conserved within upstream regions, potentially revealing interactions between the (often unknown) transcription factors that bind them.
We have created a set of web tools to superimpose the most globally conserved k-mers discovered by FastCompare to user-supplied sequences or multiple alignments. An example is shown in Figure 10a, in which the upstream regions of the STE2 gene (encoding the alpha-factor pheromone receptor) from four different yeast species were aligned using ClustalW, and the most globally conserved k-mers are highlighted. All experimentally determined sites for STE2 were also predicted to be globally conserved by FastCompare. Moreover, several other sites also appear to be conserved, both at the global level (predicted by FastCompare) and the local level (shown by the multiple alignment). In Figure 10b, the same analysis was performed on only two orthologous upstream regions instead of four. Many more sites appear to be locally conserved than when using four species, but the globally conserved sites found by FastCompare allow the efficient selection of experimentally verified and putative binding sites. These tools should be particularly useful in designing stepwise promoter deletions and site-directed mutagenesis experiments for understanding the regulatory code of specific genes.
While powerful, our approach has potential limitations. Our current approach allows matches to a given k-mer to be on different strands within pairs of orthlogous upstream regions. This flexibility substantially increases the number of k-mers that are supported by independent biological data (that is, true positives), at least for yeasts and worms (data not shown). However, it is difficult to evaluate whether this flexibility introduces more true positives than false positives. Also, transcription factors often bind several slightly distinct sites with different affinities, and it is widely acknowledged that binding-site degeneracy is better captured by using position-weight matrices (PWM) instead of k-mers or consensus patterns [74]. To evaluate whether weight matrices would display better conservation scores, we calculated a conservation score for weight matrices corresponding to 20 well characterized yeast binding sites, and compared them to the conservation scores obtained for the best k-mers that unambiguously correspond to the same binding sites. Conservation scores for weight matrices were calculated as described for k-mers in Materials and methods, except that we used the weight-matrix score thresholds that maximize the significance of the overlap between the two sets of ORFs containing matches to the weight matrices in each species. This involves progressively lowering the score threshold by small increments, and for each threshold, calculating the overlap and its hypergeometric p-value. We then choose the score threshold corresponding to the most significant p-value, and use the negative natural logarithm of this p-value as the conservation score. As shown in Table 7, only in 11 cases out of 20 did weight matrices have a higher conservation score than the corresponding k-mers. These results suggest that k-mers provide results that are almost as good as those obtained using weight matrices, when utilizing the network-level conservation criterion. One reason why, in many cases, k-mers have a higher conservation score than weight matrices may have to do with the more narrow selection of k-mers for binding sites with similar or identical affinities. In fact, we recently showed that PWM scores, widely seen as proxies for binding affinity, are statistically conserved in a comparison between S. cerevisiae and S. bayanus [6]. In the context of the present study, the different k-mers representing each transcription factor binding site may be defining affinity classes that are more strongly conserved than a looser definition of a binding site represented by a weight matrix. Recent work in bacteria has established the importance of binding affinity, especially with respect to coordinating the temporal order of events [75].
However, Table 7 shows that the conservation score for weight matrices describing very degenerate binding sites, such as RAP1, is significantly higher than the conservation score obtained for the best corresponding k-mer. This suggest that our k-mer based approach is limited in its ability to discover highly degenerate binding sites.
As shown by our inter-group analysis, many regulatory elements have remained functional across evolution, but few have remained upstream of the same genes. The network-level conservation principle thus appears less applicable to species that diverged very long ago. For example, when we compared the Drosophila and mosquito genomes (which diverged approximately 400 million years ago), we only found a handful of k-mers (interestingly including GATA-factor and Myc/Max binding sites) to have conservation scores above those obtained from randomized data.
There are also several directions in which our approach could be extended. From a methodological standpoint, the approach could be extended to take into account local over-representation of identical or nearly identical copies of the same binding sites, a well known feature in the promoter regions of higher eukaryotic species [16]. To discover highly degenerate regulatory elements, k-mers could be used to seed weight matrices whose individual weights could be optimized for network-level conservation, using stochastic optimization procedures (for example, simulated annealing; Mike Beer, personal communication). Introns and downstream noncoding regions could also be explored using our approach, as these regions are known to harbor functional regulatory elements in metazoan genomes. While our approach can deal with genomes presenting arbitrary levels of divergence and rearrangements, it would be interesting to investigate how global alignments or suboptimal and non-overlapping local alignments [76] could be used to filter out regions of non-conservation. This approach would be particularly interesting when analyzing very long upstream regions, in order to increase the signal-to-noise ratio. Finally, mRNA 3' UTRs could be compared in order to find specific downstream regulatory elements involved in post-transcriptional mRNA regulation (for example, mRNA localization, decay or translational repression).
Materials and methods
Outline of approach
First we determined orthology relationships between ORFs on the basis of reciprocal best BLAST hits (Figure 1a) and extracted the corresponding upstream regions from the genome sequences. Then, we considered every possible short DNA sequence of length k (k-mer, with k between 7 and 9) as a candidate regulatory element. For each k-mer, we found the set of ORFs whose upstream regions contain at least one exact match to the k-mer, anywhere in the upstream region, in the first genome. We did the same for the second genome, obtaining another set of ORFs. Then, we calculated the overlap between the two sets and assessed its statistical significance (Figure 1b). The statistical significance of the overlap provides a measure of conservation with which we score and rank every possible k-mer (Figure 1c). Note that our approach is very different from the classical k-tuple DNA sequence-analysis methods [77,78], which are not based on comparative genomics and are local methods; that is, they only deal with single promoters or small sets of functionally related promoters (while our approach provides a genome-level measure of conservation for candidate regulatory elements).
Sequence sources and orthology determination
Sequence data were downloaded from the Saccharomyces Genome Database (SGD) for all yeast species considered in this paper; worm (C. elegans and C. briggsae), Drosophila (D. melanogaster), human (H. sapiens) and mouse (M. musculus) sequence data were downloaded from Ensembl [79]. The D. pseudoobscura genome sequences (contigs) were downloaded from [80]. The upstream regions used in this study are immediately adjacent to the ATG codons of their downstream genes, and are 1-kb long (yeasts) or 2-kb long (worms, flies and mammals). Note that transcription-factor-binding sites generally reside in the region situated upstream of the transcription start site. Unfortunately, not all genes have well annotated transcription start sites. This problem should not, however, strongly influence the output of FastCompare, as distances between start of transcription and start of translation should be at most on the order of a few hundred base-pairs (except in certain cases, for example when 5' UTRs are interrupted by long introns). However, as gene structures become better annotated (mainly as a result of massive cDNA sequencing projects) and promoter regions become more accurately delimited, we expect that the ability of FastCompare to discover regulatory elements will be significantly improved.
Orthology information provided by Ensembl or by Kellis et al. [4] was used throughout this study, when available. Ensembl provides strong homology relationships between genes from different species, but does not provide reciprocal best matches. Therefore, we determine reciprocal best matches using the provided sequence identity between homologous genes. When orthology information is not available in Ensembl (for example, between D. melanogaster and D. pseudoobscura, or between distant species such as S. cerevisiae and C. elegans), we determine orthologs using the reciprocal best BLAST hits approach.
Motif-finding algorithm and simple clustering
Given a value of k, we first generated the set of all possible k-mers and removed half of them on the basis of reverse complementarity. We also removed k-mers with very low complexity and which are over-abundant in the intergenic regions of the genomes we analyzed (that is, those that contain k - 1 or more As or Ts), as these sequences are unlikely to be regulatory elements. Every remaining k-mer (that is, 8,170 for k = 7) is then considered as a candidate regulatory element. For each k-mer, we found the set of ORFs in the first species that have at least one exact occurrence of the k-mer in their upstream regions. We then found the set of ORFs in the second species that have at least one occurrence of the same k-mer in their upstream region. Importantly, the matches can be anywhere in the upstream regions: they do not have to be at the same positions in two orthologous upstream regions (as with multiple alignment) and can be on any strand. Since both functional and non-functional elements are expected to be conserved between two closely related species, the two sets are expected to overlap. However, under the network-level conservation principle, the extent of the overlap - and therefore its statistical significance - will be even greater for k-mers that represent functional transcription factor binding sites. The significance of the overlap can be measured using the hypergeometric distribution. The probability of two sets of size s1 and s2, drawn from a set of N elements, to have i or more elements in common is given by :
In this way, all k-mers can be ranked by their hypergeometric p-values. It is important to note that due to basal conservation (that is, conservation arising from common ancestry), the hypergeometric p-values will generally be very small for most k-mers. Therefore, we only use these p-values as relative measures of network-level conservation and focus on k-mers with the greatest conservation. For simplicity, we define the 'conservation score' to be the negative logarithm (base e) of the hypergeometric p-value obtained for a given k-mer. Therefore, the more extensive the overlap between the two sets, the higher the conservation score. Also, for the same k-mer, we call 'conserved set' the set of ORFs corresponding to the overlap between the two sets of orthologous ORFs containing at least one exact match to the k-mer in their upstream regions. Conserved sets are used throughout this study to get insights into the function of the most conserved k-mers, using functional annotation [81,82], chromatin IP [1], known motifs, and to evaluate whether these k-mers are constrained in terms of position or orientation.
The current FastCompare implementation handles k-mers with a user-specified gap (termed gapped k-mers), which is a straightforward extension of the approach described above. The conservation score returned by FastCompare is independent of the size of the patterns (that is, the value of k); therefore k-mers with different sizes, and gapped k-mers (for example, CGTNNNNNNTGA) can be compared.
We use the following strategy when applying FastCompare to pairs of genomes. First, we calculate conservation scores for all 7-mers, 8-mers and 9-mers. We then retain only the m highest-scoring 7-mers, with m chosen according to independent biological data (alternatively, m could be chosen according to the estimated number of transcription factors in the species being considered). We then replace each of the retained 7-mers by an 8-mer (if there is one) with higher conservation score for which the considered 7-mer is a substring. We also include within the final list the 8-mers which do not have any substrings within the m 7-mers. We then repeat the same process for the retained 8-mers, replacing each of them by its higher scoring 9-mer superstring if there is one, and add the 9-mers that do not have any substring within the 8-mers. This strategy thus allows the optimal length for candidate regulatory elements to be determined.
FastCompare is implemented in the C language and uses efficient data structures (hash tables and prefix trees [83]). For a given value of k, the worst-case time complexity is O(kn + 4k(p + k)), where n is the total amount of upstream sequences and p is the total number of orthologous pairs. Note that the first term is generally much larger than the second one; therefore the complexity of our approach can be seen as linear in the combined sizes of the genomes to be compared (when k is restricted to 7, 8 and 9). The calculation of hypergeometric p-values involves factorials of large integers, so we use specialized C routines, as described in [84]. FastCompare runtimes provided in the Results section are obtained using a standard desktop PC (2.0 GHz CPU, 1 GB RAM).
Discovering positional constraints for conserved regulatory elements
As described in Results, we applied FastCompare to 1 kb (yeast) or 2 kb upstream regions (worms, flies and mammals). While these lengths are reasonable, they are somewhat arbitrary, and it is known that some regulatory elements are constrained to be within specific distances (often shorter than 1 kb) from the start of transcription, reflecting mechanistic constraints for transcription factor-transcription factor or transcription factor-RNA polymerase interactions [11]. Moreover, some regulatory elements have orientation biases (see [11,12] for examples). To discover such constraints, we analyzed the most conserved k-mers found at the previous stage in the following ways.
First, for each high-scoring k-mer, we calculated the median distance to ATG (as the start of transcription is generally not known) for the set of all (non-overlapping) occurrences of this k-mer within the upstream regions of its conserved set (see previous section for a definition of the conserved set of a given k-mer). To statistically assess whether the median distance to ATG for a given k-mer is unusually small or large, we built the distribution P(d) of median distances to ATG, for the entire set of 8,170 7-mers. We first created a histogram by binning the median distances to ATG for all 7-mers into 20-bp bins, and then smoothed the histogram (using a normal kernel and a bandwidth of 50 as implemented in the ksmooth function of the R statistical software package). Then, using numerical integration, we sought the distance thresholds d0.025 and d0.975 such that P(d <d0.025) = 0.025 and P(d <d0.975) = 0.975. We then considered the median distance to ATG for a given k-mer as unusually short or long when it is less than d0.025 or greater than d0.975, respectively.
For each k-mer, we also sought the sequence window which maximizes the conservation score by progressively shortening all upstream regions (all having equal lengths) by 100 bp increments from the 5' end. Then, we did the same from the 3' end using the optimal 5' end found in the previous step. Evaluating every possible window whose length is a multiple of 100 bp almost always yields identical results. We then calculated the conserved sets for these windows, and output the orientation (strand) for each k-mer occurrence within its conserved set (palindromes were counted on both strand).
Finally, using the results of the previous step, for each k-mer, we used the binomial distribution to assess whether the proportion of occurrences of this k-mer (within its conserved sets) on one strand is significantly smaller (or larger) than 0.5. Binomial p-values less than 0.05 (after Bonferroni correction) are considered significant.
Motif interactions
It is now known that the regulatory code governing the expression of genes is combinatorial [11,85,86]. The network-level conservation principle can be trivially extended to discover interactions (that is, co-conservation) between two k-mers. To focus on heterotypic interactions, we only examined k-mers that differ by more than l nucleotides, after optimal ungapped alignment. We tested several values of l and found that l = 4 was most appropriate when using 7-, 8- and 9-mers. Then, we proceeded as described above, except that instead of seeking two sets of ORFs (one for each species) whose upstream regions contain a single k-mer, we sought the two sets of ORFs that contain the two k-mers simultaneously. Once these two sets were available, we evaluated the extent of their overlap as described above, and rank interaction pairs according to their conservation score.
Validations
We used randomized data to show that high conservation scores (obtained as described above) are unlikely to be obtained by chance, and independent biological information to assess the ability of FastCompare to predict functional regulatory elements by giving them a high conservation score. We also estimated the over-representation of predicted regulatory elements in upstream regions compared to coding regions.
Validation using randomized data
Our goal was to generate new pairs of upstream regions that are conserved at the same level of divergence as the actual sequence data. We align each pair of orthologous sequences using the Needleman-Wunsch algorithm [87], and calculate substitution frequencies between all pairs of nucleotides (A → A, A → T, and so on). Then, we reconstructed new pairs of orthologous sequences by mutating one of the sequences in each initial pair using the estimated frequencies. Generating the sequences to be mutated using locally estimated first-order Markov models yielded the same results.
Validation using independent biological information
The proportions of 7-mers supported by each type of independent data, as presented in Figures 3, 5, 7 and 8, is calculated as follows. In these figures, support for a given 7-mer is considered as binary, and depends on whether the 7-mer meets the particular validation criterion or not (or whether it is found upstream of over- or underexpressed genes, in at least one microarray condition, see below). 7-mers are first sorted by conservation score, and the proportion of supported 7-mers were calculated using a sliding window of 100 7-mers. For each window and each type of independent biological data, we simply calculated the number of 7-mers for which support is available and divided this number by 100.
Functional annotations
Yeast (S. cerevisiae), worm (C. elegans), fly (D. melanogaster) and human (H. sapiens) functional categories and corresponding ORF annotations were downloaded from the MIPS [88] and GO [89] websites. The statistical significance of the functional enrichments within sets of ORFs was evaluated using the hypergeometric distribution, as discussed above. Hypergeometric p-values for functional enrichment were not corrected for multiple testing, but only p-values smaller than 10-4 are reported, providing a slightly less stringent thresholds than Bonferroni corrections.
Known motifs
Weight matrices corresponding to known yeast motifs were obtained from Gibbs sampling-based motif finding on chromatin IP data [1], functional categories and clusters of co-expressed genes [85]. Only high-confidence binding sites (that is, sites confirmed by several sources including the literature) were included in our list of known motifs. We label a given k-mer as a known motif if it meets the following two criteria. The first is significant overlap (p < 10-4) between the conserved set for the given k-mer and the set of ORFs whose upstream regions contain at least one match to the known motif (the sets of ORFs were defined using ScanACE with the weight matrix for the known motif, and with the standard average minus two standard deviations threshold [7]). The second criterion is strong sequence similarity between the considered k-mer and the known motif weight matrix. To evaluate this similarity, we turn the considered k-mer into a weight matrix of 0s and 1s, and use CompareACE [7] to calculate the Pearson correlation between the weights of this matrix and the weights of the known motif weight matrix; correlation coefficients > 0.65 are considered significant. Finally, for a given k-mer, we report the known motif for which the above hypergeometric p-value is the smallest.
In vivo binding data (chromatin IP)
Genome-wide binding locations were previously evaluated for 106 transcription factors in S. cerevisiae [1]. For each transcription factor, we retain the set of ORFs with p-value < 0.001 (see [1] for details of the error model). To evaluate a given k-mer with respect to chromatin IP, we evaluate the statistical significance of the overlap between the conserved set of the considered k-mer and the set of ORFs defined as described above corresponding to each transcription factor. We report the most significant chromatin IP enrichment, with hypergeometric p-value < 10-4.
TRANSFAC
The 309 weight matrices and corresponding consensus patterns for known transcription factor binding sites were downloaded from [90,91]. k-mers were then simply matched to the consensus patterns. We eliminated consensus patterns that match too many k-mers, by matching each of them to all (8,170) 7-mers and removing consensus patterns that matched more than 50 7-mers.
Microarray expression data
Expression data for all species considered were downloaded from diverse sources [92,93]. Overall, we downloaded 765 microarray conditions for S. cerevisiae, 555 conditions for C. elegans, 156 conditions for D. melanogaster, and 1,384 conditions for H. sapiens. We use these expression data in the following way.
We evaluated the over-representation of each k-mer in the upstream regions of genes that are themselves over- or underexpressed in certain microarray conditions. Over- or underexpressed genes are operationally defined as having a log ratio of intensity above average plus two standard deviations, or below average minus two standard deviations, respectively (averages and standard deviations are calculated for each condition; using fold changes instead of standard deviations produced roughly the same results). To evaluate the over-representation of a given k-mer in a given microarray condition, we defined as O1 the set of overexpressed genes in this condition, and as O2 the set of ORFs whose upstream regions contain at least one occurrence of the considered k-mer, genome-wide. Then, we evaluated the significance of the overlap between O1 and O2 using the hypergeometric distribution, as described above. Overlaps whose hypergeometric p-value is smaller than 0.05 (after Bonferroni correction) were considered significant. We proceeded separately with the set of underexpressed genes in the same way. The total number of microarray conditions (overexpressed plus underexpressed) for which a k-mer was found to be significantly over-represented is reported. Note that we do not use the conserved set for the considered k-mer here, as we do not want to restrict our analysis to orthologous genes. Indeed, except for yeast, microarrays often contain only a fraction of all genes within the considered organism. In these cases, the overlap between conserved sets and over- or underexpressed genes can be very small, reducing statistical power. Using all genes, therefore, increases our power to detect significant associations, while retaining a uniform approach for all species considered.
Over-representation in upstream regions compared to coding regions
As shown in [94] for the yeast RAP1 transcription factor, some transcription factors bind intergenic regions much more frequently than they bind coding regions. Consequently, it is expected that sequences corresponding to regulatory elements are more often present in intergenic regions than in coding regions. To evaluate this bias, we calculate the ratio of the number of genes that have the k-mer in their upstream regions over the number of genes that have the k-mer in their coding regions (using only exons), and we correct this ratio using the average length of the upstream and coding regions.
Availability
The FastCompare implementation, all the sequences, and results are available on our website [9].
Acknowledgements
We thank David Stern, Mike Beer, Chang Chan, Yir-Chung Liu and two anonymous reviewers for providing helpful comments on the manuscript, Mike Beer for providing weight matrices for known transcription factors in yeast and the other members of the Tavazoie laboratory for helpful discussions. S.T. is supported in part by grants from NSF CAREER, DARPA, and NIH.
Figures and Tables
Figure 1 Overview of the FastCompare approach. (a) Determination of orthologous pairs of ORFs, and extraction of the associated upstream regions (data not shown). (b) For each k-mer (here CACGTGA), determination of the sets of ORFs that contain it in their upstream regions, in each species separately. The conservation score (hypergeometric p-values to assess the overlap between both sets) is then calculated. (c) Ranking of all k-mers on the basis of their conservation scores.
Figure 2 Distributions of conservation scores for actual (red) and randomized (black) data obtained when applying FastCompare to S. cerevisiae and S. bayanus. Both distributions were constructed using bin sizes of 5. The top portion of the figure is not shown for the purpose of presentation. The distributions show that high conservation scores are unlikely to be obtained from randomized data. Also, a large number of 7-mers on the tail of the distribution correspond to experimentally verified transcription-factor-binding sites in yeast.
Figure 3 Proportions of 7-mers supported by different types of independent biological data ((a) known motifs, (b) chromatin-IP, (c) functional enrichment, (d) under/overexpression, (e) TRANSFAC; windows of size 100 were used to construct the figures, see Materials and methods) as a function of the conservation score rank, obtained when applying FastCompare to S. cerevisiae and S. bayanus. (a-e) strongly indicate that the frequency of support increases with conservation score as calculated by FastCompare.
Figure 4 Distribution of median distances to ATG of all 7-mers, obtained when applying FastCompare to S. cerevisiae and S. bayanus. For each 7-mer, a median distance to ATG was calculated using the positions of matches upstream of S. cerevisiae genes within the conserved set for this 7-mer. The 8,170 median distances were then binned into 20-bp bins, and the resulting histogram was smoothed using a normal kernel. The median distances for several known binding sites in S. cerevisiae are also indicated (see Table 1).
Figure 5 Validation of the conservation scores obtained when applying FastCompare to C. elegans and C. briggsae. (a) Distributions of conservation scores for actual (red) and randomized (black) data, showing that high conservation scores are unlikely to be obtained by chance. Conservation scores for some known regulatory elements are also indicated. Both distributions were constructed using bin sizes of 5, and the top portion of the figure is not shown for the purpose of presentation. (b-d) Proportion of 7-mers supported by different types of independent biological data (using windows of size 100, see Materials and methods) as a function of the conservation score rank, obtained when applying FastCompare to C. elegans and C. briggsae. (b-d) indicate that the frequency of support increases with conservation score as calculated by FastCompare.
Figure 6 Distribution of median distances to ATG of all 7-mers, obtained when applying FastCompare to C. elegans and C. briggsae. For each 7-mer, a median distance to ATG was calculated using the positions of matches upstream of C. elegans genes within the conserved set for this 7-mer. The 8,170 median distances were then binned into 20-bp bins, and the resulting histogram was smoothed using a normal kernel. The median distances for several known binding sites in C. elegans are also indicated.
Figure 7 Validation of the conservation scores obtained when applying FastCompare to D. melanogaster and D. pseudoobscura. (a) Distributions of conservation scores for actual (red) and randomized (black) data, showing that high conservation scores are unlikely to be obtained from randomized data. Conservation scores for certain known regulatory elements are also indicated. Both distributions were constructed using bin sizes of 5, and the top portion of the figure is not shown for the purpose of presentation. (b, c) Proportion of 7-mers supported by different types of independent biological data (using windows of size 100, see Materials and methods) as a function of the conservation score rank, obtained when applying FastCompare to D. melanogaster and D. pseudoobscura. (b, c) strongly indicate that the frequency of support increases with conservation score as calculated by FastCompare.
Figure 8 Validation of the conservation scores obtained when applying FastCompare to H. sapiens and M. musculus. (a) Distributions of conservation scores for actual and randomized data, showing that high conservation scores are unlikely to be obtained by chance. Conservation scores for some known regulatory elements are also indicated. Both distributions were constructed using bin sizes of 5, and the top portion of the figure is not shown for the purpose of presentation. (b-d) Proportion of 7-mers supported by different types of independent biological data (using windows of size 100, see Materials and methods) as a function of the conservation score rank, obtained when applying FastCompare to H. sapiens and M. musculus. (b-d) strongly indicate that the frequency of support increases with conservation score as calculated by FastCompare.
Figure 9 Distribution of median distances to ATG of all 7-mers, obtained when applying FastCompare to H. sapiens and M. musculus. For each 7-mer, a median distance to ATG was calculated using the positions of matches upstream of H. sapiens genes within the conserved set for this 7-mer. The 8,170 median distances were then binned into 20-bp bins, and the resulting histogram was smoothed using a normal kernel. The median distances for several known binding sites in H. sapiens are also indicated.
Figure 10 Partial representation (most proximal region) of the aligned 1 kb upstream regions of the S. cerevisiae STE12 gene and its orthologs. (a) The highest scoring 7-mers found by FastCompare in a comparison between S. cerevisiae and S. bayanus are highlighted. FastCompare correctly predicts the conserved and experimentally verified binding sites for Mcm1, Matalpha2 and Ste12 (proximal) (see [8] for review). A more distal non-verified binding site for Ste12, and a RRPE site close to the distal Matalpha2 are conserved between the four species, and also predicted by FastCompare. FastCompare predicts several nonconserved sites in each species. For example, in S. cerevisiae, it identifies a Rox1-binding site overlapping with the second Ste12 site, and a putative Upc2-binding site. (b) Aligned 1 kb upstream region of the S. cerevisiae STE2 gene and its S. paradoxus ortholog only, with the same highlighted 7-mers as in (a). Since the two yeast species diverged very recently, the two upstream regions appear highly conserved. However, using the FastCompare output allows efficient selection of verified and putative binding sites. CER, S. cerevisiae; Bay, S. bayanus; Par, S. paradoxus; Mik, S. mikatae.
Table 1 Known regulatory elements obtained when applying FastCompare to S. cerevisiae and S. bayanus
Name Sequence Rank DATG WATG U/C Motif ChIP Experiment Best MIPS enrichment
Bas1 AAGAGTCA 159 307 [0;500] 1.24 BAS1 - 2(1/1) Amino-acid metabolism (p < 10-6)
Cbf1 CACGTGA 3 368 - 2.70 CBF1 CBF1 6(3/3) Amino-acid metabolism (p < 10-6)
Ecm22/Upc6 TAAACGA 59 362 [100;500] 1.36 - - 11(9/2) Lipid, fatty-acid and isoprenoid biosynthesis (p < 10-8)
Fkh1/2 TAAACAAA 88 353 - 1.73 FKH1 FKH2 2(1/1) -
Gcn4 TGACTCA 160 323.5 [0;400] 1.02 GCN4 GCN4 102(76/26) Amino acid biosynthesis (p < 10-29)
Gcr1 TGGAAGC 260 663 [600:1000] 1.24 GCR1 - 4(4/0) -
Gis1 AAGGGAT 207 402.5 [100;800] 1.31 GIS1 - 1(1/0) -
Hap4 CCAATCA 114 540 [100:700] 0.83 HAP4 HAP4 3(2/1) Respiration (p < 10-15)
Ino4 CATGTGA 177 454 [100:1000] 1.24 INO4 INO4 1(0/1) Lipid, fatty-acid and isoprenoid metabolism (p < 10-5)
Mbp1 ACGCGTC 23 225 [0;600] 3.25 MBP1 MBP1 29(18/11) DNA synthesis and replication (p < 10-11)
Met31 TGTGGCG 302 424 [100;1000] 1.35 MET31 MET31 4(4/0) -
Met4 CTGTGGC 362 500 [100;800] 1.08 MET4 MET4 1(1/0) Amino acid metabolism (p < 10-6)
Msn2/4 AAAGGGG 49 332 [0;500] 1.92 MSN2/4 - 105(93/12) -
Gln3 GATAAGA 143 434 [0;900] 1.23 - - 7(7/0) Nitrogen and sulfur metabolism (p < 10-6)
PAC GCGATGAG 4 164.5 [0;400] 6.77 PAC - 141(28/113) rRNA transcription (p < 10-10)
Pdr3 CCGCGGA 357 378 [0;500] 2.34 PDR3 - 18(15/3) -
Rap1 TGGGTGT 110 498.5 [100;900] 1.19 RAP1 - 13(1/12) -
Reb1 CGGGTAA 1 213 [0;1000] 6.48 REB1 REB1 - -
Rox1 AACAATAG 77 288.5 [0;500] 2.05 - - 1 (0/1)* -
Rpn4 TTTGCCACC 20 175.5 [0;800] 2.01 RPN4 - 10(10/0) Cytoplasmic and nuclear degradation (p < 10-31)
RRPE AAAAATTTT 2 188 [0;600] 3.04 RRPE - 167(31/136) rRNA transcription (p < 10-16)
Ste12 TGAAACA 282 477 100;1000] 1.15 STE12 STE12 5(3/2) fungal cell differentiation (p < 10-5)
Sum1/Ndt80 TGACACA 51 385 [0;600] 1.32 SUM1 SUM1 1(1/0) -
Swi4 CGCGAAA 19 261 [0;600] 3.25 SWI4 SWI4 39(22/17) -
TATA TATATAA 18 291 [100;700] 4.70 - - 49(40/9) -
Ume6 TAGCCGCC 6 457.5 - 3.92 UME6 - - Meiosis (p < 10-7)
Xbp1 CCTCGAG 219 348 [0;700] 2.41 XBP1 - 40(34/6) -
For each known regulatory element, we show the best k-mer, its rank within the set of 398 highest-scoring k-mers, the median distance to ATG (for occurrences upstream of genes within the conserved set), the optimal window, the corrected ratio of upstream/coding bias, the best known motif (see Materials and methods), the best chromatin IP (ChIP) enrichment (see Materials and methods), the total (upregulated/downregulated) number of microarray conditions in which the k-mer was found (see Materials and methods), and the best MIPS enrichment. *This sequence was the most significantly over-represented 8-mer in the upstream regions of genes that were downregulated upon overexpression of the Rox1 gene (a known repressor of hypoxia-induced genes under aerobic conditions [95]), as part of a series of microarray experiments measuring S. cerevisiae transcriptional response to various stresses [96].
Table 2 Known regulatory elements obtained when applying FastCompare to C. elegans and C. briggsae
Sequence Rank DATG WATG Orientation U/C Experiment TRANSFAC Comments
TGATAAG 5 746 [0;600] ← (p < 10-6) 1.67 103(56/47) GATA-1, GATA-2 Known GATA factor
AATCGAT 6 865.5 [0;1900] - 1.00 14(2/12) CDP, Clox Similar to DRE, embryonic development (p < 10-8)
TGACTCAT 8 708 - → (p < 10-4) 1.40 - AP-1, GCN4, NF-E2 Known AP-1 site
GTGTTTGC 9 383.5 [0;800] - 2.44 - - Known forkhead-related activator 4
CACGTGG 16 935 - - 0.73 12(9/3) Myc/Max, PHO4, USF Known Myc-Max site in Drosophila
AAGGTCA 22 882 [0;1400] - 1.52 35(16/19) ER, HNF-4 Known HRE
TGACGTC 32 858 [0;1700] - 0.94 1(1/0) CREB, ATF Known CREB site
TGTCATCA 42 879 - - 0.80 - Skn-1 Known SKN-1 site
CAGCTGG 56 1093 [100;2000] - 0.67 5(2/3) AP-4, HEN-1 Known AP-4 and MyoD/CeMyoD site
AGAGAGA 57 893 - → (p < 10-90) 1.43 4(2/2) - Known GAGA-factor site
GTAAACA 79 818 [0;400] - 2.69 28(28/0) Freac, SRY Known DAF-16 site
CCCGCCC 88 535 [0;1400] - 2.48 1(0/1) Sp1, GC box Known Sp1 site
ATCAATCA 100 911 - - 0.93 1(1/0) Pbx-1 Known Pbx-1 site
CAGGTGA 111 845 [0;200] - 2.25 - Lmo2, RAV1 Known Snail site in Drosophila
TTCGCGC 148 651.5 [0;1200] - 1.7 16(7/9) E2F Known E2F site, embryonic development (p < 10-6)
For each known regulatory element, we show the best k-mer, its rank within the set of 437 highest scoring k-mers, the median distance to ATG (for occurrences upstream of genes within the conserved set), the optimal window, the orientation bias, the corrected ratio of upstream/coding bias, the total (up-regulated/down-regulated) number of microarray conditions in which the k-mer was found (see Materials and methods), TRANSFAC matches, and the best GO enrichment.
Table 3 Novel predicted regulatory elements obtained when applying FastCompare to C. elegans and C. briggsae
Sequence Rank DATG WATG Orientation U/C Experiment Comments
CTGCGTCT 1 635.5 - → (p < 10-10) 2.70 - Unknown site
CGACACTCC 4 234 [0;1500] - 2.49 - Unknown site, positive regulation of growth (p < 10-7)
CTCCGCCC 14 440 [0;900] - 3.51 2(2/0) Unknown site, similar to Sp1
CGAGACC 20 738 [0;1900] - 1.34 30(7/23) Unknown site, embryonic development (p < 10-7)
CGCGACGC 23 457 [0;1900] - 2.34 - Unknown site
ATTTCGCAA 29 641 [0;1900] - 2.50 1(0/1) Unknown site
CGTAAATC 31 514 [0;600] - 2.78 - Unknown site
TTGCGGAC 39 253 [0;1700] - 1.43 - Unknown site
ATGATGCAA 44 600 [0;1600] - 0.88 - Unknown site
CGCGCTC 46 576 [0;900] - 2.73 2(0/2) Unknown site
TGGCGCC 49 770.5 [0;1800] - 1.01 - Unknown palindromic site
AACCGGTT 50 651 [0;1900] - 1.41 - Unknown palindromic site
TAAAGGCGC 61 524 [0;700] - 8.67 27(12/15) Unknown site
CGCGCGC 120 455 [0;600] - 5.40 11(3/8) Unknown site
CTAATCC 228 934 - → (p < 10-7) 1.20 - Unknown homeodomain site, similar to Bicoid
TACCGTA 242 975 [0;500] - 2.23 20(18/2) Unknown site
k-mers shown here were selected from the list of 437 highest scoring k-mers based on their short median distance to ATG, short optimal window, significant orientation bias, strong over-representation ratio (U/C), presence in upstream regions of over/underexpressed genes in several microarray conditions, palindromicity or resemblance to known sites in other species.
Table 4 Known and novel predicted regulatory elements, obtained when applying FastCompare to D. melanogaster and D. pseudoobscura
Sequence Rank DATG WATG Orientation U/C TRANSFAC Comments
(a) Known regulatory elements
AACAGCTG 1 373 [0;1800] - 1.64 - Known AP-4/MyoD site
ATTTGCATA 3 882 [100;2000] - 3.20 Oct-1 Known (mammalian) Oct-1 site
CACGTGC 5 825.5 - - 1.02 Myc/Max, PHO4, USF Known Myc/Max site
ATTTATGC 6 866 - - 3.52 CdxA Known CdxA site
TGACGTCA 9 825 - - 2.36 CREB Known CREB site
TGATAAG 11 760.5 [0;1100] - 2.53 GATA Known GATA site, carbohydrate metabolism (p < 10-5)
TATCGATA 12 168 [0;1900] - 5.39 - Known DRE site
TTTATGGC 14 978.5 - - 2.82 Abd-B Known Abd-B site
TAATTGA 24 907 [0;1900] - 2.58 Ubx, Athb-1 Known Antp site
GAGAGAG 26 705.5 - ← (p < 10-4) 1.87 - Known GAGA site, morphogenesis (p < 10-23)
CAGGTGC 33 1020.5 - - 0.83 Sn Known Snail site
TGACTCA 46 911 [100;2000] - 1.89 AP-1, GCN4 Known AP-1 site
ATCAATCA 51 967 [0;1900] - 1.72 Pbx-1 Known Pbx-1 site
AAGGTCA 93 1015.5 [400;1900] - 1.16 HNF-4, ER Known HRE
AACATGTG 105 994 [100;2000] - 1.62 - Known Twist site
GTAAACA 147 813 [0;1200] - 2.54 Freac, SRY Known DAF-16 site in C. elegans
(b) Novel predicted regulatory elements
ACACACAC 2 922.5 - → (p < 10-12) 1.97 - Unknown site, embryonic development (p < 10-9)
CAAGGAG 13 1091 [200;2000] ← (p < 10-8) 0.84 - Unknown site
GCACACAC 29 886 - - 1.80 - Unknown site, histogenesis (p < 10-5)
CAAGTTCA 30 920 [0;1900] - 1.23 - Unknown site
TAATTAA 31 871 [500;2000] - 3.07 Ftz Unknown palindromic homeodomain-like site
CAACAACA 42 968.5 [200;2000] - 1.22 - Unknown site, regulation of transcription (p < 10-5)
TGGCGCC 48 951 - - 0.84 - Unknown palindromic site
CCTGTTGC 111 653 [0;1800] - 0.90 - Unknown site
GTGTGACC 112 296 [0;1900] → (p < 10-5) 2.22 - Unknown site
CAGGTAG 143 924.5 [0;1700] - 0.94 - Unknown site, cell fate commitment (p < 10-8)
CACACGCA 145 968.5 - - 1.49 - Unknown site, cellular morphogenesis (p < 10-5)
GTCAACAA 169 904 - - 1.48 - Unknown site, similar to DAF-16
AAATGGCG 205 592 - - 1.54 - Unknown site
TTGACCCA 239 860 [0;1700] - 1.60 - Unknown site
TGACACAC 273 860 - - 1.83 - Unknown site
TGTCAAC 281 999 [100;1900] 1.55 - Unknown site
(a) For each known regulatory element, we show the best k-mer, its rank within the set of 469 highest scoring k-mers, the median distance to ATG (for occurrences upstream of genes within the conserved set), the optimal window, the orientation bias, the corrected ratio of upstream/coding bias, the total (up-regulated/down-regulated) number of microarray conditions in which the k-mer was found (see Method), TRANSFAC matches, and the best GO enrichment. (b) Novel predicted regulatory elements. k-mers shown here were selected from the list of 469 highest scoring k-mers based on their short median distance to ATG, short optimal window, significant orientation bias, strong over-representation ratio (U/C), presence in upstream regions of over/underexpressed genes in several microarray conditions, palindromicity or ressemblance to known sites in other species.
Table 5 Known and novel predicted regulatory elements, obtained when applying FastCompare to H. sapiens and M. musculus
Sequence Rank DATG WATG Orientation U/C Experiment TRANSFAC Comments
(a) Known regulatory sequences
CCCGCCC 1 256 - - 2.26 8(7/1) Sp1, GC box Known Sp1 site, transcription from pol II promoter (p < 10-5)
GCCCCGCCC 2 165 - - 4.64 9(9/0) Sp1, GC box Known Sp1 site, variant from above
CCGGAAG 4 160.5 [0;700] - 2.37 - Ets1, Elk1 Known Ets site, RNA metabolism (p < 10-6)
CACGTGAC 18 122.5 [0;600] - 4.90 - USF, GBP, SREBP-1 Known Myc/Max site
TGACGTCA 19 107 [0;1000] - 4.24 - CREB Known CREB site
CGCATGCG 24 132 [0;1600] - 4.26 - - Known palindromic octamer sequence (POS)
CCAATCAG 37 239 [0;700] - 2.85 4(0/4) NF-Y, CCAAT Known CAAT box and CCAAT enhancer binding protein site
CGGAAGTGA 51 94 [0;1000] - 3.96 - STAT3 Known GA-binding protein (GAB) site
CCGCCTC 78 632 [0;500] - 4.26 9(8/1) - Known insulin response element
CACGTGG 82 429.5 [0;300] - 2.09 - USF, Myc-Max Known Myc/Max site, different from above
TAATCCCAG 119 1258 [100;2000] ← (p < 10-14) 7.06 3(1/2) - Similar to Bicoid (Drosophila), RNA processing (p < 10-5)
CACCTGC 227 925 [0;600] - 1.64 1(1/0) E47, Lmo2 Known ZEB site in vertebrates, Zfh-1 in Drosophila
ATTTGCAT 234 729 [0;300] - 1.95 - Oct-1 Known Oct-1 site, chromatin assembly/disassembly (p < 10-8)
CCAAGGTCA 242 801 [0;1800] - 1.59 - - Known HRE site
GGAAGTCCC 253 124.5 [0;300] - 2.60 - NFκB Known NFκB site
CAGCTGC 256 850 [0;1600] - 1.03 - AP-4, HEN1 Known AP-4, MyoD site
TTTCGCGC 275 245 - 2.42 - E2F Known E2F site
(b) Novel predicted regulatory sequences
CGCAGGCGC 6 127 - - 2.76 - - Unknown site
GCGCCGC 13 311 [0;1900] ← (p < 10-5) 1.41 - - Unknown site
TCTCGCGA 17 116 [0;1700] - 4.45 - StuAp Unknown site, similar to E2F
TTAAAAA 52 1142 [100;2000] - 2.19 21(0/21) - Unknown site
CTCCGCCC 60 242.5 [0;1300] - 3.85 - - Unknown site, similar to Sp1
CCCCTCCC 67 563 [0;500] → (p < 10-4) 5.12 1(0/1) - Unknown site, regulation of transcription, DNA-dependent (p < 10-5)
AAGATGGCG 76 334 [0;1300] - 1.14 - - Unknown site
CTGCGCA 89 199 [0;300] - 3.63 - - Unknown site
CCAGCCTGG 123 1245 [200;2000] - 4.42 - - Unknown site
CCTGCCC 162 788 [0;1800] - 1.55 21(20/1) E47/Sp1 Unknown site
CCCTTTAAG 166 230 [0;800] → (p < 10-10) 3.45 - - Unknown site
CCCCAGC 207 785 - - 1.42 22(22/0) - Unknown site
TACAACTCC 225 154 [0;700] - 2.51 - - Unknown site
GTGAGCCAC 248 1208 - → (p < 10-6) 6.28 - - Unknown site
(a) For each known regulatory element, we show the best k-mer, its rank within the set of 284 highest scoring k-mers, the median distance to ATG (for occurrences upstream of genes within the conserved set), the optimal window, the orientation bias, the corrected ratio of upstream/coding bias, the total (upregulated/downregulated) number of microarray conditions in which the k-mer was found (see Materials and methods), TRANSFAC matches, and the best GO enrichment. (b) Novel predicted regulatory elements. k-mers shown here were selected from the list of 284 highest-scoring k-mers based on their short median distance to ATG, short optimal window, significant orientation bias, strong over-representation ratio (U/C), presence in upstream regions of over/underexpressed genes in several microarray conditions, palindromicity or resemblance to known sites in other species.
Table 6 Numbers of 7-mers and 8-mers shared between the 400 highest scoring 7-mers and 8-mers, respectively, in each pair of phylogenetic groups considered
(a) 7-mers
Y W F M
Y - 64(p < 10-17) 45(p < 10-6) 43(p < 10-6)
W - 103(p < 10-48) 43(p < 10-6)
F - 43(p < 10-6)
M -
(b) 8-mers
Y W F M
Y - 36(p < 10-19) 26(p < 10-11) 10(p = 0.025)
W - 59(p < 10-45) 23(p < 10-8)
F - 16(p < 10-4)
M -
The number of (a) 7-mers and (b) 8-mers shared between the 400 highest scoring 7-mers and 8-mers, respectively, in each pair of phylogenetic groups considered. Y, yeasts; W, worms; F, flies; M, mammals. p-values were calculated using the hypergeometric distribution, as described in Materials and Methods.
Table 7 Comparison of conservation scores between highest scoring k-mers and position weight matrices (PWM) for 20 known regulatory elements in S. cerevisiae, obtained when comparing S. cerevisiae and S. bayanus
Name Sequence Score PWM consensus Score
Bas1 AAGAGTCA 93.8* [AG][AG]NANGAGTCA 80.9
Cbf1 CACGTGA 421.3* [AG][AG]TCACGTG 406.5
Fkh1/2 TAAACAA 110.3 GTAAACAA[AT] 114.1*
Gcn4 TGACTCA 93.4 [AG][AG]TGA[CG]TCA 135.4*
Gcr1 TGGAAGC 82.7* [AG]GCTTCCT CG]T 42.7
Hap4 CCAATCA 104.2* G[AG][AG]CCAATCA 96.6
Ino4 CATGTGA 91.2* CAT[CG]TGAAAA 61.1
Mbp1 ACGCGTC 204.1 ACGCGTNA[AG]N 210.2*
Msn2/4 AAAGGGG 140.1 A[AG]GGGG 169.7*
PAC GCGATGAG 404.6 GCGATGAGNT 520.3*
Pdr3 CCGCGGA 76.9 [CG]NNTCCG[CT]GGAA 102.5*
Rap1 TGGGTGT 103.8 [AG]TGTN[CT]GG[AG]TG 253.2*
Reb1 CGGGTAA Inf [CG]CGGGTAA[CT] Inf
Rpn4 TTTGCCACC 218.6 GGTGGCAAAA 259.4*
RRPE AAAAATTT 509.9* TGAAAAATTT 388.80
Ste12 TGAAACA 81.4 ANNNTGAAACA 100.0*
Sum1/Ndt80 TGACACA 135.4* [AG][CT]G[AT]CA[CG][AT]AA[AT] 100.0
Swi4 CGCGAAA 224.1* NNNNC[AG]CGAAAA 116.6
Ume6 TAGCCGCC 377.2 TCGGCGGC[AT]A 410.0*
Xbp1 CCTCGAG 86.7 GCCTCGA[AG]G[AC]G[AG] 141.7*
*Indicates which regulatory element representation (k-mer or weight matrix) obtained the highest conservation score. Inf corresponds to very large conservation scores, obtained when taking the negative natural logarithm of near-zero hypergeometric p-values.
==== Refs
Lee TI Rinaldi NJ Robert F Odom DT Bar-Joseph Z Gerber GK Hannett NM Harbison CT Thompson CM Simon I Transcriptional regulatory networks in Saccharomyces cerevisiae. Science 2002 298 799 804 12399584 10.1126/science.1075090
Stormo GD DNA binding sites: representation and discovery. Bioinformatics 2000 16 16 23 10812473 10.1093/bioinformatics/16.1.16
Cliften P Sudarsanam P Desikan A Fulton L Fulton B Majors J Waterston R Cohen BA Johnston M Finding functional features in Saccharomyces genomes by phylogenetic footprinting. Science 2003 301 71 76 12775844 10.1126/science.1084337
Kellis M Patterson N Endrizzi M Birren B Lander ES Sequencing and comparison of yeast species to identify genes and regulatory elements. Nature 2003 423 241 254 12748633 10.1038/nature01644
Aparicio S Morrison A Gould A Gilthorpe J Chaudhuri C Rigby P Krumlauf R Brenner S Detecting conserved regulatory elements with the model genome of the Japanese puffer fish, Fugu rubripes. Proc Natl Acad Sci USA 1995 92 1684 1688 7878040
Pritsker M Liu YC Beer MA Tavazoie S Whole-genome discovery of transcription factor binding sites using network-level conservation. Genome Res 2004 14 99 108 14672978 10.1101/gr.1739204
Hughes JD Estep PW Tavazoie S Church GM Computational identification of cis -regulatory elements associated with groups of functionally related genes in Saccharomyces cerevisiae. J Mol Biol 2000 296 1205 1214 10698627 10.1006/jmbi.2000.3519
Zhu J Zhang MQ SCPD: a promoter database of the yeast Saccharomyces cerevisiae. Bioinformatics 1999 15 607 611 10487868 10.1093/bioinformatics/15.7.607
FastCompare
Yamaguchi-Iwai Y Dancis A Klausner RD AFT1: a mediator of iron regulated transcriptional control in Saccharomyces cerevisiae. EMBO J 1995 14 1231 1239 7720713
Beer MA Tavazoie S Predicting gene expression from sequence. Cell 2004 117 185 198 15084257 10.1016/S0092-8674(04)00304-6
Erives A Levine M Coordinate enhancers share common organizational features in the Drosophila genome. Proc Natl Acad Sci USA 2004 101 3851 3856 15026577 10.1073/pnas.0400611101
Sudarsanam P Pilpel Y Church GM Genome-wide co-occurrence of promoter elements reveals a cis-regulatory cassette of rRNA transcription motifs in Saccharomyces cerevisiae. Genome Res 2002 12 1723 1731 12421759 10.1101/gr.301202
Blaiseau PL Thomas D Multiple transcriptional activation complexes tether the yeast activator Met4 to DNA. EMBO J 1998 17 6327 6336 9799240 10.1093/emboj/17.21.6327
Chiang DY Moses AM Kellis M Lander ES Eisen MB Phylogenetically and spatially conserved word pairs associated with gene-expression changes in yeasts. Genome Biol 2003 4 R43 12844359 10.1186/gb-2003-4-7-r43
Davidson EH Genomic Regulatory Systems 2001 San Diego, CA: Academic Press
Coghlan A Wolfe KH Fourfold faster rate of genome rearrangement in nematodes than in Drosophila. Genome Res 2002 12 857 867 12045140 10.1101/gr.172702
Maduro MF Rothman JH Making worm guts: the gene regulatory network of the Caenorhabditis elegans endoderm. Dev Biol 2002 246 68 85 12027435 10.1006/dbio.2002.0655
Cui M Han M Cis regulatory requirements for vulval cell-specific expression of the Caenorhabditis elegans fibroblast growth factor gene egl-17. Dev Biol 2003 257 104 116 12710960 10.1016/S0012-1606(03)00033-2
Gaudet J Mango SE Regulation of organogenesis by the Caenorhabditis elegans FoxA protein PHA-4. Science 2002 295 821 825 11823633 10.1126/science.1065175
Maduro MF Meneghini MD Bowerman B Broitman-Maduro G Rothman JH Restriction of mesendoderm to a single blastomere by the combined action of SKN-1 and a GSK-3 homolog is mediated by MED-1 and -2 in C. elegans. Mol Cell 2001 7 475 485 11463373 10.1016/S1097-2765(01)00195-2
Harfe BD Fire A Muscle and nerve-specific regulation of a novel NK-2 class homeodomain factor in Caenorhabditis elegans. Development 1998 125 421 429 9425137
Jantsch-Plunger V Fire A Combinatorial structure of a body muscle-specific transcriptional enhancer in Caenorhabditis elegans. J Biol Chem 1994 269 27021 27028 7929443
Tsukiyama T Becker PB Wu C ATP-dependent nucleosome disruption at a heat-shock promoter mediated by binding of GAGA transcription factor. Nature 1994 367 525 532 8107823 10.1038/367525a0
King-Jones K Korge G Lehmann M The helix-loop-helix proteins dAP-4 and daughterless bind both in vitro and in vivo to SEBP3 sites required for transcriptional activation of the Drosophila gene Sgs-4. J Mol Biol 1999 291 71 82 10438607 10.1006/jmbi.1999.2963
Krause M Fire A Harrison SW Priess J Weintraub H CeMyoD accumulation defines the body wall muscle cell fate during C. elegans embryogenesis. Cell 1990 63 907 919 2175254 10.1016/0092-8674(90)90494-Y
Hu YF Luscher B Admon A Mermod N Tjian R Transcription factor AP-4 contains multiple dimerization domains that regulate dimer specificity. Genes Dev 1990 4 1741 1752 2123466
Blackwell TK Weintraub H Differences and similarities in DNA-binding preferences of MyoD and E2A protein complexes revealed by binding site selection. Science 1990 250 1104 1110 2174572
Krause M Park M Zhang J Yuan J Harfe B Xu S Greenwald I Cole M Paterson B Fire A A C. elegans E/Daughterless bHLH protein marks neuronal but not striated muscle development. Development 1997 124 2179 2189 9187144
Furuyama T Nakazawa T Nakano I Mori N Identification of the differential distribution patterns of mRNAs and consensus binding sequences for mouse DAF-16 homologues. Biochem J 2000 349 629 634 10880363 10.1042/0264-6021:3490629
Murphy CT McCarroll SA Bargmann CI Fraser A Kamath RS Ahringer J Li H Kenyon C Genes that act downstream of DAF-16 to influence the lifespan of Caenorhabditis elegans. Nature 2003 424 277 283 12845331 10.1038/nature01789
Lee SS Kennedy S Tolonen AC Ruvkun G DAF-16 target genes that control C. elegans life-span and metabolism. Science 2003 300 644 647 12690206 10.1126/science.1083614
Gronostajski RM Analysis of nuclear factor I binding to DNA using degenerate oligonucleotides. Nucleic Acids Res 1986 14 9117 9132 3786147
Lee W Mitchell P Tjian R Purified transcription factor AP-1 interacts with TPA-inducible enhancer elements. Cell 1987 49 741 752 3034433 10.1016/0092-8674(87)90612-X
Kockel L Homsy J Bohmann D Drosophila AP-1: lessons from an invertebrate. Oncogene 2001 20 2347 2364 11402332 10.1038/sj.onc.1204300
Karin M Liu Z Zandi E AP-1 function and regulation. Curr Opin Cell Biol 1997 9 240 246 9069263 10.1016/S0955-0674(97)80068-3
Grandori C Cowley SM James LP Eisenman RN The Myc/Max/Mad network and the transcriptional control of cell behavior. Annu Rev Cell Dev Biol 2000 16 653 699 11031250 10.1146/annurev.cellbio.16.1.653
Rice DA Mouw AR Bogerd AM Parker KL A shared promoter element regulates the expression of three steroidogenic enzymes. Mol Endocrinol 1991 5 1552 1561 1775136
Ueda H Sun GC Murata T Hirose S A novel DNA-binding motif abuts the zinc finger domain of insect nuclear hormone receptor FTZ-F1 and mouse embryonal long terminal repeat-binding protein. Mol Cell Biol 1992 12 5667 5672 1448096
Shaywitz AJ Greenberg ME CREB: a stimulus-induced transcription factor activated by a diverse array of extracellular signals. Annu Rev Biochem 1999 68 821 861 10872467 10.1146/annurev.biochem.68.1.821
Dijk MAV Voorhoeve PM Murre C Pbx1 is converted into a transcriptional activator upon acquiring the N-terminal region of E2A in pre-B-cell acute lymphoblastoid leukemia. Proc Natl Acad Sci U S A 1993 90 6061 6065 8327485
Manak JR Mathies LD Scott MP Regulation of a decapentaplegic midgut enhancer by homeotic proteins. Development 1994 120 3605 3619 7821226
Mauhin V Lutz Y Dennefeld C Alberga A Definition of the DNA-binding site repertoire for the Drosophila transcription factor SNAIL. Nucleic Acids Res 1993 21 3951 3957 8371971
Huber HE Edwards G Goodhart PJ Patrick DR Huang PS Ivey-Hoyle M Barnett SF Oliff A Heimbrook DC Transcription factor E2F binds DNA as a heterodimer. Proc Natl Acad Sci U S A 1993 90 3525 3529 8475102
Boxem M vanden Heuvel S C. elegans class B synthetic multivulva genes act in G(1) regulation. Curr Biol 2002 12 906 911 12062054 10.1016/S0960-9822(02)00844-8
Ceol CJ Horvitz HR dpl-1 DP and efl-1 E2F act with lin-35 Rb to antagonize Ras signaling in C. elegans vulval development. Mol Cell 2001 7 461 473 11463372 10.1016/S1097-2765(01)00194-0
Kwon JY Hong M Choi MS Kang S Duke K Kim S Lee S Lee J Ethanol-response genes and their regulation analyzed by a microarray and comparative genomic approach in the nematode Caenorhabditis elegans. Genomics 2004 83 600 614 15028283 10.1016/j.ygeno.2003.10.008
Lund J Tedesco P Duke K Wang J Kim SK Johnson TE Transcriptional profile of aging in C. elegans. Curr Biol 2002 12 1566 1573 12372248 10.1016/S0960-9822(02)01146-6
Ohler U Yekta S Lim LP Bartel DP Burge CB Patterns of flanking sequence conservation and a characteristic upstream motif for microRNA gene identification. RNA 2004 10 1309 1322 15317971 10.1261/rna.5206304
Celniker SE Rubin GM The Drosophila melanogaster genome. Annu Rev Genomics Hum Genet 2003 4 89 117 14527298 10.1146/annurev.genom.4.070802.110323
Matsukage A Hirose F Hayashi Y Hamada K Yamaguchi M The DRE sequence TATCGATA, a putative promoter-activating element for Drosophila melanogaster cell-proliferation-related genes. Gene 1995 166 233 236 8543167 10.1016/0378-1119(95)00586-2
Choi T Cho N Oh Y Yoo M Matsukage A Ryu Y Han K Yoon J Baek K The DNA replication-related element (DRE)-binding factor (DREF) system may be involved in the expression of the Drosophila melanogaster TBP gene. FEBS Lett 2000 483 71 77 11033359 10.1016/S0014-5793(00)02085-8
Park SY Kim YS Yang DJ Yoo MA Transcriptional regulation of the Drosophila catalase gene by the DRE/DREF system. Nucleic Acids Res 2004 32 1318 1324 14982956 10.1093/nar/gkh302
Hanes SD Brent R A genetic model for interaction of the homeodomain recognition helix with DNA. Science 1991 251 426 430 1671176
Anderson MG Perkins GL Chittick P Shrigley RJ Johnson WA Drifter, a Drosophila POU-domain transcription factor, is required for correct differentiation and migration of tracheal cells and midline glia. Genes Dev 1995 9 123 137 7828848
Bhat KM Poole SJ Schedl P The miti-mere and pdm1 genes collaborate during specification of the RP2/sib lineage in Drosophila neurogenesis. Mol Cell Biol 1995 15 4052 4063 7623801
Junger MA Rintelen F Stocker H Wasserman JD Vegh M Radimerski T Greenberg ME Hafen E The Drosophila Forkhead transcription factor FOXO mediates the reduction in cell number associated with reduced insulin signaling. J Biol 2003 2 20 12908874 10.1186/1475-4924-2-20
Erickson JW Cline TW Key aspects of the primary sex determination mechanism are conserved across the genus Drosophila. Development 1998 125 3259 3268 9671597
Waterston RH Lindblad-Toh K Birney E Rogers J Abril JF Agarwal P Agarwala R Ainscough R Alexandersson M An P Initial sequencing and comparative analysis of the mouse genome. Nature 2002 420 520 562 12466850 10.1038/nature01262
Suske G The Sp-family of transcription factors. Gene 1999 238 291 300 10570957 10.1016/S0378-1119(99)00357-1
Ramji DP Foka P CCAAT/enhancer-binding proteins: structure, function and regulation. Biochem J 2002 365 561 575 12006103
Latchman D Eukaryotic Transcription Factors 1997 London: Academic Press
Vo N Goodman RH CREB-binding protein and p300 in transcriptional regulation. J Biol Chem 2001 276 13505 13508 11279224
Bernards R Transcriptional regulation. Flipping the Myc switch. Curr Biol 1995 5 859 861 7583141 10.1016/S0960-9822(95)00173-4
Nasrin N Ercolani L Denaro M Kong XF Kang I Alexander M An insulin response element in the glyceraldehyde-3-phosphate dehydrogenase gene binds a nuclear protein induced by insulin in cultured cells and by nutritional manipulations in vivo. Proc Natl Acad Sci U S A 1990 87 5273 5277 2164673
Suzuki F Goto M Sawa C Ito S Watanabe H Sawada J Handa H Functional interactions of transcription factor human GA-binding protein subunits. J Biol Chem 1998 273 29302 29308 9792629 10.1074/jbc.273.45.29302
Zimmermann AG Wright KL Ting JP Mitchell BS Regulation of inosine-5'-monophosphate dehydrogenase type II gene expression in human T cells. Role for a novel 5' palindromic octamer sequence. J Biol Chem 1997 272 22913 22923 9278455 10.1074/jbc.272.36.22913
Gottlieb S Hanes SD Golden JA Oakey RJ Budarf ML Goosecoid-like, a gene deleted in DiGeorge and velocardiofacial syndromes, recognizes DNA with a bicoid-like specificity and is expressed in the developing mouse brain. Hum Mol Genet 1998 7 1497 1505 9700206 10.1093/hmg/7.9.1497
Singh H Sen R Baltimore D Sharp PA A nuclear factor that binds to a conserved sequence motif in transcriptional control elements of immunoglobulin genes. Nature 1986 319 154 158 3079885 10.1038/319154a0
Nie Z Mei Y Ford M Rybak L Marcuzzi A Ren H Stiles GL Ramkumar V Oxidative stress increases A1 adenosine receptor expression by activating nuclear factor kappa B. Mol Pharmacol 1998 53 663 669 9547356
Glasgow JN Wood T Perez-Polo JR Identification and characterization of nuclear factor κB binding sites in the murine bcl-x promoter. J Neurochem 2000 75 1377 1389 10987817 10.1046/j.1471-4159.2000.0751377.x
Whitfield ML Sherlock G Saldanha AJ Murray JI Ball CA Alexander KE Matese JC Perou CM Hurt MM Brown PO Botstein D Identification of genes periodically expressed in the human cell cycle and their expression in tumors. Mol Biol Cell 2002 13 1977 2000 12058064 10.1091/mbc.02-02-0030.
Rustici G Mata J Kivinen K Lio P Penkett CJ Burns G Hayles J Brazma A Nurse P Bahler J Periodic gene expression program of the fission yeast cell cycle. Nat Genet 2004 36 809 817 15195092 10.1038/ng1377
Stormo GD Fields DS Specificity, free energy and information content in protein-DNA interactions. Trends Biochem Sci 1998 23 109 113 9581503 10.1016/S0968-0004(98)01187-6
Kalir S Alon U Using a quantitative blueprint to reprogram the dynamics of the flagella gene network. Cell 2004 117 713 720 15186773 10.1016/j.cell.2004.05.010
Waterman MS Eggert M A new algorithm for best subsequence alignments with application to tRNA-rRNA comparisons. J Mol Biol 1987 197 723 728 2448477 10.1016/0022-2836(87)90478-5
Wolfertstetter F Frech K Herrmann G Werner T Identification of functional elements in unaligned nucleic acid sequences by a novel tuple search algorithm. Comput Appl Biosci 1996 12 71 80 8670622
Zhang MQ Identification of human gene core promoters in silico. Genome Res 1998 8 319 326 9521935
Curwen V Eyras E Andrews TD Clarke L Mongin E Searle SM Clamp M The ENSEMBL automatic gene annotation system. Genome Res 2004 14 942 950 15123590 10.1101/gr.1858004
Human Genome Sequencing Center at Baylor College of Medicine: Drosophila genome project
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 29 10802651 10.1038/75556
Mewes HW Amid C Arnold R Frishman D Guldener U Mannhaupt G Munsterkotter M Pagel P Strack N Stumpflen V MIPS: analysis and annotation of proteins from whole genomes. Nucleic Acids Res 2004 32 Database D41 D44 14681354 10.1093/nar/gkh092
Gusfield D Algorithms on Strings, Trees, and Sequences 1997 Cambridge, UK: Cambridge University Press
Press WH Flannery BP Teukolsky SA Vetterling WT Numerical Recipes in C: The Art of Scientific Computing 1993 Cambridge, UK: Cambridge University Press
Pilpel Y Sudarsanam P Church GM Identifying regulatory networks by combinatorial analysis of promoter elements. Nat Genet 2001 29 153 159 11547334 10.1038/ng724
Yuh CH Bolouri H Davidson EH Genomic cis -regulatory logic: experimental and computational analysis of a sea urchin gene. Science 1998 279 1896 1902 9506933 10.1126/science.279.5358.1896
Needleman SB Wunsch CD A general method applicable to the search for similarities in the amino acid sequence of two proteins. J Mol Biol 1970 48 443 453 5420325
Comprehensive yeast genome database
Gene Ontology
GenomeNet
Matys V Fricke E Geffers R Gössling E Haubrock M Hehl R Hornischer K Karas D Kel AE Kel-Margoulis OV TRANSFAC: transcriptional regulation, from patterns to profiles. Nucleic Acids Res 2003 31 374 378 12520026 10.1093/nar/gkg108
Gollub J Ball CA Binkley G Demeter J Finkelstein DB Hebert JM Hernandez-Boussard T Jin H Kaloper M Matese JC The Stanford Microarray Database: data access and quality assessment tools. Nucleic Acids Res 2003 31 94 96 12519956 10.1093/nar/gkg078
Stuart JM Segal E Koller D Kim SK A gene-coexpression network for global discovery of conserved genetic modules. Science 2003 302 249 255 12934013 10.1126/science.1087447
Lieb JD Liu X Botstein D Brown PO Promoter-specific binding of Rap1 revealed by genome-wide maps of protein-DNA association. Nat Genet 2001 28 327 334 11455386 10.1038/ng569
Balasubramanian B Lowry CV Zitomer RS The Rox1 repressor of the Saccharomyces cerevisiae hypoxic genes is a specific DNA-binding protein with a high-mobility-group motif. Mol Cell Biol 1993 13 6071 6078 8413209
Gasch AP Spellman PT Kao CM Carmel-Harel O Eisen MB Storz G Botstein D Brown PO Genomic expression programs in the response of yeast cells to environmental changes. Mol Biol Cell 2000 11 4241 4257 11102521
| 15693947 | PMC551538 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 26; 6(2):R18 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r18 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r191569394810.1186/gb-2005-6-2-r19MethodA universal method for automated gene mapping Zipperlen Peder [email protected] Knud [email protected] Ivo 2Basler Konrad 1Hafen Ernst 2Hengartner Michael 1Hajnal Alex 21 Institute of Molecular Biology, University of Zürich, Winterthurerstrasse 190, CH-8057 Zürich, Switzerland2 Institute of Zoology, University of Zürich, Winterthurerstrasse 190, CH-8057 Zürich, Switzerland2005 17 1 2005 6 2 R19 R19 9 9 2004 15 11 2004 9 12 2004 Copyright © 2005 Zipperlen 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.
A high-throughput method for genotyping by mapping InDels. This method has been used to create fragment-length polymorphism maps for Drosophila and C. elegans.
Small insertions or deletions (InDels) constitute a ubiquituous class of sequence polymorphisms found in eukaryotic genomes. Here, we present an automated high-throughput genotyping method that relies on the detection of fragment-length polymorphisms (FLPs) caused by InDels. The protocol utilizes standard sequencers and genotyping software. We have established genome-wide FLP maps for both Caenorhabditis elegans and Drosophila melanogaster that facilitate genetic mapping with a minimum of manual input and at comparatively low cost.
==== Body
Background
For humans and model organisms, such as worms and flies, the availability of high-density sequence polymorphism maps greatly facilitates the rapid mapping and cloning of genes [1-3]. Key advantages of most molecular polymorphisms are the fact that they are codominant and in general phenotypically neutral. The vast majority of sequence polymorphisms are single-nucleotide polymorphisms (SNPs).
The most direct approach for SNP detection is sequencing of a PCR product spanning the polymorphism, but this is too costly and labor intense for high-throughput genotyping. For this reason, several different strategies and methods have been developed in order to detect SNPs more efficiently. In general, assays can be grouped into strategies, where the nature of the SNP is determined by directly analyzing the primary PCR product and those that require a secondary assay performed on the primary amplification product [4-6]. An important strategy of the first group is the 5' nuclease assay, where allele-specific, dual-labeled fluorescent TaqMan probes guarantee specificity [7]. However, the need for two dual-labeled fluorescent probes, expensive specialized chemistry and specialized machinery increase the costs per assay of this approach significantly. Similarly, denaturing high-performance liquid chromatography (DHPLC) also analyses the primary amplification product [8]. This approach is based on melting differences of homo- versus heteroduplex DNA fragments under increasingly denaturing conditions and requires no specific labeling of the PCR fragments. However, conditions have to be optimized for every assay, throughput is limited and specialized equipment is required. DHPLC has been used in small-scale genotyping projects in Drosophila melanogaster [9].
Of the methods that detect the SNP in a secondary assay, restriction fragment length polymorphism (RFLP) analysis are very popular [10]. For this purpose, only those SNPs that alter a restriction site are analyzed. A great advantage of RFLP analysis is that no specialized equipment is needed and it can be carried out in every laboratory. RFLP maps recently established for Caenorhabditis elegans and Drosophila are used regularly in genotyping projects [2,3,11]. However, RFLP analysis requires significant manual input. Moreover, the use of different restriction enzymes with different reaction requirements adds another level of complexity that makes this method difficult to automate. Primer-extension-based technologies have also gained some prominence [12]. Here, a primer that anneals right next to the polymorphism is extended by one polymorphism-specific terminator nucleotide. Extension products are analyzed by size or, alternatively, by differences in the behavior of incorporated versus non-incorporated terminator nucleotides under polarized fluorescent light [13]. Swan and colleagues [14] have developed a set of fluorescence polarization-template directed incorporation (FP-TDI) assays for C. elegans. However, this approach is labor intensive and requires specialized chemistry and equipment. Using DNA microarrays, large numbers of SNPs can be analyzed in parallel, but the number of individuals that can be analyzed is low because of the high cost per chip [15,16].
Besides SNPs, short tandem repeats (STRs) or microsatellites represent another class of sequence polymorphisms used for genotyping [17-21]. STRs result in fragment length differences that are either detected on gel-based or capillary sequencers or high-resolution hydrogels (Elchrom Scientific Inc.). One advantage of STRs over SNPs is that they are highly polymorphic and are thus ideal for measuring the degree of variability in natural populations. STRs are, however, present at much lower density than SNPs and are therefore not suitable for high-resolution mapping of genes.
Interestingly, a significant proportion of the currently available polymorphisms are caused by small insertions or deletions (InDels). Weber et al. [22] identified a genome-wide set of about 2,000 human InDel polymorphisms and estimated that InDels comprise at least 8% and up to 20% of all human polymorphisms. This is in line with the findings of Berger and co-workers [2] who found that 16.2% of polymorphisms in Drosophila are of the InDel type. Also, two independent studies in C. elegans found that InDels constitute between 25% and 28% of all polymorphisms [3,14]. In addition, those studies found that the vast majority of InDels are due to 1-2 base-pair (bp) differences (65% in Drosophila [2], 84% in C. elegans [3]).
To take full advantage of this class of small InDel polymorphisms, we have developed a strategy that allows us to detect most, if not all, InDels by analyzing the lengths of primary PCR products on a capillary sequencer at single base-pair resolution. We call these assays fragment length polymorphism (FLP) assays. Importantly, this approach can easily be automated on standard robotic pipetting platforms as it involves a simple PCR reaction setup. Furthermore, allele calling is performed automatically using the Applied Biosystems GeneMapper software commonly used for genotyping STRs (Materials and methods).
To demonstrate the feasibility of this strategy, we have validated 112 evenly spaced FLP assays at 3 centimorgan (cM) resolution in C. elegans (one every 0.9 megabase-pair (Mbp)) and 54 FLP assays at 4 cM resolution for the Drosophila autosomes. This set of FLP assays allows us to rapidly map mutations to small chromosomal subregions with a minimum of manual input. Furthermore, we provide a list of predicted InDels for which additional assays can be readily designed in the chromosomal subregion of interest. Those non-validated FLPs enhance the resolution of the map by a factor of 5.6 and 17.9, respectively.
We show the usefulness of this approach by identifying novel alleles of previously characterized genes. In summary, we have taken advantage of a publicly available dataset to adapt a technology widely used for STR analysis to genetic mapping. Thanks to the complete automation of genotyping, this approach is considerably faster, more reliable and cheaper than previously used mapping strategies in C. elegans or Drosophila.
Results and discussion
Detection of fragment length polymorphisms (FLPs)
To detect a FLP, the region of interest is amplified in a standard PCR reaction with one fluorescently labeled primer, and the PCR products are directly analyzed on a capillary sequencer. Fragment sizes are determined automatically relative to an internal size standard with AppliedBiosystem's GeneMapper software (for details see Materials and methods). The software then allocates fragment sizes to previously calibrated genotypes.
Taq polymerase has the tendency to catalyze the addition of adenosine (A) to the 3' end of PCR products. This activity could make it difficult to achieve the single base-pair resolution required to assay all available InDels and may hamper allele-calling [23]. However, we have found that the sensitivity of a capillary sequencer and the genotyping software is sufficient to allow for unambiguous allele assignment even for 'difficult' sequences exhibiting 3' A addition. The examples shown in Figure 1a-d illustrate that robust genotyping is feasible for 1-bp InDels even when 3' A addition occurs. Another problem is the stuttering of the polymerase when it encounters poly(N) stretches. However, larger InDels are reliably detected by the software in poly(N) stretches (Figure 1f), and in a few difficult cases visual inspection can even resolve and unambiguously assign 'stuttering' 1-bp InDels according to the location and number of peaks (Figure 1e).
Genotyping with FLP assays is extremely accurate. In a control experiment, we genotyped all 96 samples of the fly strains FRT42B and EP0755 for the 1-bp InDel 2R090 and 231 samples homo- and heterozygous for the C. elegans Bristol and Hawaii backgrounds, respectively, for the 1-bp InDel ZH5-16. 2R090 exhibits both stuttering and A addition and hence is especially difficult to resolve (see Additional data file 8). The genotype was correctly and automatically assigned by GeneMapper in all 423 assays. Thus, automated genotyping based on FLPs is sensitive down to single base-pair resolution and is extremely robust. The accuracy of FLP mapping is comparable to other methods such as TaqMan (error rate less than 1 in 2,000 [24]), minisequencing (99.5% [25]), and pyrosequencing (97.3 % [25]).
C. elegans and Drosophila FLP maps
In C. elegans, genetic experiments are performed almost exclusively in the background of the standard wild-type strain N2 (C. elegans variety Bristol) [26]. For gene mapping experiments, the polymorphic strain CB4856 (C. elegans, variety Hawaii) has proved extremely useful [3]. When compared to N2, CB4856 contains on average one SNP every 840 bp and approximately 25% of all polymorphisms are InDels [14]. Starting from the dataset previously published by Wicks et al. [3], 112 FLPs that are evenly spaced on the physical map of C. elegans were validated to date (Figure 2a). The confirmation rate of the predicted InDels was 88% (n = 169). Most failures to detect a FLP are probably due to original sequencing errors. The average distance between neighboring FLP assays is about 0.9 Mbp. This physical distance corresponds to about 3 cM, assuming 300 kb per map unit, and encompasses between 100 and a maximum of 500 genes (Figure 2a). The length of the amplicons ranges from 100 to 444 bp, and the fragment length differences are between 1 and 21 bp (Additional data file 9). If necessary, another 2,454 predicted InDels are available to increase the mapping resolution down to 50 kbp on average (Additional data files 12-17).
To establish a Drosophila FLP map, a set of 54 FLP assays (12 to 17 per arm of the two major autosomes) was validated from the list of polymorphisms identified by Berger et al. [2] (Figure 2b, and Additional data file 10); high-resolution X-chromosomal SNP and FLP maps have yet to be established. Similarly to C. elegans, the confirmation rate of the predicted Drosophila InDels was 80% (n = 30). Furthermore, another 509 InDels are predicted at 248 sites for which an assay can be established to discriminate between EP and FRT strains (Additional data file 18). The validated Drosophila FLP assays were evenly spaced on the genetic map with an average distance between neighboring assays of about 4 cM, corresponding to an average resolution of 1.77 Mbp on the physical map encompassing 95,55 Mbp [27,28]. Taking into account the non-validated InDels, the maximal average resolution is currently 314 kb or 0.7 cM. On the left arm of chromosome 3, where the genetic map is inexact, FLPs were spaced on the physical map assuming colinearity between the two maps. The length of amplicons ranges from 99 to 365 bp, and the size difference ranges from 1 to 54 bp (Additional data file 9).
Our Drosophila FLP assays are in part derived from a set of InDels of size difference 7 bp or more (termed PLPs by Berger et al. [2]). However, since 86.8% of all Drosophila InDels exhibit a length difference of one to six nucleotides [2], so far only a small subset of the available InDels has been covered. The approach presented here significantly increases the number of possible FLP assays for genotyping and offers a greater flexibility and higher resolution.
FLP mapping of C. elegans genes
To demonstrate the usefulness of the C. elegans FLP map, we mapped three previously characterized mutations on chromosome II that exhibit diverse phenotypes. Those were the centrally located let-23(sy1) allele that causes an 80% penetrant vulvaless phenotype [29], rol-1(e91) in the middle of the left chromosome arm, which causes the animals to roll around their body axis [30], and the unc-52(e444) mutation located at the right end of the chromosome, which results in a paralyzed phenotype [31]. Mutant hermaphrodites were crossed with CB4856 males, and wild-type F1 cross-progeny was selected (F1 self-progeny would exhibit a mutant phenotype). Finally, mutant self-progeny was isolated in the F2 generation and used for genotyping (Figure 3a). To minimize the number of PCR reactions, we pursued a two-step strategy. First, we determined chromosomal linkage by analyzing 16 individual F2 animals (corresponding to 32 chromosomes in total) with one centrally located FLP assay per chromosome (Tier 1, Figure 2a). This allowed us to establish clear linkage to chromosome 2 for all three mutations (Additional data file 2). Surprisingly, the rol-1(e91) mutation showed linkage to the X chromosome of N2 in addition to chromosome II. This pseudo-linkage could be due to a suppressor of the Rol phenotype present on the CB4856 X chromosome. In a second step, 48 F2 animals for each mutation were analyzed with eight FLP assays along chromosome 2 (Tier 2, Figure 2a). In this way, we could narrow down the three mutations to the correct chromosomal subregions (Additional data files 3-5). We used the same strategy to map the zh41 mutation that was identified in a forward genetic screen for mutants exhibiting a loss of egl-17::gfp expression in the vulval cell linage ([32] and I. Rimann and A. Hajnal, unpublished work). Analysis with Tier 1 established linkage to chromosome 1 (Figure 3b), and Tier 2 narrowed down the candidate region to an interval of 2.2 Mbp containing 498 genes (Figure 3c). The phenotype of zh41 animals is similar to the phenotype caused by loss-of-function mutations in lin-11, which maps to the same interval in the center of chromosome I [33]. Like lin-11 mutants, zh41 animals exhibit a penetrant protruding vulva (Pvl) phenotype, and staining of the adherens junctions with the MH27 antibody showed defects in the formation of the vulval torroid rings (Figure 3d) [33]. Subsequent sequencing of the lin-11 locus in zh41 animals revealed a point mutation that results in a change of leucine 274 to phenylalanine. Furthermore, zh41 failed to complement lin-11(n389), indicating that the zh41 mutation in the lin-11 open reading frame (ORF) is responsible for the vulval phenotype.
In cases where a mutation maps to an interval that contains no obvious candidate gene, we first screen for additional informative recombinants by FLP analysis and then refine the map position by extracting more FLPs from our set of non-validated InDels (Additional data files 12-17) and by genotyping existing SNPs in the candidate interval [3]. In many cases, this resolution is sufficient to identify the affected gene through RNA interference (RNAi) analysis of the genes in the corresponding interval [34]. (See Additional data file 6 for a detailed flowchart of the mapping process).
In summary, FLP mapping in C. elegans allows us to rapidly map a mutation down to a small region containing, on average, 200 candidate genes by crossing a mutant strain to CB4856 and analyzing 48 F2 animals with 300 to 500 PCR reactions.
Genotyping Drosophila strains with FLP assays
In contrast to the well defined genetic backgrounds used for C. elegans, zebrafish (Danio rerio) or Arabidopsis genetics, Drosophila strains are very heterogeneous and of ill-defined origin [2,9,11]. In this respect, gene mapping in Drosophila resembles human genetics in that standard inbred lines do not exist and the genotypes of the parental lines have to be determined first. As genome-wide polymorphism databases for reference strains are available [2,11], a line of interest can be crossed with two reference strains, such as EP and FRT (see below). Owing to the codominant character of sequence polymorphisms, at least one of the two respective crosses will distinguish between the mutant and the mapping chromosomes. To further facilitate mapping with our set of FLP assays, we genotyped several common laboratory lines such as two 'wild-type' yw strains for the whole set, four FRT-Minute or FRT-cell-lethal strains at the relevant autosomal arms [35], as well as the FRT and EP reference strains at both relevant autosomal arms (Figure 2b). Surprisingly, the FRT and EP lines are largely not of FRT or EP genotype on the chromosome arm for which they have not been calibrated. Overall, we found novel alleles for 18 of the 48 assays, and in an extreme case, we even observed five different alleles in five examined strains (2R017, Figure 2b). This result further highlights the heterogeneity of Drosophila strains (see Additional data file 1 for further details on FLP calibration and fly genetics).
FLP mapping in Drosophila
In a genetic screen devised to isolate genes that regulate growth and are situated on chromosome 2R, we found a complementation group characterized by a mild overgrowth phenotype (Figure 4b (2), and C. Rottig and E.H., unpublished work). From a cross between allele VI.29 and EP0755 we recovered three types of recombinant chromosomes: recombinants with a crossover proximal or distal to the mutation, respectively, and double-crossovers (Figure 4a, see also Additional data file 1 for further details on the crossing scheme). The mutation could be placed 16.9 cM proximal to EP0755 and 38.7 cM distal to FRT42D. The FLPs in the recombinant flies were directly analyzed without backcrossing the recombinant chromosome into a parental strain background. DNA was prepared from recombinants by a novel high-throughput protocol (see Materials and methods). We genotyped 34 distal crossover events, 40 proximal crossovers, and eight double-crossovers. This analysis placed the mutation between markers 2R096 and 2R109 (Figure 4c). This interval includes the tumor suppressor hippo [36], and subsequent complementation analysis confirmed VI.29 as a weak hippo allele (data not shown). Furthermore, data from this and other FLP mappings in this region allowed us to further refine the genetic map (Additional data file 11). This kind of experimental data is helpful to space new FLP assays more evenly on the genetic map should the available map turn out to be inexact.
If the resolution of the validated FLP map is too low to identify a candidate gene, we further refine the map position by several approaches. First, we design novel FLP-assays in the region of interest and genotype the most informative recombinants from the first round of FLP mapping (Additional data file 18). Second, we genotype recombinants with SNPs available in the region of interest and resolve them by RFLP, sequencing or DHPLC [2,9]. Third, we perform complementation analysis with recently established Drosophila lines with molecularly defined deletions [37,38]. (See Additional data file 7 for a detailed flowchart illustrating the mapping process.)
Conclusions
We have developed an automated method to detect most naturally occurring InDel polymorphisms at single base-pair resolution. Since a significant fraction of polymorphisms are caused by InDels of only a few base pairs (for example, 8% to 20% in humans [22]) the resolution of the medium-density FLP maps can be greatly increased where necessary, for example during the positional cloning of genes. We are therefore continually designing new FLP assays according to our specific needs using the predicted FLPs (Additional data files 12-18). The full automation of the genotyping has three main advantages when compared to manual methods. First, the error rate (the number of wrongly assigned genotypes) is extremely low, as it was not measurable in 432 assays. Second, genotyping can be done very rapidly and at a high-throughput with little manpower. The automatic allele-calling, in particular, saves much time. As the identification of informative recombinants is usually the rate-limiting step, FLP mapping is very helpful in extracting the few relevant recombinants from a large number of samples. Third, thanks to the standardized conditions, the low error rate and the absence of a secondary assay, FLP mapping is considerably cheaper than the previously published 'manual' mapping methods [2,3]. Unlike other high-throughput methods like TaqMan, Pyrosequencing, DHPLC, fluorescence polarization or primer-extension assays, FLP mapping does not require any investment in specialized equipment. It can be done in any molecular biology lab with access to a sequencing facility equipped with a capillary- or gel-based system, which usually includes the genotyping software. PCR costs are marginally higher because of the use of fluorescently labeled primers, but there are no added expenses for secondary enzymatic assays.
It seems likely that in most organisms the frequency of polymorphisms caused by InDels is in the same range as found in humans, C. elegans or Drosophila. For example, 7.3% of the Arabidopsis sequence polymorphisms are InDels [39]. Thus, FLP mapping can easily be adapted to any organism for which polymorphism maps have been established, as there is no conceptual difference between human, Arabidopsis, C. elegans or Drosophila FLPs.
Materials and methods
C. elegans and Drosophila culture techniques and alleles
Culturing and crossing of C. elegans was done according to standard procedures described in [26]. C. elegans alleles used were: LG I: lin-11(zh41), lin-11(n389); LG II: rol-1(e91), let-23(sy1), unc-52(e444). Drosophila strains and the genetic screen have been described previously [9,35,40-42].
Single worm DNA extraction
Adult worms were collected in 10 μl lysis buffer (50 mM KCl, 10 mM Tris pH 8.2, 2.5 mM MgCl2, 0.45% NP-40, 0.45% Tween-20, 100 μg/ml freshly added proteinase K) and incubated for 60 min at 65°C followed by heat-inactivation of proteinase K at 95°C for 10 min. Before PCR, 90 μl double-distilled H2O (ddH2O) was added to obtain a total volume of 100 μl per lysate.
Fly DNA extraction
DNA from recombinant flies was extracted in bulk by squishing flies through mechanical force in a vibration mill (Retsch MM30) programmed to shake for 20 sec at 20 strokes per second [43]. Single flies were placed into wells of a 96-well format deep-well plate with each well filled with 200 μl squishing buffer (10 mM Tris-Cl pH 8.2, 1 mM EDTA, 0.2% Triton X-100, 25 mM NaCl, 200 μg/ml freshly added proteinase K) and a tungsten carbide bead (Qiagen). The deep-well plate was then sealed with a rubber mat (Eppendorf) and clamped into the vibration mill. (Tungsten carbide beads can be recycled: after an overnight incubation in 0.1 M HCl and thorough washing in ddH2O the beads are virtually free of contaminating DNA.) Debris was allowed to settle for about 5 min, and 50 μl of each supernatant were transferred into a 96-well PCR plate. The reactions were incubated in a thermo-cycler for 30 min at 37°C and finally for 10 min at 95°C to heat-inactivate proteinase K. Before PCR amplification, the crude DNA extracts were diluted 20-fold to reduce the concentration of proteins that might be harmful for the capillary sequencer.
PCR and FLP fragment analysis
Diluted single-worm lysates (2 μl samples) or single fly extracts were added to 23 μl PCR reaction mix. Final concentrations in the PCR reaction were: 0.4 μM forward/reverse primer, 0.2 mM dNTPs, 2 mM MgCl2, 1x PCR reaction buffer, 0.25 U EuroTaq polymerase (Euroclone). PCR reaction setup was done in 96-well plates using a Tecan Genesis pipetting robot with disposable tips. PCR was carried out in two MJR thermo-cyclers that are integrated into the robot. The current setup allows for the sequential processing of six 96-well plates at a time. Cycling parameters were 2 min 95°C, 20 sec 95°C, 20 sec 61°C (-0.5°C for each cycle), 45 sec 72°C (for 10 cycles) followed by 24 cycles of 20 sec 95°C, 20 sec 56°C, 45 sec 72°C and a 10 min 72°C final extension. Following PCR, reactions were diluted 1:100 in water, and 2 μl diluted PCR products were mixed with 10 μl HiDi formamide containing 0.025 μl LIZ500 size standard (Applied Biosystems). This dilution before analysis on the capillary sequencer is necessary to reduce signal intensity because too strong signals compromise data analysis. In addition, sample dilution reduces the risk of damaging the capillaries with proteins or lipids present in the crude lysates. The dilution was done with standard tips using the Tecan Genesis pipetting station. Carryover of fragments was prohibited by a simple wash step with H2O. Fragments were analyzed on an ABI3730 capillary sequencer using POP7 polymer according to standard procedures. Data were analyzed using AppliedBiosystems GeneMapper software and raw data were treated further with Microsoft Excel.
Additional data files
The following additional data are available with the online version of this article. Additional data file 1 contains general information on fly genetics.
Further C. elegans mapping results are given in Additional data files 2,3,4 and 5. Detailed flowcharts illustrating the FLP mapping process are shown in Additional data files 6 and 7. Additional data file 8 contains electropherograms demonstrating the accuracy of allele-calling. Additional data files 9 and 10 contain tables of primer and sequence data of experimentally verified FLP assays in C. elegans and Drosophila, respectively. Additional data file 11 contains a table of the refined genetic distances for FLP assays on the right arm of Drosophila chromosome 2. Additional non-validated FLPs can be found in Additional data files 12,13,14,15,16 and 17 (C. elegans) and Additional data file 18 (Drosophila).
Supplementary Material
Additional data file 1
General information on fly genetics
Click here for additional data file
Additional data file 2
Proof-of-principle for chromosomal linkage
with 3 known mutations on chromosome 2. Assays used to assess linkage
were ZH1-01, ZH2-01, ZH3-05a, ZH4-03, ZH5-01 and ZHX-02
Click here for additional data file
Additional data file 3
Mapping of let-23 to its subchromosomal region (C. elegans)
Click here for additional data file
Additional data file 4
Mapping of rol-1 to its subchromosomal region (C. elegans)
Click here for additional data file
Additional data file 5
Mapping of unc-52 to its subchromosomal region (C. elegans)
Click here for additional data file
Additional data file 6
C. elegans FLP mapping flow chart
Click here for additional data file
Additional data file 7
Drosophila FLP mapping flow chart
Click here for additional data file
Additional data file 8
Electropherograms demonstrating the accuracy of allele-calling
Click here for additional data file
Additional data file 9
Tables of primer and sequence data of experimentally verified FLP assays in C. elegans
Click here for additional data file
Additional data file 10
Tables of primer and sequence data of experimentally verified FLP assays in Drosophila
Click here for additional data file
Additional data file 11
A table of the refined genetic distances for FLP assays on the right arm of Drosophila chromosome 2
Click here for additional data file
Additional data file 12
Additional non-validated FLPs (predicted C. elegans InDels LGI)
Click here for additional data file
Additional data file 13
Additional non-validated FLPs (predicted C. elegans InDels LGII)
Click here for additional data file
Additional data file 14
Additional non-validated FLPs (predicted C. elegans InDels LGIII)
Click here for additional data file
Additional data file 15
Additional non-validated FLPs (predicted C. elegans InDels LGIV)
Click here for additional data file
Additional data file 16
Additional non-validated FLPs (predicted C. elegans InDels LGV)
Click here for additional data file
Additional data file 17
Additional non-validated FLPs (predicted C. elegans InDels LGX)
Click here for additional data file
Additional data file 18
Additional non-validated FLPs (Drosophila)
Click here for additional data file
Acknowledgements
We are grateful to Carmen Rottig for providing us with the novel hippo mutant and to DJ Pan for the hpo42-20 mutation. Angela Baer is acknowledged for excellent technical assistance. This work was funded by projects from the Swiss National Science Foundation and the Kanton Zürich.
Figures and Tables
Figure 1 FLP detection of InDels of various sizes in homozygotes and heterozygotes. In each panel the top two graphs show the homozygotes and the bottom graph the heterozygote. Gray shaded areas mark the defined expected allele lengths and red lines indicate the borders of a predefined window of expected allele lengths. (a-c) Detection of InDels in C. elegans that show increasing levels of adenosine (A) addition. (a) 3-bp InDel ZH1-01 with no A addition; (b) 12-bp InDel ZH2-01 with A addition; (c) 2-bp InDel ZH3-05a with A addition. (d) 1-bp InDel ZH3-23 in C. elegans with A addition. An unambiguous allele-call can be made, irrespectively of the level of A addition: both homozygous samples consist of two peaks at different positions, whereas the heterozygous animal exhibits three peaks. (e) The 1-bp InDel 3R160 in Drosophila runs over a 12-13 nucleotide poly(T) stretch and exhibits stutter bands. Even in this case, a clear allele-call can be made (three peaks in homozygous and four peaks in heterozygous animals). (f) The 6-bp InDel ZHX-22 in C. elegans occurs in a poly(C) stretch and the FLP graph displays stutter bands. As expected, the longer fragment exhibits a higher degree of stuttering.
Figure 2 C. elegans and Drosophila FLP maps. (a) The C. elegans FLP map. Marker names comprise a ZH prefix followed by the chromosome number and a unique identifier number. Markers used in first-level assays (Tier 1) for determination of chromosomal linkage are in red, those used for second-level assays (Tier 2) for higher resolution mapping are in black. (b) The Drosophila FLP map of chromosomes 2 and 3. The FRT sites and EP elements are symbolized by blue and green triangles, respectively. The strains that were genotyped are shown below each chromosome. Green indicates the EP genotype, blue the FRT genotypes and new alleles are shown in other colors.
Figure 3 FLP mapping in C. elegans. (a) Crossing scheme used to map mutations generated in the N2 Bristol background. The different classes of recombinants recovered in the F2 generation are shown. (b) Analysis of the zh41 mutation with Tier 1 assays establishes linkage to chromosome I. (c) Analysis with Tier 2 places zh41 between assays ZH1-01 and ZH1-15. ND, no data as a result of PCR reaction failure. (d) Ventral views of the vulva in wild-type and zh41 L4 larvae stained with the adherens junction antibody MH27 [44]. In the wild type, the vulval cells have fused to generate the torroids that appear as concentric rings. zh41 mutants exhibit the same fusion defects observed in other lin-11 alleles [33].
Figure 4 FLP mapping in Drosophila. (a) Crossing scheme used to map mutations generated in the FRT background and recombined with an EP line. The different classes of recombinants recovered in the F2 generation are shown. (b) Big head phenotypes of the hippo null allele hpo42-20 (1) and the VI.29 mutation (2). A wild-type control is shown in (3). (c) FLP mapping of the VI.29 mutation on chromosome 2R. Analysis of the different classes of recombinants places the mutation between markers 2R096 and 2R109 (dashed red line). Informative recombinants are boxed in red. ND, not determined or no data as a result of PCR reaction failure.
==== Refs
Sachidanandam R Weissman D Schmidt SC Kakol JM Stein LD Marth G Sherry S Mullikin JC Mortimore BJ Willey DL A map of human genome sequence variation containing 1.42 million single nucleotide polymorphisms. Nature 2001 409 928 933 11237013 10.1038/35057149
Berger J Suzuki T Senti KA Stubbs J Schaffner G Dickson BJ Genetic mapping with SNP markers in Drosophila. Nat Genet 2001 29 475 481 11726933 10.1038/ng773
Wicks SR Yeh RT Gish WR Waterston RH Plasterk RH Rapid gene mapping in Caenorhabditis elegans using a high density polymorphism map. Nat Genet 2001 28 160 164 11381264 10.1038/88878
Kwok PY Chen X Detection of single nucleotide polymorphisms. Curr Issues Mol Biol 2003 5 43 60 12793528
Syvanen AC Accessing genetic variation: genotyping single nucleotide polymorphisms. Nat Rev Genet 2001 2 930 942 11733746 10.1038/35103535
Chen X Sullivan PF Single nucleotide polymorphism genotyping: biochemistry, protocol, cost and throughput. Pharmacogenomics J 2003 3 77 96 12746733 10.1038/sj.tpj.6500167
Livak KJ Allelic discrimination using fluorogenic probes and the 5' nuclease assay. Genet Anal 1999 14 143 149 10084106
Wolford JK Blunt D Ballecer C Prochazka M High-throughput SNP detection by using DNA pooling and denaturing high performance liquid chromatography (DHPLC). Hum Genet 2000 107 483 487 11140946 10.1007/s004390000396
Nairz K Stocker H Schindelholz B Hafen E High-resolution SNP mapping by denaturing HPLC. Proc Natl Acad Sci USA 2002 99 10575 10580 12149455 10.1073/pnas.162136299
White R Lalouel JM Chromosome mapping with DNA markers. Sci Am 1988 258 40 48 2903549
Hoskins RA Phan AC Naeemuddin M Mapa FA Ruddy DA Ryan JJ Young LM Wells T Kopczynski C Ellis MC Single nucleotide polymorphism markers for genetic mapping in Drosophila melanogaster. Genome Res 2001 11 1100 1113 11381036 10.1101/gr.GR-1780R
Syvanen AC From gels to chips: "minisequencing" primer extension for analysis of point mutations and single nucleotide polymorphisms. Hum Mutat 1999 13 1 10 9888384
Kwok PY High-throughput genotyping assay approaches. Pharmacogenomics 2000 1 95 100 11258600 10.1517/14622416.1.1.95
Swan KA Curtis DE McKusick KB Voinov AV Mapa FA Cancilla MR High-throughput gene mapping in Caenorhabditis elegans. Genome Res 2002 12 1100 1105 12097347
Sosnowski RG Tu E Butler WF O'Connell JP Heller MJ Rapid determination of single base mismatch mutations in DNA hybrids by direct electric field control. Proc Natl Acad Sci USA 1997 94 1119 1123 9037016 10.1073/pnas.94.4.1119
Gilles PN Wu DJ Foster CB Dillon PJ Chanock SJ Single nucleotide polymorphic discrimination by an electronic dot blot assay on semiconductor microchips. Nat Biotechnol 1999 17 365 370 10207885 10.1038/7921
Weissenbach J Microsatellite polymorphisms and the genetic linkage map of the human genome. Curr Opin Genet Dev 1993 3 414 417 8353415 10.1016/0959-437X(93)90114-5
A comprehensive genetic linkage map of the human genome. NIH/CEPH Collaborative Mapping Group. Science 1992 258 67 86 1439770
McCouch SR Chen X Panaud O Temnykh S Xu Y Cho YG Huang N Ishii T Blair M Microsatellite marker development, mapping and applications in rice genetics and breeding. Plant Mol Biol 1997 35 89 99 9291963 10.1023/A:1005711431474
Knapik EW Goodman A Ekker M Chevrette M Delgado J Neuhauss S Shimoda N Driever W Fishman MC Jacob HJ A microsatellite genetic linkage map for zebrafish (Danio rerio). Nat Genet 1998 18 338 343 9537415 10.1038/ng0498-338
Collins JR Stephens RM Gold B Long B Dean M Burt SK An exhaustive DNA micro-satellite map of the human genome using high performance computing. Genomics 2003 82 10 9 12809672 10.1016/S0888-7543(03)00076-4
Weber JL David D Heil J Fan Y Zhao C Marth G Human diallelic insertion/deletion polymorphisms. Am J Hum Genet 2002 71 854 862 12205564 10.1086/342727
Smith JR Carpten JD Brownstein MJ Ghosh S Magnuson VL Gilbert DA Trent JM Collins FS Approach to genotyping errors caused by nontemplated nucleotide addition by Taq DNA polymerase. Genome Res 1995 5 312 317 8593617
Ranade K Chang MS Ting CT Pei D Hsiao CF Olivier M Pesich R Hebert J Chen YD Dzau VJ High-throughput genotyping with single nucleotide polymorphisms. Genome Res 2001 11 1262 1268 11435409
Chen DC Saarela J Nuotio I Jokiaho A Peltonen L Palotie A Comparison of GenFlex Tag array and pyrosequencing in SNP genotyping. J Mol Diagn 2003 5 243 249 14573784
Brenner S The genetics of Caenorhabditis elegans. Genetics 1974 77 71 94 4366476
Lindsley DL Zimm GG The genome of Drosophila melanogaster 1992 New York: Academic Press
Adams MD Celniker SE Holt RA Evans CA Gocayne JD Amanatides PG Scherer SE Li PW Hoskins RA Galle RF The genome sequence of Drosophila melanogaster. Science 2000 287 2185 2195 10731132 10.1126/science.287.5461.2185
Aroian RV Sternberg PW Multiple functions of let-23, a Caenorhabditis elegans receptor tyrosine kinase gene required for vulval induction. Genetics 1991 128 251 267 2071015
Higgins BJ Hirsh D Roller mutants of the nematode Caenorhabditis elegans. Mol Gen Genet 1977 150 63 72 834177
Mackenzie JM Garcea RL JrZengel JM Epstein HF Muscle development in Caenorhabditis elegans: mutants exhibiting retarded sarcomere construction. Cell 1978 15 751 762 728988 10.1016/0092-8674(78)90261-1
Burdine RD Branda CS Stern MJ EGL-17(FGF) expression coordinates the attraction of the migrating sex myoblasts with vulval induction in C. elegans. Development 1998 125 1083 1093 9463355
Gupta BP Wang M Sternberg PW The C. elegans LIM homeobox gene lin-11 specifies multiple cell fates during vulval development. Development 2003 130 2589 2601 12736204 10.1242/dev.00500
Kamath RS Fraser AG Dong Y Poulin G Durbin R Gotta M Kanapin A LeBot N Moreno S Sohrmann M Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 2003 421 231 237 12529635 10.1038/nature01278
Newsome TP Asling B Dickson BJ Analysis of Drosophila photoreceptor axon guidance in eye-specific mosaics. Development 2000 127 851 860 10648243
Ryoo HD Steller H Hippo and its mission for growth control. Nat Cell Biol 2003 5 853 855 14523394 10.1038/ncb1003-853
Ryder E Blows F Ashburner M Bautista-Llacer R Coulson D Drummond J Webster J Gubb D Gunton N Johnson G The DrosDel collection: a set of P-element insertions for generating custom chromosomal aberrations in Drosophila melanogaster. Genetics 2004 167 797 813 15238529 10.1534/genetics.104.026658
Parks AL Cook KR Belvin M Dompe NA Fawcett R Huppert K Tan LR Winter CG Bogart KP Deal JE Systematic generation of high-resolution deletion coverage of the Drosophila melanogaster genome. Nat Genet 2004 36 288 292 14981519 10.1038/ng1312
Schmid KJ Sorensen TR Stracke R Torjek O Altmann T Mitchell-Olds T Weisshaar B Large-scale identification and analysis of genome-wide single-nucleotide polymorphisms for mapping in Arabidopsis thaliana. Genome Res 2003 13 1250 1257 12799357 10.1101/gr.728603
Xu T Rubin GM Analysis of genetic mosaics in developing and adult Drosophila tissues. Development 1993 117 1223 1237 8404527
Rorth P A modular misexpression screen in Drosophila detecting tissue-specific phenotypes. Proc Natl Acad Sci USA 1996 93 12418 12422 8901596 10.1073/pnas.93.22.12418
St Johnston D The art and design of genetic screens: Drosophila melanogaster. Nat Rev Genet 2002 3 176 188 11972155 10.1038/nrg751
Nairz K Zipperlen P Dearolf C Basler K Hafen E A reverse genetic screen in Drosophila using a deletion-inducing mutagen. Genome Biol 2004 5 R83 15461801 10.1186/gb-2004-5-10-r83
Francis R Waterston RH Muscle cell attachment in Caenorhabditis elegans. J Cell Biol 1991 114 465 479 1860880 10.1083/jcb.114.3.465
| 15693948 | PMC551539 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Jan 17; 6(2):R19 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r19 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r201569394910.1186/gb-2005-6-2-r20MethodGenomic analysis of early murine mammary gland development using novel probe-level algorithms Master Stephen R [email protected] Alexander J [email protected] L Charles [email protected] Tien-Chi [email protected] Katherine D [email protected] Lewis A [email protected] Department of Cancer Biology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA2 Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA3 Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA4 Abramson Family Cancer Research Institute, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA5 Department of Pediatrics, Children's Hospital of Philadelphia, Philadelphia, PA 19104, USA2005 1 2 2005 6 2 R20 R20 25 8 2004 1 10 2004 8 12 2004 Copyright © 2005 Master 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.
A novel algorithm (ChipStat) is presented for detecting gene-expression changes from Affymetrix microarray data. The method is used to identify changes in murine mammary development.
We describe a novel algorithm (ChipStat) for detecting gene-expression changes utilizing probe-level comparisons of replicate Affymetrix oligonucleotide microarray data. A combined detection approach is shown to yield greater sensitivity than a number of widely used methodologies including SAM, dChip and logit-T. Using this approach, we identify alterations in functional pathways during murine neonatal-pubertal mammary development that include the coordinate upregulation of major urinary proteins and the downregulation of loci exhibiting reciprocal imprinting.
==== Body
Background
The widespread use of DNA microarrays to measure transcript abundance from a significant fraction of the genome has proven to be a valuable tool for identifying functional cellular pathways as well as for capturing the global state of a biological system [1-4]. These arrays have typically been constructed by spotting large, pre-synthesized strands of nucleic acid on an appropriate surface [5] or by directly synthesizing smaller oligonucleotides in situ at defined locations [6]. The latter technique has been implemented in Affymetrix oligonucleotide microarrays designed for expression analysis. Because hybridization to short (25-mer) oligonucleotides is used to measure expression, Affymetrix arrays contain multiple, independent oligonucleotides designed to bind a unique transcript. In this way, specificity and a high signal-to-noise ratio can be maintained despite the noise due to the hybridization itself. When the intensity of hybridization to a given oligonucleotide designed to detect the transcript (a 'perfect match' probe, PM) is corrected by its corresponding (single base-pair 'mismatch', MM) control, an estimate of gene expression (PM - MM) is derived. This probe pair value is then combined with values from the other, independent, oligonucleotides designed to bind the same transcript (together designated the probe set) to obtain a more robust estimate of transcript abundance [7].
The ability to sensitively detect changes in gene expression is crucial for a transcript-level analysis of developmental processes and other processes involving changes in the relative sizes of cellular compartments. Early attempts to limit the false-positive rate of microarray studies focused on the magnitude of fold-change in gene expression (see, for example [1]). For studying purified cell populations, where a substantial change in gene expression is more likely to reflect biologically relevant function, such a crude limitation was acceptable. However, adequate studies of complex tissues require a substantially more sensitive method of detection. For example, a small yet reproducible change in gene expression within a whole organ may reflect a substantial expansion or regulatory change within a subpopulation of cells that overexpress a given gene relative to the surrounding tissue. Thus, a method for identifying such small, statistically significant changes in gene expression is required.
Because of the variety of techniques used to measure gene expression, it has become commonplace to utilize simple, numerical estimates of gene expression as the starting point for such identification. One major drawback to this approach has been that individual probe cell information from Affymetrix microarrays is routinely discarded. This issue has only recently begun to be addressed [8-10], and it appears that a substantial amount of useful information can be obtained from probe-level analysis.
An additional compromise has been driven by the practical difficulties of performing large numbers of microarray experiments. Given limited samples, permutation of the existing experimental dataset, rather than use of independent sets of control samples, has been widely used to estimate the statistical significance of differential gene expression [11]. Although this technique has been useful given the historically high cost of performing microarray analysis, it may inherently limit the sensitivity of the results obtained. As such, a test for differential gene expression that utilizes a 'gold standard' negative-control dataset would have clear advantages.
The impetus for the work described here is the desire to sensitively identify coherent patterns of gene expression during mammary gland development. At 2 weeks of age, the female FVB mouse mammary gland exists as a rudimentary epithelial tree embedded at one end of a fat pad composed of adipose tissue and fibroblasts. Previous work has demonstrated a fundamental transition in the composition of the mammary adipose compartment from brown fat to white fat during early development [4]. By 3 weeks of age, the onset of puberty heralds the beginning of the process of ductal morphogenesis, which results in the formation of the branching epithelial tree of the adult gland. The onset of puberty results not only in the rapid growth of a ductal epithelial tree but also the appearance of specialized, highly proliferative structures known as terminal end buds that elaborate this tree via branching morphogenesis [12,13]. Furthermore, puberty is known to be a time of increased susceptibility to carcinogenesis [14,15]. Thus, a detailed examination of transcriptional changes during this period would be of substantial use.
We describe here a novel algorithm for sensitively detecting gene-expression changes using information derived from individual probe cell hybridizations to Affymetrix oligonucleotide microarrays. In addition to modeling the predicted behavior of this algorithm, we have generated an independent cohort of control samples derived from the murine mammary gland that can be used to empirically calibrate its statistical behavior. We have then used this algorithm to analyze a biological transition in early murine mammary gland development in order to compare the sensitivity of this approach to other commonly used algorithms. In conjunction with a second novel algorithm, we have developed an aggregate approach to the reliable detection of differential gene expression that yields substantially improved sensitivity across a range of false-positive rates and have applied this approach to the analysis of early murine mammary gland development.
Results
A variety of traditional statistical methods, such as the t test, have been used in conjunction with microarray datasets to detect changes in gene expression (see for example [16]). Given the large numbers of genes tested, it is widely recognized that a stringent threshold for statistical significance is necessary in order to reduce the number of false positive changes. For example, a threshold of statistical significance of P < 0.001 would be expected to yield around 100 false positives on a typical array measuring 10,000 genes. Some algorithms, such as significance analysis for microarrays (SAM) [11], explicitly control the number of expected false-positive results using permutations of the existing dataset. Regardless of the method utilized, statistical differences are typically calculated on the basis of an aggregate measure of gene expression (a gene signal). However, a fundamental difficulty with these methods is that they often do not have the requisite statistical power to sensitively detect changes in gene expression after correction for multiple hypothesis testing. We reasoned that utilizing the multiple hybridizations to independent oligonucleotides on the Affymetrix platform might allow us to develop a method for detecting expression changes with substantially greater statistical power.
To test this approach, we developed a novel analytical algorithm that is based on identifying individual differences at a given statistical significance between corresponding probe pairs. To a first approximation, the signal on any given probe cell can be modeled as:
S = M + E(b) + E(p) + E(h), E ~ N
Where S is the signal detected on the microarray, M is the average message level in a given experimental state, E(b) is noise due to biological variation between animals or animal pools, E(p) is the noise due to variations in sample measurement, and E(h) is the noise inherent in hybridization to oligonucleotide features on the array. The goal of our analysis was to identify a method that would allow us to reliably distinguish significant differences in M under particular experimental conditions.
Given this model, we reasoned that the relative magnitude of E(b) + E(p) (the experimental noise) compared with E(h) (the hybridization noise) should determine whether comparisons between individual probe pairs would be useful. If the bulk of noise in our microarray data was due to factors influencing the level of transcript available for measurement (that is, E(b) + E(p) >> E(h)), then individual probe-pair measurements should only reflect the pre-hybridization bias in transcript availability. In this case, the t-test or other measurement based on the average of the probe set would be expected to perform as well as an algorithm based on individual probe-pair comparisons. In contrast, if most noise in the measurement of true transcript level exists at the level of hybridization to a given oligonuclotide (E(b) + E(p) << E(h)), then the independent measurements of probe-pair differences more closely approximate independent measurements of differences in gene expression. In the most extreme case - if E(h) is sufficiently larger than E(b) + E(p) - each oligonucleotide in the probe set could be considered as an independent measurement of gene expression and the probability of observing a given number of probe pairs changing under the null hypothesis would be determined by the binomial distribution.
To explore this possibility, we implemented an algorithm, hereafter designated ChipStat, that takes corresponding probe pairs across two comparison groups and tests them for statistical significance with P less than a fixed value (hereafter denoted pps). To avoid making assumptions about equal variance in both groups, a heteroscedastic t-test is used. We would expect that probe sets in which larger numbers of individual probe pairs show a significant change in the same direction are more likely to be measuring differentially regulated genes. Thus, for any given probe set, the number of probe pairs (0-16) changing in a given direction with P less than pps is tabulated and used as a measure of the significance of change in gene expression. We simulated the expected behavior of this algorithm under the null hypothesis (no difference in gene expression) across various ratios of E(b) + E(p) and E(h) (see Materials and methods for details). Results are shown in Figure 1.
Validation and optimization of the ChipStat algorithm
Although this approach provides a statistical methodology for identifying changes in gene expression, it is only possible to directly calculate a P value associated with this change in limiting cases. If E(h) >> E(b) + E(p), the binomial distribution can be used to calculate the resulting significance (given the number of changes, total number of probe pairs, and pps); however, the relative contributions of E(h), E(b), and E(p) to the total error function are not known a priori.
To empirically measure the null distribution for three-sample versus three-sample comparisons, a cohort of independent control samples for our experimental system was generated. To do this, the third, fourth and fifth mammary glands were harvested from 18 age-matched 5-week-old control female mice. After extraction of RNA, groups of three animals were pooled to create six initial RNA samples. Biotinylated cRNA was then independently prepared from these pooled RNA samples and hybridized to Affymetrix MG_U74Av2 oligonucleotide microarrays, yielding six datasets. All possible three by three combinations were compared across 11,820 probe sets (corresponding to all probe sets on the MG_U74Av2 that contain exactly 16 probe pairs), and the cumulative distribution of false positives as a function of pps and the number of probe pairs changed was tabulated. Results are shown for pps = 0.05 (Figure 2). It is notable that very few false positives are associated with large numbers (more than 10/16) of probe pairs changing. While the number of false-positive probe sets does not decline as rapidly as the binomial distribution, the overall curve is consistent with a large component of hybridization noise (compare Figures 1 and 2), suggesting the utility of a probe-level approach. Likelihood maximization of our initial statistical model (E ~ N, ignoring probe-specific effects) using results for low numbers of probe pairs (0 to 6) changing suggests that E(h) (hybridization noise) is approximately 2.5 times greater than E(b) + E(p) (experimental noise). We note, however, that the empirically derived null distribution can be used to derive a valid test of significance for ChipStat regardless of the validity of the underlying model and without any direct calculation of relative noise contributions by E(h), E(b) and E(p).
An ideal method for identifying differentially regulated genes would maximize the number of genes identified while maintaining a low fixed number of expected false positives. We have previously shown the utility of testing the statistical overlap of discrete gene lists with biologically relevant annotation in order to identify functional pathways during murine mammary gland development [4]. This maximization is therefore of particular experimental interest. To evaluate the ChipStat algorithm from this perspective, we performed triplicate microarray measurements of RNA derived from the mammary glands of independent pools (more than 10 animals per pool) of wild-type female FVB mice harvested at 2 or 5 weeks of postnatal development. We wished to determine the number of statistically significant increases in gene expression from 2 to 5 weeks of age, a period of postnatal development that encompasses the rapid epithelial proliferation that accompanies ductal morphogenesis in the mammary gland at the onset of puberty [17].
ChipStat was used to analyze differences between the 2- and 5-week mammary gland samples (pps = 0.05), and the number of statistically significant increases was measured as a function of the number of genes expected to appear on the list by chance. Results are shown in Figure 3a. The number of expected false positives was empirically obtained from the negative-control dataset described previously. Thus, for example, under conditions pps = 0.05 with 8/16 probe pairs increasing, where around five genes are expected to be identified by chance, we find that the measured number of differentially regulated genes is around 160. This corresponds to a false-positive rate of approximately 3% (or, conversely, a true-positive rate of approximately 97%). It is also apparent (Figure 3a) that the sensitivity of detection can be 'tuned' on the basis of the number of false positives that are deemed acceptable.
To determine whether the sensitivity of this algorithm could be further optimized, similar analyses were performed at various values of pps (Figure 3b). These data suggest that relative sensitivity as a function of false-positive rate is maximized at pps approximately equal to 0.04-0.05 (note the similarity of these curves in Figure 3b). Furthermore, while certain other values of pps yield increased sensitivity at specific points (for example, pps = 0.03 at around four genes expected by chance; data not shown), values of 0.04-0.05 appear appropriate across most highly-significant P values. A marked decrease in sensitivity for a given false-positive rate is noted both at low (0.01) and high (0.1, 0.15) values of pps.
Although the use of negative-control samples provides a definitive method for evaluating the behavior of our statistical algorithms, we independently verified these results using northern blot hybridization. Genes differentially expressed (6/16 probe pairs increasing, pps = 0.04) from 2 to 5 weeks of mammary gland development were identified, and analysis of the control data suggested that fewer than 10 increases would be expected by chance at this significance level (corresponding to P < 7.7 × 10-4). Manual inspection of the resulting list revealed the presence of a number of genes known to be upregulated during this developmental transition, including cytokeratin 19 (Krt1-19), cytokeratin 8 (Krt2-8), and κ casein (Csnk). However, to avoid bias toward previously studied genes or known genes with high fold change, genes were randomly selected from subsets of this list corresponding to high-stringency (P < 2.2 × 10-4), low-stringency with high fold change (2.2 × 10-4 <P < 7.7 × 10-4, ≥ 1.8-fold change), and low-stringency with low fold change (2.2 × 10-4 <P < 7.7 × 10-4, < 1.8-fold change). Results from northern blot analyses using probes for these randomly selected genes are shown in Table 1. Of nine genes selected, eight were shown to change significantly via northern blot analysis.
Of note, the single gene that did not show a significant change (Ldh1) was from the low-stringency group and was predicted to show only a 1.37-fold change. In contrast, northern hybridization confirmed the differential expression of other genes with only modest fold-changes (for example, Sqstm1, 1.48-fold change from 2 to 5 weeks). As the genes tested were not biased toward higher fold change (only 2/75 genes with fold change > 3 were randomly selected for northern confirmation), our data demonstrate the ability of ChipStat to reliably detect the types of small, reproducible changes in gene expression that are necessary for whole-organ analysis.
Comparison of ChipStat with other analytical methods
Other methods of detecting differential gene expression have been widely utilized, including SAM [11] and dChip [8]. As previously discussed, SAM utilizes an aggregate (probe-set-level) estimate of gene expression as its analytical starting point. Similarly, although dChip utilizes probe-cell-level analysis to determine the level and statistical bounds of gene expression, it does not explicitly make use of probe-level comparisons for identifying differentially regulated genes. More recently, the logit-T algorithm, which in contrast to SAM and dChip utilizes probe-pair-level comparisons for statistical testing, has been shown to improve differential expression testing performance in a variety of Latin square datasets reflecting technical replicates of samples with spiked-in transcripts [10]. We therefore wished to determine the performance of the ChipStat algorithm relative to these methodologies. Further, as our control dataset incorporates biological and experimental variability in addition to sample preparation and hybridization noise, we reasoned that it would provide a more appropriate estimate of the performance of these algorithms when analyzing data from an experimentally plausible animal model.
SAM, dChip, the t-test and logit-T all provide a P value estimating statistical significance in the absence of an empirical measurement of the underlying null distribution; Figure 3c shows a comparison with ChipStat when using these estimated P values. However, as ChipStat requires the additional information provided by this empirical distribution for statistical calibration, the inherent performance of other algorithms may be underestimated if they are not similarly calibrated. To correct for this difference, the significance of SAM, dChip and logit-T values were assessed using all three by three combinations of the null dataset (given the permutation-based calibration of false-discovery rate utilized by SAM, note that SAM values are not predicted to improve significantly using this method of calibration). Results are shown in Figure 3d. In the case of the t-test, results obtained using calculated P values are generally within 5% of comparable results using empirically calibrated P values. Logit-T and dChip appear much less sensitive when using reported P values, although both of these techniques show improvement when calibrated using the control dataset. Of particular note, logit-T performs only slightly less well than ChipStat when calibrated against our control distribution, consistent with the fact that it was the only other algorithm considered that performs probe-pair-level comparisons when testing for differential gene expression.
Design and validation of the Intersector algorithm
Although the Affymetrix Microarray Suite (MAS) software utilizes probe-level information in identifying differentially expressed genes, its use has been restricted to single-array comparisons. As a result, it has been widely recognized that this approach generates an unacceptably high number of false-positive results. The use of replicate samples, however, might be expected to lower the false-positive rate while achieving a higher sensitivity. We therefore combined pairwise comparisons between triplicate data points in two different groups (that is, nine comparisons in total) and determined differential expression based on the Affymetrix call (for example, increases + marginal increases) for these comparisons. A similar technique, in which a simple majority cutoff (5/9 changes) was considered to denote significant change, has recently been described [18]. Although this approach involves N2 comparisons in general for equal groups of N arrays, it is easily feasible for three-sample versus three-sample comparisons. We have designated this approach Intersector. Significantly, the control data previously generated to calibrate ChipStat also allow us to determine the empirical false-positive rate for Intersector as a function of the number of 'increase' calls and to perform direct comparisons with other algorithms.
The performance of the Intersector algorithm in comparing 2- versus 5-week mammary gland gene expression is shown in Figure 4a. Interestingly, the Intersector algorithm is able to achieve a slightly improved sensitivity at a given false-positive rate when compared with ChipStat. To determine whether the particular version of the MAS algorithm influences this result, all analyses were run using difference calls from both MAS 4.0 and MAS 5.0 (see Figure 4a). Although the number of changes required to achieve similar sensitivity was different, the Intersector results from MAS 4.0 and MAS 5.0 are comparable at a given false-positive rate.
Given substantial differences between the types of probe-pair comparisons performed by ChipStat and MAS, we next wished to ascertain if these algorithms identify the same sets of upregulated genes. Direct comparison requires that the analyses result in comparable false-detection rates. We therefore compared the lists at thresholds corresponding to approximately 2.5 genes expected by chance, and the closest available threshold with each algorithm was chosen. The resulting thresholds were Intersector (MAS4) 7/9 (1.75 expected by chance), Intersector (MAS5) 8/9 (2.8 expected by chance), and ChipStat (.04) 8/16 (2.68 expected by chance). Notably, examination of these lists demonstrates that each algorithm (Intersector with MAS 4.0 data, Intersector with MAS 5.0 data and ChipStat) detects a discrete set of genes that are not detected by the others (Figure 4b). This is particularly intriguing since empirically estimated false positive rates suggest that these groups of genes are not likely to reflect chance fluctuations alone. Thus, in addition to identifying a core set of regulated genes, the Intersector and ChipStat algorithms each detect sets of complementary, nonoverlapping genes that change significantly.
To confirm this result, five out of the 13 genes uniquely identified by ChipStat were randomly chosen for confirmation. One of these genes was undetectable by northern blot hybridization, and the remaining 4/4 showed differential expression in the predicted direction (5 weeks > 2 weeks) (Table 1, and data not shown). This demonstrates that, at comparable levels of statistical stringency, ChipStat correctly identifies differentially expressed genes that are not identified by Intersector. Further, having directly tested approximately 40% of all genes in this category, no false positives were identified. Examination of lower stringency lists (9.5 expected by chance from ChipStat, 7.4 expected by chance from Intersector using MAS5) also revealed sets of genes identified by ChipStat or Intersector alone. For example, the 'Intersector only' list created at this lower stringency contains α-, β-, and γ-casein; previous work in our lab has demonstrated that these genes are differentially regulated with expression at 5 weeks greater than that at 2 weeks (data not shown).
Development of a hybrid approach
Given the presence of genes uniquely identified by Intersector or ChipStat at a given false positive rate and the feasibility of performing Intersector analysis on small numbers of replicates, we next explored whether a combination of these approaches could further improve overall detection. To test this, all possible pairwise threshold combinations of ChipStat (pps = 0.05, 0/16 to 16/16 probe pairs changing) and Intersector (0/9 to 9/9 increases or marginal increases) were combined, and aggregate lists of genes identified by both algorithms were tabulated (see Additional data file 1). The results demonstrate that a combination of these two approaches can lower the expected false positive rate while maintaining a high sensitivity. For example, the combination of ChipStat (pps = 0.05, 6/16 probe pairs increasing) and Intersector (7/9 increases + marginal increases) detects 209 increasing probe sets with only 3.4 expected to increase by chance (expected false-positive rate less than 2%). A comparison of the false-positive rates for single (ChipStat or Intersector alone) and combined (ChipStat and Intersector) approaches is shown in Figure 4c. Note that the total number of probe sets detected by the combined approach shown in Figure 4c is greater than the number detected by the single approach with a comparable false-detection rate (209 probe sets and 173 probe sets, respectively). The behavior of optimal combinations with respect to the number of genes detected is shown in Figure 4d.
One additional feature of this combined approach is the ability to 'fine-tune' the number of expected false positives. That is, while Intersector (MAS5) allows no choice between approximately three and approximately seven expected false positives (2.8 and 7.35, corresponding to 8/9 or 7/9 changes, respectively), the combined approach provides a smoother continuum of values. More important, these data show that, for certain targeted numbers of expected false positives, a combination of ChipStat and Intersector can provide improved performance in gene detection compared with either algorithm alone.
Genomic characterization of early mammary gland development
The goal of these methodological developments has been the elucidation of biological mechanisms underlying mammary gland development and carcinogenesis. We therefore used the hybrid ChipStat/Intersector lists representing early mammary gland development as a basis for further exploration of developmental processes during this time period. A complete list of genes differentially expressed between 2- and 5-week murine mammary gland was compiled using the techniques described above. The results are listed in Additional data file 2.
To identify coherent functional patterns of gene expression during neonatal development through the onset of puberty, statistically significant associations between Gene Ontology (GO) categories [19] and lists of up- and downregulated genes were identified using EASE [20]. Multiple testing correction was performed using within-system bootstrapping, and a corrected significance threshold of P less than 0.05 was used. Results are shown in Table 2. Upregulated genes were associated with a total of 22 GO categories, and downregulated genes with 10 categories. In addition, this approach provides a convenient test of whether the increased sensitivity of ChipStat/Intersector yields corresponding power in identifying patterns of biological activity. To test this directly, lists of differentially expressed genes with the same number of expected false positives (empirically calibrated as previously) were identified using dChip and logit-T. These lists were then tested for association with GO annotation, and the results are shown (Table 1, Figure 5). Of note, ChipStat/Intersector lists were associated with a greater number of GO categories than were dChip or logit-T, and this was true for both up- and downregulated gene lists. Consistent with our suggestion that logit-T should be most similar to ChipStat/Intersector because of its use of probe-pair-level comparisons, logit-T also generated lists that are statistically associated with a larger number of GO categories than did dChip (Figure 5), although it did not outperform ChipStat/Intersector. ChipStat/Intersector identified 22/22 of categories associated with any of the list of upregulated genes and 10/11 categories identified using any of the lists of downregulated genes. A single downregulated category ('cellular component: extracellular') was associated only with the logit-T list.
To provide a crude check on the reliability of these results in addition to the confirmation previously performed, gene lists were examined for association with previously described biological processes. In addition to individual genes that are consistent with epithelial proliferation and differentiation (discussed above), several statistically associated categories represent pathways that have been previously described in the mammary gland during this developmental window [4]. These include 'blood vessel development' and 'mitochondrial inner membrane'. The latter category reflects the previously reported decrease in brown adipose tissue at the end of the neonatal period and the corresponding decrease in the capability of the mouse to utilize adaptive thermogenesis to maintain body temperature. Brown adipose tissue is not only rich in mitochondria, but the fatty-acid metabolic pathways necessary for adequate thermogenic activity are also spatially localized at the inner mitochondrial membrane. Of note, this category only reached statistical significance using the ChipStat/Intersector list.
Interestingly, 'pheromone binding' and 'odorant binding' categories are also associated with upregulated expression at the onset of puberty. Genes within these categories are primarily members of the major urinary protein (MUP) gene family, and MUP transcripts (Mup1, Mup3, Mup4, Mup5) account for four of the five most highly upregulated genes from 2 to 5 weeks. Large quantities of MUPs are synthesized in the male liver and excreted in the urine, where they bind pheromone and play a role in signaling for complex behavioral traits [21,22]. MUP levels are upregulated during puberty in the liver, although expression levels are much higher in males than in females. While MUP expression within the mammary gland has previously been reported [23,24], its expression was considered to be detectable only with the onset of pregnancy. Our data show that MUPs are highly upregulated in the female mammary gland during the 2- to 5-week transition. Interestingly, Slp (sex-limited protein), which also shows sex-restricted expression in the male liver and - like Mup expression - is normally repressed by Rsl [25], is also significantly upregulated during this period.
Additional examination of these gene lists revealed an interesting transcriptional pattern that is not reflected in the current GO hierarchy. The nontranslated RNA transcript Meg3/Gtl2 is significantly downregulated from 2 to 5 weeks of development, and its reciprocally imprinted neighbor Dlk1 [26] shows a similar decrease. This is noteworthy because two other genes with decreasing expression, H19 (nontranslated RNA) and Igf2, are also reciprocally imprinted neighbors, suggesting the possibility of a common regulatory mechanism for altering expression from loci exhibiting this genomic organizational structure (see [27]).
Discussion
The ability to reliably detect changes in gene expression is critical for the analysis of experimental microarray data. This problem assumes particular importance when analyzing complex mixtures of cells, such as those derived from a whole organ during ontogeny. The challenge can be most clearly seen by considering a small subpopulation of cells that demonstrate a marked change in gene expression. If the expression of this gene is uniform and low throughout the rest of the tissue, the biologically relevant change within a few cells will appear as a low fold change in organ-wide gene expression. A variety of such nonabundant yet developmentally critical cell types have been described. For example, the proliferative capacity of small structures in the mammary gland known as terminal end buds gives rise to the extensive ductal structure that is elaborated during puberty [17]. More recently, the characteristics of mammary stem cells have been described, and these cells have been suggested to serve as targets for carcinogenesis [28,29]. To facilitate the study of such subpopulations within a whole-organ context, therefore, we have developed a novel approach to the analysis of Affymetrix oligonucleotide microarray data.
A variety of nonparametric and parametric statistical tests, including variants of Student's t-test, have been used to identify significant changes in gene expression using replicate microarray data. Given the substantial economic investment required for large microarray experiments, attempts have also been made to improve detection of differentially regulated genes through better estimates of the null distribution using permutation analysis; the use of software incorporating such methods, such as SAM [11], has become widespread. A different approach to improved detection (dChip, see [8]) has attempted to use probe-level information to derive an improved estimate of relative gene expression before assessing differential regulation.
While much work has focused on such use of probe-level analysis for estimating gene expression [8,9], the analysis of replicate data at the probe level for identifying differentially expressed genes has only recently become a focus [10,30]. In particular, if hybridization noise contributes a substantial portion of the overall noise inherent in microarray measurements, the use of multiple probe pairs devoted to measuring a single gene suggests a potential approach to overcoming this noise.
The ChipStat algorithm uses heteroscedastic t-test comparisons between probe pairs, and the number of probe pairs that change greater than a significance threshold are tabulated. A greater number of consistently changing probe pairs should indicate that the difference is less likely to be due to hybridization noise, and thus this number relates the overall probability that the probe set is measuring a true change in gene expression. The processing time for the ChipStat algorithm scales as a linear function of the number of replicates processed (O(N)), and thus it is feasible to apply this approach to much larger numbers of samples.
To assess the statistical significance of ChipStat results, it was necessary to empirically measure the underlying null distribution. While the recent availability of a number of publicly available Latin square datasets representing measurements of spiked-in control samples has greatly facilitated measurements of this sort [31], these datasets reflect technical replicates without biological noise. As we have demonstrated, the behavior of the ChipStat algorithm would be expected to change depending on the relative contributions of biological/experimental noise and probe-level hybridization noise. Thus, a set of negative control samples reflecting an experimental system that include biological noise was required.
To generate these samples, mammary glands from six independent cohorts of mice were harvested. These data provide a true, gold-standard negative control within a representative mammalian experimental system, and we anticipate that their public availability will be similarly useful to the broader scientific community in analytical development and validation. Furthermore, the use of this dataset as an empirical calibration control for ChipStat argues that these results will be valid independent of the adequacy of the statistical noise model used. It is worth noting that the use of pooled groups of animals is likely an important parameter, as single-animals groups, for example, would be expected to exhibit increased biological variability and thus decrease the proportional contribution of hybridization noise.
Given empirical measurements of the expected number of false positives for a given set of analytical parameters, it was possible to assess the relative sensitivity of a variety of algorithms using a positive control dataset (2-week versus 5-week murine mammary gland) known to contain a substantial number of increasing transcripts. Consistent with our hypothesis that probe-level comparison analysis should improve sensitivity, ChipStat was able to substantially outperform a variety of methods (t-test, SAM, and dChip) based on aggregate gene-expression measures (Figures 3c,d). Furthermore, this remained the case even when the statistical significance of dChip was recalibrated using a negative control dataset.
Recently, Lemon et al. have described a method (logit-T) that is also based on probe-level t-test comparisons for identifying differentially expressed genes [10]. The logit-T algorithm estimates statistical significance using the median result of t-tests performed on log-transformed PM probe data. ChipStat differs from this approach in several significant respects. These include the use of a fixed P value threshold for pairwise probe comparisons and the use of the degree of reproducibility across the entire probe set as an indication of statistical significance. Results from the empirical control data suggest that ChipStat performs slightly better than logit-T in most cases within our biological system. Interestingly, however, the advantage of ChipStat over logit-T was more modest than the advantage over SAM, dChip, and the t-test; as logit-T also uses probe-level comparisons, this result is consistent with our overall observations regarding the increased power of probe-based analysis. It is also worth noting that the nominal P values derived from both logit-T and dChip substantially underestimated statistical significance prior to correction with our control data, suggesting that, for example, the median P value cannot be used to directly assess significance without such correction.
One additional difference between ChipStat and logit-T stems from the use of mismatch (MM) probe cells (ChipStat) and log-transformed data (logit-T). As currently implemented, the ChipStat algorithm compares differences in probe pair (PM - MM) values rather than in PM values alone. Interestingly, the use of PM values within the ChipStat algorithm does not result in superior performance (data not shown), and log(PM) data yield performance that is roughly comparable to PM - MM (data not shown). Further work will be required to determine if the log(PM) approach can be adapted to improve the performance of ChipStat.
The Intersector algorithm tabulates MAS calls from all pairwise comparisons across replicate groups. As we have shown, this algorithm provides the most sensitive method for detecting gene expression changes at low false-detection rates. However, it suffers from several substantial drawbacks. First, the proprietary nature of the Affymetrix algorithm and its associated decision matrices limits the ability to automate the analytical process. Additionally, because N2 pairwise comparisons are required for equal groups of N replicates (that is, O(N2)), this method is not easily scalable to larger numbers of samples. In contrast, ChipStat scales linearly with N, and the use of the heteroscedastic t-test also makes it possible to precompute results for a (potentially large) baseline control population against which multiple comparisons will be performed. While both approaches are feasible for triplicate comparisons, extension of Intersector to much larger numbers is unlikely to be practical.
A third disadvantage to the Intersector approach stems from the lack of a detailed model for its underlying statistical framework. Both ChipStat and Intersector, as currently described, require the use of control samples to generate an estimate of statistical significance. Thus, extension of these results to encompass either a substantially different experimental system or larger numbers of replicates will require the generation of new empirical significance curves. In the case of large numbers of replicates, however, the cost of generating such data is likely to remain prohibitive at least in the near future. Statistical simulations of ChipStat behavior may, however, provide a mechanism for extending the current control curves for larger datasets. This approach would rely on the relative estimates of E(b) + E(p) and E(h) obtained by fitting the current empirical curve derived from six samples. While a small number of probe sets change more often than would be predicted by simulation, it should be possible to conservatively estimate an upper bound to the P value curve by overestimating the relative contribution of E(b) + E(p) vs. E(h). Further experimental work will be required to confirm this possibility.
One caveat to this approach is that the simplified statistical model that we have used to illustrate the theoretical advantages of probe-level comparisons does not account for the variability in the behavior of specific probe cells that exist on Affymetrix arrays. In contrast, the model-based approach to signal estimation implemented in dChip explicitly incorporates such variability and has been shown to provide a good fit to empirical measurements of array data [8,32]. It is likely that incorporation of probe-specific parameters in this way would improve the ability to predict the theoretical behavior of ChipStat and provide a better estimate of hybridization noise from our empirical data. Given this likelihood, current estimates of the relative contributions of E(b) + E(p) and E(h) should be taken as provisional.
In this context, it is worth noting that the ability to perform our current validation and to tune both Intersector and ChipStat was critically dependent on the gene-expression datasets derived from our independent cohorts of control animals. One might naively assume, in the absence of true negative control data, that robust changes in gene expression should be detected by simply taking the Affymetrix MAS calls and requiring that they consistently demonstrate increases for all pairwise comparisons. In contrast, our data show that the Intersector algorithm can achieve increased sensitivity while retaining an appropriately low (and defined) false-positive rate. Both the Intersector and ChipStat algorithms can be tuned using negative control data for sensitivity versus false-positive rate, depending on the type of analysis and application-specific tolerance for false-positive calls. Furthermore, these algorithms can be combined to further improve their sensitivity. As we have demonstrated that each of these algorithms detects a population of probe sets not identified by the other at a comparable stringency, this combined approach may yield the best result. Given these considerations, we favor the use of the hybrid ChipStat/Intersector approach for small number of replicates (around three), with ChipStat alone being useful for large numbers of replicates. Although ChipStat shows greater sensitivity than logit-T at moderate numbers of false positives (more than five expected false positives out of 12,488 probe sets), their comparable performance at high stringency (less than five expected false positives) suggests that the overlap in genes identified by these two techniques may also be of interest.
An additional piece of evidence for the utility of our approach is provided by the statistical association of GO annotation with lists derived from ChipStat/Intersector, dChip or logit-T. At the level of significance tested (3.4 genes per list expected by chance), ChipStat/Intersector lists were statistically associated with a greater number of GO terms than were lists derived from dChip or logit-T. Furthermore, as would be predicted from the fact that logit-T is also a probe pair-level comparison method, logit-T lists are associated with GO terms at a level that is intermediate between dChip and ChipStat/Intersector. One of the terms associated only with the ChipStat/Intersector list of downregulated genes is 'mitochondrial inner membrane'; this example is particularly noteworthy in light of previous work demonstrating a presumptive role for enzymes of fatty acid oxidation in adaptive thermogenesis during the neonatal period [4].
It should be noted that these results depend on the level of significance chosen for generation of the original lists, and an increase in the total number of differentially expressed genes identified may actually decrease the statistical significance of a given association if it does not result in the detection of more genes within the category in question. Despite some caveats as to the generalizability of these results, our data demonstrate that the improved sensitivity of ChipStat/Intersector can measurably influence the ability to interpret patterns of biological activity.
Early murine mammary gland development
For the FVB murine mammary gland, the period from 2 to 5 weeks of age encompasses critical developmental milestones that include the suckling-weaning transition as well as the profound hormonal changes that characterize the onset of puberty and its consequent rapid ductal epithelial proliferation. Our present work has more completely characterized changes in the transcriptome that occur during early murine mammary gland development than have previous reports. A total of 213 upregulated and 130 downregulated probe sets were identified under conditions designed to yield a low expected false-positive rate (3.4 probe sets expected to change by chance per list).
Four out of five of the most highly upregulated transcripts through the onset of puberty are members of the MUP family of odorant-binding proteins. MUPs are lipocalins that can bind hydrophobic molecules such as pheromones, and they have previously been shown to play a role both in the delivery of signals within the urine as well as in the reception of these signals on the nasal epithelium ([33,34]; see [35] for review). Isoform-specific MUP expression has also previously been reported in a number of secretory glands, including the murine mammary gland [23,24]. However, detectable expression has previously been reported only beginning with the first pregnancy [23].
Our results demonstrate a striking increase in the expression of a variety of MUP isoforms as the mammary gland makes the transition from the neonatal period through the beginning of puberty. This expression pattern is noteworthy given the known effect of puberty on MUP expression in the liver [36]. Interestingly, however, expression in the liver is markedly greater in the male and has been causally linked to the male pattern of growth-hormone pulses [36]. As this male-specific pattern of expression has been shown to be Stat5b-dependent, the availability of Stat5b -/- mice should allow future determination of whether mammary expression is mediated via a similar signaling pathway. Regardless of this, despite an interval of over two decades since the first description of MUP expression in the mammary gland, a functional role has still not been elucidated. Although it has long been assumed that MUP synthesis occurs in the secretory epithelium of expressing organs, our observation that these molecules are upregulated during puberty (with a corresponding approximately threefold downregulation following puberty, data not shown) suggests that their functional role may not be limited to the secretory function of the gland.
Delta-like kinase (Dlk1) is a member of the epidermal growth factor (EGF) superfamily [37] that is encoded on murine chromosome 12 [38]. Dlk1 is one of several genes showing substantial (greater than fivefold) downregulation from 2 to 5 weeks of murine mammary gland development. As this gene was first identified as a preadipocyte transcript that is downregulated during subsequent differentiation [38], we hypothesize that its relatively high expression during the neonatal period reflects ongoing differentiation of the mammary fat pad. This kinase has also been shown to have a role in other developmental contexts, specifically within neuroendocrine tissues. Further work will be required to elucidate its specific role in the mammary gland. Notable, however, is the corresponding downregulation (more than 10-fold) of Meg3/Gtl2, a noncoding RNA that is reciprocally imprinted with Dlk1 [26]. This Dlk1-Meg3/Gtl2 regulation has been compared with Igf2-H19, another tandem pair of reciprocally imprinted genes in which one member produces a noncoding RNA [27,39]. Interestingly, both Igf2 and H19 are also downregulated during this time period, suggesting the hypothesis that a common regulatory mechanism exists for the tandem control of both imprinted genes at these loci. It will be particularly important to determine whether there is functional significance to this Igf2-H19 regulation, or whether it reflects the epiphenomenal byproduct of a mechanism designed to downregulate Dlk1 during adipocyte development.
Conclusions
We have developed two novel algorithms for the analysis of Affymetrix oligonucleotide microarray data. We have validated these algorithms by using empirically derived distributions from control animals to calibrate their statistical significance. These control data, which reflect both experimental and biological sources of variability likely to be representative of many mammalian experimental systems, should facilitate further work in this area. For triplicate samples, Intersector appears to provide the most sensitivity at a given threshold of statistical significance, and its performance is substantially superior to other widely used methods including the t-test, SAM, dChip, and logit-T. However, its lack of scalability, along with the baseline time required for processing, make it unsuitable for larger numbers of replicates. ChipStat, in contrast, provides comparable sensitivity with triplicate samples and has the capability of handling much larger numbers of replicates in order to improve the reliable dectection of small changes in gene expression. Both algorithms provide a substantial increase in the ability to sensitively detect statistically significant changes in gene expression within the context of the whole mammary gland.
We have applied these techniques to the analysis of genomic patterns during early murine mammary gland development. In addition to detecting patterns reflecting known biology, we have noted the coordinate upregulation of a class of molecules not previously known to be differentially regulated in the mammary gland. We also suggest that peri-pubertal changes in the mammary gland may utilize mechanisms for tandem upregulation of multiple imprinted regions. Our observations suggest a variety of future directions for functional validation and demonstrate the utility of coupling sensitive detection of differential gene expression with pathway analysis for the elucidation of biological patterns during organogenesis.
Materials and methods
Animals, RNA isolation, and northern blot hybridization
The third, fourth and fifth mammary glands were harvested from FVB mice at the indicated time points. Samples from 2 and 5 weeks of age reflect triplicate pools of 10 animals at each time point (total 60 animals). In addition, tissue from 18 control animals was harvested when they were 6 weeks and 4 days old. These control animals also carry a transgenic construct consisting of the murine mammary tumor virus (MMTV) promoter upstream of the reverse tetracycline transactivator (rtTA) and had been given 2.0 mg/ml doxycycline in drinking water for 96 h before harvest. This line (previously designated MTB) has been previously described, and no developmental abnormalities have been noted [40]. All animal experimentation was conducted in accord with accepted standards of humane care, and protocols for animal work were approved by the University of Pennsylvania institutional committee on animal care.
All tissue was snap frozen after removal of the lymph node present in the fourth gland, and total RNA was isolated by homogenization in guanidinium isothiocyanate and subsequent centrifugation through a cesium chloride cushion as previously described [41]. Northern blot hybridization was performed as previously described [42].
Arrays and hybridization
Approximately 15-20 μg total RNA was used for each hybridization. RNA was visualized by gel electrophoresis to ensure its integrity before analysis. Biotinylated cRNA was generated and hybridized to Affymetrix MG_U74Av2 arrays according to the manufacturer's instructions. To scale between chips, these expression values were rank ordered, and the median approximately 96% were averaged. Chips were scaled relative to each other to equalize this average value. All Affymetrix control probe sets were eliminated from analysis, yielding data from a total of 12,422 probe sets. Datasets are publicly available as CEL files designated MTB_ [1-6] (Additional data files 3-8 available with the online version of this paper), 2wk_G0P0_ [1-3] (Additional data files 9-11) and 5wk_G0P0_ [1-3] (Additional data files 12-14) containing results derived from control cohorts, 2-week nulliparous cohorts, and 5-week nulliparous cohorts respectively.
Algorithms and software
To detect differentially regulated genes, we implemented an algorithm (ChipStat) that takes identical probe pairs across two comparison groups and performs a heteroscedastic t-test. The number of probe pairs within a probe set that are significantly different (P <pps where pps is a fixed value) was tabulated. We consider that a greater number of probe pairs changing in a given direction indicates a greater probability that the gene detected by the probe set is differentially expressed. If the bulk of the noise within the array data derives from pre-hybridization experimental factors (that is, E(b) + E(p); see Results section for definition), the expectation is that all probe pairs would change coordinately. That is, if there are 16 probe pairs in the probe set, we would expect (for E(b) + E(p) >> E(h)) that under the null hypothesis (no change in gene expression) either 0/16 or 16/16 probe pairs should change significantly (at frequencies of approximately 1 - pps and approximately pps, respectively). Conversely, if the bulk of the noise derives from hybridization to individual probe cells (that is, if E(b) + E(p) << E(h)), then the number of probe pairs r that change within a given probe set of size t can be approximated by the binomial distribution:
However, under experimentally realistic conditions, neither of these limiting cases is likely to apply. Therefore, to empirically determine the null distribution using six independent, biologically identical control populations, all pairwise three by three combinations were compared and the number of probe pairs changing was tabulated. To determine the expected number of changes per probe set when fewer than 16 probe pairs are available, these analyses were repeated after randomly discarding 1, 2...15 probe pairs. In this way, a similar statistical estimate was obtained for the 602 probe sets on the MG_U74Av2 array that have fewer than 16 probe pairs per probe set. A conservative simplification of these data was performed by rounding up the significance of changes in these 602 probe sets to the nearest appropriate bin in the 16 probe pair per probe set curve. A Microsoft Windows-compatible application implementing the ChipStat algorithm is freely available for academic use [43].
On the basis of the simplified statistical model described, a Monte Carlo simulation was implemented to determine the number of expected false-positive values as a function of pps for various relative proportions of E(b) + E(p) and E(h). Briefly, a random test dataset was generated in which equal gene expression was perturbed by Gaussian noise (representing E(b) + E(p)). Each expression value was then independently perturbed 16 times (representing 16 probe pairs/probe set) by another Gaussian noise function (representing E(h)), and comparisons were tabulated using the ChipStat algorithm. This simulation was implemented in C and the source code is available [43]. All values reported reflect the mean of 100 trials, where each trial simulates 11,820 probe sets with 16 probe pairs each. The relative contributions of E(b) + E(p) and E(h) were estimated by maximizing the likelihood function:
with respect to (E(b) + E(p)) / E(h) where xi is the number of times i probe pairs increased significantly and μi and σi represent the mean and standard deviation from the Monte Carlo simulations.
A separate algorithm (Intersector) uses pairwise calls of differential gene expression derived from Affymetrix Microarray Suite (MAS) analysis. All pairwise comparisons were performed (that is, 3 × 3 = 9 comparisons for a 3- vs 3-replicate comparison) using the manufacturer's default settings, and the number of 'increases' or 'marginal increases' was tabulated. Similarly to the ChipStat method described above, the null distribution was generated by tabulating results from all 20 distinguishable 3 vs 3 combinations of the six control samples. Results were obtained using both MAS version 4 (MAS4) and MAS version 5 (MAS5), as indicated in the text and figures.
Tests for differential gene expression using a homoscedastic t-test or SAM [11] were performed using signal values derived from MAS5. SAM results were obtained using software obtained from its authors [44]. Because the analyses described are reported as a function of the number of genes expected to increase by chance (essentially a one-tailed test of significance), the false-discovery rate reported by SAM was multiplied by 0.5 to derive a corrected false-positive rate (false-increase rate). dChip analysis [8] was performed using software available from its authors [45], and a PM-only expression model was constructed. Logit-T analysis [10] was performed using software provided by its authors and compiled to run locally on an AMD Linux server. Both dChip and Logit-T significance values were empirically calibrated by analyzing all possible 3 vs 3 combinations of control arrays (20 total) and tabulating the average number of false positives as a function of the reported significance.
Association with biological annotation
Associations between GO [19] annotation and lists of differentially expressed genes were identified using EASE [20]. Multiple testing correction was performed using within-system bootstrapping, and a final cutoff of P < 0.05 was used to identify statistically significant associations.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 contains a table showing ChipStat and Intersector in combination. For each level of stringency available, the pairwise intersection of ChipStat (CS, pps = 0.05) and Intersector (IT, MAS5) lists of significantly increasing probe sets was generated. Rows indicate the threshold number of probe pairs (0-16) significantly increasing from ChipStat, and columns indicate the threshold number of Increase or Marginal Increase calls (0-9) identified by Intersector. (a) Number of increasing probe sets in 2- vs 5-week murine mammary gland. Selected results correspond to values plotted on the y axis of Figure 4d (number of probe sets increasing). (b) Average number of increasing probe sets using all 3 × 3 combinations of 6 negative control samples. Selected results correspond to values plotted on the x axis of Figure 4d (expected number of probe sets increasing by chance). Additional data file 2 contains a table showing differential gene expression in 2- vs 5-week murine mammary gland using a hybrid ChipStat/Intersector approach. The criteria ChipStat pps = 0.05, 6/16 probe pairs increasing and Intersector 7/9 increases + marginal increases, were used to identify lists of probe sets that are up- and downregulated from 2 to 5 weeks of FVB female murine mammary gland development. Additional data files 3,4,5,6,7 and 8 contain six control files containing CEL file data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment. Additional data files 9,10 and 11 contain three CEL files of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age. Additional data files 12,13 and 14 contain three CEL files of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age.
Supplementary Material
Additional data file 1
A table showing ChipStat and Intersector in combination
Click here for additional data file
Additional data file 2
A table showing differential gene expression in 2- vs 5-week murine mammary gland using a hybrid ChipStat/Intersector approach
Click here for additional data file
Additional data file 3
Control file 1 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 4
Control file 2 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 5
Control file 3 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 6
Control file 4 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 7
Control file 5 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 8
Control file 6 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 9
CEL file 1 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age
Click here for additional data file
Additional data file 10
CEL file 2 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age
Click here for additional data file
Additional data file 11
CEL file 3 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age
Click here for additional data file
Additional data file 12
CEL file 1 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age
Click here for additional data file
Additional data file 13
CEL file 2 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age
Click here for additional data file
Additional data file 14
CEL file 3 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age
Click here for additional data file
Acknowledgements
This research was supported in part by NIH Grants CA94393, CA92910, and CA93719 from the National Cancer Institute, NRSA HL007150-27 (L.C.B.), and by US Army Breast Cancer Research Program Grant DAMD17-01-1-0364.
Figures and Tables
Figure 1 ChipStat behavior using simulated biological/experimental + hybridization noise model. The behavior of the ChipStat algorithm was evaluated (pps = 0.05, 16 probe pairs per probe set) using a Monte Carlo model in which the ratio of biological + experimental noise (E(b) + E(p)) to hybridization noise (E(h)) is constant (see text for further details). Results are shown for E(h) = 0 (Exp noise only; blue), E(h) = E(b) + E(p) (Hyb noise = Exp noise; red), E(h) = 2 × (E(b) + E(p)) (Hyb noise = 2 × Exp noise; green), and E(b) + E(p) = 0 (Hyb noise only; yellow). The total number of probe sets simulated (11,820) was chosen to match the number of probe sets containing 16 probe pairs per probe set on the Affymetrix MG_U74Av2 array. The number of probe pairs increasing by chance is shown on the x axis, and the fraction of total probe sets simulated is shown on the y axis. This simulation was repeated 100×, and the average of these results is shown. (a) Probability of the indicated number of probe pairs increasing. (b) Cumulative P value (equal to or greater than the indicated number of probe pairs changing).
Figure 2 Empirical measurement of the ChipStat null distribution. Mammary gland tissue was harvested from six separate, biologically identical pools of FVB (MTB) mice, and hybridization data to Affymetrix MG_U74Av2 microarrays was obtained. Comparisons of all possible three versus three combinations (total 20) were performed using ChipStat (pps = 0.05), and the number of significant increases was tabulated for all probe sets containing 16 probe pairs per probe set (total = 11,820). The cumulative average probability is shown as a function of the number of probe pairs that increase within the probe set.
Figure 3 Relative detection sensitivity of differential gene expression. The number of probe sets shown to increase from 2 to 5 weeks of murine mammary gland development was tabulated as a function of the number of probe sets expected to increase by chance. (a) ChipStat (pps = 0.05), vs t-test. (b) Optimization of ChipStat sensitivity as a function of pps. (c) ChipStat vs other techniques: reported P values. For ChipStat, the number of probe sets expected to increase by chance was empirically estimated from negative control data. For the t-test, SAM, dChip and logit-T, reported P values from the 2-week vs 5-week mammary gland comparison were used. (d) ChipStat vs other techniques: empirical P values. The number of probe sets expected to increase by chance was empirically estimated for ChipStat, t-test, SAM, dChip and logit-T (representative points).
Figure 4 Intersector and ChipStat performance. (a) The number of probe sets shown to increase from 2 to 5 weeks of murine mammary gland development was tabulated as a function of the number of probe sets expected to increase by chance, and a comparison of ChipStat (pps = 0.05), Intersector (MAS 5.0 change calls), and Intersector (MAS 4.0 change calls) is shown. (b) Venn diagram showing distinct probe sets identified by ChipStat and Intersector. The number of genes shown to be differentially expressed at the indicated expected false-positive levels is shown for ChipStat (CS) (pps = 0.04), Intersector (IT) with MAS 5.0 calls, and Intersector (IT) with MAS 4.0 calls. (c) False-positive rates for ChipStat (CS 6/16: pps = 0.05, 6/16 probe pairs increasing; CS 9/16: pps = 0.05, 9/16 probe pairs increasing), Intersector (MAS5) (IT 7/9: 7/9 increases or marginal increases; IT 8/9: 8/9 increases or marginal increases), or ChipStat and Intersector together (Combined: intersection of CS 6/16 and IT 7/9) are shown. (d) Combined performance of ChipStat and Intersector. Increases from 2 to 5 weeks of mammary gland development are shown for ChipStat alone (pps = 0.05), Intersector alone (MAS 5.0), and optimized intersections of ChipStat and Intersector (see Additional data file 1).
Figure 5 Quantitative association with GO categories. The number of GO terms found to be statistically associated (P < 0.05 using within-system bootstrap to account for multiple testing) with lists of differentially regulated genes (2 vs 5 weeks of murine mammary gland development) is shown. Lists of up- and downregulated genes were generated using dChip (DC), logit-T (LT) and a ChipStat/Intersector hybrid (CS/IT) that were matched in stringency to give equivalent numbers of expected false-positive genes.
Table 1 Northern blot validation of differential gene expression
Probe set ID Accession number Gene Fold change Probe pairs increasing Differential expression confirmed
99067_at X59846 Gas6 3.41 16/16 x
100064_f_at M63801 Gja1 1.67 12/16 x
102016_at M61737 Fsp27 2.07 11/16 x
93996_at X01026 Cyp2e1 11.6 10/16 x
97507_at X67809 Ppicap 2.85 9/16 x
101995_at U40930 Sqstm1 1.48 8/16 x
93096_at AA986050 3010002H13Rik 2.65 7/16 x
102791_at U22033 Psmb8 1.65 7/16 x
96072_at M17516 Ldh1 1.37 6/16
Genes identified as being differentially expressed were randomly chosen for verification by northern blot hybridization (see text for description). Gene identifiers are shown along with fold changes, numbers of probe pairs increasing (as identified by ChipStat with pps = 0.04), and confirmation of differential expression.
Table 2 Association with GO annotation
System Gene category CS LT DC
(a) Upregulated genes
GO Biological Process Defense response x x x
GO Cellular Component Extracellular space x x
GO Cellular Component Extracellular x x
GO Biological Process Response to biotic stimulus x x x
GO Biological Process Immune response x x x
GO Biological Process Response to external stimulus x x x
GO Biological Process Organismal physiological process x x x
GO Biological Process Antigen presentation x x
GO Biological Process Response to stimulus x x x
GO Biological Process Antigen presentation\, endogenous antigen x
GO Molecular Function MHC class I receptor activity x
GO Biological Process Antigen processing x x
GO Biological Process Complement activation x x
GO Biological Process Antigen processing, endogenous antigen via MHC class I x
GO Biological Process Response to pest/pathogen/parasite x x x
GO Biological Process Humoral defense mechanism (sensu Vertebrata) x
GO Molecular Function Pheromone binding x x
GO Molecular Function Oxidoreductase activity x
GO Molecular Function Oxidoreductase activity, acting on the aldehyde or oxo group of donors x
GO Molecular Function Odorant binding x x
GO Molecular Function Transmembrane receptor activity x
GO Biological Process Humoral immune response x
(b) Downregulated genes
GO Cellular Component Mitochondrion x x
GO Biological Process Main pathways of carbohydrate metabolism x x
GO Biological Process Tricarboxylic acid cycle x x
GO Biological Process Energy derivation by oxidation of organic compounds x x
GO Biological Process Energy pathways x x
GO Cellular Component Mitochondrial membrane x
GO Biological Process Carbohydrate metabolism x x
GO Cellular Component Inner membrane x
GO Biological Process Blood vessel development x
GO Cellular Component Mitochondrial inner membrane x
GO Cellular Component Extracellular x
Lists of differentially expressed genes derived from a hybrid ChipStat/Intersector approach (ChipStat: pps = 0.05, 6/16 probe pairs increasing AND Intersector: 7/9 increases + marginal increases), logit-T, and dChip were associated with GO terms using EASE [20]. Individual terms are annotated according to whether association with the given annotation group was statistically significant (P < 0.05 using within-system bootstrap to account for multiple testing) using lists derived from ChipStat/Intersector (CS), logit-T (LT), or dChip (DC). (a) Association with lists of upregulated genes. (b) Association with lists of downregulated genes.
==== Refs
Coller HA Grandori C Tamayo P Colbert T Lander ES Eisenman RN Golub TR Expression analysis with oligonucleotide microarrays reveals that MYC regulates genes involved in growth, cell cycle, signaling, and adhesion. Proc Natl Acad Sci USA 2000 97 3260 3265 10737792 10.1073/pnas.97.7.3260
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JC Sabet H Tran T Yu X Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000 403 503 511 10676951 10.1038/35000501
van't Veer LJ Dai H van de Vijver MJ He YD Hart AA Mao M Peterse HL van der Kooy K Marton MJ Witteveen AT Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002 415 530 536 11823860 10.1038/415530a
Master SR Hartman JL D'Cruz CM Moody SE Keiper EA Ha SI Cox JD Belka GK Chodosh LA Functional microarray analysis of mammary organogenesis reveals a developmental role in adaptive thermogenesis. Mol Endocrinol 2002 16 1185 1203 12040007 10.1210/me.16.6.1185
Schena M Shalon D Davis RW Brown PO Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 1995 270 467 470 7569999
Lockhart DJ Dong H Byrne MC Follettie MT Gallo MV Chee MS Mittmann M Wang C Kobayashi M Horton H Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 1996 14 1675 1680 9634850 10.1038/nbt1296-1675
Lipshutz RJ Fodor SP Gingeras TR Lockhart DJ High density synthetic oligonucleotide arrays. Nat Genet 1999 21 1 Suppl 20 24 9915496 10.1038/4447
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
Irizarry RA Hobbs B Collin F Beazer-Barclay YD Antonellis KJ Scherf U Speed TP Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003 4 249 264 12925520 10.1093/biostatistics/4.2.249
Lemon WJ Liyanarachchi S You M A high performance test of differential gene expression for oligonucleotide arrays. Genome Biol 2003 4 R67 14519202 10.1186/gb-2003-4-10-r67
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001 98 5116 5121 11309499 10.1073/pnas.091062498
Williams JM Daniel CW Mammary ductal elongation: differentiation of myoepithelium and basal lamina during branching morphogenesis. Dev Biol 1983 97 274 290 6852366 10.1016/0012-1606(83)90086-6
Daniel CW Silberstein GB Neville MC, Daniel CW Postnatal development of the rodent mammary gland. The Mammary Gland: Development, Regulation, and Function 1987 New York: Plenum Press 3 36
Dao TL Mammary cancer induction by 7,12-dimethylbenz[a]anthracene: Relation to age. Science 1969 165 810 811 5796556
Ip C Mammary tumorigenesis and chemoprevention studies in carcinogen-treated rats. J Mammary Gland Biol Neoplasia 1996 1 37 47 10887479
Rogge L Bianchi E Biffi M Bono E Chang SY Alexander H Santini C Ferrari G Sinigaglia L Seiler M Transcript imaging of the development of human T helper cells using oligonucleotide arrays. Nat Genet 2000 25 96 101 10802665 10.1038/75671
Richert MM Schwertfeger KL Ryder JW Anderson SM An atlas of mouse mammary gland development. J Mammary Gland Biol Neoplasia 2000 5 227 241 11149575 10.1023/A:1026499523505
Rajagopalan D A comparison of statistical methods for analysis of high density oligonucleotide array data. Bioinformatics 2003 19 1469 1476 12912826 10.1093/bioinformatics/btg202
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000 25 25 29 10802651 10.1038/75556
Hosack DA Dennis G JrSherman BT Lane HC Lempicki RA Identifying biological themes within lists of genes with EASE. Genome Biol 2003 4 R70 14519205 10.1186/gb-2003-4-10-r70
Sharrow SD Vaughn JL Zidek L Novotny MV Stone MJ Pheromone binding by polymorphic mouse major urinary proteins. Protein Sci 2002 11 2247 2256 12192080 10.1110/ps.0204202
Mucignat-Caretta C Modulation of exploratory behavior in female mice by protein-borne male urinary molecules. J Chem Ecol 2002 28 1853 1863 12449511 10.1023/A:1020521420271
Shaw PH Held WA Hastie ND The gene family for major urinary proteins: expression in several secretory tissues of the mouse. Cell 1983 32 755 761 6831559 10.1016/0092-8674(83)90061-2
Shahan K Denaro M Gilmartin M Shi Y Derman E Expression of six mouse major urinary protein genes in the mammary, parotid, sublingual, submaxillary, and lachrymal glands and in the liver. Mol Cell Biol 1987 7 1947 1954 3600653
Tullis KM Krebs CJ Leung JY Robins DM The regulator of sex-limitation gene, rsl, enforces male-specific liver gene expression by negative regulation. Endocrinology 2003 144 1854 1860 12697692 10.1210/en.2002-0190
Schmidt JV Matteson PG Jones BK Guan XJ Tilghman SM The Dlk1 and Gtl2 genes are linked and reciprocally imprinted. Genes Dev 2000 14 1997 2002 10950864
Takada S Paulsen M Tevendale M Tsai CE Kelsey G Cattanach BM Ferguson-Smith AC Epigenetic analysis of the Dlk1-Gtl2 imprinted domain on mouse chromosome 12: implications for imprinting control from comparison with Igf2-H19. Hum Mol Genet 2002 11 77 86 11773001 10.1093/hmg/11.1.77
Smith GH Mammary cancer and epithelial stem cells: a problem or a solution? Breast Cancer Res 2002 4 47 50 11879561 10.1186/bcr420
Welm BE Tepera SB Venezia T Graubert TA Rosen JM Goodell MA Sca-1(pos) cells in the mouse mammary gland represent an enriched progenitor cell population. Dev Biol 2002 245 42 56 11969254 10.1006/dbio.2002.0625
Zhang L Wang L Ravindranathan A Miles MF A new algorithm for analysis of oligonucleotide arrays: application to expression profiling in mouse brain regions. J Mol Biol 2002 317 225 235 11902839 10.1006/jmbi.2001.5350
Liu WM Mei R Di X Ryder TB Hubbell E Dee S Webster TA Harrington CA Ho MH Baid J Analysis of high density expression microarrays with signed-rank call algorithms. Bioinformatics 2002 18 1593 1599 12490443 10.1093/bioinformatics/18.12.1593
Li C Wong WH Model-based analysis of oligonucleotide arrays: model validation, design issues and standard error application. Genome Biol 2001 2 research0032.1 0032.11 11532216 10.1186/gb-2001-2-8-research0032
Utsumi M Ohno K Kawasaki Y Tamura M Kubo T Tohyama M Expression of major urinary protein genes in the nasal glands associated with general olfaction. J Neurobiol 1999 39 227 236 10235677 10.1002/(SICI)1097-4695(199905)39:2<227::AID-NEU7>3.0.CO;2-4
Mucignat-Caretta C Caretta A Cavaggioni A Acceleration of puberty onset in female mice by male urinary proteins. J Physiol 1995 486 517 522 7473215
Cavaggioni A Mucignat C Tirindelli R Pheromone signalling in the mouse: role of urinary proteins and vomeronasal organ. Arch Ital Biol 1999 137 193 200 10349497
Norstedt G Palmiter R Secretory rhythm of growth hormone regulates sexual differentiation of mouse liver. Cell 1984 36 805 812 6323022 10.1016/0092-8674(84)90030-8
Laborda J Sausville EA Hoffman T Notario V dlk, a putative mammalian homeotic gene differentially expressed in small cell lung carcinoma and neuroendocrine tumor cell line. J Biol Chem 1993 268 3817 3820 8095043
Smas CM Sul HS Pref-1, a protein containing EGF-like repeats, inhibits adipocyte differentiation. Cell 1993 73 725 734 8500166 10.1016/0092-8674(93)90252-L
Wylie AA Murphy SK Orton TC Jirtle RL Novel imprinted DLK1/GTL2 domain on human chromosome 14 contains motifs that mimic those implicated in IGF2/H19 regulation. Genome Res 2000 10 1711 1718 11076856 10.1101/gr.161600
Gunther EJ Belka GK Wertheim GB Wang J Hartman JL Boxer RB Chodosh LA A novel doxycycline-inducible system for the transgenic analysis of mammary gland biology. FASEB J 2002 16 283 292 11874978 10.1096/fj.01-0551com
Marquis ST Rajan JV Wynshaw-Boris A Xu J Yin GY Abel KJ Weber BL Chodosh LA The developmental pattern of Brca1 expression implies a role in differentiation of the breast and other tissues. Nat Genet 1995 11 17 26 7550308 10.1038/ng0995-17
D'Cruz CM Moody SE Master SR Hartman JL Keiper EA Imielinski MB Cox JD Wang JY Ha SI Keister BA Persistent parity-induced changes in growth factors, TGF-beta3, and differentiation in the rodent mammary gland. Mol Endocrinol 2002 16 2034 2051 12198241 10.1210/me.2002-0073
ChipStat software
Significance Analysis of Microarrays
DNA-Chip Analyzer (dChip)
| 15693949 | PMC551540 | CC BY | 2021-01-04 16:05:36 | no | Genome Biol. 2005 Feb 1; 6(2):R20 | utf-8 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r20 | oa_comm |
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-2-r201569394910.1186/gb-2005-6-2-r20MethodGenomic analysis of early murine mammary gland development using novel probe-level algorithms Master Stephen R [email protected] Alexander J [email protected] L Charles [email protected] Tien-Chi [email protected] Katherine D [email protected] Lewis A [email protected] Department of Cancer Biology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA2 Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA3 Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA4 Abramson Family Cancer Research Institute, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA5 Department of Pediatrics, Children's Hospital of Philadelphia, Philadelphia, PA 19104, USA2005 1 2 2005 6 2 R20 R20 25 8 2004 1 10 2004 8 12 2004 Copyright © 2005 Master 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.
A novel algorithm (ChipStat) is presented for detecting gene-expression changes from Affymetrix microarray data. The method is used to identify changes in murine mammary development.
We describe a novel algorithm (ChipStat) for detecting gene-expression changes utilizing probe-level comparisons of replicate Affymetrix oligonucleotide microarray data. A combined detection approach is shown to yield greater sensitivity than a number of widely used methodologies including SAM, dChip and logit-T. Using this approach, we identify alterations in functional pathways during murine neonatal-pubertal mammary development that include the coordinate upregulation of major urinary proteins and the downregulation of loci exhibiting reciprocal imprinting.
==== Body
Background
The widespread use of DNA microarrays to measure transcript abundance from a significant fraction of the genome has proven to be a valuable tool for identifying functional cellular pathways as well as for capturing the global state of a biological system [1-4]. These arrays have typically been constructed by spotting large, pre-synthesized strands of nucleic acid on an appropriate surface [5] or by directly synthesizing smaller oligonucleotides in situ at defined locations [6]. The latter technique has been implemented in Affymetrix oligonucleotide microarrays designed for expression analysis. Because hybridization to short (25-mer) oligonucleotides is used to measure expression, Affymetrix arrays contain multiple, independent oligonucleotides designed to bind a unique transcript. In this way, specificity and a high signal-to-noise ratio can be maintained despite the noise due to the hybridization itself. When the intensity of hybridization to a given oligonucleotide designed to detect the transcript (a 'perfect match' probe, PM) is corrected by its corresponding (single base-pair 'mismatch', MM) control, an estimate of gene expression (PM - MM) is derived. This probe pair value is then combined with values from the other, independent, oligonucleotides designed to bind the same transcript (together designated the probe set) to obtain a more robust estimate of transcript abundance [7].
The ability to sensitively detect changes in gene expression is crucial for a transcript-level analysis of developmental processes and other processes involving changes in the relative sizes of cellular compartments. Early attempts to limit the false-positive rate of microarray studies focused on the magnitude of fold-change in gene expression (see, for example [1]). For studying purified cell populations, where a substantial change in gene expression is more likely to reflect biologically relevant function, such a crude limitation was acceptable. However, adequate studies of complex tissues require a substantially more sensitive method of detection. For example, a small yet reproducible change in gene expression within a whole organ may reflect a substantial expansion or regulatory change within a subpopulation of cells that overexpress a given gene relative to the surrounding tissue. Thus, a method for identifying such small, statistically significant changes in gene expression is required.
Because of the variety of techniques used to measure gene expression, it has become commonplace to utilize simple, numerical estimates of gene expression as the starting point for such identification. One major drawback to this approach has been that individual probe cell information from Affymetrix microarrays is routinely discarded. This issue has only recently begun to be addressed [8-10], and it appears that a substantial amount of useful information can be obtained from probe-level analysis.
An additional compromise has been driven by the practical difficulties of performing large numbers of microarray experiments. Given limited samples, permutation of the existing experimental dataset, rather than use of independent sets of control samples, has been widely used to estimate the statistical significance of differential gene expression [11]. Although this technique has been useful given the historically high cost of performing microarray analysis, it may inherently limit the sensitivity of the results obtained. As such, a test for differential gene expression that utilizes a 'gold standard' negative-control dataset would have clear advantages.
The impetus for the work described here is the desire to sensitively identify coherent patterns of gene expression during mammary gland development. At 2 weeks of age, the female FVB mouse mammary gland exists as a rudimentary epithelial tree embedded at one end of a fat pad composed of adipose tissue and fibroblasts. Previous work has demonstrated a fundamental transition in the composition of the mammary adipose compartment from brown fat to white fat during early development [4]. By 3 weeks of age, the onset of puberty heralds the beginning of the process of ductal morphogenesis, which results in the formation of the branching epithelial tree of the adult gland. The onset of puberty results not only in the rapid growth of a ductal epithelial tree but also the appearance of specialized, highly proliferative structures known as terminal end buds that elaborate this tree via branching morphogenesis [12,13]. Furthermore, puberty is known to be a time of increased susceptibility to carcinogenesis [14,15]. Thus, a detailed examination of transcriptional changes during this period would be of substantial use.
We describe here a novel algorithm for sensitively detecting gene-expression changes using information derived from individual probe cell hybridizations to Affymetrix oligonucleotide microarrays. In addition to modeling the predicted behavior of this algorithm, we have generated an independent cohort of control samples derived from the murine mammary gland that can be used to empirically calibrate its statistical behavior. We have then used this algorithm to analyze a biological transition in early murine mammary gland development in order to compare the sensitivity of this approach to other commonly used algorithms. In conjunction with a second novel algorithm, we have developed an aggregate approach to the reliable detection of differential gene expression that yields substantially improved sensitivity across a range of false-positive rates and have applied this approach to the analysis of early murine mammary gland development.
Results
A variety of traditional statistical methods, such as the t test, have been used in conjunction with microarray datasets to detect changes in gene expression (see for example [16]). Given the large numbers of genes tested, it is widely recognized that a stringent threshold for statistical significance is necessary in order to reduce the number of false positive changes. For example, a threshold of statistical significance of P < 0.001 would be expected to yield around 100 false positives on a typical array measuring 10,000 genes. Some algorithms, such as significance analysis for microarrays (SAM) [11], explicitly control the number of expected false-positive results using permutations of the existing dataset. Regardless of the method utilized, statistical differences are typically calculated on the basis of an aggregate measure of gene expression (a gene signal). However, a fundamental difficulty with these methods is that they often do not have the requisite statistical power to sensitively detect changes in gene expression after correction for multiple hypothesis testing. We reasoned that utilizing the multiple hybridizations to independent oligonucleotides on the Affymetrix platform might allow us to develop a method for detecting expression changes with substantially greater statistical power.
To test this approach, we developed a novel analytical algorithm that is based on identifying individual differences at a given statistical significance between corresponding probe pairs. To a first approximation, the signal on any given probe cell can be modeled as:
S = M + E(b) + E(p) + E(h), E ~ N
Where S is the signal detected on the microarray, M is the average message level in a given experimental state, E(b) is noise due to biological variation between animals or animal pools, E(p) is the noise due to variations in sample measurement, and E(h) is the noise inherent in hybridization to oligonucleotide features on the array. The goal of our analysis was to identify a method that would allow us to reliably distinguish significant differences in M under particular experimental conditions.
Given this model, we reasoned that the relative magnitude of E(b) + E(p) (the experimental noise) compared with E(h) (the hybridization noise) should determine whether comparisons between individual probe pairs would be useful. If the bulk of noise in our microarray data was due to factors influencing the level of transcript available for measurement (that is, E(b) + E(p) >> E(h)), then individual probe-pair measurements should only reflect the pre-hybridization bias in transcript availability. In this case, the t-test or other measurement based on the average of the probe set would be expected to perform as well as an algorithm based on individual probe-pair comparisons. In contrast, if most noise in the measurement of true transcript level exists at the level of hybridization to a given oligonuclotide (E(b) + E(p) << E(h)), then the independent measurements of probe-pair differences more closely approximate independent measurements of differences in gene expression. In the most extreme case - if E(h) is sufficiently larger than E(b) + E(p) - each oligonucleotide in the probe set could be considered as an independent measurement of gene expression and the probability of observing a given number of probe pairs changing under the null hypothesis would be determined by the binomial distribution.
To explore this possibility, we implemented an algorithm, hereafter designated ChipStat, that takes corresponding probe pairs across two comparison groups and tests them for statistical significance with P less than a fixed value (hereafter denoted pps). To avoid making assumptions about equal variance in both groups, a heteroscedastic t-test is used. We would expect that probe sets in which larger numbers of individual probe pairs show a significant change in the same direction are more likely to be measuring differentially regulated genes. Thus, for any given probe set, the number of probe pairs (0-16) changing in a given direction with P less than pps is tabulated and used as a measure of the significance of change in gene expression. We simulated the expected behavior of this algorithm under the null hypothesis (no difference in gene expression) across various ratios of E(b) + E(p) and E(h) (see Materials and methods for details). Results are shown in Figure 1.
Validation and optimization of the ChipStat algorithm
Although this approach provides a statistical methodology for identifying changes in gene expression, it is only possible to directly calculate a P value associated with this change in limiting cases. If E(h) >> E(b) + E(p), the binomial distribution can be used to calculate the resulting significance (given the number of changes, total number of probe pairs, and pps); however, the relative contributions of E(h), E(b), and E(p) to the total error function are not known a priori.
To empirically measure the null distribution for three-sample versus three-sample comparisons, a cohort of independent control samples for our experimental system was generated. To do this, the third, fourth and fifth mammary glands were harvested from 18 age-matched 5-week-old control female mice. After extraction of RNA, groups of three animals were pooled to create six initial RNA samples. Biotinylated cRNA was then independently prepared from these pooled RNA samples and hybridized to Affymetrix MG_U74Av2 oligonucleotide microarrays, yielding six datasets. All possible three by three combinations were compared across 11,820 probe sets (corresponding to all probe sets on the MG_U74Av2 that contain exactly 16 probe pairs), and the cumulative distribution of false positives as a function of pps and the number of probe pairs changed was tabulated. Results are shown for pps = 0.05 (Figure 2). It is notable that very few false positives are associated with large numbers (more than 10/16) of probe pairs changing. While the number of false-positive probe sets does not decline as rapidly as the binomial distribution, the overall curve is consistent with a large component of hybridization noise (compare Figures 1 and 2), suggesting the utility of a probe-level approach. Likelihood maximization of our initial statistical model (E ~ N, ignoring probe-specific effects) using results for low numbers of probe pairs (0 to 6) changing suggests that E(h) (hybridization noise) is approximately 2.5 times greater than E(b) + E(p) (experimental noise). We note, however, that the empirically derived null distribution can be used to derive a valid test of significance for ChipStat regardless of the validity of the underlying model and without any direct calculation of relative noise contributions by E(h), E(b) and E(p).
An ideal method for identifying differentially regulated genes would maximize the number of genes identified while maintaining a low fixed number of expected false positives. We have previously shown the utility of testing the statistical overlap of discrete gene lists with biologically relevant annotation in order to identify functional pathways during murine mammary gland development [4]. This maximization is therefore of particular experimental interest. To evaluate the ChipStat algorithm from this perspective, we performed triplicate microarray measurements of RNA derived from the mammary glands of independent pools (more than 10 animals per pool) of wild-type female FVB mice harvested at 2 or 5 weeks of postnatal development. We wished to determine the number of statistically significant increases in gene expression from 2 to 5 weeks of age, a period of postnatal development that encompasses the rapid epithelial proliferation that accompanies ductal morphogenesis in the mammary gland at the onset of puberty [17].
ChipStat was used to analyze differences between the 2- and 5-week mammary gland samples (pps = 0.05), and the number of statistically significant increases was measured as a function of the number of genes expected to appear on the list by chance. Results are shown in Figure 3a. The number of expected false positives was empirically obtained from the negative-control dataset described previously. Thus, for example, under conditions pps = 0.05 with 8/16 probe pairs increasing, where around five genes are expected to be identified by chance, we find that the measured number of differentially regulated genes is around 160. This corresponds to a false-positive rate of approximately 3% (or, conversely, a true-positive rate of approximately 97%). It is also apparent (Figure 3a) that the sensitivity of detection can be 'tuned' on the basis of the number of false positives that are deemed acceptable.
To determine whether the sensitivity of this algorithm could be further optimized, similar analyses were performed at various values of pps (Figure 3b). These data suggest that relative sensitivity as a function of false-positive rate is maximized at pps approximately equal to 0.04-0.05 (note the similarity of these curves in Figure 3b). Furthermore, while certain other values of pps yield increased sensitivity at specific points (for example, pps = 0.03 at around four genes expected by chance; data not shown), values of 0.04-0.05 appear appropriate across most highly-significant P values. A marked decrease in sensitivity for a given false-positive rate is noted both at low (0.01) and high (0.1, 0.15) values of pps.
Although the use of negative-control samples provides a definitive method for evaluating the behavior of our statistical algorithms, we independently verified these results using northern blot hybridization. Genes differentially expressed (6/16 probe pairs increasing, pps = 0.04) from 2 to 5 weeks of mammary gland development were identified, and analysis of the control data suggested that fewer than 10 increases would be expected by chance at this significance level (corresponding to P < 7.7 × 10-4). Manual inspection of the resulting list revealed the presence of a number of genes known to be upregulated during this developmental transition, including cytokeratin 19 (Krt1-19), cytokeratin 8 (Krt2-8), and κ casein (Csnk). However, to avoid bias toward previously studied genes or known genes with high fold change, genes were randomly selected from subsets of this list corresponding to high-stringency (P < 2.2 × 10-4), low-stringency with high fold change (2.2 × 10-4 <P < 7.7 × 10-4, ≥ 1.8-fold change), and low-stringency with low fold change (2.2 × 10-4 <P < 7.7 × 10-4, < 1.8-fold change). Results from northern blot analyses using probes for these randomly selected genes are shown in Table 1. Of nine genes selected, eight were shown to change significantly via northern blot analysis.
Of note, the single gene that did not show a significant change (Ldh1) was from the low-stringency group and was predicted to show only a 1.37-fold change. In contrast, northern hybridization confirmed the differential expression of other genes with only modest fold-changes (for example, Sqstm1, 1.48-fold change from 2 to 5 weeks). As the genes tested were not biased toward higher fold change (only 2/75 genes with fold change > 3 were randomly selected for northern confirmation), our data demonstrate the ability of ChipStat to reliably detect the types of small, reproducible changes in gene expression that are necessary for whole-organ analysis.
Comparison of ChipStat with other analytical methods
Other methods of detecting differential gene expression have been widely utilized, including SAM [11] and dChip [8]. As previously discussed, SAM utilizes an aggregate (probe-set-level) estimate of gene expression as its analytical starting point. Similarly, although dChip utilizes probe-cell-level analysis to determine the level and statistical bounds of gene expression, it does not explicitly make use of probe-level comparisons for identifying differentially regulated genes. More recently, the logit-T algorithm, which in contrast to SAM and dChip utilizes probe-pair-level comparisons for statistical testing, has been shown to improve differential expression testing performance in a variety of Latin square datasets reflecting technical replicates of samples with spiked-in transcripts [10]. We therefore wished to determine the performance of the ChipStat algorithm relative to these methodologies. Further, as our control dataset incorporates biological and experimental variability in addition to sample preparation and hybridization noise, we reasoned that it would provide a more appropriate estimate of the performance of these algorithms when analyzing data from an experimentally plausible animal model.
SAM, dChip, the t-test and logit-T all provide a P value estimating statistical significance in the absence of an empirical measurement of the underlying null distribution; Figure 3c shows a comparison with ChipStat when using these estimated P values. However, as ChipStat requires the additional information provided by this empirical distribution for statistical calibration, the inherent performance of other algorithms may be underestimated if they are not similarly calibrated. To correct for this difference, the significance of SAM, dChip and logit-T values were assessed using all three by three combinations of the null dataset (given the permutation-based calibration of false-discovery rate utilized by SAM, note that SAM values are not predicted to improve significantly using this method of calibration). Results are shown in Figure 3d. In the case of the t-test, results obtained using calculated P values are generally within 5% of comparable results using empirically calibrated P values. Logit-T and dChip appear much less sensitive when using reported P values, although both of these techniques show improvement when calibrated using the control dataset. Of particular note, logit-T performs only slightly less well than ChipStat when calibrated against our control distribution, consistent with the fact that it was the only other algorithm considered that performs probe-pair-level comparisons when testing for differential gene expression.
Design and validation of the Intersector algorithm
Although the Affymetrix Microarray Suite (MAS) software utilizes probe-level information in identifying differentially expressed genes, its use has been restricted to single-array comparisons. As a result, it has been widely recognized that this approach generates an unacceptably high number of false-positive results. The use of replicate samples, however, might be expected to lower the false-positive rate while achieving a higher sensitivity. We therefore combined pairwise comparisons between triplicate data points in two different groups (that is, nine comparisons in total) and determined differential expression based on the Affymetrix call (for example, increases + marginal increases) for these comparisons. A similar technique, in which a simple majority cutoff (5/9 changes) was considered to denote significant change, has recently been described [18]. Although this approach involves N2 comparisons in general for equal groups of N arrays, it is easily feasible for three-sample versus three-sample comparisons. We have designated this approach Intersector. Significantly, the control data previously generated to calibrate ChipStat also allow us to determine the empirical false-positive rate for Intersector as a function of the number of 'increase' calls and to perform direct comparisons with other algorithms.
The performance of the Intersector algorithm in comparing 2- versus 5-week mammary gland gene expression is shown in Figure 4a. Interestingly, the Intersector algorithm is able to achieve a slightly improved sensitivity at a given false-positive rate when compared with ChipStat. To determine whether the particular version of the MAS algorithm influences this result, all analyses were run using difference calls from both MAS 4.0 and MAS 5.0 (see Figure 4a). Although the number of changes required to achieve similar sensitivity was different, the Intersector results from MAS 4.0 and MAS 5.0 are comparable at a given false-positive rate.
Given substantial differences between the types of probe-pair comparisons performed by ChipStat and MAS, we next wished to ascertain if these algorithms identify the same sets of upregulated genes. Direct comparison requires that the analyses result in comparable false-detection rates. We therefore compared the lists at thresholds corresponding to approximately 2.5 genes expected by chance, and the closest available threshold with each algorithm was chosen. The resulting thresholds were Intersector (MAS4) 7/9 (1.75 expected by chance), Intersector (MAS5) 8/9 (2.8 expected by chance), and ChipStat (.04) 8/16 (2.68 expected by chance). Notably, examination of these lists demonstrates that each algorithm (Intersector with MAS 4.0 data, Intersector with MAS 5.0 data and ChipStat) detects a discrete set of genes that are not detected by the others (Figure 4b). This is particularly intriguing since empirically estimated false positive rates suggest that these groups of genes are not likely to reflect chance fluctuations alone. Thus, in addition to identifying a core set of regulated genes, the Intersector and ChipStat algorithms each detect sets of complementary, nonoverlapping genes that change significantly.
To confirm this result, five out of the 13 genes uniquely identified by ChipStat were randomly chosen for confirmation. One of these genes was undetectable by northern blot hybridization, and the remaining 4/4 showed differential expression in the predicted direction (5 weeks > 2 weeks) (Table 1, and data not shown). This demonstrates that, at comparable levels of statistical stringency, ChipStat correctly identifies differentially expressed genes that are not identified by Intersector. Further, having directly tested approximately 40% of all genes in this category, no false positives were identified. Examination of lower stringency lists (9.5 expected by chance from ChipStat, 7.4 expected by chance from Intersector using MAS5) also revealed sets of genes identified by ChipStat or Intersector alone. For example, the 'Intersector only' list created at this lower stringency contains α-, β-, and γ-casein; previous work in our lab has demonstrated that these genes are differentially regulated with expression at 5 weeks greater than that at 2 weeks (data not shown).
Development of a hybrid approach
Given the presence of genes uniquely identified by Intersector or ChipStat at a given false positive rate and the feasibility of performing Intersector analysis on small numbers of replicates, we next explored whether a combination of these approaches could further improve overall detection. To test this, all possible pairwise threshold combinations of ChipStat (pps = 0.05, 0/16 to 16/16 probe pairs changing) and Intersector (0/9 to 9/9 increases or marginal increases) were combined, and aggregate lists of genes identified by both algorithms were tabulated (see Additional data file 1). The results demonstrate that a combination of these two approaches can lower the expected false positive rate while maintaining a high sensitivity. For example, the combination of ChipStat (pps = 0.05, 6/16 probe pairs increasing) and Intersector (7/9 increases + marginal increases) detects 209 increasing probe sets with only 3.4 expected to increase by chance (expected false-positive rate less than 2%). A comparison of the false-positive rates for single (ChipStat or Intersector alone) and combined (ChipStat and Intersector) approaches is shown in Figure 4c. Note that the total number of probe sets detected by the combined approach shown in Figure 4c is greater than the number detected by the single approach with a comparable false-detection rate (209 probe sets and 173 probe sets, respectively). The behavior of optimal combinations with respect to the number of genes detected is shown in Figure 4d.
One additional feature of this combined approach is the ability to 'fine-tune' the number of expected false positives. That is, while Intersector (MAS5) allows no choice between approximately three and approximately seven expected false positives (2.8 and 7.35, corresponding to 8/9 or 7/9 changes, respectively), the combined approach provides a smoother continuum of values. More important, these data show that, for certain targeted numbers of expected false positives, a combination of ChipStat and Intersector can provide improved performance in gene detection compared with either algorithm alone.
Genomic characterization of early mammary gland development
The goal of these methodological developments has been the elucidation of biological mechanisms underlying mammary gland development and carcinogenesis. We therefore used the hybrid ChipStat/Intersector lists representing early mammary gland development as a basis for further exploration of developmental processes during this time period. A complete list of genes differentially expressed between 2- and 5-week murine mammary gland was compiled using the techniques described above. The results are listed in Additional data file 2.
To identify coherent functional patterns of gene expression during neonatal development through the onset of puberty, statistically significant associations between Gene Ontology (GO) categories [19] and lists of up- and downregulated genes were identified using EASE [20]. Multiple testing correction was performed using within-system bootstrapping, and a corrected significance threshold of P less than 0.05 was used. Results are shown in Table 2. Upregulated genes were associated with a total of 22 GO categories, and downregulated genes with 10 categories. In addition, this approach provides a convenient test of whether the increased sensitivity of ChipStat/Intersector yields corresponding power in identifying patterns of biological activity. To test this directly, lists of differentially expressed genes with the same number of expected false positives (empirically calibrated as previously) were identified using dChip and logit-T. These lists were then tested for association with GO annotation, and the results are shown (Table 1, Figure 5). Of note, ChipStat/Intersector lists were associated with a greater number of GO categories than were dChip or logit-T, and this was true for both up- and downregulated gene lists. Consistent with our suggestion that logit-T should be most similar to ChipStat/Intersector because of its use of probe-pair-level comparisons, logit-T also generated lists that are statistically associated with a larger number of GO categories than did dChip (Figure 5), although it did not outperform ChipStat/Intersector. ChipStat/Intersector identified 22/22 of categories associated with any of the list of upregulated genes and 10/11 categories identified using any of the lists of downregulated genes. A single downregulated category ('cellular component: extracellular') was associated only with the logit-T list.
To provide a crude check on the reliability of these results in addition to the confirmation previously performed, gene lists were examined for association with previously described biological processes. In addition to individual genes that are consistent with epithelial proliferation and differentiation (discussed above), several statistically associated categories represent pathways that have been previously described in the mammary gland during this developmental window [4]. These include 'blood vessel development' and 'mitochondrial inner membrane'. The latter category reflects the previously reported decrease in brown adipose tissue at the end of the neonatal period and the corresponding decrease in the capability of the mouse to utilize adaptive thermogenesis to maintain body temperature. Brown adipose tissue is not only rich in mitochondria, but the fatty-acid metabolic pathways necessary for adequate thermogenic activity are also spatially localized at the inner mitochondrial membrane. Of note, this category only reached statistical significance using the ChipStat/Intersector list.
Interestingly, 'pheromone binding' and 'odorant binding' categories are also associated with upregulated expression at the onset of puberty. Genes within these categories are primarily members of the major urinary protein (MUP) gene family, and MUP transcripts (Mup1, Mup3, Mup4, Mup5) account for four of the five most highly upregulated genes from 2 to 5 weeks. Large quantities of MUPs are synthesized in the male liver and excreted in the urine, where they bind pheromone and play a role in signaling for complex behavioral traits [21,22]. MUP levels are upregulated during puberty in the liver, although expression levels are much higher in males than in females. While MUP expression within the mammary gland has previously been reported [23,24], its expression was considered to be detectable only with the onset of pregnancy. Our data show that MUPs are highly upregulated in the female mammary gland during the 2- to 5-week transition. Interestingly, Slp (sex-limited protein), which also shows sex-restricted expression in the male liver and - like Mup expression - is normally repressed by Rsl [25], is also significantly upregulated during this period.
Additional examination of these gene lists revealed an interesting transcriptional pattern that is not reflected in the current GO hierarchy. The nontranslated RNA transcript Meg3/Gtl2 is significantly downregulated from 2 to 5 weeks of development, and its reciprocally imprinted neighbor Dlk1 [26] shows a similar decrease. This is noteworthy because two other genes with decreasing expression, H19 (nontranslated RNA) and Igf2, are also reciprocally imprinted neighbors, suggesting the possibility of a common regulatory mechanism for altering expression from loci exhibiting this genomic organizational structure (see [27]).
Discussion
The ability to reliably detect changes in gene expression is critical for the analysis of experimental microarray data. This problem assumes particular importance when analyzing complex mixtures of cells, such as those derived from a whole organ during ontogeny. The challenge can be most clearly seen by considering a small subpopulation of cells that demonstrate a marked change in gene expression. If the expression of this gene is uniform and low throughout the rest of the tissue, the biologically relevant change within a few cells will appear as a low fold change in organ-wide gene expression. A variety of such nonabundant yet developmentally critical cell types have been described. For example, the proliferative capacity of small structures in the mammary gland known as terminal end buds gives rise to the extensive ductal structure that is elaborated during puberty [17]. More recently, the characteristics of mammary stem cells have been described, and these cells have been suggested to serve as targets for carcinogenesis [28,29]. To facilitate the study of such subpopulations within a whole-organ context, therefore, we have developed a novel approach to the analysis of Affymetrix oligonucleotide microarray data.
A variety of nonparametric and parametric statistical tests, including variants of Student's t-test, have been used to identify significant changes in gene expression using replicate microarray data. Given the substantial economic investment required for large microarray experiments, attempts have also been made to improve detection of differentially regulated genes through better estimates of the null distribution using permutation analysis; the use of software incorporating such methods, such as SAM [11], has become widespread. A different approach to improved detection (dChip, see [8]) has attempted to use probe-level information to derive an improved estimate of relative gene expression before assessing differential regulation.
While much work has focused on such use of probe-level analysis for estimating gene expression [8,9], the analysis of replicate data at the probe level for identifying differentially expressed genes has only recently become a focus [10,30]. In particular, if hybridization noise contributes a substantial portion of the overall noise inherent in microarray measurements, the use of multiple probe pairs devoted to measuring a single gene suggests a potential approach to overcoming this noise.
The ChipStat algorithm uses heteroscedastic t-test comparisons between probe pairs, and the number of probe pairs that change greater than a significance threshold are tabulated. A greater number of consistently changing probe pairs should indicate that the difference is less likely to be due to hybridization noise, and thus this number relates the overall probability that the probe set is measuring a true change in gene expression. The processing time for the ChipStat algorithm scales as a linear function of the number of replicates processed (O(N)), and thus it is feasible to apply this approach to much larger numbers of samples.
To assess the statistical significance of ChipStat results, it was necessary to empirically measure the underlying null distribution. While the recent availability of a number of publicly available Latin square datasets representing measurements of spiked-in control samples has greatly facilitated measurements of this sort [31], these datasets reflect technical replicates without biological noise. As we have demonstrated, the behavior of the ChipStat algorithm would be expected to change depending on the relative contributions of biological/experimental noise and probe-level hybridization noise. Thus, a set of negative control samples reflecting an experimental system that include biological noise was required.
To generate these samples, mammary glands from six independent cohorts of mice were harvested. These data provide a true, gold-standard negative control within a representative mammalian experimental system, and we anticipate that their public availability will be similarly useful to the broader scientific community in analytical development and validation. Furthermore, the use of this dataset as an empirical calibration control for ChipStat argues that these results will be valid independent of the adequacy of the statistical noise model used. It is worth noting that the use of pooled groups of animals is likely an important parameter, as single-animals groups, for example, would be expected to exhibit increased biological variability and thus decrease the proportional contribution of hybridization noise.
Given empirical measurements of the expected number of false positives for a given set of analytical parameters, it was possible to assess the relative sensitivity of a variety of algorithms using a positive control dataset (2-week versus 5-week murine mammary gland) known to contain a substantial number of increasing transcripts. Consistent with our hypothesis that probe-level comparison analysis should improve sensitivity, ChipStat was able to substantially outperform a variety of methods (t-test, SAM, and dChip) based on aggregate gene-expression measures (Figures 3c,d). Furthermore, this remained the case even when the statistical significance of dChip was recalibrated using a negative control dataset.
Recently, Lemon et al. have described a method (logit-T) that is also based on probe-level t-test comparisons for identifying differentially expressed genes [10]. The logit-T algorithm estimates statistical significance using the median result of t-tests performed on log-transformed PM probe data. ChipStat differs from this approach in several significant respects. These include the use of a fixed P value threshold for pairwise probe comparisons and the use of the degree of reproducibility across the entire probe set as an indication of statistical significance. Results from the empirical control data suggest that ChipStat performs slightly better than logit-T in most cases within our biological system. Interestingly, however, the advantage of ChipStat over logit-T was more modest than the advantage over SAM, dChip, and the t-test; as logit-T also uses probe-level comparisons, this result is consistent with our overall observations regarding the increased power of probe-based analysis. It is also worth noting that the nominal P values derived from both logit-T and dChip substantially underestimated statistical significance prior to correction with our control data, suggesting that, for example, the median P value cannot be used to directly assess significance without such correction.
One additional difference between ChipStat and logit-T stems from the use of mismatch (MM) probe cells (ChipStat) and log-transformed data (logit-T). As currently implemented, the ChipStat algorithm compares differences in probe pair (PM - MM) values rather than in PM values alone. Interestingly, the use of PM values within the ChipStat algorithm does not result in superior performance (data not shown), and log(PM) data yield performance that is roughly comparable to PM - MM (data not shown). Further work will be required to determine if the log(PM) approach can be adapted to improve the performance of ChipStat.
The Intersector algorithm tabulates MAS calls from all pairwise comparisons across replicate groups. As we have shown, this algorithm provides the most sensitive method for detecting gene expression changes at low false-detection rates. However, it suffers from several substantial drawbacks. First, the proprietary nature of the Affymetrix algorithm and its associated decision matrices limits the ability to automate the analytical process. Additionally, because N2 pairwise comparisons are required for equal groups of N replicates (that is, O(N2)), this method is not easily scalable to larger numbers of samples. In contrast, ChipStat scales linearly with N, and the use of the heteroscedastic t-test also makes it possible to precompute results for a (potentially large) baseline control population against which multiple comparisons will be performed. While both approaches are feasible for triplicate comparisons, extension of Intersector to much larger numbers is unlikely to be practical.
A third disadvantage to the Intersector approach stems from the lack of a detailed model for its underlying statistical framework. Both ChipStat and Intersector, as currently described, require the use of control samples to generate an estimate of statistical significance. Thus, extension of these results to encompass either a substantially different experimental system or larger numbers of replicates will require the generation of new empirical significance curves. In the case of large numbers of replicates, however, the cost of generating such data is likely to remain prohibitive at least in the near future. Statistical simulations of ChipStat behavior may, however, provide a mechanism for extending the current control curves for larger datasets. This approach would rely on the relative estimates of E(b) + E(p) and E(h) obtained by fitting the current empirical curve derived from six samples. While a small number of probe sets change more often than would be predicted by simulation, it should be possible to conservatively estimate an upper bound to the P value curve by overestimating the relative contribution of E(b) + E(p) vs. E(h). Further experimental work will be required to confirm this possibility.
One caveat to this approach is that the simplified statistical model that we have used to illustrate the theoretical advantages of probe-level comparisons does not account for the variability in the behavior of specific probe cells that exist on Affymetrix arrays. In contrast, the model-based approach to signal estimation implemented in dChip explicitly incorporates such variability and has been shown to provide a good fit to empirical measurements of array data [8,32]. It is likely that incorporation of probe-specific parameters in this way would improve the ability to predict the theoretical behavior of ChipStat and provide a better estimate of hybridization noise from our empirical data. Given this likelihood, current estimates of the relative contributions of E(b) + E(p) and E(h) should be taken as provisional.
In this context, it is worth noting that the ability to perform our current validation and to tune both Intersector and ChipStat was critically dependent on the gene-expression datasets derived from our independent cohorts of control animals. One might naively assume, in the absence of true negative control data, that robust changes in gene expression should be detected by simply taking the Affymetrix MAS calls and requiring that they consistently demonstrate increases for all pairwise comparisons. In contrast, our data show that the Intersector algorithm can achieve increased sensitivity while retaining an appropriately low (and defined) false-positive rate. Both the Intersector and ChipStat algorithms can be tuned using negative control data for sensitivity versus false-positive rate, depending on the type of analysis and application-specific tolerance for false-positive calls. Furthermore, these algorithms can be combined to further improve their sensitivity. As we have demonstrated that each of these algorithms detects a population of probe sets not identified by the other at a comparable stringency, this combined approach may yield the best result. Given these considerations, we favor the use of the hybrid ChipStat/Intersector approach for small number of replicates (around three), with ChipStat alone being useful for large numbers of replicates. Although ChipStat shows greater sensitivity than logit-T at moderate numbers of false positives (more than five expected false positives out of 12,488 probe sets), their comparable performance at high stringency (less than five expected false positives) suggests that the overlap in genes identified by these two techniques may also be of interest.
An additional piece of evidence for the utility of our approach is provided by the statistical association of GO annotation with lists derived from ChipStat/Intersector, dChip or logit-T. At the level of significance tested (3.4 genes per list expected by chance), ChipStat/Intersector lists were statistically associated with a greater number of GO terms than were lists derived from dChip or logit-T. Furthermore, as would be predicted from the fact that logit-T is also a probe pair-level comparison method, logit-T lists are associated with GO terms at a level that is intermediate between dChip and ChipStat/Intersector. One of the terms associated only with the ChipStat/Intersector list of downregulated genes is 'mitochondrial inner membrane'; this example is particularly noteworthy in light of previous work demonstrating a presumptive role for enzymes of fatty acid oxidation in adaptive thermogenesis during the neonatal period [4].
It should be noted that these results depend on the level of significance chosen for generation of the original lists, and an increase in the total number of differentially expressed genes identified may actually decrease the statistical significance of a given association if it does not result in the detection of more genes within the category in question. Despite some caveats as to the generalizability of these results, our data demonstrate that the improved sensitivity of ChipStat/Intersector can measurably influence the ability to interpret patterns of biological activity.
Early murine mammary gland development
For the FVB murine mammary gland, the period from 2 to 5 weeks of age encompasses critical developmental milestones that include the suckling-weaning transition as well as the profound hormonal changes that characterize the onset of puberty and its consequent rapid ductal epithelial proliferation. Our present work has more completely characterized changes in the transcriptome that occur during early murine mammary gland development than have previous reports. A total of 213 upregulated and 130 downregulated probe sets were identified under conditions designed to yield a low expected false-positive rate (3.4 probe sets expected to change by chance per list).
Four out of five of the most highly upregulated transcripts through the onset of puberty are members of the MUP family of odorant-binding proteins. MUPs are lipocalins that can bind hydrophobic molecules such as pheromones, and they have previously been shown to play a role both in the delivery of signals within the urine as well as in the reception of these signals on the nasal epithelium ([33,34]; see [35] for review). Isoform-specific MUP expression has also previously been reported in a number of secretory glands, including the murine mammary gland [23,24]. However, detectable expression has previously been reported only beginning with the first pregnancy [23].
Our results demonstrate a striking increase in the expression of a variety of MUP isoforms as the mammary gland makes the transition from the neonatal period through the beginning of puberty. This expression pattern is noteworthy given the known effect of puberty on MUP expression in the liver [36]. Interestingly, however, expression in the liver is markedly greater in the male and has been causally linked to the male pattern of growth-hormone pulses [36]. As this male-specific pattern of expression has been shown to be Stat5b-dependent, the availability of Stat5b -/- mice should allow future determination of whether mammary expression is mediated via a similar signaling pathway. Regardless of this, despite an interval of over two decades since the first description of MUP expression in the mammary gland, a functional role has still not been elucidated. Although it has long been assumed that MUP synthesis occurs in the secretory epithelium of expressing organs, our observation that these molecules are upregulated during puberty (with a corresponding approximately threefold downregulation following puberty, data not shown) suggests that their functional role may not be limited to the secretory function of the gland.
Delta-like kinase (Dlk1) is a member of the epidermal growth factor (EGF) superfamily [37] that is encoded on murine chromosome 12 [38]. Dlk1 is one of several genes showing substantial (greater than fivefold) downregulation from 2 to 5 weeks of murine mammary gland development. As this gene was first identified as a preadipocyte transcript that is downregulated during subsequent differentiation [38], we hypothesize that its relatively high expression during the neonatal period reflects ongoing differentiation of the mammary fat pad. This kinase has also been shown to have a role in other developmental contexts, specifically within neuroendocrine tissues. Further work will be required to elucidate its specific role in the mammary gland. Notable, however, is the corresponding downregulation (more than 10-fold) of Meg3/Gtl2, a noncoding RNA that is reciprocally imprinted with Dlk1 [26]. This Dlk1-Meg3/Gtl2 regulation has been compared with Igf2-H19, another tandem pair of reciprocally imprinted genes in which one member produces a noncoding RNA [27,39]. Interestingly, both Igf2 and H19 are also downregulated during this time period, suggesting the hypothesis that a common regulatory mechanism exists for the tandem control of both imprinted genes at these loci. It will be particularly important to determine whether there is functional significance to this Igf2-H19 regulation, or whether it reflects the epiphenomenal byproduct of a mechanism designed to downregulate Dlk1 during adipocyte development.
Conclusions
We have developed two novel algorithms for the analysis of Affymetrix oligonucleotide microarray data. We have validated these algorithms by using empirically derived distributions from control animals to calibrate their statistical significance. These control data, which reflect both experimental and biological sources of variability likely to be representative of many mammalian experimental systems, should facilitate further work in this area. For triplicate samples, Intersector appears to provide the most sensitivity at a given threshold of statistical significance, and its performance is substantially superior to other widely used methods including the t-test, SAM, dChip, and logit-T. However, its lack of scalability, along with the baseline time required for processing, make it unsuitable for larger numbers of replicates. ChipStat, in contrast, provides comparable sensitivity with triplicate samples and has the capability of handling much larger numbers of replicates in order to improve the reliable dectection of small changes in gene expression. Both algorithms provide a substantial increase in the ability to sensitively detect statistically significant changes in gene expression within the context of the whole mammary gland.
We have applied these techniques to the analysis of genomic patterns during early murine mammary gland development. In addition to detecting patterns reflecting known biology, we have noted the coordinate upregulation of a class of molecules not previously known to be differentially regulated in the mammary gland. We also suggest that peri-pubertal changes in the mammary gland may utilize mechanisms for tandem upregulation of multiple imprinted regions. Our observations suggest a variety of future directions for functional validation and demonstrate the utility of coupling sensitive detection of differential gene expression with pathway analysis for the elucidation of biological patterns during organogenesis.
Materials and methods
Animals, RNA isolation, and northern blot hybridization
The third, fourth and fifth mammary glands were harvested from FVB mice at the indicated time points. Samples from 2 and 5 weeks of age reflect triplicate pools of 10 animals at each time point (total 60 animals). In addition, tissue from 18 control animals was harvested when they were 6 weeks and 4 days old. These control animals also carry a transgenic construct consisting of the murine mammary tumor virus (MMTV) promoter upstream of the reverse tetracycline transactivator (rtTA) and had been given 2.0 mg/ml doxycycline in drinking water for 96 h before harvest. This line (previously designated MTB) has been previously described, and no developmental abnormalities have been noted [40]. All animal experimentation was conducted in accord with accepted standards of humane care, and protocols for animal work were approved by the University of Pennsylvania institutional committee on animal care.
All tissue was snap frozen after removal of the lymph node present in the fourth gland, and total RNA was isolated by homogenization in guanidinium isothiocyanate and subsequent centrifugation through a cesium chloride cushion as previously described [41]. Northern blot hybridization was performed as previously described [42].
Arrays and hybridization
Approximately 15-20 μg total RNA was used for each hybridization. RNA was visualized by gel electrophoresis to ensure its integrity before analysis. Biotinylated cRNA was generated and hybridized to Affymetrix MG_U74Av2 arrays according to the manufacturer's instructions. To scale between chips, these expression values were rank ordered, and the median approximately 96% were averaged. Chips were scaled relative to each other to equalize this average value. All Affymetrix control probe sets were eliminated from analysis, yielding data from a total of 12,422 probe sets. Datasets are publicly available as CEL files designated MTB_ [1-6] (Additional data files 3-8 available with the online version of this paper), 2wk_G0P0_ [1-3] (Additional data files 9-11) and 5wk_G0P0_ [1-3] (Additional data files 12-14) containing results derived from control cohorts, 2-week nulliparous cohorts, and 5-week nulliparous cohorts respectively.
Algorithms and software
To detect differentially regulated genes, we implemented an algorithm (ChipStat) that takes identical probe pairs across two comparison groups and performs a heteroscedastic t-test. The number of probe pairs within a probe set that are significantly different (P <pps where pps is a fixed value) was tabulated. We consider that a greater number of probe pairs changing in a given direction indicates a greater probability that the gene detected by the probe set is differentially expressed. If the bulk of the noise within the array data derives from pre-hybridization experimental factors (that is, E(b) + E(p); see Results section for definition), the expectation is that all probe pairs would change coordinately. That is, if there are 16 probe pairs in the probe set, we would expect (for E(b) + E(p) >> E(h)) that under the null hypothesis (no change in gene expression) either 0/16 or 16/16 probe pairs should change significantly (at frequencies of approximately 1 - pps and approximately pps, respectively). Conversely, if the bulk of the noise derives from hybridization to individual probe cells (that is, if E(b) + E(p) << E(h)), then the number of probe pairs r that change within a given probe set of size t can be approximated by the binomial distribution:
However, under experimentally realistic conditions, neither of these limiting cases is likely to apply. Therefore, to empirically determine the null distribution using six independent, biologically identical control populations, all pairwise three by three combinations were compared and the number of probe pairs changing was tabulated. To determine the expected number of changes per probe set when fewer than 16 probe pairs are available, these analyses were repeated after randomly discarding 1, 2...15 probe pairs. In this way, a similar statistical estimate was obtained for the 602 probe sets on the MG_U74Av2 array that have fewer than 16 probe pairs per probe set. A conservative simplification of these data was performed by rounding up the significance of changes in these 602 probe sets to the nearest appropriate bin in the 16 probe pair per probe set curve. A Microsoft Windows-compatible application implementing the ChipStat algorithm is freely available for academic use [43].
On the basis of the simplified statistical model described, a Monte Carlo simulation was implemented to determine the number of expected false-positive values as a function of pps for various relative proportions of E(b) + E(p) and E(h). Briefly, a random test dataset was generated in which equal gene expression was perturbed by Gaussian noise (representing E(b) + E(p)). Each expression value was then independently perturbed 16 times (representing 16 probe pairs/probe set) by another Gaussian noise function (representing E(h)), and comparisons were tabulated using the ChipStat algorithm. This simulation was implemented in C and the source code is available [43]. All values reported reflect the mean of 100 trials, where each trial simulates 11,820 probe sets with 16 probe pairs each. The relative contributions of E(b) + E(p) and E(h) were estimated by maximizing the likelihood function:
with respect to (E(b) + E(p)) / E(h) where xi is the number of times i probe pairs increased significantly and μi and σi represent the mean and standard deviation from the Monte Carlo simulations.
A separate algorithm (Intersector) uses pairwise calls of differential gene expression derived from Affymetrix Microarray Suite (MAS) analysis. All pairwise comparisons were performed (that is, 3 × 3 = 9 comparisons for a 3- vs 3-replicate comparison) using the manufacturer's default settings, and the number of 'increases' or 'marginal increases' was tabulated. Similarly to the ChipStat method described above, the null distribution was generated by tabulating results from all 20 distinguishable 3 vs 3 combinations of the six control samples. Results were obtained using both MAS version 4 (MAS4) and MAS version 5 (MAS5), as indicated in the text and figures.
Tests for differential gene expression using a homoscedastic t-test or SAM [11] were performed using signal values derived from MAS5. SAM results were obtained using software obtained from its authors [44]. Because the analyses described are reported as a function of the number of genes expected to increase by chance (essentially a one-tailed test of significance), the false-discovery rate reported by SAM was multiplied by 0.5 to derive a corrected false-positive rate (false-increase rate). dChip analysis [8] was performed using software available from its authors [45], and a PM-only expression model was constructed. Logit-T analysis [10] was performed using software provided by its authors and compiled to run locally on an AMD Linux server. Both dChip and Logit-T significance values were empirically calibrated by analyzing all possible 3 vs 3 combinations of control arrays (20 total) and tabulating the average number of false positives as a function of the reported significance.
Association with biological annotation
Associations between GO [19] annotation and lists of differentially expressed genes were identified using EASE [20]. Multiple testing correction was performed using within-system bootstrapping, and a final cutoff of P < 0.05 was used to identify statistically significant associations.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 contains a table showing ChipStat and Intersector in combination. For each level of stringency available, the pairwise intersection of ChipStat (CS, pps = 0.05) and Intersector (IT, MAS5) lists of significantly increasing probe sets was generated. Rows indicate the threshold number of probe pairs (0-16) significantly increasing from ChipStat, and columns indicate the threshold number of Increase or Marginal Increase calls (0-9) identified by Intersector. (a) Number of increasing probe sets in 2- vs 5-week murine mammary gland. Selected results correspond to values plotted on the y axis of Figure 4d (number of probe sets increasing). (b) Average number of increasing probe sets using all 3 × 3 combinations of 6 negative control samples. Selected results correspond to values plotted on the x axis of Figure 4d (expected number of probe sets increasing by chance). Additional data file 2 contains a table showing differential gene expression in 2- vs 5-week murine mammary gland using a hybrid ChipStat/Intersector approach. The criteria ChipStat pps = 0.05, 6/16 probe pairs increasing and Intersector 7/9 increases + marginal increases, were used to identify lists of probe sets that are up- and downregulated from 2 to 5 weeks of FVB female murine mammary gland development. Additional data files 3,4,5,6,7 and 8 contain six control files containing CEL file data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment. Additional data files 9,10 and 11 contain three CEL files of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age. Additional data files 12,13 and 14 contain three CEL files of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age.
Supplementary Material
Additional data file 1
A table showing ChipStat and Intersector in combination
Click here for additional data file
Additional data file 2
A table showing differential gene expression in 2- vs 5-week murine mammary gland using a hybrid ChipStat/Intersector approach
Click here for additional data file
Additional data file 3
Control file 1 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 4
Control file 2 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 5
Control file 3 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 6
Control file 4 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 7
Control file 5 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 8
Control file 6 containing CEL data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to RNA from the third to fifth mammary glands harvested from independent pools of three female MTB transgenic mice at 6 weeks 4 days old after 96 hours of doxycycline treatment
Click here for additional data file
Additional data file 9
CEL file 1 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age
Click here for additional data file
Additional data file 10
CEL file 2 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age
Click here for additional data file
Additional data file 11
CEL file 3 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized with mammary gland RNA from independent pools of 10 female FVB mice harvested at 2 weeks of age
Click here for additional data file
Additional data file 12
CEL file 1 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age
Click here for additional data file
Additional data file 13
CEL file 2 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age
Click here for additional data file
Additional data file 14
CEL file 3 of data from Affymetrix MG_U74Av2 oligonucleotide microarrays hybridized to mammary gland RNA from independent pools of 10 female FVB mice harvested at 5 weeks of age
Click here for additional data file
Acknowledgements
This research was supported in part by NIH Grants CA94393, CA92910, and CA93719 from the National Cancer Institute, NRSA HL007150-27 (L.C.B.), and by US Army Breast Cancer Research Program Grant DAMD17-01-1-0364.
Figures and Tables
Figure 1 ChipStat behavior using simulated biological/experimental + hybridization noise model. The behavior of the ChipStat algorithm was evaluated (pps = 0.05, 16 probe pairs per probe set) using a Monte Carlo model in which the ratio of biological + experimental noise (E(b) + E(p)) to hybridization noise (E(h)) is constant (see text for further details). Results are shown for E(h) = 0 (Exp noise only; blue), E(h) = E(b) + E(p) (Hyb noise = Exp noise; red), E(h) = 2 × (E(b) + E(p)) (Hyb noise = 2 × Exp noise; green), and E(b) + E(p) = 0 (Hyb noise only; yellow). The total number of probe sets simulated (11,820) was chosen to match the number of probe sets containing 16 probe pairs per probe set on the Affymetrix MG_U74Av2 array. The number of probe pairs increasing by chance is shown on the x axis, and the fraction of total probe sets simulated is shown on the y axis. This simulation was repeated 100×, and the average of these results is shown. (a) Probability of the indicated number of probe pairs increasing. (b) Cumulative P value (equal to or greater than the indicated number of probe pairs changing).
Figure 2 Empirical measurement of the ChipStat null distribution. Mammary gland tissue was harvested from six separate, biologically identical pools of FVB (MTB) mice, and hybridization data to Affymetrix MG_U74Av2 microarrays was obtained. Comparisons of all possible three versus three combinations (total 20) were performed using ChipStat (pps = 0.05), and the number of significant increases was tabulated for all probe sets containing 16 probe pairs per probe set (total = 11,820). The cumulative average probability is shown as a function of the number of probe pairs that increase within the probe set.
Figure 3 Relative detection sensitivity of differential gene expression. The number of probe sets shown to increase from 2 to 5 weeks of murine mammary gland development was tabulated as a function of the number of probe sets expected to increase by chance. (a) ChipStat (pps = 0.05), vs t-test. (b) Optimization of ChipStat sensitivity as a function of pps. (c) ChipStat vs other techniques: reported P values. For ChipStat, the number of probe sets expected to increase by chance was empirically estimated from negative control data. For the t-test, SAM, dChip and logit-T, reported P values from the 2-week vs 5-week mammary gland comparison were used. (d) ChipStat vs other techniques: empirical P values. The number of probe sets expected to increase by chance was empirically estimated for ChipStat, t-test, SAM, dChip and logit-T (representative points).
Figure 4 Intersector and ChipStat performance. (a) The number of probe sets shown to increase from 2 to 5 weeks of murine mammary gland development was tabulated as a function of the number of probe sets expected to increase by chance, and a comparison of ChipStat (pps = 0.05), Intersector (MAS 5.0 change calls), and Intersector (MAS 4.0 change calls) is shown. (b) Venn diagram showing distinct probe sets identified by ChipStat and Intersector. The number of genes shown to be differentially expressed at the indicated expected false-positive levels is shown for ChipStat (CS) (pps = 0.04), Intersector (IT) with MAS 5.0 calls, and Intersector (IT) with MAS 4.0 calls. (c) False-positive rates for ChipStat (CS 6/16: pps = 0.05, 6/16 probe pairs increasing; CS 9/16: pps = 0.05, 9/16 probe pairs increasing), Intersector (MAS5) (IT 7/9: 7/9 increases or marginal increases; IT 8/9: 8/9 increases or marginal increases), or ChipStat and Intersector together (Combined: intersection of CS 6/16 and IT 7/9) are shown. (d) Combined performance of ChipStat and Intersector. Increases from 2 to 5 weeks of mammary gland development are shown for ChipStat alone (pps = 0.05), Intersector alone (MAS 5.0), and optimized intersections of ChipStat and Intersector (see Additional data file 1).
Figure 5 Quantitative association with GO categories. The number of GO terms found to be statistically associated (P < 0.05 using within-system bootstrap to account for multiple testing) with lists of differentially regulated genes (2 vs 5 weeks of murine mammary gland development) is shown. Lists of up- and downregulated genes were generated using dChip (DC), logit-T (LT) and a ChipStat/Intersector hybrid (CS/IT) that were matched in stringency to give equivalent numbers of expected false-positive genes.
Table 1 Northern blot validation of differential gene expression
Probe set ID Accession number Gene Fold change Probe pairs increasing Differential expression confirmed
99067_at X59846 Gas6 3.41 16/16 x
100064_f_at M63801 Gja1 1.67 12/16 x
102016_at M61737 Fsp27 2.07 11/16 x
93996_at X01026 Cyp2e1 11.6 10/16 x
97507_at X67809 Ppicap 2.85 9/16 x
101995_at U40930 Sqstm1 1.48 8/16 x
93096_at AA986050 3010002H13Rik 2.65 7/16 x
102791_at U22033 Psmb8 1.65 7/16 x
96072_at M17516 Ldh1 1.37 6/16
Genes identified as being differentially expressed were randomly chosen for verification by northern blot hybridization (see text for description). Gene identifiers are shown along with fold changes, numbers of probe pairs increasing (as identified by ChipStat with pps = 0.04), and confirmation of differential expression.
Table 2 Association with GO annotation
System Gene category CS LT DC
(a) Upregulated genes
GO Biological Process Defense response x x x
GO Cellular Component Extracellular space x x
GO Cellular Component Extracellular x x
GO Biological Process Response to biotic stimulus x x x
GO Biological Process Immune response x x x
GO Biological Process Response to external stimulus x x x
GO Biological Process Organismal physiological process x x x
GO Biological Process Antigen presentation x x
GO Biological Process Response to stimulus x x x
GO Biological Process Antigen presentation\, endogenous antigen x
GO Molecular Function MHC class I receptor activity x
GO Biological Process Antigen processing x x
GO Biological Process Complement activation x x
GO Biological Process Antigen processing, endogenous antigen via MHC class I x
GO Biological Process Response to pest/pathogen/parasite x x x
GO Biological Process Humoral defense mechanism (sensu Vertebrata) x
GO Molecular Function Pheromone binding x x
GO Molecular Function Oxidoreductase activity x
GO Molecular Function Oxidoreductase activity, acting on the aldehyde or oxo group of donors x
GO Molecular Function Odorant binding x x
GO Molecular Function Transmembrane receptor activity x
GO Biological Process Humoral immune response x
(b) Downregulated genes
GO Cellular Component Mitochondrion x x
GO Biological Process Main pathways of carbohydrate metabolism x x
GO Biological Process Tricarboxylic acid cycle x x
GO Biological Process Energy derivation by oxidation of organic compounds x x
GO Biological Process Energy pathways x x
GO Cellular Component Mitochondrial membrane x
GO Biological Process Carbohydrate metabolism x x
GO Cellular Component Inner membrane x
GO Biological Process Blood vessel development x
GO Cellular Component Mitochondrial inner membrane x
GO Cellular Component Extracellular x
Lists of differentially expressed genes derived from a hybrid ChipStat/Intersector approach (ChipStat: pps = 0.05, 6/16 probe pairs increasing AND Intersector: 7/9 increases + marginal increases), logit-T, and dChip were associated with GO terms using EASE [20]. Individual terms are annotated according to whether association with the given annotation group was statistically significant (P < 0.05 using within-system bootstrap to account for multiple testing) using lists derived from ChipStat/Intersector (CS), logit-T (LT), or dChip (DC). (a) Association with lists of upregulated genes. (b) Association with lists of downregulated genes.
==== Refs
Coller HA Grandori C Tamayo P Colbert T Lander ES Eisenman RN Golub TR Expression analysis with oligonucleotide microarrays reveals that MYC regulates genes involved in growth, cell cycle, signaling, and adhesion. Proc Natl Acad Sci USA 2000 97 3260 3265 10737792 10.1073/pnas.97.7.3260
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JC Sabet H Tran T Yu X Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000 403 503 511 10676951 10.1038/35000501
van't Veer LJ Dai H van de Vijver MJ He YD Hart AA Mao M Peterse HL van der Kooy K Marton MJ Witteveen AT Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002 415 530 536 11823860 10.1038/415530a
Master SR Hartman JL D'Cruz CM Moody SE Keiper EA Ha SI Cox JD Belka GK Chodosh LA Functional microarray analysis of mammary organogenesis reveals a developmental role in adaptive thermogenesis. Mol Endocrinol 2002 16 1185 1203 12040007 10.1210/me.16.6.1185
Schena M Shalon D Davis RW Brown PO Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 1995 270 467 470 7569999
Lockhart DJ Dong H Byrne MC Follettie MT Gallo MV Chee MS Mittmann M Wang C Kobayashi M Horton H Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 1996 14 1675 1680 9634850 10.1038/nbt1296-1675
Lipshutz RJ Fodor SP Gingeras TR Lockhart DJ High density synthetic oligonucleotide arrays. Nat Genet 1999 21 1 Suppl 20 24 9915496 10.1038/4447
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
Irizarry RA Hobbs B Collin F Beazer-Barclay YD Antonellis KJ Scherf U Speed TP Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003 4 249 264 12925520 10.1093/biostatistics/4.2.249
Lemon WJ Liyanarachchi S You M A high performance test of differential gene expression for oligonucleotide arrays. Genome Biol 2003 4 R67 14519202 10.1186/gb-2003-4-10-r67
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001 98 5116 5121 11309499 10.1073/pnas.091062498
Williams JM Daniel CW Mammary ductal elongation: differentiation of myoepithelium and basal lamina during branching morphogenesis. Dev Biol 1983 97 274 290 6852366 10.1016/0012-1606(83)90086-6
Daniel CW Silberstein GB Neville MC, Daniel CW Postnatal development of the rodent mammary gland. The Mammary Gland: Development, Regulation, and Function 1987 New York: Plenum Press 3 36
Dao TL Mammary cancer induction by 7,12-dimethylbenz[a]anthracene: Relation to age. Science 1969 165 810 811 5796556
Ip C Mammary tumorigenesis and chemoprevention studies in carcinogen-treated rats. J Mammary Gland Biol Neoplasia 1996 1 37 47 10887479
Rogge L Bianchi E Biffi M Bono E Chang SY Alexander H Santini C Ferrari G Sinigaglia L Seiler M Transcript imaging of the development of human T helper cells using oligonucleotide arrays. Nat Genet 2000 25 96 101 10802665 10.1038/75671
Richert MM Schwertfeger KL Ryder JW Anderson SM An atlas of mouse mammary gland development. J Mammary Gland Biol Neoplasia 2000 5 227 241 11149575 10.1023/A:1026499523505
Rajagopalan D A comparison of statistical methods for analysis of high density oligonucleotide array data. Bioinformatics 2003 19 1469 1476 12912826 10.1093/bioinformatics/btg202
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000 25 25 29 10802651 10.1038/75556
Hosack DA Dennis G JrSherman BT Lane HC Lempicki RA Identifying biological themes within lists of genes with EASE. Genome Biol 2003 4 R70 14519205 10.1186/gb-2003-4-10-r70
Sharrow SD Vaughn JL Zidek L Novotny MV Stone MJ Pheromone binding by polymorphic mouse major urinary proteins. Protein Sci 2002 11 2247 2256 12192080 10.1110/ps.0204202
Mucignat-Caretta C Modulation of exploratory behavior in female mice by protein-borne male urinary molecules. J Chem Ecol 2002 28 1853 1863 12449511 10.1023/A:1020521420271
Shaw PH Held WA Hastie ND The gene family for major urinary proteins: expression in several secretory tissues of the mouse. Cell 1983 32 755 761 6831559 10.1016/0092-8674(83)90061-2
Shahan K Denaro M Gilmartin M Shi Y Derman E Expression of six mouse major urinary protein genes in the mammary, parotid, sublingual, submaxillary, and lachrymal glands and in the liver. Mol Cell Biol 1987 7 1947 1954 3600653
Tullis KM Krebs CJ Leung JY Robins DM The regulator of sex-limitation gene, rsl, enforces male-specific liver gene expression by negative regulation. Endocrinology 2003 144 1854 1860 12697692 10.1210/en.2002-0190
Schmidt JV Matteson PG Jones BK Guan XJ Tilghman SM The Dlk1 and Gtl2 genes are linked and reciprocally imprinted. Genes Dev 2000 14 1997 2002 10950864
Takada S Paulsen M Tevendale M Tsai CE Kelsey G Cattanach BM Ferguson-Smith AC Epigenetic analysis of the Dlk1-Gtl2 imprinted domain on mouse chromosome 12: implications for imprinting control from comparison with Igf2-H19. Hum Mol Genet 2002 11 77 86 11773001 10.1093/hmg/11.1.77
Smith GH Mammary cancer and epithelial stem cells: a problem or a solution? Breast Cancer Res 2002 4 47 50 11879561 10.1186/bcr420
Welm BE Tepera SB Venezia T Graubert TA Rosen JM Goodell MA Sca-1(pos) cells in the mouse mammary gland represent an enriched progenitor cell population. Dev Biol 2002 245 42 56 11969254 10.1006/dbio.2002.0625
Zhang L Wang L Ravindranathan A Miles MF A new algorithm for analysis of oligonucleotide arrays: application to expression profiling in mouse brain regions. J Mol Biol 2002 317 225 235 11902839 10.1006/jmbi.2001.5350
Liu WM Mei R Di X Ryder TB Hubbell E Dee S Webster TA Harrington CA Ho MH Baid J Analysis of high density expression microarrays with signed-rank call algorithms. Bioinformatics 2002 18 1593 1599 12490443 10.1093/bioinformatics/18.12.1593
Li C Wong WH Model-based analysis of oligonucleotide arrays: model validation, design issues and standard error application. Genome Biol 2001 2 research0032.1 0032.11 11532216 10.1186/gb-2001-2-8-research0032
Utsumi M Ohno K Kawasaki Y Tamura M Kubo T Tohyama M Expression of major urinary protein genes in the nasal glands associated with general olfaction. J Neurobiol 1999 39 227 236 10235677 10.1002/(SICI)1097-4695(199905)39:2<227::AID-NEU7>3.0.CO;2-4
Mucignat-Caretta C Caretta A Cavaggioni A Acceleration of puberty onset in female mice by male urinary proteins. J Physiol 1995 486 517 522 7473215
Cavaggioni A Mucignat C Tirindelli R Pheromone signalling in the mouse: role of urinary proteins and vomeronasal organ. Arch Ital Biol 1999 137 193 200 10349497
Norstedt G Palmiter R Secretory rhythm of growth hormone regulates sexual differentiation of mouse liver. Cell 1984 36 805 812 6323022 10.1016/0092-8674(84)90030-8
Laborda J Sausville EA Hoffman T Notario V dlk, a putative mammalian homeotic gene differentially expressed in small cell lung carcinoma and neuroendocrine tumor cell line. J Biol Chem 1993 268 3817 3820 8095043
Smas CM Sul HS Pref-1, a protein containing EGF-like repeats, inhibits adipocyte differentiation. Cell 1993 73 725 734 8500166 10.1016/0092-8674(93)90252-L
Wylie AA Murphy SK Orton TC Jirtle RL Novel imprinted DLK1/GTL2 domain on human chromosome 14 contains motifs that mimic those implicated in IGF2/H19 regulation. Genome Res 2000 10 1711 1718 11076856 10.1101/gr.161600
Gunther EJ Belka GK Wertheim GB Wang J Hartman JL Boxer RB Chodosh LA A novel doxycycline-inducible system for the transgenic analysis of mammary gland biology. FASEB J 2002 16 283 292 11874978 10.1096/fj.01-0551com
Marquis ST Rajan JV Wynshaw-Boris A Xu J Yin GY Abel KJ Weber BL Chodosh LA The developmental pattern of Brca1 expression implies a role in differentiation of the breast and other tissues. Nat Genet 1995 11 17 26 7550308 10.1038/ng0995-17
D'Cruz CM Moody SE Master SR Hartman JL Keiper EA Imielinski MB Cox JD Wang JY Ha SI Keister BA Persistent parity-induced changes in growth factors, TGF-beta3, and differentiation in the rodent mammary gland. Mol Endocrinol 2002 16 2034 2051 12198241 10.1210/me.2002-0073
ChipStat software
Significance Analysis of Microarrays
DNA-Chip Analyzer (dChip)
| 15693950 | PMC551541 | CC BY | 2021-01-04 16:05:37 | no | Genome Biol. 2005 Jan 14; 6(2):R21 | latin-1 | Genome Biol | 2,005 | 10.1186/gb-2005-6-2-r21 | oa_comm |
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-171570748510.1186/1465-9921-6-17ResearchChronic pneumonia with Pseudomonas aeruginosa and impaired alveolar fluid clearance Boyer Sophie [email protected] Karine [email protected] Florence [email protected] Marie Odile [email protected] Eric [email protected] Thierry [email protected] Xavier [email protected] Benoit P [email protected] Laboratoire de recherche en Pathologie Infectieuse, EA 2689. Faculté de Médecine de Lille, 59031 Lille Cedex, France2 Laboratoire de Bactériologie; Hôpital Calmette, CHRU de Lille, Lille, France3 Laboratoire de Biophysique, CHRU, Lille, France4 Laboratoire d'anatomo-pathologie, CHRU Lille, France2005 11 2 2005 6 1 17 17 21 10 2004 11 2 2005 Copyright © 2005 Boyer et al; licensee BioMed Central Ltd.2005Boyer 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
While the functional consequences of acute pulmonary infections are widely documented, few studies focused on chronic pneumonia. We evaluated the consequences of chronic Pseudomonas lung infection on alveolar function.
Methods
P. aeruginosa, included in agar beads, was instilled intratracheally in Sprague Dawley rats. Analysis was performed from day 2 to 21, a control group received only sterile agar beads. Alveolar-capillary barrier permeability, lung liquid clearance (LLC) and distal alveolar fluid clearance (DAFC) were measured using a vascular (131I-Albumin) and an alveolar tracer (125I-Albumin).
Results
The increase in permeability and LLC peaked on the second day, to return to baseline on the fifth. DAFC increased independently of TNF-α or endogenous catecholamine production. Despite the persistence of the pathogen within the alveoli, DAFC returned to baseline on the 5th day. Stimulation with terbutaline failed to increase DAFC. Eradication of the pathogen with ceftazidime did not restore DAFC response.
Conclusions
From these results, we observe an adequate initial alveolar response to increased permeability with an increase of DAFC. However, DAFC increase does not persist after the 5th day and remains unresponsive to stimulation. This impairment of DAFC may partly explain the higher susceptibility of chronically infected patients to subsequent lung injury.
==== Body
Introduction
Pseudomonas aeruginosa is a Gram negative bacteria producing a wide array of virulence factors frequently responsible for chronic airway infections in cystic fibrosis (CF) or chronic obstructive pneumonia disease (COPD) patients, as well as acute nosocomial airway infections in intensive care units [1-3].
In acute P. aeruginosa pneumonia, the functional consequences, and particularly lung fluid movements, have been studied extensively. Lung fluid balance is the result of fluid movements following active ion transport by functional alveolar cells, and permeability of the alveolar capillary barrier. In P. aeruginosa-induced acute lung injury (ALI), distal airspace fluid clearance (DAFC) is typically increased at 24 hours through a TNF-α pathway [4]. Studies have also shown that the capacity of maintaining alveolar active fluid transport is correlated with patient outcome in ALI [5,6]. Lung liquid clearance (LLC) is another functional marker reflecting the capacity of the lung to evacuate fluid instilled in the alveoli outside the lung, LLC involves DAFC, epithelial and endothelial permeabilities [7]. We previously showed that, even though DAFC is upregulated, LLC is decreased at both 4 and 24 hours in ALI [7] reflecting a major endothelial injury overwhelming the alveolar response.
In chronic infection, these functional consequences on lung fluid balance are less clear. In the 70's, Cash developed an experimental model of chronic pneumonia by intra tracheal injection of P. aeruginosa embedded in agar beads [8]. Most of the work performed with this model has focused on immunological, inflammatory, or nutritional aspects [9-12]. To the best of our knowledge, no previous work has tried to evaluate alveolar permeability and lung fluid transport in P. aeruginosa chronic lung infection. In order to elucidate these functional aspects we studied lung fluid transport in an experimental model of chronic P. aeruginosa lung infection in the rat. After the validation of the experimental model, we studied alveolar function: alveolar-capillary barrier permeability, lung liquid clearance, distal airspace fluid clearance and its pharmacologic stimulation.
Materials and Methods
Animals
Specific pathogen-free Sprague Dawley rats (n = 280) (230–270 g), (Depre, St Doulchard, France) were housed in the Lille University Animal Care Facility and allowed food and water ad lib. All experiments were performed with approval of the Lille Institutional Animal Care and Use Committee.
Preparation of the bacterial inoculum
The methodology was adapted from Cash et al [8]. Briefly, P. aeruginosa (PAO1 strain) was incubated in 125 ml of tryptic soy broth at 37°C in a rotating shaking water bath for 8 hours. The culture was then washed twice, and resuspended in phosphate-buffered saline. The resulting bacterial suspension was 1 × 109 CFU/ml. A sample of 1 mL of this suspension was mixed in agarose and mineral oil (Sigma Diagnoses, St Louis, USA) at 56°C. The resulting oil-agar emulsion was cooled to obtain agar beads. Dilutions of the final suspension were cultured to determine the size of the final inoculum.
Experimental infection
Under a short general anesthesia with ether (Mallinkrodt, Paris, France), with sterile surgical conditions, a small midline incision was made on the neck ventral surface after swabbing it with ethanol. The trachea was exposed by blunt dissection. Using a 28-gauge needle, 0.1 mL of agar beads followed by 0.5 mL of air were inoculated intra-tracheally.
Quantitative bacteriological analysis
After exsanguination of the animal, the lungs were isolated and homogenized in 2 mL of sterile isotonic saline. Bacterial culture after serial dilutions was performed and bacterial colonies counted after 12 h at 37°C.
Antimicrobial therapy
In a subgroup of animals, ceftazidime (GlaxoSmithKline, Marly-le-Roi, France), 100 mg/kg, was administered in the peritoneal cavity every 8 hours during 72 hours. Lungs were harvested, homogenized and cultures were performed to confirm bacterial eradication. Serum ceftazidime levels were measured in HPLC.
Broncho-alveolar lavage (BAL)
Broncho-alveolar lavage (BAL) was performed by cannulating the trachea. Lungs from each experimental group were lavaged with a total of 20 ml in 5-ml aliquots of PBS with EDTA (3 mM). BAL fluid samples were filtered and immediately frozen at -80°C. A cell count was performed directly. Cellular monolayers were prepared with a cytocentrifuge and stained with Wright-Giemsa stain. Cellular morphotype differential was obtained by counting 200 cells/sample and expressing each type of cell as a percentage of the total number counted. Protein concentration in the BAL was measured with an automated analyzer (Hitachi 917, Japan).
Histological study
After a vascular flushing with sterile isotonic saline through the pulmonary artery, the lungs were removed. Samples were fixed by intratracheal instillation of paraformaldehyde 10 %. Samples were included in paraffin and sections of 5 μm were realized. Analysis was performed after coloration with Hematoxyline-Eosine-Safran (Zeiss, LEO 906).
Serum and BAL TNF-α measurement
Levels of tumor necrosis factor α (TNF-α), in the serum, and the BAL fluid, were determined by use of commercial immunoassay kits (ELISA) specific for rat cytokines (Quantikine Murine rat TNFα, R&D Systems, Abingdon OX, UK). The reading was performed with a microplate reader Digiscan (Spectracount Packard Instrument Company; Meriden CT USA).
BAL and serum measurement of epinephrine and nor-epinephrine
Blood and broncho-alveolar lavage fluid were collected on heparin/Na-metabisulfite coated tubes. The samples were centrifuged (2500 g, 4°C), supernatants were frozen (-80°C).
Catecholamines are specifically fixed on alumina (pH = 8.7), the eluent is analyzed with an inversed phase H.P.L.C (Coulochem II ESA). The results are expressed in μg/L.
Functional study
Surgical preparation
Sprague-Dawley male rats were anesthetized with pentobarbital (Sanofi, Libourne, France). A catheter (PE-50) was inserted into the left carotid artery in order to monitor systemic arterial pressure (Acqknowledge Software v 3.7.1, Biopac systems, Santa Barbara, CA, USA) and obtain blood samples. An endotracheal tube (PE-220) was inserted through a tracheostomy. The rats were ventilated with a constant volume pump (Harvard Apparatus, South Natick, MA) with an inspired O2 fraction of 1.0, a peak airway pressure of 8–12 cmH2O, and a positive end expiratory pressure of 2 cmH2O. The animals were placed in left decubitus position until the end of the protocol. The body temperature was maintained at 37°C.
Preparation of the instillate
The test solution, used for alveolar instillation, was prepared as follows : briefly, a 5% bovine albumin solution was prepared using Ringer lactate and was adjusted with NaCl to be isoosmolar with the rat circulating plasma [13,14]. A sample of the instilled solution was saved for total protein measurement, and water to dry weight ratio measurements. In different experimental groups, terbutaline (10-4 M) (Sigma Aldrich, St Quentin Fallavier, France) was added to the instillate or injected intra-peritoneally to the animals.
General Protocol
For all ventilated rats experiments, the following general protocol was used. After the surgical preparation, heart rate and blood pressure were allowed to stabilize for 1 hour. To calculate the flux of plasma protein into the lung interstitium, a vascular tracer, 1 μCi of 131I-labeled human albumin, was injected into the bloodstream [14,15]. 131I-HSA was prepared in our institution according to a standardized technique. Administration of the instillate (3 ml/kg) was performed into the left lung over a 2-min period, using a 1-ml syringe and polypropylene tube (PE 50, Intramedic, Becton Dickinson, Sparks, MD, USA)[13].
One hour after the beginning of the alveolar instillation, the rat was exanguinated. The lungs were removed, and fluid from the distal airspaces was obtained (aspirate). The total protein concentration and the radioactivity of the liquid sampled were measured. Right and left lungs were homogenized separately for water to dry weight ratio measurements and radioactivity counts.
Measurements
• Hemodynamics, pulmonary gas exchange, and protein concentration
Systemic arterial pressure and airway pressures were measured continuously. Arterial blood gases were measured at one hour intervals. The arterial PO2 was used to quantify the oxygenation deficit [13,14]. Samples from instillated protein solution, final distal airspace fluid, and from initial and final blood were collected to measure total protein concentration with an automated analyzer (Hitachi 917, Japan).
• Albumin flux across endothelial and epithelial barriers
The flux of albumin across the lung endothelial and epithelial barriers was used to evaluate the permeability. This method requires measurement of the vascular protein tracer, 131I-albumin, in the alveolar and extravascular spaces of the lungs. Endothelial permeability was assessed by measuring the ratio of 131-iodine radioactivity in the aspirate to the radioactivity obtained in the plasma (Asp/plasma), it reflects the leak of the vascular tracer in the alveolar compartment. We estimated the quantity of plasma that entered the instilled lungs by measuring the transfer of the vascular protein tracer, 131I-albumin, into the extravascular spaces of the instilled lung using the equation of plasma equivalents previously described [7,13,14].
• Extravascular lung water (EVLW) and lung liquid clearance (LLC)
The EVLW was estimated by gravimetry: 300 μL of the lung homogenate were weighed, to determine the wet weight, and dessicated at 45°C during 7 days, to obtain the dry weight. The blood fraction was calculated from the homogenate hemoglobin supernatant content. The wet to dry weight ratio (W/D) was estimated using the values of the right lung which was not instilled [7,14,16]. Lung liquid clearance was calculated as previously described [7].
• Distal Airspace Fluid Clearance (DAFC):
A change of native bovine albumin concentration over the study period (1 h) was used to measure alveolar fluid movement. DAFC was calculated from the ratio of the final unlabeled alveolar protein concentration, compared to the initial instilled alveolar protein concentration.
Experimental groups
15 experimental groups were constituted for the study:
- A control group (Ctr), which received an intratracheal instillation of sterile saline at the beginning of the protocol
- 7 Sterile groups (St) received an intratracheal instillation of sterile beads and were studied at different days after inoculation: St 1, St 2, St 5, St 8, St 15, St 21 and St 28.
- 7 Pneumonic groups (Pn) received an intratracheal instillation of Pseudomonas containing beads and were studied at different days after inoculation: Pn 1, Pn 2, Pn 5, Pn 8, Pn 15, Pn 21, Pn 28.
Statistical analysis
Comparisons between two groups were made using an unpaired, two tailed Student's t-test. Comparisons between more than two groups were made using a one way analysis of variance with post hoc test for multiple comparisons. A value of p < 0.05 was considered as significant. The data are expressed as means ± SD.
Results
Pseudomonas beads instillation is associated with the development of a chronic infection
Clinically, a major weight loss was observed from the second day in P. aeruginosa beads infected animals compared to the sterile beads groups (Figure 1). 5% of the infected animals died within the first 48 hours after inoculation, none did in the sterile groups.
Figure 1 Evolution of animals' weight during the four weeks of the analysis. An initial weight loss is observed for the infected animals compared to the sterile beads group. Footnote: Data are mean (± SD). Comparisons between groups were made using analysis of variance. *p < 0.05 vs the Pn group. Pn: Pneumonic animals, St: animals which received only sterile beads
Prior to the instillation, the size of the inoculum was 7.9 105 ± 1.5 105 CFU/mL. Lung bacterial load reached a peak on the second day of the infection; from the 5th day, a progressive decrease occurred to finally remain steady between the 15th day (8.25 ± 5.2 104 CFU/mL) and the 3rd week (1.67 ± 1.63 105 CFU/mL).
Total broncho-alveolar lavage (BAL) cells slightly increased in the sterile beads group, the difference was however not statistically significant compared to the control group, the analysis showed that the number of cells peaked on the second day and was constituted, at that time, of 25% polymorphonuclear cells and 75% macrophages. The results were not statistically different over time and therefore pooled in Table 1. In the infected groups, alveolar cellularity was maximum on the 2nd day mostly polymorphonuclear's neutrophils (PMN). From the 8th day, the relative number of PMN progressively decreased as alveolar macrophages increased. All the results are summarized in Table 1.
Table 1 Analysis of the bronchoalveolar lavage All the animals who received sterile beads were included in the sterile group and compared to the control and pneumonic groups at respectively 2, 5, 8, 15 and 21 days post instillation.
Total cells (× 106)/mL PMNs (%) Macrophages (%)
Ctr 0.4 ± 0.1 0.5 ± 0.4 98.5 ± 0.5
St 3.2 ± 0.6 5.6 ± 4.4 92.9 ± 4.4
Pn 2 10.5 ± 2.9* 79.8 ± 5.2* 19.0 ± 4.6*
Pn 5 7.9 ± 1.7* 19.0 ± 8.0 79.3 ± 8.4
Pn 8 4.9 ± 1.0 3.8 ± 0.7 95.5 ± 1.0
Pn 15 4.0 ± 0.8 1.2 ± 0.6 98.8 ± 0.6
Pn 21 4.9 ± 1.8 2.0 ± 0.6 97.2 ± 0.6
Footnote: Data are mean (± SD). Comparisons between groups were made using analysis of variance with post hoc for multiple comparisons. *p < 0.05 vs the other groups.
PMNs: Polymorphonuclear neutrophils, Ctr: Control group, St: Sterile group, Pn: Pneumonic group
Histologically, in the infected groups, from the 2nd day, large numbers of PMNs were observed, mostly centered on the alveoli (Figure 2C–D). Agar beads were clearly observed in the Pn2 group (Figure 2D). With time, increased extracellular material became more prominent (Figure 2G–L). The lung architecture of animals inoculated with sterile beads remained strictly normal (Figure 2A–B).
Figure 2 Histological analysis of the different groups, controls and sterile beads instilled animals are compared to pneumonic rats from the second to the 21st day post instillation. Coloration was performed with Hematoxyline-Eosine-Safran. A: Control group; B: Sterile beads; C-D: Pneumonia on the 2nd day (the arrow on panel D underlines infected beads); E-F: Pneumonia on the 5th day; G-H: Pneumonia on the 8th day; I-J: Pneumonia on the 15th day; K-L: Pneumonia on the 21st day.
A transient increase of alveolar-capillary barrier permeability is observed on the second day post infection
No variation in permeability or clearance was observed between St groups, so all the results were included in a single group (St) for the analysis (at least 5 animals were included in each time point). Alveolar-capillary barrier permeability, evaluated by the leakage of the vascular marker into the alveoli (Asp/plasma ratio), was increased in infected animals on the second day compared to the control group (0.59 ± 0.08 vs 0.11 ± 0.02). This ratio came back to control values from the fifth to the 28th day. In the St group a moderate but significant increase of the Asp/plasma ratio was observed throughout the study (0.31 ± 0.04).
Both lung liquid clearance and DAFC increased on the 2nd day post infection; DAFC increase is not related to a TNF-α or catecholamine dependent mechanism
• Extra-vascular lung water and Lung liquid clearance (LLC)
As shown in Table 2, no difference in wet to dry lung weight ratio was observed between the groups. LLC increased in the pneumonic group on the second day after the infection (p = 0.02) to return to baseline on the 5th day. A moderate but not statistically significant increase was observed in the Pn15 group (p = 0.13).
Table 2 Lung liquid clearance (LLC) and lung wet to dry weight ratio (W/D). LLC increases on the second day post instillation and returns to baseline on the fifth day. W/D remains constant over time.
W/D LLC (%)
Ctr 4.33 ± 0.87 22.24 ± 3.65
St 4.29 ± 0.24 36.53 ± 4.95
Pn 2 4.66 ± 0.51 45.51 ± 4.26 *
Pn 5 4.03 ± 0.27 20.99 ± 5.94
Pn 8 3.47 ± 0.81 23.01 ± 2.80
Pn 15 3.92 ± 0.29 36.21 ± 8.23
Pn 21 4.31 ± 0.07 22.37 ± 2.56
Footnote: Data are mean (± SD). Comparisons between groups were made using analysis of variance with post hoc for multiple comparisons. *p < 0,05 vs the other groups.
Ctr: Control group, St: Sterile group, Pn: Pneumonic groups from the 2nd to the 21st days.
• Distal alveolar fluid clearance
Distal alveolar fluid clearance increased in the Pn2 group (Figure 3). This ratio decreased back to baseline on the 5th day and remained comparable to both the St and the Ctr groups.
Figure 3 Evolution of the DAFC over time in sterile and infected beads injected groups. We observe an increase on the 2nd day post infection, the clearance returns to a basal level on the 5th day. Footnote: Data are mean (± SD). Comparisons between groups were made using analysis of variance with post hoc for multiple comparisons. *p < 0.05 vs the other groups. DAFC: distal alveolar fluid clearance, Ctr: Control group, St: sterile beads injected group, Pn: pneumonic groups from the 2nd to the 21st day.
We tested whether the increase in DAFC observed at 48 hours was related to a TNF-α or a catecholamine dependent mechanism. No TNF-α was detected between the 2nd and 21st days in the serum or the alveolar compartment. Similarly, neither epinephrine nor nor-epinephrine could be detected in the alveolar compartment at 48 hours. The levels recovered in the plasma were comparable between control and pneumonic animals on the 2nd and the 5th days (Table 3).
Table 3 Plasma catecholamines measurement Plasma catecholamines were measured in pneumonic animals on the 2nd and the 5th day post instillation compared to the control group. No statistically significant difference could be observed.
Ctr Pn2 Pn5
Epinephrine (μg/L) 8.5 ± 2.1 11.2 ± 4.9 14.2 ± 3.5
Norepinephrine (μg/L) 5.8 ± 0.8 7.0 ± 2.2 8.2 ± 1.9
Footnote: Data are mean (± SD). Comparisons between groups were made using analysis of variance with post hoc for multiple comparisons. Ctr: Control group, St: Sterile group, Pn: Pneumonic groups.
Distal airspace fluid clearance cannot be stimulated on the 5th day post infection even after bacterial eradication
Even though DAFC returned to baseline values on the fifth day post infection, alveolar function was not normal in these chronically infected animals. First of all, since bacterial load persisted in the alveoli at least a modest increase of DAFC would have been expected in response to this stimulus. This absence of the expected response led us to test the DAFC response, in each group, to well known pharmacological stimuli.
• Terbutaline
The administration of terbutaline is associated with an increase in DAFC in controls. Stimulation with terbutaline intratracheally could not increase DAFC on the 5th day post infection, the intraperitoneal injection also failed to increase DAFC (Figure 4).
Figure 4 Evaluation of DAFC in the control compared to the pneumonic groups on the fifth day post instillation at baseline and after stimulation with terbutaline. A last group received terbutaline after bacterial eradication with ceftazidime administered intraperitoneally. None of the pneumonic groups could increase DAFC after either stimulation or bacterial eradication. Footnote: Data are mean (± SD). Comparisons between groups were made using analysis of variance with post hoc for multiple comparisons. *p < 0.05, statistically different from the control group. DAFC: distal alveolar fluid clearance. Ctr: Control group, Terbut: Control instilled with terbutaline (10-4 M), St + Terbut: Sterile beads instilled with terbutaline (10-4 M), Cefta + Terbut: Control group treated with ceftazidime (100 mg/kg/8 h for 72 h) and instilled with terbutaline (10-4 M), Pn5: Pneumonic group on the 5th day, Pn5 + Terbut: Pneumonic group on the 5th day instilled with terbutaline (10-4 M), Pn5 + Terbut IP: Pneumonic group on the 5th day with an intraperitoneal injection of terbutaline, Pn5 + Cefta+ Terbut: Pneumonic group on the 5th day treated with ceftazidime (100 mg/kg/8 h for 72 h) instilled with terbutaline (10-4 M).
• Terbutaline after bacterial eradication
In order to eliminate the possibility of a direct bacterial effect inhibiting the expected response in the chronically infected animals, we performed a comparable stimulation with terbutaline on 10 animals treated with ceftazidime initiated 24 hours after the infection. On the 5th day, all lungs were sterilized and measurement of ceftazidime levels showed a steady state level at 46.3 ± 4.8 μg/mL. However, even after the eradication of the pathogen, DAFC remained unresponsive to beta-adrenergic stimulation (Figure 4).
Discussion
In our study we validated an experimental model allowing us to explore alveolar function in chronic P. aeruginosa lung infection through measurements of lung liquid movements. In this model of chronic P. aeruginosa lung infection, after observing an initial increase of both alveolar permeability and lung fluid movements, we characterized an impairment of DAFC where, even though DAFC returned to baseline, it remained unresponsive to pharmacological stimuli.
In the first part of our work, we validated, on several parameters, the chronic infection model previously described by Cash et al [8]. After reaching a peak on the second day of the infection and decreasing from the 5th to the 15th day, lung bacterial load persisted for 3 weeks. These results, as well as the analysis of the BAL and the histological features, are consistent with the literature [8,11,17,18].
Since, in this model, P. aeruginosa is associated with agar beads, we performed, as control groups, instillation of sterile agar beads. Sterile agar bead instilled rats did not show any evidence of weight loss and although they did present an increase in BAL cellularity, there were no PMN's except a slight increase on the second day which failed to reach a statistical significance (data not shown). This result is consistent with the literature, Nacucchio et al showed that agar beads alone could not reproduce the same level of injury than P. aeruginosa in agar beads [19].
From this first part of our work, we concluded that the model of chronic infection with P. aeruginosa is adequate, based on clinical, bacteriological, cytological and histological data.
Although a clinical study has reported increased lung permeability in COPD patients infected by P. aeruginosa [20], few studies have focused on the consequences of chronic lung infection on alveolar function and particularly fluid movements. In our study, lung fluid movements were maximal on the 2nd day post infection. We observed an increase of alveolar-capillary barrier permeability, DAFC and overall lung liquid clearance. A normal lung wet to dry weight ratio was a consequence of this adequate alveolar response. This contrasts sharply with the data we obtained in an acute lung injury model where LLC dramatically decreased and W/D weight ratio increased at 4 and 24 hours after Pseudomonas instillation [7].
In our chronic model, following the increase in both permeability and lung liquid clearance, we observed an improvement in permeability with a return to baseline of these 2 parameters on the 5th day.
The St group presented a moderate increase in permeability (Asp/plasma ratio: 0.31 ± 0.04), it has previously been reported that agar beads could alone be responsible for a moderate increase in permeability [19]. However, taking into account the association of the other parameters validating the model (clinical, bacteriological, cytological and histological), this effect does not challenge the model.
Our results showed an increase of the DAFC at 48 hours post infection. In acute lung injury, the initial alveolar response is usually towards an increase of DAFC which many authors have documented in septic shock [21], or after endotoxin administration [22]. In septic shock, this increase was related to the release of endogenous catecholamines. In acute P. aeruginosa pneumonia, increased DAFC can be related to either Pseudomonas exoproducts [15] or to a TNF-α dependent mechanism during the first 24 hours of the infection [4]. We tested in our model whether TNF-α or catecholamines could explain our results. TNF-α was not detectable and systemic endogenous epinephrine or nor-epinephrine not different from controls on the 2nd or the 5th day. TNF-α is produced during the early phase of pneumonia, and its short half life probably explains the absence of detectable levels at 48 hours. A dynamic evaluation of TNF-α production with serial samples or antibody neutralization experiments would be helpful to precisely study the role of TNF-α. We therefore did not rule out that TNF-α may have triggered an inflammatory response which could be responsible for the increased DAFC. Other potential mechanisms such as Transforming Growth Factor β remain to be investigated [23].
Surprisingly, on the fifth day, DAFC returned to baseline along with the improvement in permeability. Although it is logical to see an improvement in permeability, consistent with a decrease of the bacterial burden and an adequate host response, DAFC was expected to remain increased. The persisting presence of the pathogen in the alveoli and many factors only related to its presence would normally lead to a persistent increase of DAFC [15]. We therefore decided to evaluate if a normal increase in DAFC could be elicited on the 5th day post infection in response to known pharmacological stimuli [24,25]. In the normal lung, intra-alveolar administration of terbutaline generates a DAFC increase of approximately 30% [26]. We observed comparable results in our study in control animals as well as animals which received only sterile beads. In our model, on the 5th day, terbutaline intratrachéal administration did not change DAFC. However the lack of effect may be due to airway inflammation and an inability to adequately deliver the drug, we therefore decided to use intraperitoneal administration with the same agent. Our results also show the absence of DAFC increase. We then hypothesized that the absence of response to the stimulation might be related to the persistence of the pathogen in the alveoli. To test this hypothesis, we injected the animals with ceftazidime to sterilize the lungs on the 5th day. Sterilization was achieved but failed to restore DAFC stimulation with terbutaline. To explain this impairment of DAFC, different hypotheses still remain to be investigated concerning these agonist's receptors and their regulation. Other authors have shown in different situations that either an internalization or a decrease of affinity of the receptors [27] could be observed. Another hypothesis could be a lost of sensitization through a decrease of the AMPc dependent signal transmission. It was shown, in vitro, on tracheal cells that a continuous or repeated exposure to isoproterenol could lead to a lost of sensitization [28].
If this unresponsiveness exists in patients, the absence of an adapted DAFC response in chronic lung infection could lead to major damage in the presence of any new lung injury. Although chronic lung infection has not been isolated, per se, as an aggravating factor associated to mortality in COPD patients admitted in an intensive care unit, a pre-existing underlying pathology is associated with a worsening of the prognosis in community and nosocomial pneumonia [29,30]. DAFC impairment might be part of the answer to this effect of underlying disease.
In conclusion, chronic P. aeruginosa pneumonia is characterized initially at 48 hours by an increased alveolar-capillary barrier permeability and an adapted host response with an increased DAFC and LLC preserving a normal lung wet to dry weight ratio. On the 5th day, DAFC remains non responsive to pharmacological stimulation even after bacterial elimination. This impairment of DAFC could represent one of the factors responsible for the increased susceptibility of chronically infected patients to other respiratory insults.
Authors' contributions
SB and FA were responsible for the acquisition of the data. KF and MOH made substantial contributions to the drafting of the manuscript and the analysis of the data. TP performed the radioactive labelling of the albumin (I131). EK was involved in the revision of the manuscript and the English editing. XL performed all the histological analysis. BG was involved in the acquisition of the data, the design and the conception of the study as well as the drafting of the article. All the authors read and approved the final manuscript.
==== Refs
Gibson RL Burns JL Ramsey BW Pathophysiology and management of pulmonary infections in cystic fibrosis Am J Respir Crit Care Med 2003 168 918 951 14555458 10.1164/rccm.200304-505SO
Chastre J Fagon JY Ventilator-associated pneumonia Am J Respir Crit Care Med 2002 165 867 903 11934711
Fagon JY Chastre J Domart Y Trouillet JL Gibert C Mortality due to ventilated-associated pneumonia or colonization with Pseudomonas or Acinetobacter species : assessment by quantitative culture of samples obtained by a protected specimen brush Clin Infect Dis 1996 23 538 542 8879777
Rezaiguia S Garat C Delclaux C Fleury J Legrand P Matthay MA Jayr C Acute bacterial pneumonia in rats increases alveolar epithelial fluid clearance by a tumor necrosis factor-alpha-dependent mechanism J Clin Invest 1997 99 325 335 9006001
Matthay MA Wiener-Kronish JP Intact epithelial barrier function is critical for the resolution of alveolar edema in humans Am Rev Respir Dis 1990 142 1250 1257 2252240
Ware LB Matthay MA Alveolar fluid clearance is impaired in the majority of patients with acute lung injury and the acute respiratory distress syndrome Am J Respir Crit Care Med 2001 163 1376 1383 11371404
Viget N Guery B Ader F Nevière R Alfandari S Creusy C Roussel-Delvallez M Foucher C Mason CM Beaucaire G Pittet JF Keratinocyte Growth Factor protects against Pseudomonas aeruginosa-induced lung injury Am J Physiol Lung Cell Mol Physiol 2000 279 L1199 L1209 11076810
Cash HA Woods DE McCullough B Johanson WGJ Bass JA A rat model of chronic respiratory infection with Pseudomonas aeruginosa Am Rev Respir Dis 1979 119 453 459 109021
Amano H Oishi K Sonoda F Senba M Wada A Nakagawa H Nagatake T Role of cytokine-induced neutrophil chemoattractant-2 (CINC-2) alpha in a rat model of chronic bronchopulmonary infections with Pseudomonas aeruginosa Cytokine 2000 12 1662 1668 11052817 10.1006/cyto.2000.0771
Morissette C Skamene E Gervais F Endobronchial inflammation following Pseudomonas aeruginosa infection in resistant and susceptible strains of mice Infect Immun 1995 63 1718 1724 7729877
van Heeckeren AM Tscheikuna J Walenga RW Konstan MW Davis PB Erokwu B Haxhiu MA Ferkol TW Effect of Pseudomonas infection on weight loss, lung mechanics, and cytokines in mice Am J Respir Crit Care Med 2000 161 271 279 10619831
van Heeckeren AM Schluchter MD Murine models of chronic Pseudomonas aeruginosa lung infection Lab Anim 2002 36 291 312 12144741 10.1258/002367702320162405
McElroy MC Wiener-Kronish JP Miyazaki H Sawa T Modelska K Dobbs LG Pittet JF Nitric oxide attenuates lung endothelial injury caused by sublethal hyperoxia in rats Am J Physiol 1997 272 L631 L638 9142935
Modelska K Matthay MA McElroy MC Pittet JF Upregulation of alveolar liquid clearance after fluid resuscitation for hemorrhagic shock in rats Am J Physiol 1997 273 L305 L314 9277441
Pittet J Hashimoto S Pian M McElroy MC Nitenberg G Wiener-Kronish JP Exotoxin A stimulates fluid reabsorption from distal airspaces of lung in anesthetized rats Am J Physiol 1996 270 L232 L241 8779992
Jayr C Garat C Meignan M Pittet J Harf A Matthay MA Basal and stimulated alveolar and lung liquid clearance in ventilated, anesthetized rats J Appl Physiol 1994 76 2636 2642 7928894 10.1063/1.357560
Graham LM Vasil A Vasil ML Voelkel NF Stenmark KR Decreased pulmonary vasoreactivity in an animal model of chronic Pseudomonas pneumonia Am Rev Respir Dis 1990 142 221 229 2368972
Johansen HK Espersen F Pedersen SS Hougen HP Rygaard J Hoiby N Chronic Pseudomonas aeruginosa lung infection in normal and athymic rats APMIS 1993 101 207 225 8507458
Nacucchio MC Cerquetti MC Meiss RP Sordelli DO Short communication. Role of agar beads in the pathogenicity of Pseudomonas aeruginosa in the rat respiratory tract Pediatr Res 1984 18 295 296 6728563
Ishihara H Honda I Shimura S Sasaki H Takishima T Role of chronic Pseudomonas aeruginosa infection in airway mucosal permeability Chest 1991 100 1607 1613 1959404
Pittet J Wiener-Kronish JP McElroy MC Folkesson HG Matthay MA Stimulation of lung epithelial liquid clearance by endogenous release of catecholamines in septic shock in anesthetized rats J Clin Invest 1994 94 663 671 8040320
Garat C Rezaiguia S Meignan M D'Ortho MP Harf A Matthay MA Jayr C Alveolar endotoxin increases alveolar liquid clearance in rats J Appl Physiol 1995 79 2021 2028 8847269
Folkesson HG Pittet JF Nitenberg G Matthay MA Transforming growth factor-alpha increases alveolar liquid clearance in anesthetized ventilated rats Am J Physiol 1996 271 L236 L244 8770062
Sakuma T Folkesson HG Suzuki S Okaniwa G Fujimura S Matthay MA Beta-adrenergic agonist stimulated alveolar fluid clearance in ex vivo human and rat lungs Am J Respir Crit Care Med 1997 155 506 512 9032186
Crandall E Heming TA Palombo RL Goodman B Effects of terbutaline on sodium transport in isolated perfused rat lung J Appl Physiol 1986 60 289 294 3944038
Sakuma T Okaniwa G Nakada T Nishimura T Fujimura S Matthay MA Alveolar fluid clearance in the resected human lung Am J Respir Crit Care Med 1994 150 305 310 8049807
Nishikawa M Mak JC Shirasaki H Harding SE Barnes PJ Long-term exposure to norepinephrine results in down-regulation and reduced mRNA expression of pulmonary beta-adrenergic receptors in guinea pigs Am J Respir Cell Mol Biol 1994 10 91 99 8292387
Kume H Takagi K Inhibition of beta-adrenergic desensitization by KCa channels in human trachealis Am J Respir Crit Care Med 1999 159 452 460 9927357
Georges H Leroy O Guery B Alfandari S Beaucaire G Predisposing factors for nosocomial pneumonia in patients receiving mechanical ventilation and requiring tracheotomy Chest 2000 118 767 774 10988201 10.1378/chest.118.3.767
Leroy O Devos P Guery B Georges H Vandenbussche C Coffinier C Thevenin D Beaucaire G Simplified prediction rule for prognosis of patients with severe community-acquired pneumonia in ICUs Chest 1999 116 157 165 10424520 10.1378/chest.116.1.157
| 15707485 | PMC551591 | CC BY | 2021-01-04 16:36:27 | no | Respir Res. 2005 Feb 11; 6(1):17 | utf-8 | Respir Res | 2,005 | 10.1186/1465-9921-6-17 | oa_comm |
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-191572370310.1186/1465-9921-6-19ReviewAnaphylatoxin C3a receptors in asthma Ali Hydar [email protected] Reynold A [email protected] Department of Pathology, School of Dental Medicine, University of Pennsylvania, 240 South 40th Street, Philadelphia, PA, 19104, USA2 Pulmonary Allergy and Critical Care Division, Department of Medicine, University of Pennsylvania, BRBII/III, 421 Curie Boulevard, Philadelphia PA 19104, USA2005 21 2 2005 6 1 19 19 10 2 2005 21 2 2005 Copyright © 2005 Ali and Panettieri; licensee BioMed Central Ltd.2005Ali and Panettieri; 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 complement system forms the central core of innate immunity but also mediates a variety of inflammatory responses. Anaphylatoxin C3a, which is generated as a byproduct of complement activation, has long been known to activate mast cells, basophils and eosinophils and to cause smooth muscle contraction. However, the role of C3a in the pathogenesis of allergic asthma remains unclear. In this review, we examine the role of C3a in promoting asthma. Following allergen challenge, C3a is generated in the lung of subjects with asthma but not healthy subjects. Furthermore, deficiency in C3a generation or in G protein coupled receptor for C3a abrogates allergen-induced responses in murine models of pulmonary inflammation and airway hyperresponsiveness. In addition, inhibition of complement activation or administration of small molecule inhibitors of C3a receptor after sensitization but before allergen challenge inhibits airway responses. At a cellular level, C3a stimulates robust mast cell degranulation that is greatly enhanced following cell-cell contact with airway smooth muscle (ASM) cells. Therefore, C3a likely plays an important role in asthma primarily by regulating mast cell-ASM cell interaction.
==== Body
Role of complement system in the development of asthma
Asthma, a complex airway inflammatory disease, is characterized by bronchoconstriction, airway hyperresponsiveness (AHR) and remodeling. Current consensus suggests that TH2 cytokine producing T cells, mast cells and ASM cells play central roles in the pathogenesis of asthma [1-7]. The complement system forms an important part of innate immunity against bacteria and other pathogens. As a system of 'pattern recognition molecules', foreign surface antigens and immune complexes initiate a proteolytic pathway leading to the formation of a lytic membrane attack complex. The anaphylatoxins C3a and C5a are released as byproducts of complement activation and modulate innate immunity. Accordingly, C5a is involved in a number of inflammatory diseases such as immune-complex-mediated lung injury and sepsis [8,9]. A role for C3a in innate or adaptive immunity, however, has only been recently recognized [10].
C3a levels are elevated in bronchoalveolar lavage (BAL) fluid after segmental allergen challenge in asthmatic but not healthy subjects [11-14]. Furthermore, plasma C3a is also elevated in acute exacerbations of asthma [11]. Additionally, single nucleotide polymorphisms in C3 and C3a receptor genes increases susceptibility to asthma [15]. Collectively, these findings suggest that C3a and the cognate G protein coupled receptor (C3aR) may play a role in the development of airway hyperresponsiveness (AHR) and inflammation.
C3a receptors in models of Airway Hyperresponsiveness
Studies with animal models provided compelling evidence for C3aR activation in the development of AHR and inflammation. Humbles et al., [12], showed that C3aR (-/-) mice in BALB/c strain are protected from AHR in response to aerosolized ovalbumin challenge following intraperitoneal sensitization with ovalbumin [12]. However, C3aR (-/-) mice developed normal airway inflammatory response with no difference in TH2 cytokine production and eosinophil recruitment in BAL when compared to wild-type mice. Additionally, guinea pigs with a natural defect in C3aR expression were also protected from AHR in response ovalbumin to challenge with no effect on airway inflammation [16]. These initial findings suggested that C3a modulates AHR, perhaps, via a direct action on airway smooth muscle cells [12,17,18].
Recent studies using C3aR (-/-) mice provided new insights on the role of C3a on both AHR and airway inflammation [19]. When sensitized intraperitoneally with extracts of Aspergillus fumigatous and challenged intranasally with allergen, these mice experienced substantial decreases in both AHR and airway eosinophilia relative to wild-type mice. Furthermore, BAL levels of TH2 cytokines (IL-4, IL-5, IL-13), IgE titres and mucous production were all significantly reduced in C3aR (-/-) mice. Allergen-challenged C3 (-/-) mice also display diminished AHR, lung eosinophilia and TH2 cytokine production when compared to wild-type mice [20]. These findings support a role of C3a receptors in the development of AHR and inflammation. However, the effect of C3aR on different phases of AHR models may depend on the nature of the allergen, method of sensitization and the strain of mice used.
C3a generation in asthma
Increased level of C3a in BAL of subjects with asthma implies a potential role for this apaphylatoxin in promoting airway inflammation. However, the cells responsible for C3a generation and the airway effector cells stimulated by C3a remain unknown. Plausibly, antibody generated during antigen sensitization may interact with allergen to activate the classical complement pathway. Additionally, airway epithelial cells and pulmonary macrophages secrete both C3 and several components of the alternate pathway of complement (factors B, H, and I and properdin) [21-23]. Thus, activation of alternative or the lectin pathway on the allergen may also lead to the generation of C3a. It is noteworthy that house dust mite protease, allergenic extracts of Aspergillus fumigatous and mast cell tryptase also activate the complement pathway directly [13,24-26]. Thus, combination of different pathways likely generates C3a in the airway of individuals with asthma (Figure 1).
Figure 1 Model for C3a generation in individuals with asthma. C3 may be secreted from pulmonary resident cells (e.g. epithelial cells and macrophages) or derived from plasma leakage. Antibody (IgG) present in the serum of sensitized individual can form a complex with allergen to activate complement via the classical pathway. Proteases derived from allergen or released from activated mast cells are able to cleave C3 to generate C3a. Activation of alternative or the lectin pathway on the allergen together with factors B, H, and I and properdin released resident cells may generate C3a.
C3a has little effect on allergen sensitization in models of AHR
Both antigen-presenting cells (APCs) and activated T cells express C3aR [27-30], raising the possibility C3a may regulate sensitization phase of allergic asthma. Kawamoto et al [31], recently used wild-type and C3aR-/- mice to characterize the immune response to C3a. Convincingly, C3aR deficiency had little effect on TH2 cytokine response to intraperitoneal ovalbumin sensitization. Furthermore, C3a had no effect on TH2 cytokine production in response to T cell receptor ligation. Further, Taube et al., [32] showed that administration of complement inhibitor in mice after sensitization but before allergen challenge prevented the development of AHR and blocked TH2 cytokine production and lung inflammation. Additionally, a small molecule antagonist of C3a receptor, when administered after sensitization but before challenge also caused significant inhibition of airway inflammation [33]. These findings suggest that the effect of C3a on the development of allergic AHR may not involve modulation of the sensitization phase of the disease.
Relationship between C3aR and FcεRI in mast cell activation in asthma
Mast cells appear to play a pivotal role in the development of AHR and inflammation [34]. The ability of allergen to cross-link high affinity IgE receptors (FcεRI) on mast cells to induce degranulation and leukotriene generation is well documented [35,36]. Surprisingly, the role of C3a in mast cell activation remains controversial and appears to depend on the mast cells subtype. For example, murine bone marrow-derived mast cells and a rat basophilic leukemia, RBL-2H3 cells, which have been used extensively as mast cell models, do not express C3a receptors [37]. In contrast, C3a receptors are expressed in human CD34+-derived primary mast cell cultures [38,39], human mast cell lines HMC-1 [40,41] and LAD 2 [39] as well as murine pulmonary mast cells (Thangam, B and Ali, H, unpublished data). Interestingly, C3a is one of the most potent mast cell chemoattractants known [42,43]. C3a also induces robust mast cell degranulation [38,39] and leukotriene C4 generation (Thangam, B and Ali, H, unpublished data). These findings suggest that allergen induces mast cell degranulation by at least two mechanisms: cross-linking of FcεRI and via C3a generation following complement activation by allergen protease (Figure 2). Mast cell proteases also activate the complement pathway to generate C3a [26]. Therefore, C3a generation following FcεRI aggregation may amplify mast cell mediator release (Figure 2).
Figure 2 Proposed interaction between FcεRI and C3aR leading to mast cell activation. Allergen cross-links FcεRI on mast cells to induce degranulation. Allergen can also activate complement pathway (see Fig. 1) to generate C3a, which in turn activates its cognate G protein coupled receptors on mast cells to induce degranulation. Mast cell proteases also activates complement cascade to generate C3a. This C3a may serve to amplify mast cell mediator release.
Mast cell-ASM interaction in asthma
Recent studies with immunohistological analysis of bronchial biopsy specimens from subjects with asthma and those from patients with eosinophilic bronchitis provided important insight on the role of mast cell-ASM cell interaction in the development of AHR in asthma [4,44,45]. Asthma and eosinophilic bronchitis are characterized by similar inflammatory infiltrates in the submucosa of the lower airway. However, ASM infiltration by mast cells is a feature of asthma and not eosinophilic bronchitis. This difference in mast cell recruitment in asthma is associated with AHR, which is absent in in eosiniphilic bronchitis [6]. Furthermore, degranulated mast cells are detected in greater number in ASM bundles of patients who died from asthma when compared to non-asthmatic control [46]. Based on these findings, new hypothesis suggests that increased mast cell recruitment and interaction with ASM may promote release of mast cell-derived mediators that modulate resident airway cell function is asthma [4,5,44].
ASM is not only a contractile tissue that responds to mast cell-derived mediators in asthma, but also modulates mast cell function and airway inflammation. ASM cells express stem cell factor (SCF), which induce mast cell chemotaxis, survival and differentiation [47,48]. Interleukin-1β, tumor necrosis factor (TNF) and TH2 cytokines IL-4 and IL-13 derived cytokines also stimulate ASM to express a large number of chemokines and cytokines [49-52]. Thus, activated ASM cells secrete chemokines and cytokines that may recruit and retain mast cells into the ASM.
C3a receptors and mast cell-ASM cell interaction
C3a has long been recognized as an agent that evokes force generation in smooth muscle. In guinea pigs, C3a-induced contraction of lung parenchyma may involve indirect effects of histamine and arachidonic acid metabolites [53]. In mice, C3a does not cause shortening of isolated tracheal strips [10]. Furthermore, C3a fails to induce AHR after intratracheal instillation in naïve mice [10]. In contrast, in mice immunized with house dust mite, subsequent intratracheal administration of C3a stimulates both AHR and airway inflammation [10]. These findings suggest that C3a-induced AHR and bronchoconstriction requires enhanced infiltration and activation of inflammatory cells, likely mast cells.
Recently, investigations showed that human mast cells but not human or murine ASM express C3aR [54]. Interestingly, incubation of mast cells with human ASM cells, but not its culture supernatant, significantly enhanced C3a-induced mast cell degranulation. Although stem cell factor (SCF) and its receptor c-kit are constitutively expressed on ASM cells and mast cells respectively, neutralizing antibodies to SCF and c-kit failed to inhibit ASM cell-mediated enhancement of mast cell degranulation. Dexamethasone-treated ASM cells however normally express cell surface SCF but were significantly less effective in enhancing C3a-induced mast cell degranulation when compared to untreated cells. Collectively, these findings suggest that cell-cell interaction between ASM cells and mast cells, via a SCF-c-kit independent but dexamethasone-sensitive mechanism, enhances C3a-induced mast cell degranulation, which likely regulates ASM function and may contribute to the pathogenesis of asthma.
While mast cells and ASM cell interaction plays a role in AHR, airway inflammation in asthma is strongly linked to TH2 lymphocyte and their cytokines IL-4, IL-5 and IL-13. These cytokines play key roles in the recruitment and activation of eosinophil, mucous production and IgE synthesis. Allergen challenge of sensitized C3 (-/-) and C3aR (-/-) mice decreased production of TH2 cytokines in BAL and substantially reduced recruitment of T cells, eosinophils and neutrophils in lung tissue [19,20]. Furthermore, inhibition of complement activation or administration of C3aR antagonist during the effector phase of asthma substantially inhibited airway inflammation [32,33]. These findings suggest activation of C3aR is required for TH2 effector function in murine model of allergen-induced inflammation. Accordingly, in human mast cells, C3a stimulates the production of MCP-1, RANTES [39], IL-8 and IL-13 (Thangam, B and Ali, H, unpublished data)-cytokines and chemokines are responsible for the recruitment of T lymphocytes, eosinophils and neutrophils into the airway. Further, C3aR are expressed on basophils, eosinophils and bronchial epithelial cells [18,54-57]. Thus, interaction of a number of inflammatory and resident cells likely regulate C3a-dependent TH2 cytokine and chemokine production in asthma (Figure 3).
Figure 3 Model for the role C3a in AHR and airway inflammation in asthma. C3a generated in individuals with asthma (see Fig. 1) induces mast degranulation (Fig. 2) to promote ASM force generation. Chemokines and cytokines expressed by ASM recruit and retain mast cells into the ASM layer resulting in further smooth muscle dysfunction. TH2 cytokines and chemokines generated from mast cells (and possibly eosinophils and bronchial epithelial cells) regulate AHR and airway inflammation.
Conclusion
Accumulating evidence suggests that C3a may play an important role in the pathogenesis of asthma. In murine models of allergic AHR and inflammation, inhibition of complement activation or small molecule antagonists of C3a receptor after sensitization but before allergen challenge inhibits airway responses. Furthermore, cell-cell interaction between ASM cells and mast cells enhances C3a-induced mast cell degranulation, which likely regulates ASM function, thus contributing to the pathogenesis of asthma. Further investigations on cellular and molecular mechanisms by which C3a modules mast cell-ASM interactions may offer novel therapeutic approaches to the treatment of asthma and airway inflammation.
List of Abbreviations used
C3aR, C3a receptor; AHR, airway hyperresponsiveness: ASM, airway smooth muscle; BAL, bronchoalveolar lavage.
Competing interests
The author(s) declare that they have no competing interests.
Acknowledgements
This work was supported by National Institutes of Health grants 1RO1-HL63372 and 2RO1-HL-55301. We are grateful to Drs. Asifa K Zaidi and E. Berla Thangam for critical review of this manuscript.
==== Refs
Amrani Y Panettieri RA Airway smooth muscle: contraction and beyond Int J Biochem Cell Biol 2003 35 272 276 12531237 10.1016/S1357-2725(02)00259-5
Howarth PH Knox AJ Amrani Y Tliba O Panettieri RAJ Johnson M Synthetic responses in airway smooth muscle J Allergy Clin Immunol 2004 114 S32 50 15309017 10.1016/j.jaci.2004.04.041
Panettieri RAJ Airway smooth muscle: immunomodulatory cells that modulate airway remodeling? Respir Physiol Neurobiol 2003 137 277 293 14516732 10.1016/S1569-9048(03)00153-8
Robinson DS The role of the mast cell in asthma: induction of airway hyperresponsiveness by interaction with smooth muscle? J Allergy Clin Immunol 2004 114 58 65 15241345 10.1016/j.jaci.2004.03.034
Brightling CE Symon FA Holgate ST Wardlaw AJ Pavord ID Bradding P Interleukin-4 and -13 expression is co-localized to mast cells within the airway smooth muscle in asthma Clin Exp Allergy 2003 33 1711 1716 14656359 10.1111/j.1365-2222.2003.01827.x
Brightling CE Bradding P Symon FA Holgate ST Wardlaw AJ Pavord ID Mast-cell infiltration of airway smooth muscle in asthma N Engl J Med 2002 346 1699 1705 12037149 10.1056/NEJMoa012705
Rivera J Molecular adapters in Fc(epsilon)RI signaling and the allergic response Curr Opin Immunol 2002 14 688 693 12413516 10.1016/S0952-7915(02)00396-5
Huber-Lang MS Riedeman NC Sarma JV Younkin EM McGuire SR Laudes IJ Lu KT Guo RF Neff TA Padgaonkar VA Lambris JD Spruce L Mastellos D Zetoune FS Ward PA Protection of innate immunity by C5aR antagonist in septic mice Faseb J 2002 16 1567 1574 12374779 10.1096/fj.02-0209com
Shushakova N Skokowa J Schulman J Baumann U Zwirner J Schmidt RE Gessner JE C5a anaphylatoxin is a major regulator of activating versus inhibitory FcgammaRs in immune complex-induced lung disease J Clin Invest 2002 110 1823 1830 12488432 10.1172/JCI200216577
Hawlisch H Wills-Karp M Karp CL Kohl J The anaphylatoxins bridge innate and adaptive immune responses in allergic asthma Mol Immunol 2004 41 123 131 15159057 10.1016/j.molimm.2004.03.019
Nakano Y Morita S Kawamoto A Suda T Chida K Nakamura H Elevated complement C3a in plasma from patients with severe acute asthma J Allergy Clin Immunol 2003 112 525 530 13679811 10.1016/S0091-6749(03)01862-1
Humbles AA Lu B Nilsson CA Lilly C Israel E Fujiwara Y Gerard NP Gerard C A role for the C3a anaphylatoxin receptor in the effector phase of asthma Nature 2000 406 998 1001 10984054 10.1038/35023175
Castro FF Schmitz-Schumann M Rother U Kirschfink M Complement activation by house dust: reduced reactivity of serum complement in patients with bronchial asthma Int Arch Allergy Appl Immunol 1991 96 305 310 1809688
Krug N Tschernig T Erpenbeck VJ Hohlfeld JM Kohl J Complement factors C3a and C5a are increased in bronchoalveolar lavage fluid after segmental allergen provocation in subjects with asthma Am J Respir Crit Care Med 2001 164 1841 1843 11734433
Hasegawa K Tamari M Shao C Shimizu M Takahashi N Mao XQ Yamasaki A Kamada F Doi S Fujiwara H Miyatake A Fujita K Tamura G Matsubara Y Shirakawa T Suzuki Y Variations in the C3, C3a receptor, and C5 genes affect susceptibility to bronchial asthma Hum Genet 2004 115 295 301 15278436 10.1007/s00439-004-1157-z
Bautsch W Hoymann HG Zhang Q Meier-Wiedenbach I Raschke U Ames RS Sohns B Flemme N Meyer Zu Vilsendorf A Grove M Klos A Kohl J Cutting edge: guinea pigs with a natural C3a-receptor defect exhibit decreased bronchoconstriction in allergic airway disease: evidence for an involvement of the C3a anaphylatoxin in the pathogenesis of asthma [In Process Citation] J Immunol 2000 165 5401 5405 11067890
Gerard NP Gerard C Complement in allergy and asthma Curr Opin Immunol 2002 14 705 708 12413519 10.1016/S0952-7915(02)00410-7
Drouin SM Kildsgaard J Haviland J Zabner J Jia HP McCray PBJ Tack BF Wetsel RA Expression of the complement anaphylatoxin C3a and C5a receptors on bronchial epithelial and smooth muscle cells in models of sepsis and asthma J Immunol 2001 166 2025 2032 11160252
Drouin SM Corry DB Hollman TJ Kildsgaard J Wetsel RA Absence of the complement anaphylatoxin C3a receptor suppresses Th2 effector functions in a murine model of pulmonary allergy J Immunol 2002 169 5926 5933 12421977
Drouin SM Corry DB Kildsgaard J Wetsel RA Cutting edge: the absence of C3 demonstrates a role for complement in Th2 effector functions in a murine model of pulmonary allergy J Immunol 2001 167 4141 4145 11591733
Vandermeer J Sha Q Lane AP Schleimer RP Innate immunity of the sinonasal cavity: expression of messenger RNA for complement cascade components and toll-like receptors Arch Otolaryngol Head Neck Surg 2004 130 1374 1380 15611395 10.1001/archotol.130.12.1374
Varsano S Kaminsky M Kaiser M Rashkovsky L Generation of complement C3 and expression of cell membrane complement inhibitory proteins by human bronchial epithelium cell line Thorax 2000 55 364 369 10770816 10.1136/thorax.55.5.364
Strunk RC Eidlen DM Mason RJ Pulmonary alveolar type II epithelial cells synthesize and secrete proteins of the classical and alternative complement pathways J Clin Invest 1988 81 1419 1426 2966814
Nagata S Glovsky MM Activation of human serum complement with allergens. I. Generation of C3a, C4a, and C5a and induction of human neutrophil aggregation J Allergy Clin Immunol 1987 80 24 32 3496373
Maruo K Akaike T Ono T Okamoto T Maeda H Generation of anaphylatoxins through proteolytic processing of C3 and C5 by house dust mite protease J Allergy Clin Immunol 1997 100 253 260 9275149
Schwartz LB Kawahara MS Hugli TE Vik D Fearon DT Austen KF Generation of C3a anaphylatoxin from human C3 by human mast cell tryptase J Immunol 1983 130 1891 1895 6339618
Gutzmer R Lisewski M Zwirner J Mommert S Diesel C Wittmann M Kapp A Werfel T Human monocyte-derived dendritic cells are chemoattracted to C3a after up-regulation of the C3a receptor with interferons Immunology 2004 111 435 443 15056381 10.1111/j.1365-2567.2004.01829.x
Kirchhoff K Weinmann O Zwirner J Begemann G Gotze O Kapp A Werfel T Detection of anaphylatoxin receptors on CD83+ dendritic cells derived from human skin Immunology 2001 103 210 217 11412308 10.1046/j.1365-2567.2001.01197.x
Soruri A Kiafard Z Dettmer C Riggert J Kohl J Zwirner J IL-4 down-regulates anaphylatoxin receptors in monocytes and dendritic cells and impairs anaphylatoxin-induced migration in vivo J Immunol 2003 170 3306 3314 12626590
Werfel T Kirchhoff K Wittmann M Begemann G Kapp A Heidenreich F Gotze O Zwirner J Activated human T lymphocytes express a functional C3a receptor J Immunol 2000 165 6599 6605 11086104
Kawamoto S Yalcindag A Laouini D Brodeur S Bryce P Lu B Humbles AA Oettgen H Gerard C Geha RS The anaphylatoxin C3a downregulates the Th2 response to epicutaneously introduced antigen J Clin Invest 2004 114 399 407 15286806 10.1172/JCI200419082
Taube C Rha YH Takeda K Park JW Joetham A Balhorn A Dakhama A Giclas PC Holers VM Gelfand EW Inhibition of complement activation decreases airway inflammation and hyperresponsiveness Am J Respir Crit Care Med 2003 168 1333 1341 14500265 10.1164/rccm.200306-739OC
Baelder R Fuchs B Bautsch W Zwirner J Kohl J Hoymann HG Glaab T Erpenbeck V Krug N Braun A Pharmacological Targeting of Anaphylatoxin Receptors during the Effector Phase of Allergic Asthma Suppresses Airway Hyperresponsiveness and Airway Inflammation J Immunol 2005 174 783 789 15634899
Taube C Wei X Swasey CH Joetham A Zarini S Lively T Takeda K Loader J Miyahara N Kodama T Shultz LD Donaldson DD Hamelmann EH Dakhama A Gelfand EW Mast cells, FcepsilonRI, and IL-13 are required for development of airway hyperresponsiveness after aerosolized allergen exposure in the absence of adjuvant J Immunol 2004 172 6398 6406 15128831
Andrade MV Hiragun T Beaven MA Dexamethasone suppresses antigen-induced activation of phosphatidylinositol 3-kinase and downstream responses in mast cells J Immunol 2004 172 7254 7262 15187100
Furumoto Y Nunomura S Terada T Rivera J Ra C The FcepsilonRIbeta immunoreceptor tyrosine-based activation motif exerts inhibitory control on MAPK and IkappaB kinase phosphorylation and mast cell cytokine production J Biol Chem 2004 279 49177 49187 15355979 10.1074/jbc.M404730200
Erdei A Andrasfalvy M Peterfy H Toth G Pecht I Regulation of mast cell activation by complement-derived peptides Immunol Lett 2004 92 39 42 15081525 10.1016/j.imlet.2003.11.019
Woolhiser MR Brockow K Metcalfe DD Activation of human mast cells by aggregated IgG through FcgammaRI: additive effects of C3a Clin Immunol 2004 110 172 180 15003814 10.1016/j.clim.2003.11.007
Venkatesha RT Thangam EB Zaidi AK Ali H Distinct regulation of C3a-induced MCP-1/CCL2 and RANTES/CCL5 production in human mast cells by extracellular signal regulated kinase and PI3 kinase Mol Immunol 2005 42 581 587 15607817 10.1016/j.molimm.2004.09.009
Ali H Ahamed J Hernandez-Munain C Baron JL Krangel MS Patel DD Chemokine production by G protein-coupled receptor activation in a human mast cell line: roles of extracellular signal-regulated kinase and NFAT J Immunol 2000 165 7215 7223 11120854
Ahamed J Venkatesha RT Thangam EB Ali H C3a Enhances Nerve Growth Factor-Induced NFAT Activation and Chemokine Production in a Human Mast Cell Line, HMC-1 J Immunol 2004 172 6961 6968 15153516
Nilsson G Johnell M Hammer CH Tiffany HL Nilsson K Metcalfe DD Siegbahn A Murphy PM C3a and C5a are chemotaxins for human mast cells and act through distinct receptors via a pertussis toxin-sensitive signal transduction pathway J Immunol 1996 157 1693 1698 8759757
Hartmann K Henz BM Kruger-Krasagakes S Kohl J Burger R Guhl S Haase I Lippert U Zuberbier T C3a and C5a stimulate chemotaxis of human mast cells Blood 1997 89 2863 2870 9108406
Page S Ammit AJ Black JL Armour CL Human mast cell and airway smooth muscle cell interactions: implications for asthma Am J Physiol Lung Cell Mol Physiol 2001 281 L1313 23 11704524
Berger P Girodet PO Begueret H Ousova O Perng DW Marthan R Walls AF Tunon de Lara JM Tryptase-stimulated human airway smooth muscle cells induce cytokine synthesis and mast cell chemotaxis Faseb J 2003 17 2139 2141 14500550
Carroll NG Mutavdzic S James AL Distribution and degranulation of airway mast cells in normal and asthmatic subjects Eur Respir J 2002 19 879 885 12030728 10.1183/09031936.02.00275802
Kassel O Schmidlin F Duvernelle C Gasser B Massard G Frossard N Human bronchial smooth muscle cells in culture produce stem cell factor Eur Respir J 1999 13 951 954 10414388 10.1034/j.1399-3003.1999.13e04.x
Tsujimura T Role of c-kit receptor tyrosine kinase in the development, survival and neoplastic transformation of mast cells Pathol Int 1996 46 933 938 9110344
Ammit AJ Lazaar AL Irani C O'Neill GM Gordon ND Amrani Y Penn RB Panettieri RAJ Tumor necrosis factor-alpha-induced secretion of RANTES and interleukin-6 from human airway smooth muscle cells: modulation by glucocorticoids and beta-agonists Am J Respir Cell Mol Biol 2002 26 465 474 11919083
Oltmanns U Issa R Sukkar MB John M Chung KF Role of c-jun N-terminal kinase in the induced release of GM-CSF, RANTES and IL-8 from human airway smooth muscle cells Br J Pharmacol 2003 139 1228 1234 12871843 10.1038/sj.bjp.0705345
Song R Ning W Liu F Ameredes BT Calhoun WJ Otterbein LE Choi AM Regulation of IL-1beta -induced GM-CSF production in human airway smooth muscle cells by carbon monoxide Am J Physiol Lung Cell Mol Physiol 2003 284 L50 6 12388337
Faffe DS Whitehead T Moore PE Baraldo S Flynt L Bourgeois K Panettieri RA Shore SA IL-13 and IL-4 promote TARC release in human airway smooth muscle cells: role of IL-4 receptor genotype Am J Physiol Lung Cell Mol Physiol 2003 285 L907 14 12871855
Stimler NP Bloor CM Hugli TE C3a-induced contraction of guinea pig lung parenchyma: role of cyclooxygenase metabolites Immunopharmacology 1983 5 251 257 6403485 10.1016/0162-3109(83)90031-0
Thangam BE Venkatesha RT Zaidi AK Jordan-Sciutto KL Goncharov DA Krymskaya VP Amrani Y Panettierijn RA Ali H Airway smooth muscle cells enhance C3a-induced mast cell degranulation following cell-cell contact Faseb J 2005 In Press
Daffern PJ Pfeifer PH Ember JA Hugli TE C3a is a chemotaxin for human eosinophils but not for neutrophils. I. C3a stimulation of neutrophils is secondary to eosinophil activation J Exp Med 1995 181 2119 2127 7760001 10.1084/jem.181.6.2119
Elsner J Oppermann M Czech W Dobos G Schopf E Norgauer J Kapp A C3a activates reactive oxygen radical species production and intracellular calcium transients in human eosinophils Eur J Immunol 1994 24 518 522 8125125
Bischoff SC de Weck AL Dahinden CA Interleukin 3 and granulocyte/macrophage-colony-stimulating factor render human basophils responsive to low concentrations of complement component C3a Proc Natl Acad Sci U S A 1990 87 6813 6817 1697689
| 15723703 | PMC551592 | CC BY | 2021-01-04 16:36:27 | no | Respir Res. 2005 Feb 21; 6(1):19 | utf-8 | Respir Res | 2,005 | 10.1186/1465-9921-6-19 | oa_comm |
==== Front
J Transl MedJournal of Translational Medicine1479-5876BioMed Central London 1479-5876-3-91572370510.1186/1479-5876-3-9ResearchA phase I study of dexosome immunotherapy in patients with advanced non-small cell lung cancer Morse Michael A [email protected] Jennifer [email protected] Takuya [email protected] Shubi [email protected] Amy [email protected] Timothy M [email protected] Nancy [email protected] Revati [email protected] Mary Ann [email protected] Alain [email protected] Di-Hwei [email protected] Pecq Jean-Bernard [email protected] H Kim [email protected] Department of Medicine, Duke University Medical Center, Durham, NC, USA2 Department of Surgery, Duke University of Medical Center, Durham, NC, USA3 Anosys Inc., Menlo Park, CA, USA4 Currently at Genentech, Inc., South San Francisco, CA, USA2005 21 2 2005 3 9 9 30 12 2004 21 2 2005 Copyright © 2005 Morse et al; licensee BioMed Central Ltd.2005Morse 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
There is a continued need to develop more effective cancer immunotherapy strategies. Exosomes, cell-derived lipid vesicles that express high levels of a narrow spectrum of cell proteins represent a novel platform for delivering high levels of antigen in conjunction with costimulatory molecules. We performed this study to test the safety, feasibility and efficacy of autologous dendritic cell (DC)-derived exosomes (DEX) loaded with the MAGE tumor antigens in patients with non-small cell lung cancer (NSCLC).
Methods
This Phase I study enrolled HLA A2+ patients with pre-treated Stage IIIb (N = 4) and IV (N = 9) NSCLC with tumor expression of MAGE-A3 or A4. Patients underwent leukapheresis to generate DC from which DEX were produced and loaded with MAGE-A3, -A4, -A10, and MAGE-3DPO4 peptides. Patients received 4 doses of DEX at weekly intervals.
Results
Thirteen patients were enrolled and 9 completed therapy. Three formulations of DEX were evaluated; all were well tolerated with only grade 1–2 adverse events related to the use of DEX (injection site reactions (N = 8), flu like illness (N = 1), and peripheral arm pain (N = 1)). The time from the first dose of DEX until disease progression was 30 to 429+ days. Three patients had disease progression before the first DEX dose. Survival of patients after the first DEX dose was 52–665+ days. DTH reactivity against MAGE peptides was detected in 3/9 patients. Immune responses were detected in patients as follows: MAGE-specific T cell responses in 1/3, increased NK lytic activity in 2/4.
Conclusion
Production of the DEX vaccine was feasible and DEX therapy was well tolerated in patients with advanced NSCLC. Some patients experienced long term stability of disease and activation of immune effectors
==== Body
Introduction
Vaccine immunotherapy as an approach to cancer treatment has evolved over the last 10 years as the basic biology of the immune response has been elucidated. Tumor-associated antigens that are capable of eliciting cytotoxic T cell responses have been identified. Among the most frequently expressed across many malignancies are the MAGE antigens, originally described in melanoma, but expressed by other tumors including non-small cell lung cancer (NSCLC) [1-3]. Immune responses to MAGE 3 have been correlated with clinical outcome in melanoma patients [4]. This has lead to many tumor antigen-specific strategies for the treatment of cancer, including the use of an immunodominant peptide alone, protein or peptide-pulsed dendritic cells, and antigen/co-stimulatory fusion proteins expressed from viral vectors. Although each strategy has its proponents, none have achieved the goal of activating significant immune responses that correlate with clinical responses in a majority of patients; therefore, there is a continued need to develop even more effective strategies. Recently, a novel platform for delivering high levels of antigen in conjunction with costimulatory molecules has been described, called exosomes, cell-derived lipid vesicles that express high levels of a narrow spectrum of cell proteins.
A variety of cells have been shown to release exosomes including dendritic cells [5], B lymphocytes [6], T lymphocytes [7], mast cells [8], platelets [9], and tumor cells [10]. The small (60–90 mm) vesicles form within the late endosomes or multivesicular bodies and have biologic functions dependent on the cell type from which they were secreted [11-13]. Originally described as vesicles released from reticulocytes containing proteins (transferring receptor) that were no longer required in the mature red blood cell [14], they have subsequently been demonstrated to play a role in activation of the immune response. Exosomes derived from B lymphocytes were able to stimulate CD4+ T cells in an antigen/MHC class II restricted manner [6] and have been demonstrated to be the source of MHC class II molecules on follicular dendritic cells [15]. In addition, tumor cells release vesicles that function in cross-priming by transferring a protein antigen from the tumor cell to a dendritic cell for immune presentation [10,16]. Importantly, dendritic cells release vesicles (named "dexosomes") that have been demonstrated to prime specific T cells in vitro and eradicate established murine tumors [5]. In vitro, dexosomes have the capacity to present antigen to naïve CD8+ cytolytic T cells and CD4+ T cells [17,18].
Human dexosomes are enriched in the components necessary to function as an antigen-presenting entity. Extensive electron microscopic and protein characterization has revealed that dexosomes contain a specific set of proteins that differentiate them from other plasma membrane derived vesicles (such as apoptotic cells for example), including MHC class I and II molecules and CD1a, b, c, d molecules, as well as the co-stimulatory molecule CD86 and several tetraspan proteins (CD9, CD37, CD53, CD63, CD81, and CD82) [Anosys unpublished data, [19,20]].
Dexosomes have been demonstrated to participate in antigen presentation in the following way [21,22]. After capturing antigens at the periphery, DC incorporate MHC-antigenic peptide complexes in dexosomes with immunostimulating factors. Released dexosomes subsequently transfer MHC-antigenic peptide complexes and associated proteins to antigen-naïve DC in the regional lymph nodes. The latter thereby acquire the ability to stimulate CD4+ and CD8+ T cells. Thus, dexosomes appear to act as a vehicle for disseminating antigen amongst DC, representing a potentially important mechanism of immune response amplification. This hypothesis forms the rationale for the potential use of dexosomes as a therapeutic cancer immunotherapy.
Dexosomes have demonstrated significant antitumor activity in a mouse tumor model, suggesting that the use of dexosomes derived from dendritic cells may result in improved efficacy relative to the ex vivo dendritic cell approach for eradication of advanced cancer. Purified dexosomes were shown to be effective in both suppressing tumor growth and eradicating an established tumor in this model. Furthermore, the effect of the dendritic cell-derived dexosome was greater than that of the dendritic cell from which it was produced [5]. Therefore, we hypothesized that dendritic cell-derived dexosomes would be an effective platform for activating tumor antigen-specific immune responses in humans.
We performed this study to investigate the safety, feasibility, and efficacy of administering autologous dexosomes loaded with tumor antigens (subsequently referred to as DEX) to patients with advanced NSCLC. We also evaluated the immunologic responses in selected patients and monitored the clinical outcomes.
Methods
Patients
This phase I clinical protocol was approved by the Duke University Medical Center Institutional Review Board and conducted in compliance with the Helsinki Declaration and under an IND from the United States Food and Drug Administration held by Anosys Corporation. All subjects provided written informed consent. Patients were eligible for enrollment if they had histologically confirmed, unresectable Stage III A or B or Stage IV NSCLC, were HLA A*0201 positive, at least 18 years of age, and had adequate organ function and a Karnofsky performance status of at least 80%. Patients were required to have been treated with at least one prior standard chemotherapy regimen and have measurable disease. In addition, patients were required to have tumor expressing MAGE A3 or MAGE A4. To avoid performing repeat biopsies, this was achieved by detecting MAGE A3 or MAGE A4 expression in peripheral blood tumor cells by RT-PCR using established methods.
The main exclusion criteria were: prior therapy within 4 weeks of the leukapheresis, CNS disease, history of autoimmune disease, concurrent use of systemic steroids, presence of HIV infection or acute or chronic viral hepatitis B or C. Pregnant or lactating women were also excluded.
Manufacture of DEX
Dexosomes were manufactured from peripheral blood mononuclear cells (PBMCs) as previously described [23]. Briefly, PBMCs were obtained from the patient during a 2-blood volume leukapheresis and shipped overnight to Anosys, Inc., Menlo Park CA. The cells were washed, adhered to plastic to isolate monocytes and placed in a 7-day serum-free culture at 37°C in a humidified 5% CO2 atmosphere in the presence of 50 ng/mL GM-CSF (Immunex, Seattle, Washington) and 10 ng/mL of IL-4 (Schering-Plough, Kennilworth, NJ). On the 7th day of culture, the supernatant of the resulting dendritic cell preparation was harvested, filtered, and concentrated. Dexosomes were then isolated by ultracentrifugation on a D2O/sucrose cushion. As described in table 1, the final dexosome product (DEX) consisted of one of three different formulations based on different methods for loading the following peptides onto the dexosomes: MAGE-derived, HLA-A2 restricted Class I peptides KVAELVHFL (MAGE-A3(112–120)), GVYDGREHTV (MAGE-A4(230–239)) and GLYDGMEHL (MAGE-A10(254–262)); MAGE derived HLA-DP04 restricted Class II peptide TQHFVQENYLEY (MAGE-A3(247–258)) [24]; and the control peptides, the cytomegalovirus (CMV) pp65-derived, HLA-A2 restricted Class I peptide NLVPMVATV and the tetanus toxoid-derived, promiscuous HLA-DR Class II peptide QYIKANSKFIGITE (produced by Multiple Peptide Systems, San Diego, CA). Peptides were loaded either "directly" onto dexosomes (i.e., following purification of dexosomes from the DC culture) or "indirectly" (i.e. onto cultured DCs that are the source of the dexosomes). The quantity of DEX prepared from a single leukapheresis was measured by ELISA as previously described [23]. The measured number of MHC class II molecules present in the DEX product was utilized for the purpose of dosing. The final DEX product was diluted in 0.9% normal saline for injection, sterile filtered, and stored at -80°C; subsequently the DEX product was shipped overnight to the investigative site, and maintained in its frozen state until 1 hour before use.
Table 1 Dose Groups and Product Formulations
Dose Cohorts Number of patients in Cohort (Patient number) Peptides loaded/HLA class Peptide loading method and concentration DEX dose (expressed as numbers of MHC class II molecules)
A 3 (DU 5, 6, 8) MAGE-A3 (112–120)/class I
MAGE-A4 (230–239)/class I
MAGE-A10 (254–262)/class I
CMV pp65/class II
Tetanus toxoid/class II Indirect (10 μg/mL)
Indirect (10 μg/mL)
Indirect (10 μg/mL)
Indirect (10 μg/mL)
Indirect (10 μg/mL) 0.13 × 10 14
B 5 (DU 24, 39, 44, 50, 63) MAGE-A3 (112–120)/class I
MAGE-A4 (230–239)/class I
MAGE-A10 (254–262)/class I
CMV pp65/class I
MAGE-A3 (247–258)/class II
Tetanus toxoid/class II Direct (10 μg/mL)
Direct (10 μg/mL)
Direct (10 μg/mL)
Direct (10 μg/mL)
Indirect (10 μg/mL)
Indirect (10 μg/mL) 0.13 × 10 14
C 4 (DU 49, 73, 81, 83) MAGE-A3 (112–120)/class I
MAGE-A4 (230–239)/class I
MAGE-A10 (254–262)/class I
MAGE-A3 (247–258)/class II Direct (100 μg/mL)
Direct (100 μg/mL)
Direct (100 μg/mL)
Indirect (10 μg/mL) 0.13 × 10 14
The amino acid sequences of the peptides used for loading the dexosomes are: MAGE-A3(112–120) = KVAELVHFL; MAGE-A4(230–239) = GVYDGREHTV; MAGE-A10(254–262) = GLYDGMEHL; MAGE-A3(247–258) = TQHFVQENYLEY; CMV pp6 = NLVPMVATV; tetanus toxoid = QYIKANSKFIGITE
Note: DU39, DU44, and DU83 were not treated.
Treatment and Follow-up Schedule
Patients were enrolled into three cohorts (A,B,C) that varied in the method of MHC Class I peptide loading and concentration as described in Table 1. The quantity of DEX administered to the patients in each cohort was identical: 1.3 × 1013 MHC class II molecules in a volume of 3 mL (divided into twoinjections given at two sites on opposite sides of the body) as a combination of subcutaneous (90% of the volume) and intradermal (10%) injections weekly for 4 weeks. No retreatment was allowed. Vital signs were monitored for 1 hour after each injection. Clinical responses were assessed by RECIST criteria. CT scans of the chest through the upper abdomen were obtained at baseline, 1 month following the last dose of DEX and every 3 months after last dose of DEX for 1 year, but scans to confirm responses were not required in this phase I study. All surviving patients have been followed every 6 months for assessment of vital status.
Delayed type hypersensitivity (DTH) testing
Prior to the initial leukapheresis and 1 week after the last dose of DEX, the following peptides were injected intradermally, in addition to the standard recall antigen panel of Candida, Mumps, and tetanus: MAGE-A3(112–120), MAGE-A4(230–239), MAGE-A10(254–262), and MAGE-A3(247–258), each at 10 μg in 0.1 mL saline. The diameter of the induration and erythema was measured 48 hours following the peptide injection.
ELISPOT testing
Immune response was evaluated at baseline and 1 week following last dose of DEX with cryopreserved PBMCs obtained by leukapheresis. The ELISPOT assay was performed by ImmunoSite, Inc. (Pittsburgh, PA) according to previously reported methods [25] using both direct assessment of thawed PBMCs, and when possible, following in vitro stimulation of PBMCs with autologous DCs pulsed with the MAGE-A3(112–120), MAGE-A4(230–239), and MAGE-A10(254–262). The number of spots (interferon-gamma-secreting T cells) per 20,000 responding PBMC was reported. The background number of spots against an irrelevant antigen was subtracted from the number of spots for the experimental conditions.
Natural killer cell activity
NK cells were isolated from cryopreserved PBMCs using an NK Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's instruction. NK cell purity was checked by flow cytometry using anti-CD3-FITC, anti-CD45-PerCP, and anti-CD56-APC antibodies (BD Bioscience, San Jose, CA). Isolated NK cells and NK cells activated for 40 hours by IL-2 (Proleukin, Chiron) 600 Units/ml were incubated at various effector to target rations with chromium-51 labeled K562 cells, an NK target, for 4 hours at 37°C and cytotoxicity was assessed by the amount of radiolabeled chromium released. Cytotoxicity was calculated as follows: percentage of target cell lysis = 100 × (counts per minute (cpm) of experimental release - cpm of spontaneous release) / (cpm of maximum release - cpm of spontaneous release).
Statistics
The primary endpoints of this study were safety and feasibility, with secondary endpoints of clinical and immunologic response rates. The incidence, type and severity of adverse events were recorded during the study treatment through 30 days following the last dose of DEX. Descriptive statistics were used to present the data. Adverse events were coded using MEDDRA version 5.0. Survival and time to progression were measured from the date of the first injection to the date of documented disease progression or death. For patients who progressed, the time to disease progression was determined by the interval from the first injection+ 1 day to the last evaluation of disease staging. For patients who did not progress or die during the two year follow up period, the time of disease progression and survival was determined by the interval between the first dose of DEX and the date of last evaluation of disease staging + 1 day and concatenated with the '+' sign.
Results
Patient Characteristics
Thirteen patients, (8 female, 5 male) median age 62 years (range 44–72 years) with unresectable pretreated Stage III or IV NSCLC were enrolled. The median time from original diagnosis to study entry was 9.9 months (range 2–61 months) and the median Karnofsky score of the patient population was 80% (range 80–100). DEX therapy was administered to 9/13 (6 female and 3 male) patients. Of the 9 dosed patients, 5 patients had Stage IV and 4 patients had Stage IIIB disease. Six patients had stable disease and 3 patients had progressive disease at study entry. Two patients had squamous cell carcinoma, 4 patients had adenocarcinoma, 2 patients had large cell carcinoma, and in 1 case the histological type was not reported. All patients had received prior chemotherapy (median number of cycles: 6.5, range: 3–30), 6/9 patients had received prior radiotherapy and 4/9 patients had prior surgery for cancer treatment. Four patients did not receive DEX for the following reasons: manufacturing failure in 2 cases (DU39, DU83), one of whom had received chemotherapy 13 days prior to leukapheresis and one of whom (DU39) also had rapid disease progression at the time of leukapheresis; delay in shipment in one instance (DU14); and, rapid disease progression prior to planed dosing with DEX in one case (DU44). The characteristics of all dosed patients are listed in Table 2 (see separate file for Table 2).
Dexosome manufacture
The dose of DEX that was selected corresponded to the maximum dose that could be achieved from healthy donors. We confirmed that this dose could be generated in all but two patients with NSCLC. The mean dexosome generation consisted of a total class II number of 3.14 × 1014 (range 4.1 × 1012 to 9.1 × 1014). This quantity of dexosomes in our advanced NSCLC patients is similar in quantity to that generated from healthy donors (mean for 111 healthy donors was 3.9 × 1014 total class II).
Toxicity
The DEX immunotherapy was generally well tolerated without evidence of serious toxicity. The most frequently reported adverse events causally related to the use of DEX were mild (Grade 1–2) in severity and included: Injection site reactions (erythema, contusion, induration and edema) in 8 patients; flu like syndrome (1 patient); and, peripheral edema and pain in the arm (1 patient). There were no significant organ or laboratory toxicities attributable to the vaccine. No autoimmune reactions were observed.
Immunologic Response
DTH analysis
All 9 dosed patients underwent DTH testing with individual tumor-associated peptides prior to and following all doses of DEX. There was no DTH response to the specific peptide antigens prior to DEX therapy. Three patients (DU06, DU24 and DU49) had a positive response of at least 5 mm erythema or induration in the longest dimension 48 hours after skin testing with one of the MAGE peptides. Specifically, DU06 had 5 mm induration and erythema with MAGE-A4(230–239), DU24 had 6 mm induration and erythema with MAGE-A10(254–262) and DU49 had 5 mm induration and erythema with MAGE-A3(112–120), respectively.
In vitro immunologic analysis
The peptide-specific immune response to MAGE and CMV was analyzed using ELISPOT in 5 of 9 dosed patients (DU24, DU49, DU50, DU63, DU81). One patient (DU49) exhibited detectable increases in T cell precursor frequency to MAGE-A10(254–262) following in-vitro stimulation (an increase of 12 MAGE-A10-specific cells/20,000 responders). Assays for DU50 and DU63 could not be completed because of poor viability. Robust responses to anti-CD3 and to the control peptide CMV pp65 were observed in DU24, DU81, and DU49, but no MAGE-specific responses were detected.
Since most patients did not exhibit a significant increase in antigen-specific T cell activity, we hypothesized that regulatory influences such as CD4+CD25+ regulatory T cell populations might inhibit augmentation of the T cell response. In 2/3 patients who had analyzable specimens available, an increase in CD4+CD25+ T cells as a percentage of CD4+ T cells was observed following completion of DEX therapy when compared with baseline values (DU05: increase from 19.49 to 26.64%, DU08: minimal change from 20.39 to 23.42%, DU50: increase from 17.45 to 31.81%). The small number of samples available for this analysis precludes any conclusions but does suggest that CD4+CD25+ T cell analyses should accompany future studies of DEX immunotherapy.
During the study, new data from Escudier et al (manuscript submitted) suggested that the immunologic activity of DEX might be due to activation of NK cells. We therefore explored the hypothesis that NK cells may be activated following DEX therapy. This was not planned as part of the initial analysis and therefore specimens of PBMC were limiting in all but 4 patients (DU05, DU08, DU24, DU50). Although there was no consistent change in NK percentage before and after immunization (Table 1: DU05: 10.7 to 9.2%; DU08: 6.0 to 5.4%; DU24: 8.8 to 8.5%; DU50: 9.9 to 13.9%), NK activity as determined by the ability to lyse K562 target cells was observed to increase in 2/4 patients following immunizations (Fig 1 -see additional file Fig 1). Short-term culture with IL-2 was required to activate the NK cells in vitro as there was very low activity in the absence of IL-2. Although addition of IL-2 increased the NK cell activity, it did not change the relative pattern of activity, i.e., in no instance did the order of the results change as a result of IL-2 stimulation.
Figure 1 Cytolytic activity of NK cells. Cytolytic activity of NK cells isolated from the PBMC of 4 patients (DU05, DU08, DU24, DU50) pre (squares) and post (circles) immunization and cultured with (dark shapes) or without (open shapes) IL-2 was determined. The percentage lysis of the NK target (K562) cells is reported at effector to target ratios of 0.2:1 to 25:1.
Clinical Outcomes
At approximately 2 years of follow up, survival from the first dose of DEX ranged from 52 to 309 days for cohort A, 280 to 665+ days for cohort B, and 244 to 502 days for cohort C (Table 1). In order to obtain preliminary data on response rate, CT scans were obtained prior to immunization and at 1 month and 3 month intervals following completion of the immunizations, but additional scans were not obtained to confirm responses. Of the two patients (DU05, DU08) with disease progression at study entry, DU05 was stable at the end of the immunizations but was felt to have clinically progressed at day 88 shortly before death. DU08 also had stable disease at the end of the immunizations and on every three month follow-up until having progression at day 302. Of four additional patients (DU06, DU24, DU63, DU81) who began the study with stable disease, two (DU24 and DU63) have remained without progression for greater than 12 months. DU06 was stable at the post-immunization CT but subsequently died unexpectedly of unknown etiology and without a follow-up scan, and DU81 was stable at the post-immunization CT but had progressed by the next CT scan at the three month follow-up. The remainder of the patients had progressed at the post-immunization CT scan including DU73 who had disease progression prior to the first dose of DEX. The time until progressive disease, as documented from the first dose of DEX, ranged from 30+ to 302 days for cohort A, 40 to 429+ days for cohort B, and 51 to 166 days for cohort C.
Discussion
The objective of this study was to show that DEX could be manufactured from NSCLC patients and could be safely administered. We demonstrated the feasibility of producing dexosomes loaded with specific MAGE and other peptides and demonstrated that this form of immunotherapy was well tolerated in patients with advanced NSCLC. Leukapheresis products could be shipped to a central processing facility with good cell viability after transport in a majority of cases, in contrast to other autologous therapies involving tissues where logistics of tissue harvest and processing are complex. The dexosome product was successfully manufactured and loaded with multiple peptides in the majority of patients. This suggests that different panels of tumor antigen-derived peptides could be successfully loaded onto dexosomes. Using multiple peptide panels may allow for targeting various tumor types and larger patient populations, and decreases the likelihood of antigenic escape.
We observed increases in systemic immune responses against MAGE by DTH reactivity in 3/9 patients who had no reactivity to the MAGE peptides prior to immunization and activation of NK cells, but found minimal increases in antigen-specific T cell activity in in vitro assays performed circulating PBMCs. Possible explanations include nonoptimized or low-sensitivity assays, inadequate antigen presentation, counter-regulatory mechanisms that dampen immune responses, or the lack of persistence of antigen-specific Tcells in the circulation (i.e., the T cell may have migrated to tumor tissue or lymph nodes). The possible role of negative regulatory mechanisms was suggested by the presence of elevated levels of CD4+CD25+ regulatory T cells following immunization in some patients.
An intriguing immunologic observation was the increase in NK activity following immunization in 2/4 patients analyzed. Although DEX are intended to activate antigen-specific, MHC-restricted T cell responses, it is possible that cytokines released in response to DEX therapy could cause activation of NK cells or that DEX could directly activate NK cells. DEX therapy may stimulate both innate and adaptive arms of the immune response and thereby provide a rationale for maximizing the anti-tumor effect of this approach, even in cases where tumors have lost Class I antigens, a common finding as cancers become more advanced [26]. Indeed, in a phase I study in melanoma patients, DEX loaded with MAGE peptides were well tolerated and associated with both clinical response and increased NK activity (Escudier, manuscript submitted to J. Trans Med).
Despite the small sample size and the fact that 3/9 dosed patients had disease progression at the time of initiation of DEX treatment, we observed prolonged disease stabilization in some patients. Large clinical trials in patients with advanced NSCLC have generally reported median time to progression of 3–5 months in patients with advanced NSCLC treated with systemic chemotherapy regimens [27-30]. The lack of toxicity and interesting clinical and immunologic observations support further investigation of DEX immunotherapy as a treatment approach for both advanced and early stage NSCLC and other tumors. Phase II clinical studies in non-small cell lung cancer and other tumor types are planned to continue to explore the efficacy of this novel immunotherapy.
Conclusion
DEX therapy was well. Immune activation and stability of disease was observed in some immunized patients with advanced NSCLC.
Competing Interests
Michael Morse received funding from NIH 5R21CA89957-02. Additionally, portions of this study were funded by Anosys, Inc.
Nancy Valente, Revati Shreeniwas, Mary Ann Sutton, Alain Delcayre, Di-Hwei Hsu, and Jean BernardLe Pecq held stock and were employees in Anosys,
H. Kim Lyerly was a consultant for Anosys, Inc.
Authors' contributions
MAM was the principal investigator of the study and oversaw all aspects including protocol development, patient management, data collection and analysis, and manuscript preparation.
JG enrolled patients to the study and managed their care and participated in data analysis.
Takuya Osada performed the NK assays and analyzed the data.
SK enrolled patients to the study and managed their care.
AH performed in vitro immunologic assays and analyzed the data.
TMC oversaw the immunologic analyses performed at Duke University and analyzed the data.
NV, RS, and MAS oversaw development of the protocol, data collection and analysis, and manuscript preparation.
AD developed and oversaw the MAGE screening for patient eligibility.
D-H H oversaw portions of the immunologic analysis and data analysis.
J-B L provided scientific direction regarding generation of the dexosomes, protocol development, and data analysis and manuscript preparation
HKL provided consultation on immunologic assay development All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Table 2 (DOC) presents the remainder of clinical and immunological data from all patients
Click here for file
Acknowledgements
The authors thank Allyson Gattis for helping coordinate the long-term follow-up of patients. Statistical support was provided by Carol Zhao, PhD, ICON Clinical Research, 555 Twin Dolphin Drive, Redwood Shores, CA.
This study was supported by NIH grant 5R21CA089957-01 (Morse, Michael), and a grant from Anosys Inc., Menlo Park, CA, and was performed in the Duke General Clinical Research Unit (M01RR00030).
This work was presented as an abstract at the American Society of Clinical Oncology in May 2002 and published in the abstract book as: Morse MA, Garst J, Khan S, et al. Preliminary results of a phase I/II study of active immunotherapy with autologous dexosomes loaded with MAGE peptides in HLA A2+ patients with stage III/IV non-small cell lung cancer. Proc Am Soc Clin Oncol 2002;11a (abstract 42).
==== Refs
Gotoh K Yatabe Y Sugiura T Takagi K Ogawa M Takahashi T Takahashi T Mitsudomi T Frequency of MAGE-3 gene expression in HLA-A2 positive patients with non-small cell lung cancer Lung Cancer 1998 20 117 25 9711530 10.1016/S0169-5002(98)00017-8
Shichijo S Hayashi A Takamori S Tsunosue R Hoshino T Sakata M Kuramoto T Oizumi K Itoh K Detection of MAGE-4 protein in lung cancers Int J Cancer 1995 64 158 65 7622303
Park JW Kwon TK Kim IH Sohn SS Kim YS Kim CI Bae OS Lee KS Lee KD Lee CS Chang HK Choe BK Ahn SY Jeon CH A new strategy for the diagnosis of MAGE-expressing cancers J Immunol Methods 2002 266 79 86 12133624 10.1016/S0022-1759(02)00105-9
Reynolds SR Zeleniuch-Jacquotte A Shapiro RL Roses DF Harris MN Johnston D Bystryn JC Vaccine-induced CD8+ T-cell responses to MAGE-3 correlate with clinical outcome in patients with melanoma Clin Cancer Res 2003 9 657 62 12576432
Zitvogel L Regnault A Lozier A Wolfers J Flament C Tenza D Ricciardi-Castagnoli P Raposo G Amigorena S Eradication of established murine tumors using a novel cell-free vaccine: dendritic cell-derived exosomes Nat Med 1998 4 594 600 9585234 10.1038/nm0598-594
Raposo G Nijman HW Stoorvogel W Liejendekker R Harding CV Melief CJ Geuze HJ B lymphocytes secrete antigen-presenting vesicles J Exp Med 1996 183 1161 1172 8642258 10.1084/jem.183.3.1161
Martinez-Lorenzo MJ Anel A Gamen S Monle n I Lasierra P Larrad L Pineiro A Alava MA Naval J human T cells release bioactive Fas ligand and APO2 ligand in microvesicles J Immunol 1999 163 1274 81 10415024
Skokos D Le Panse S Villa I Rousselle JC Peronet R David B Namane A Mecheri S Mast cell-dependent B and T lymphocyte activation is mediated by the secretion of immunologically active exosomes J Immunol 2001 166 868 76 11145662
Heijnen HF Schiel AE Fijnheer R Geuze HJ Sixma JJ Activated platelets release two types of membrane vesicles: microvesicles by surface shedding and exosomes derived from exocytosis of multivesicular bodies and alpha-granules Blood 1999 94 3791 9 10572093
Wolfers J Lozier A Raposo G Regnault A Thery C Masurier C Flament C Pouzieux S Faure F Tursz T Angevin E Amigorena S Zitvogel L Tumor-derived exosomes are a source of shared tumor rejection antigens for CTL cross-priming Nat Med 2001 7 297 303 11231627 10.1038/85438
Farsad K Exosomes: novel organelles implicated in immunomodulation and apoptosis Yale J Biol Med 2002 75 95 101 12230314
Denzer K Kleijmeer MJ Heijnen HF Stoorvogel W Geuze HJ Exosome: from internal vesicle of the multivesicular body to intercellular signaling device J Cell Sci 2000 113 3365 74 10984428
Schartz NE Chaput N Andre F Zitvogel L From the antigen-presenting cell to the antigen-presenting vesicle: the exosomes Curr Opin Mol Ther 2002 4 372 81 12222875
Johnstone RM The Jeanne Manery-Fisher Memorial Lecture 1991. Maturation of reticulocytes: formation of exosomes as a mechanism for shedding membrane proteins Biochem Cell Biol 1992 70 179 90 1515120
Denzer K van Eijk M Kleijmeer MJ Jakobson E de Groot C Geuze HJ Follicular dendritic cells carry MHC class II-expressing microvesicles at their surface J Immunol 2000 165 1259 65 10903724
Andre F Schartz NE Movassagh M Flament C Pautier P Morice P Pomel C Lhomme C Escudier B Le Chevalier T Tursz T Amigorena S Raposo G Angevin E Zitvogel L Malignant effusions and immunogenic tumor derived-exosomes Lancet 2002 360 295 305 12147373 10.1016/S0140-6736(02)09552-1
Thery C Duban L Segura E Veron P Lantz O Amigorena S Indirect activation of naive CD4+ T cells by dendritic cell-derived exosomes Nat Immunol 2002 3 1156 62 12426563 10.1038/ni854
Hsu DH Paz P Villaflor G Rivas A Mehta-Damani A Angevin E Zitvogel L Le Pecq JB Exosomes as a tumor vaccine: enhancing potency through direct loading of antigenic peptides J Immunother 2003 26 440 50 12973033 10.1097/00002371-200309000-00007
Thery C Boussac M Veron P Ricciardi-Castagnoli P Raposo G Garin J Amigorena S Proteomic analysis of dendritic cell-derived exosomes: a secreted subcellular compartment distinct from apoptotic vesicles J Immunol 2001 166 7309 18 11390481
Thery C Regnault A Garin J Wolfers J Zitvogel L Ricciardi-Castagnoli P Raposo G Amigorena S Molecular characterization of dendritic cell-derived exosomes. Selective accumulation of the heat shock protein hsc73 J Cell Biol 1999 147 599 610 10545503 10.1083/jcb.147.3.599
Thery C Zitvogel L Amigorena S Exosomes: Composition, biogenesis and function Nat Rev Immunol 2002 2 569 579 12154376
André F Chaput N Schartz NEC Exosomes as potent cell-free peptide-based vaccine. I. Dendritic cell-derived exosomes transfer functional MHC ClassI/peptide complexes to dendritic cells J Immunol 2004 172 2126 2136 14764678
Lamparski H Metha-Damani A Yao J Patel S Hsu D Ruegg C Le Pecq J Production and characterization of clinical grade exosomes derived from dendritic cells J Immunol Methods 2002 270 211 12379326
Schultz ES Lethé B Cambiaso CL A Mage-A3 peptide presented by HLA-DP4 is recognized on tumor cells by CD4+ cytolytic T lymphocytes Cancer Res 2000 60 6272 6275 11103782
Whiteside TL Immunologic monitoring of clinical trials in patients with cancer: technology versus common sense Immunol Invest 2000 29 149 162 10854184
Garcia-Lora A Algarra I Garrido F MHC class I antigens, immune surveillance and tumor immune escape J Cell Physiol 2003 195 346 351 12704644 10.1002/jcp.10290
Schiller JH Harrington D Belani CP Comparison of four chemotherapy regimens for advanced non-small-cell lung cancer N Engl J Med 2002 346 92 98 11784875 10.1056/NEJMoa011954
Fossella FV DeVore R Kerr RN Randomized phase III trial of docetaxel versus vinorelbine or ifosfamide in patients with advanced non-small-cell lung cancer previously treated with platinum-containing chemotherapy regimens. The TAX 320 Non-Small Cell Lung Cancer Study Group J Clin Oncol 2000 18 2354 2362 10856094
Massarelli E Andre F Liu DD A retrospective analysis of the outcome of patients who have received two prior chemotherapy regimens including platinum and docetaxel for recurrent non-small-cell lung cancer Lung Cancer 2003 39 55 61 12499095 10.1016/S0169-5002(02)00308-2
Pfister DG Johnson DH Azzoli CG American Society of Clinical Oncology treatment of unresectable non-small-cell lung cancer guideline: update 2003 J Clin Oncol 2004 22 330 353 14691125 10.1200/JCO.2004.09.053
| 15723705 | PMC551593 | CC BY | 2021-01-04 16:39:28 | no | J Transl Med. 2005 Feb 21; 3:9 | utf-8 | J Transl Med | 2,005 | 10.1186/1479-5876-3-9 | oa_comm |
==== Front
Nutr Metab (Lond)Nutrition & Metabolism1743-7075BioMed Central London 1743-7075-2-61572536010.1186/1743-7075-2-6Brief CommunicationThe role of the vomeronasal system in food preferences of the gray short-tailed opossum, Monodelphis domestica Halpern Mimi [email protected] Yasmine [email protected] Ido [email protected] Department of Anatomy and Cell Biology, SUNY Downstate Medical Center, 450 Clarkson Avenue, Brooklyn, NY 11203 USA2005 22 2 2005 2 6 6 3 2 2005 22 2 2005 Copyright © 2005 Halpern et al; licensee BioMed Central Ltd.2005Halpern 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.
Although feeding deficits have been reported in snakes and lizards following vomeronasal system disruption, no deficit has been previously reported in a mammal. We tested gray short-tailed opossums with items from four different food categories prior to occluding access to the vomeronasal organ. Preoperatively, opossums preferred meat to fruit or vegetables. Following occlusion of the nasopalatine canal, but not after control treatment, opossums failed to demonstrate food preferences.
==== Body
The vomeronasal system (VNS) is a nasal chemosensory system usually considered primarily a pheromone-detecting system [[1-4] for recent reviews]. The system, present in most terrestrial vertebrates, is comprised of the peripherally situated vomeronasal organ (VNO), whose sensory neurons project their axons to the accessory olfactory bulb (Figure 1), and the further projections from the accessory bulb into the limbic forebrain [1-4].
Figure 1 Head of an opossum as viewed from the side illustrating the position of the vomeronasal organ, olfactory epithelium and their respective connections to the accessory olfactory bulb and main olfactory bulb.
In snakes and lizards, in addition to its role in pheromone detection, the VNS is important for feeding behavior [5-7]. To date, no study has reported a feeding deficit in a mammal deprived of a functional VNS. It is possible that the failure to report feeding deficits in rats, mice and hamsters deprived of a functional VNS is a result of the fact that these animals have been fed laboratory diets over many generations. It is also conceivable that with the emphasis on the pheromonal function of the VNS, the issue of the role of the VNS in feeding behavior was overlooked.
Gray short-tailed opossums, Monodelphis domestica, are members of the Didelphidae or American opossums. The didelphids are the most ancient marsupial family and thought to be the origin of all other New World and Australian marsupials [8,9]. M. domestica was recently introduced into laboratories [10] and has, therefore, not been fed a laboratory diet over many generations. As these opossums respond to a variety of foods [11-13], they are good models to investigate the role of the VNS in food preferences in mammals.
Six male and seven female gray short-tailed opossums, 12–16 months old, were used as test subjects in this study. The opossums were progeny of animals originally purchased from the Southwest Foundation for Biomedical Research (San Antonio, Texas, USA) and were offspring of different parents. The opossums were housed individually in plastic cages (42 × 21 × 20 cm) with wood shavings as bedding, and provided with cylindrical plastic containers for nesting. The animals had dry fox food (Milk Specialties Co., New Holstein, WI, USA) and water available ad libitum.
To determine which foods to use in the formal experiment, a preliminary test was conducted in which 24 different foods, representing meats, fruits and vegetables were presented to the opossums. Each opossum was presented with one food per day, selected at random, and its consumption of the food noted. The animals were allowed three minutes to investigate and eat the food, which was placed on an 8 × 20 cm cardboard tray at the front of the home cage. Each animal received two trials with each food. Of the 24 foods tested 16 were selected for the formal experiment based on the animals' observed preferences.
During testing the opossums were simultaneously presented with four foods one from each of the following food groups: fruits (apples, oranges, peaches, cantaloupes), meats (mealworms, chicken, pork, crickets), processed vegetables (Raisin Bran, Cheerios, whole wheat bread, bagel) and unprocessed vegetables (corn, peppers, carrots, broccoli). The animals were tested daily with one of 16 food combinations (4 foods × 4 categories). Four sterile Petri dishes, each 3.5 × 1.0 cm, were placed equidistant from each other and glued to an 8 × 20 cm cardboard tray. One piece (0.21 cm3) of food from each food category, was placed in a dish.
Before each test session, feeding troughs containing fox food were removed from each opossum's cage. At the beginning of a trial the food tray was placed in the front of the animal's home cage and the behavior of the animal videotaped for three minutes using a digital video-camera (Sony DCR-VX2000, 30 frames/sec) placed at a distance of 3 m from the test cage. Trays, dishes and unconsumed food were discarded after each trial. After testing, food troughs containing fox food were replaced in the animals' home cages. Fresh disposable gloves were used when handling dishes, trays and foods to prevent transfer of food odors.
Videotapes of food trials were analyzed using a videocassette recorder (JVC SR-VS30U), BTV Pro 5.4.1 and "Videoanalyzer" software (designed by John L. Kubie, Downstate Medical Center).
For control surgery and occlusion of the nasopalatine canal, opossums were lightly anesthetized with Ketaset (0.25 cc/100 g, i.m.) and atropine sulfate (.05 cc/100 g, i.m.) Access to the vomeronasal organ (VNO) was blocked with gel foam and Crazy Glue™ (Elmer's Products, Inc., Columbus, OH) applied to the roof of the mouth, covering the opening to the nasopalatine canal. Cresyl violet crystals were added to the Crazy Glue to facilitate visualization of the block. Opossums were visually checked daily to insure that the VNO block remained in place. Control surgery consisted of placing Crazy Glue on the roof of the mouth to either side of the nasopalatine canal. Opossums were allowed two days to recover before postoperative testing.
The VNO block remained in place for only one test day in all thirteen opossums. Therefore, pre-post operative and control comparisons were only made for trials identical in composition (foods tested) and order (position on the tray) to the first postoperative test day. Control postoperative trials were run as described above. All 13 animals were tested under both control and experimental (nasopalatine canal blocked) conditions, with the order of surgical treatment randomized.
The first food selected and consumed on each trial was identified as the animal's food choice. Statistical analyses utilized a chi-square test of the frequencies of choices of each food type for pre-operative, control and experimental trials. Figure 2 depicts the results in terms of the percentage of choices of each food type under the three conditions.
Figure 2 Percentage of food choice responses (first food type selected and consumed) by opossums prior to surgery (Preop), after control occlusion (Control) and after occlusion of the nasopalatine canal (Postop).
Preoperatively, opossums preferred meats to fruits and fruits to processed and unprocessed vegetables (Figure 2). The most preferred food within the meat category was crickets, within the fruit category was cantaloupe, within the unprocessed vegetable category was corn and within the processed vegetable category was whole wheat bread.
Whereas the preoperative comparison trials on the thirteen opossums revealed significant preferences among the food categories (χ2 = 10.08, df = 3, p < .02), postoperatively, no significant difference in food category preference was observed (χ2 = 0.85, df = 3, p > .05). During control trials the opossums continued to demonstrate significant food preferences (χ2 = 9.46, df = 3, p < .05,). Meat was again, the most preferred food category (Figure 2).
One deficiency of this study is that, because the opossums removed the glue blocks after the first day, we were unable to verify that the nasopalatine occlusion had, indeed, prevented access to the vomeronasal organ. However, we had previously [14] demonstrated, using the identical technique, that this method prevented access of substances to the VNO. Furthermore, the failure of opossums with nasopalatine canal occlusion to demonstrate the food preferences observed during preoperative and under control conditions, strongly suggests that the block was effective.
This study suggests that without a functional VNS, the food preferences of gray short-tailed opossums are significantly impaired. Previous studies on the VNS of mammals have not addressed the issue of changes in feeding behavior. It remains to be seen whether other mammals, newly introduced into the laboratory, or bred over many generations in the laboratory, also demonstrate a food preference deficit when deprived of a functional VNS. Studies comparing wild populations with laboratory-reared populations might contribute information resolving this issue.
It is not surprising that food preferences might be influenced by vomeronasal stimulation since the vomeronasal organ of many vertebrates, including opossums, is directly accessible from the oral cavity [2] via the nasopalatine duct. Thus, food in the mouth could be sensed by the vomeronasal system. Whether this mechanism is, in fact, utilized in food selection is not known and would have to be the subject of future investigation. It is likely, however, that the expression of food preference is a result of the interaction of multiple sensory systems including taste, olfaction, vision, trigeminal and vomeronasal.
Authors' contributions
MH conceived of and designed the experiment and wrote the manuscript
YD conducted the experiment, extracted the data from the videotapes and analyzed the data statistically
IZ supervised YD in the conduct of the experiment and instructed YD in the experimental procedures and data analysis
Acknowledgements
This research was supported by NIDCD grant #DC0745. We thank Linda Bartoshuk for advice during the planning of this experiment.
==== Refs
Brennan PA The vomeronasal system, Cell Mol Life Sci 2001 58 546 555 11361090
Halpern M The organization and function of the vomeronasal system Annu Rev Neurosci 1987 10 325 362 3032065 10.1146/annurev.ne.10.030187.001545
Halpern M Martínez-Marcos A Structure and Function of the Vomeronasal System: An Update Prog Neurobiol 2003 70 245 318 12951145 10.1016/S0301-0082(03)00103-5
Wysocki CJ Meredith M Finger TE, Silver WL The vomeronasal system Neurobiology of Taste and Smell 1987 New York: John Wiley & Sons 125 150
Graves BM Halpern M Roles of vomeronasal organ chemoreception in tongue flicking, exploratory and feeding behaviour of the lizard, Calcides ocellatus Anim Behav 1990 39 692 698
Halpern M Frumin N Roles of the vomeronasal and olfactory systems in prey attack and feeding in adult garter snakes, Physiol Behav 1979 22 1183 1189 573911 10.1016/0031-9384(79)90274-9
Kubie JL Halpern M The chemical senses involved in garter snake prey trailing J Comp Physiol Psych 1979 93 648 667
Clemens WA Origin and early evolution of marsupials Evolution 1968 22 1 18
Keast A Stonehouse F, Gilmore D Historical biogeography of the marsupials Biology of Marsupials 1977 Baltimore: University Park Press 69 95
VandeBerg JL The gray short-tailed opossums: a new laboratory animal ILARNews 1983 26 9 12
Cothran EG Aivaliotis MJ VandeBerg JL The effects of diet on growth and reproduction in gray short-tailed opossums (Monodelphis domestica) J Exp Zool 1985 236 103 114 4056701
Fadem BH Trupin GL Maliniak E VandeBerg JL Hayssen V Care and breeding of the gray, short-tailed opossum (Monodelphis domestica) Lab Anim Sci 1982 32 405 409 7144118
Streilein KE Mares MAR, Genoways HH Behavior, ecology and distribution of South American marsupials, Mammalian Biology of South America 1982 Philadelphia: U. of Pittsburg 231 250 [Special publication series #6.]
Poran NS Vandoros A Halpern M Nuzzling in the gray short-tailed opossum I: Delivery of odors to vomeronasal organ Physiol Behav 1993 53 959 967 8511213 10.1016/0031-9384(93)90275-K
| 15725360 | PMC551594 | CC BY | 2021-01-04 16:37:46 | no | Nutr Metab (Lond). 2005 Feb 22; 2:6 | utf-8 | Nutr Metab (Lond) | 2,005 | 10.1186/1743-7075-2-6 | oa_comm |
==== Front
Cancer Cell IntCancer Cell International1475-2867BioMed Central London 1475-2867-5-31571004410.1186/1475-2867-5-3Primary ResearchMelanoma Inhibitory Activity (MIA) increases the invasiveness of pancreatic cancer cells El Fitori Jamael [email protected] Jörg [email protected] Nathalia A [email protected] Ahmed [email protected] Anja K [email protected]üchler Markus W [email protected] Helmut [email protected] Department of General Surgery, University of Heidelberg, Germany2 Institute of Pathology, University of Regensburg, Germany2005 14 2 2005 5 3 3 21 11 2004 14 2 2005 Copyright © 2005 El Fitori et al; licensee BioMed Central Ltd.2005El Fitori 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
Melanoma inhibitory activity (MIA) is a small secreted protein that interacts with extracellular matrix proteins. Its over-expression promotes the metastatic behavior of malignant melanoma, thus making it a potential prognostic marker in this disease. In the present study, the expression and functional role of MIA was analyzed in pancreatic cancer by quantitative real-time PCR (QRT-PCR), immunohistochemistry, immunoblot analysis and ELISA. To determine the effects of MIA on tumor cell growth and invasion, MTT cell growth assays and modified Boyden chamber invasion assays were used.
Results
The mRNA expression of MIA was 42-fold increased in pancreatic cancers in comparison to normal pancreatic tissues (p < 0.01). In contrast, MIA serum levels were not significantly different between healthy donors and pancreatic cancer patients. In pancreatic tissues, MIA was predominantly localized in malignant cells and in tubular complexes of cancer specimens, whereas normal ductal cells, acinar cells and islets were devoid of MIA immunoreactivity. MIA significantly promoted the invasiveness of cultured pancreatic cancer cells without influencing cell proliferation.
Conclusion
MIA is over-expressed in pancreatic cancer and has the potential of promoting the invasiveness of pancreatic cancer cells.
MIApancreatic cancerinvasionmetastasistumor cell invasion
==== Body
Background
Despite improvements in diagnosis and treatment, pancreatic cancer remains one of the most common causes of cancer-related deaths in the world [1]. One of the reasons for the dismal prognosis is the propensity of pancreatic cancer cells to invade surrounding tissues and to metastasize. Melanoma inhibitory activity (MIA) is a small secreted protein normally expressed in cartilage and also produced by malignant melanoma cells and to a lesser degree by breast, colon cancer, and glioblastoma cells [2-4]. The exact biological functions of MIA are still unclear, but recent evidence indicates an important role of MIA in tumor progression and metastasis. MIA has been shown to interact with the components of the extracellular matrix, such as fibronectin and laminin, possibly via the binding motif for integrins. For example, MIA inhibits the attachment of suspended melanoma cells to surfaces coated with laminin or fibronectin [5]. In addition, overexpression of MIA in melanoma cells induces an aggressive tumor type by enhancing the metastatic potential. It has been shown that there is a correlation between increased plasma levels of MIA and a more advanced metastatic disease state in malignant melanoma patients [6]. Previously, using DNA array technology, we have demonstrated an increase of MIA mRNA expression in pancreatic cancer in comparison with the normal pancreas [7]. In the present study, we have further investigated the expression and localization of MIA, and its functional role in pancreatic cancer.
Results
To quantify the mRNA expression of MIA in pancreatic tissues, QRT-PCR was performed using RNA from pancreatic cancer tissues (n = 23) and normal pancreatic tissue samples (n = 17). The analysis revealed a 42 ± 28-fold increase (p = 0.0013) of MIA mRNA levels in pancreatic cancer tissues compared to the normal pancreas (Fig. 1A). To determine the exact localization of MIA in pancreatic tissues, immunostaining was carried out in 32 primary pancreatic cancers and in 17 normal pancreatic tissue samples. In the normal pancreas, MIA immunoreactivity was absent in ductal, acinar and islets cells, but was observed in the muscular layer of vessels (Fig. 2A–C). In contrast, in pancreatic cancer tissues, MIA immunoreactivity was moderate in the cancer cells and in tubular complexes in CP-like lesions adjacent to the tumor mass (Fig. 2D–F), and in blood vessels and nerve ganglia (Fig. 2G,H). Further analysis to correlate different tumor stages and grades with MIA immunoreactivity in pancreatic cancer tissues was performed. No difference between tumor stages, grades and MIA immunostaining was observed. To ensure the specificity of the antibody used, a malignant melanoma metastasis to the peritoneum was also analyzed. Strong cytoplasmic MIA staining of the malignant melanoma cells but not the surrounding tissues could be detected (Fig. 2I). Since a correlation between increased serum levels of MIA in malignant melanoma and more advanced metastatic disease has been reported previously, we next investigated MIA serum levels in pancreatic cancer patients and healthy donors. The mean values of MIA serum levels were 8.3 ± 3.56 ng/ml in pancreatic cancer patients (n = 50) and 8.82 ± 2.01 ng/ml in control subjects (n = 14) (n.s.) (Fig. 1B). Further analysis revealed that there was also no significant difference between patient groups with different tumor stages or grades.
Figure 1 MIA mRNA expression and serum levels. A: MIA mRNA values in normal pancreatic tissues and pancreatic ductal adenocarcinoma (cancer) tissues by real-time quantitative polymerase chain reaction, as described in the Methods section. Values were normalized to housekeeping genes (cyclophilin B and hypoxanthine guanine phosphoribosyltransferase). B: ELISA was carried out as described in the Methods section. Fifty pancreatic cancer sera samples and 14 healthy donor sera samples were analyzed. The horizontal lines represent the mean expression levels.
Figure 2 Localization of MIA in pancreatic tissues. MIA localization in pancreatic tissues: immunohistochemistry using a MIA specific antibody was carried out as described in the Methods section. A-C: normal pancreatic tissues; D-H: pancreatic cancers; I: malignant melanoma metastasis to the peritoneum. Note that in melanoma metastasis the signal is red (using HistoMark Red phosphatase system) to differentiate the staining from the brown pigment in melanoma cells.
To determine the functional role of MIA in pancreatic cancer, we first investigated MIA expression in pancreatic cell lines. QRT-PCR analysis revealed relatively high MIA mRNA levels in Mia PaCa-2, Panc-1, and SU8686 pancreatic cancer cells compared to the other cell lines (Fig. 3A). Immunoblot analysis was employed to evaluate MIA protein levels in pancreatic cancer cell lines. This analysis revealed a band of 15 kDa corresponding to the known size of MIA in the control melanoma cell line B16 (B78/H1), whereas it was only weakly present in Mia Paca-2, Panc-1, and SU8686 pancreatic cancer cell lines, and below the level of detection in the other tested cell lines (Fig 3B). To confirm the specificity of the signal, immunoprecipitation was performed. This demonstrated an immunospecific band at 15 kDa in B78/H1, Mia PaCa-2, and SU8686 cell lines (Fig. 3C).
Figure 3 Expression and effects of MIA in cultured pancreatic cancer cells. A: MIA mRNA levels in indicated pancreatic cancer cell lines were determined by real-time quantitative polymerase chain reaction, as described in the Methods section. Data are presented as median + SD of MIA mRNA copies per μl of input cDNA normalized to housekeeping genes CPB and HPRT. B: 30 μg protein lysates of the indicated cell lines were subjected to immunoblotting analysis using a specific MIA antibody, as described in the Methods section. C: Immunoprecipitation analysis was carried out as described in the Methods section. D: An in vitro cell invasion assay was performed using 8 μM filters coated with Matrigel, as described in the Methods section. 4 × 105of the indicated pancreatic cancer cells were seeded onto the filters in 10% serum overnight, and then treated as indicated for 24 h. Invaded cells were stained and counted. The values shown are the mean ± SEM obtained from three independent experiments.
In order to determine the effects of MIA on the proliferation of pancreatic cancer cells, MTT cell growth assays were performed next. This analysis revealed no significant effects of MIA on the proliferation of pancreatic cancer cells (data not shown). To analyze whether MIA may promote invasiveness of pancreatic cancer cells, Matrigel-based invasion assays using a modified Boyden chamber were carried out in T3M4 and Aspc-1 pancreatic cancer cell lines that exhibited relatively low MIA expression levels according to QRT-PCR and Western blot data. Added to the top chamber as described in Methods section, MIA significantly increased the invasion of Aspc-1 by 2.4-fold (p < 0.001) and the invasion of T3M4 by 3.1-fold (p < 0.0001) (Fig. 3D). To investigate possible mechanisms of MIA overexpression in pancreatic cancer cells, we analyzed whether micro-environmental changes may modulate MIA expression. Neither TGF-β1 nor hypoxia was able to alter MIA mRNA expression, according to the QRT-PCR analysis of correspondingly treated pancreatic cancer cell lines (data not shown).
Discussion
One of the most devastating aspects of malignant growth is the emergence of cancer foci in organs distant from the primary tumor, with most cancer mortality being related to metastases. Thus, understanding the molecular mechanisms underlying the metastatic process is one of the most important issues in cancer research.
Pancreatic cancer is characterized by aggressive local tumor growth and early systemic tumor spread [8-10]. Many factors are involved in transforming pancreatic cancer into a highly aggressive and metastatic disease, such as alterations in cell-cell interaction [11], deregulated expression of extracellular proteases [12], and metastasis-associated genes such as KAI-1, heparanase [13,14] and a number of other molecules.
Another important aspect of the metastasis process is neo-angiogenesis. Angiogenesis itself encompasses a cascade of sequential processes emanating from microvascular endothelial cells, which are stimulated to proliferate, degrade the endothelial basement membranes of parental vessels, migrate, penetrate host stroma, and initiate a capillary sprout [15]. Numerous angiogenic factors are overexpressed in pancreatic cancer, including vascular endothelial growth factor (VEGF), bFGF, and angiogenin, as well as members of the TGF-β, and FGF gene families [16-20]. In order to migrate and metastasize, cancer cells have to overcome and move through natural barriers created by cell-cell and cell-extracellular matrix (ECM) adhesion structures. Any destruction in cell-cell and ECM networks will facilitate motility and allow the cancer cells to migrate and metastasize. The invasion and metastatic potential of cancer cells depends on their intrinsic properties and the host microenvironment [21].
Melanoma inhibitory activity (MIA) increases cell motility by decreasing the attachment of the cells to the extracellular matrix (ECM). Overexpression of MIA leads to increased metastasis of malignant melanoma cells by enhancing invasion and extravasation [22,23]. In the present study we show by QRT-PCR and immunohistochemistry that MIA is significantly over-expressed in pancreatic cancer in comparison with normal pancreatic tissues. These data are in agreement with findings in malignant melanoma and breast cancer, in which MIA is also highly expressed [6,24,25].
In contrast to observations in malignant melanoma, where MIA has been established as a reliable marker for prognosis [6,24], we could not detect either a significant difference of MIA serum levels between pancreatic cancer patients and donors or a significant difference between patients at different stages of pancreatic cancer. Therefore, MIA cannot serve as a diagnostic or prognostic marker in pancreatic cancer.
The reason why high MIA mRNA levels lead to high serum levels in malignant melanoma, but not in pancreatic cancer, is currently not known. As to the possible role of MIA in pancreatic cancer pathogenesis, MIA had no effect on the proliferation of pancreatic cancer cells, similar to previous experiments employing fibroblasts, keratinocytes, endothelial cells and lymphocytes. The only cellular system in which MIA has been found to influence cell growth is melanoma cells; in these cells, MIA exerts anti-proliferative effects [26].
Although MIA did not affect the growth of pancreatic cancer cells in vitro, its impact on the invasion was striking. Interestingly, promotion of invasiveness of pancreatic cancer contrasts the previously demonstrated decrease in the invasion of MIA-treated malignant melanoma cells [26]. A possible explanation for this dissimilarity might be that MIA has no effect in detachment of pancreatic cancer cells from the ECM, like in malignant melanoma [26]. Alternatively, the status of MIA-interacting partners or downstream targets in the MIA signaling pathway may differ between pancreatic and melanoma cancer cells. So far, a possible correlation between MIA and other proteins known to be involved in pancreatic cancer metastasis – such as extracellular proteases, MMP, VEGF and bFGF – has not been studied. Therefore, the mechanism responsible for the invasion-promoting action of MIA requires further evaluation.
Conclusion
In conclusion, our study shows a striking overexpression of MIA in cancerous pancreatic tissues without consequent elevation of MIA in the circulation. Involvement of MIA in regulation of invasiveness of pancreatic cancer cells indicates that this protein may serve as a novel therapeutic target in the search for anti-metastatic drugs.
Methods
Patients and tissues
Pancreatic cancer tissue samples were collected from 55 patients who underwent pancreatic cancer resection in the Department of General Surgery at the University of Heidelberg, Germany, and in the Department of Visceral and Transplantation Surgery at the University of Bern, Switzerland. Twelve cases were stage I, 13 cases stage II, 23 cases stage III, and 7 cases stage IV pancreatic adenocarcinomas, according to the Union International Contre Le Cancer (UICC) system. According to routine pathological grading, 16 cases were well-differentiated, 22 moderately differentiated, and 17 poorly differentiated. Normal pancreatic tissues were collected from 34 healthy organ donors. Pancreatic tissues were either frozen in liquid nitrogen and stored at -80°C (for RNA and protein extraction) or immediately fixed in 4% paraformaldehyde solution and subsequently embedded in paraffin. In order to determine MIA serum concentrations, sera from 50 pancreatic cancer patients (35 male, 15 female; median age 59 years; range 29–80 years) and healthy volunteers (14 male; median age 27 years; range 25–35 years) were collected at the Department of General Surgery, University of Heidelberg, Germany. Written informed consent was obtained from all patients. The study was approved by the Ethics Committees of the Universities of Bern, Switzerland, and Heidelberg, Germany.
Cell lines and culture conditions
Mia PaCa-2, T3M4, Aspc-1, Bxpc-3, Capan-1, Colo-357, SU8686 and Panc-1 pancreatic cancer cells and B16 (cloneB78/H1) mouse melanoma cells were grown in RPMI 1640 medium containing 10% FBS (fetal bovine serum), 100 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen, Karlsruhe, Germany). Cells were maintained in a 37°C humidified atmosphere saturated with 5% CO2. For TGF-β1 induction experiments, pancreatic cancer cells were seeded in 10 cm dishes in 10% FBS growth medium and allowed to attach for 12 hrs. Growth medium was replaced by serum-reduced medium (0.5% FBS), supplemented with 200 pM TGF-β1 for the indicated time periods. For experimental hypoxia, cells were subjected to a hypoxic microenvironment by one hour-long flushing in a special incubator chamber with an anoxic gas mixture (89.25% N2, 10% CO2, 0.75%O2) and sealing of the unit.
Real-time quantitative polymerase chain reaction (QRT-PCR)
All reagents and equipment for mRNA/cDNA preparation were purchased from Roche Applied Science (Mannheim, Germany). mRNA was prepared by automated isolation using MagNA Pure LC instrument and isolation kits I (for cells) and II (for tissue). cDNA was prepared using a 1st strand cDNA synthesis kit for RT-PCR according to the manufacturer's instructions. Real-time PCR was performed with the Light Cycler Fast Start DNA SYBR Green kit [27]. The number of specific transcripts was normalized to housekeeping genes (cyclophilin B and hypoxanthine guanine phosphoribosyltransferase, HPRT). All primers were obtained from Search-LC (Heidelberg, Germany).
Immunohistochemistry
Briefly, consecutive paraffin-embedded tissue sections (5 μm thick) were deparaffinized and rehydrated. Antigen retrieval was performed by pretreatment of the slides in citrate buffer (pH 6.0) in a microwave oven for 10 min. Thereafter, slides were cooled to room temperature in deionized water for 5 min. After blocking of endogenous peroxidase activity with 0.3% hydrogen peroxide and washing in deionized water 3 times for 10 min, the sections were blocked for 1 h at room temperature with normal rabbit serum (DAKO, Hamburg, Germany), then incubated with primary goat polyclonal anti-MIA antibody (A-20, Santa Cruz Biotechnology, Santa Cruz, CA; dilution 1:35 in normal rabbit serum) overnight at 4°C. The slides were rinsed with washing buffer (Tris-buffered saline with 0.1% BSA) and incubated with secondary rabbit anti-goat HRPO-labeled IgG (Sigma-Aldrich, Taufkirchen, Germany), diluted 1:200 for 45 min at room temperature. After color reaction, tissues were counterstained with Mayer's hematoxylin. For negative control, appropriately diluted goat IgG was used instead of the primary antibody.
Enzyme-linked immunosorbent assay (ELISA)
The amount of secreted MIA protein in cell culture supernatants and serum samples was determined using a one-step MIA ELISA (Roche Diagnostic GmbH, Mannheim, Germany) according to the manufacturer's instructions.
Immunoblot
Cells were washed with ice-cold PBS and collected in lysis buffer (50 mM Tris-HCl, 100 mM NaCl, 2 mM EDTA, 1% SDS) containing the Complete mini-EDTA-free protease inhibitor cocktail tablets from Roche (Roche Applied Science, Mannheim, Germany). Lysates were centrifuged at 13,000 rpm at 4°C for 30 min, the supernatants were collected, and protein concentrations were measured with the BCA protein assay (Pierce Chemical Co., Rockford, IL, USA) using BSA as protein standard. 20 μg of protein were mixed with loading buffer, heated at 95°C for 5 min, separated on 12% SDS polyacrylamide gels, and transferred onto nitrocellulose membrane at 100 V for 90 min. Membranes were blocked in 5% non-fat milk in TBS-T (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20) for 1 h, incubated overnight at 4°C with anti-MIA antibody (A-20, Santa Cruz) and exposed to secondary HRPO-labeled donkey anti-goat antibody (Santa Cruz) for 1 h at room temperature. The signal detection was performed using the ECL system (Amersham Life Science, Amersham, UK).
Immunoprecipitation
For immunoprecipitation, pancreatic cell lines (Mia PaCa-2 and SU8686) were suspended in lysis buffer (50 mM Tris, 150 mM NaCl, 1% Triton X-100, 25 mM NaF, 10% glycerol, 1 mM PMSF) supplemented with the Complete-TM mixture of proteinase inhibitors (Roche Diagnostic, Mannheim, Germany) and incubated for 30 min on ice. After centrifugation, the supernatant was transferred into a fresh vial, pre-cleared with protein A-Sepharose beads (Santa Cruz) and incubated with 50 μl anti-MIA antibody (A-20, Santa Cruz) overnight at 4°C. Following addition of 30 μl of protein A-Sepharose for 1 h at 4°C, the mixture was pelleted, washed three times with lysis buffer, and resuspended in Laemmli sample buffer.
MTT cell growth assays
Cell growth experiments were performed using the 3-(4, 5-methylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay. Pancreatic cancer cells were seeded at a density of 5000 cells/well in 96-well plates, grown overnight, and then exposed to different concentrations of recombinant MIA protein as indicated. After 24 h, MTT was added (50 μg/well) for 4 hours. Formazan products were solubilized with acidic isopropanol, and the optical density was measured at 570 nm.
Invasion assays
Invasion assays were performed in a BD Biocoat Matrigel Invasion Chamber with 8-μm pore size (BD Biosciences, Heidelberg, Germany) according to the manufacturer's instructions. The Matrigel was rehydrated with 500 μl DMEM (serum-free) and incubated in a 37°C, 5% CO2 atmosphere for 2 h. 5 × 104 cells were incubated for 24 h and subsequently treated with MIA (100 ng/ml) [26], which was added to the top chamber and incubated for 24 h. The non-invading cells were removed from the upper surface of the membrane with cotton-tipped swabs. Cells adhering to the lower surface were fixed with 75% methanol mixed with 25% acetone and stained with 1% toluidine blue (Sigma-Aldrich, Taufkirchen, Germany). The whole membrane was scanned using the software of the Zeiss KS300 and Zeiss AxioCam HR system (Jena, Germany). To calculate the total number of all invading cells, the cells were counted in every cut-out of the mosaic image of the whole membrane using the same software. The assays were performed in duplicate and repeated three times.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contribution
JEF, NG, and AG carried out all the experiments, and participated in data analysis and interpretation. JK, MWB, and HF conceived of the study, and participated in its design and coordination. JK, NG, and AKB analyzed and interpreted the data and drafted the manuscript. All authors read and approved the final version.
==== Refs
Jemal A Tiwari RC Murray T Ghafoor A Samuels A Ward E Feuer EJ Thun MJ Cancer statistics, 2004 CA Cancer J Clin 2004 54 8 29 14974761
Bosserhoff AK Moser M Hein R Landthaler M Buettner R In situ expression patterns of melanoma-inhibiting activity (MIA) in melanomas and breast cancers J Pathol 1999 187 446 454 10398105 10.1002/(SICI)1096-9896(199903)187:4<446::AID-PATH267>3.0.CO;2-Y
de Vries TJ Fourkour A Punt CJ Diepstra H Ruiter DJ van Muijen GN Melanoma-inhibiting activity (MIA) mRNA is not exclusively transcribed in melanoma cells: low levels of MIA mRNA are present in various cell types and in peripheral blood Br J Cancer 1999 81 1066 1070 10576666 10.1038/sj.bjc.6690808
Dietz UH Sandell LJ Cloning of a retinoic acid-sensitive mRNA expressed in cartilage and during chondrogenesis J Biol Chem 1996 271 3311 3316 8621736 10.1074/jbc.271.6.3311
Bosserhoff AK Kondo S Moser M Dietz UH Copeland NG Gilbert DJ Jenkins NA Buettner R Sandell LJ Mouse CD-RAP/MIA gene: structure, chromosomal localization, and expression in cartilage and chondrosarcoma Dev Dyn 1997 208 516 525 9097023 10.1002/(SICI)1097-0177(199704)208:4<516::AID-AJA7>3.0.CO;2-L
Bosserhoff AK Kaufmann M Kaluza B Bartke I Zirngibl H Hein R Stolz W Buettner R Melanoma-inhibiting activity, a novel serum marker for progression of malignant melanoma Cancer Res 1997 57 3149 3153 9242442
Friess H Ding J Kleeff J Fenkell L Rosinski JA Guweidhi A Reidhaar-Olson JF Korc M Hammer J Buchler MW Microarray-based identification of differentially expressed growth- and metastasis-associated genes in pancreatic cancer Cell Mol Life Sci 2003 60 1180 1199 12861384
DiMagno EP Reber HA Tempero MA AGA technical review on the epidemiology, diagnosis, and treatment of pancreatic ductal adenocarcinoma. American Gastroenterological Association Gastroenterology 1999 117 1464 1484 10579989
Neoptolemos JP Dunn JA Stocken DD Almond J Link K Beger H Bassi C Falconi M Pederzoli P Dervenis C Fernandez-Cruz L Lacaine F Pap A Spooner D Kerr DJ Friess H Buchler MW Adjuvant chemoradiotherapy and chemotherapy in resectable pancreatic cancer: a randomised controlled trial Lancet 2001 358 1576 1585 11716884 10.1016/S0140-6736(01)06651-X
Warshaw AL Fernandez-del CC Pancreatic carcinoma N Engl J Med 1992 326 455 465 1732772
Karayiannakis AJ Syrigos KN Polychronidis A Simopoulos C Expression patterns of alpha-, beta- and gamma-catenin in pancreatic cancer: correlation with E-cadherin expression, pathological features and prognosis Anticancer Res 2001 21 4127 4134 11911306
Coussens LM Werb Z Matrix metalloproteinases and the development of cancer Chem Biol 1996 3 895 904 8939708 10.1016/S1074-5521(96)90178-7
Friess H Guo XZ Berberat P Graber HU Zimmermann A Korc M Buchler MW Reduced KAI1 expression in pancreatic cancer is associated with lymph node and distant metastases Int J Cancer 1998 79 349 355 9699525 10.1002/(SICI)1097-0215(19980821)79:4<349::AID-IJC7>3.0.CO;2-V
Koliopanos A Friess H Kleeff J Shi X Liao Q Pecker I Vlodavsky I Zimmermann A Buchler MW Heparanase expression in primary and metastatic pancreatic cancer Cancer Res 2001 61 4655 4659 11406531
Holmgren L O'Reilly MS Folkman J Dormancy of micrometastases: balanced proliferation and apoptosis in the presence of angiogenesis suppression Nat Med 1995 1 149 153 7585012 10.1038/nm0295-149
Buchler P Reber HA Buchler MW Friess H Hines OJ VEGF-RII influences the prognosis of pancreatic cancer Ann Surg 2002 236 738 749 12454512 10.1097/00000658-200212000-00006
Friess H Yamanaka Y Buchler M Berger HG Kobrin MS Baldwin RL Korc M Enhanced expression of the type II transforming growth factor beta receptor in human pancreatic cancer cells without alteration of type III receptor expression Cancer Res 1993 53 2704 2707 8389240
Itakura J Ishiwata T Friess H Fujii H Matsumoto Y Buchler MW Korc M Enhanced expression of vascular endothelial growth factor in human pancreatic cancer correlates with local disease progression Clin Cancer Res 1997 3 1309 1316 9815813
Kisker O Onizuka S Banyard J Komiyama T Becker CM Achilles EG Barnes CM O'Reilly MS Folkman J Pirie-Shepherd SR Generation of multiple angiogenesis inhibitors by human pancreatic cancer Cancer Res 2001 61 7298 7304 11585769
Yamanaka Y Friess H Buchler M Beger HG Uchida E Onda M Kobrin MS Korc M Overexpression of acidic and basic fibroblast growth factors in human pancreatic cancer correlates with advanced tumor stage Cancer Res 1993 53 5289 5296 7693336
Fidler IJ Critical factors in the biology of human cancer metastasis: twenty-eighth G.H.A. Clowes memorial award lecture Cancer Res 1990 50 6130 6138 1698118
Bosserhoff AK Echtenacher B Hein R Buettner R Functional role of melanoma inhibitory activity in regulating invasion and metastasis of malignant melanoma cells in vivo Melanoma Res 2001 11 417 421 11479431 10.1097/00008390-200108000-00013
Guba M Bosserhoff AK Steinbauer M Abels C Anthuber M Buettner R Jauch KW Overexpression of melanoma inhibitory activity (MIA) enhances extravasation and metastasis of A-mel 3 melanoma cells in vivo Br J Cancer 2000 83 1216 1222 11027436 10.1054/bjoc.2000.1424
Bosserhoff AK Lederer M Kaufmann M Hein R Stolz W Apfel R Bogdahn U Buettner R MIA, a novel serum marker for progression of malignant melanoma Anticancer Res 1999 19 2691 2693 10470221
van Groningen JJ Bloemers HP Swart GW Identification of melanoma inhibitory activity and other differentially expressed messenger RNAs in human melanoma cell lines with different metastatic capacity by messenger RNA differential display Cancer Res 1995 55 6237 6243 8521420
Bosserhoff AK Stoll R Sleeman JP Bataille F Buettner R Holak TA Active detachment involves inhibition of cell-matrix contacts of malignant melanoma cells by secretion of melanoma inhibitory activity Lab Invest 2003 83 1583 1594 14615412 10.1097/01.LAB.0000097191.12477.5D
Giese NA Raykov Z DeMartino L Vecchi A Sozzani S Dinsart C Cornelis JJ Rommelaere J Suppression of metastatic hemangiosarcoma by a parvovirus MVMp vector transducing the IP-10 chemokine into immunocompetent mice Cancer Gene Ther 2002 9 432 442 11961666 10.1038/sj.cgt.7700457
| 15710044 | PMC551595 | CC BY | 2021-01-04 16:40:10 | no | Cancer Cell Int. 2005 Feb 14; 5:3 | utf-8 | Cancer Cell Int | 2,005 | 10.1186/1475-2867-5-3 | oa_comm |
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-81570307910.1186/1742-4690-2-8ResearchCritical role of hnRNP A1 in HTLV-1 replication in human transformed T lymphocytes Kress Elsa [email protected] Hicham Hachem [email protected] Françoise [email protected] Louis [email protected] Dodon Madeleine [email protected] Virologie Humaine INSERM-U412, Ecole Normale Supérieure de Lyon, IFR 128 Biosciences Lyon-Gerland, 46 allée d'ltalie 69364 Lyon Cedex 07, France2 Laboratory of Microbiology, University of Brussels, 1 Avenue E. Gryson, 1070 Brussels, Belgium2005 9 2 2005 2 8 8 1 10 2004 9 2 2005 Copyright © 2005 Kress et al; licensee BioMed Central Ltd.2005Kress 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 this study, we have examined the role of heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) in viral gene expression in T lymphocytes transformed by HTLV-1.
Results
We have previously observed that hnRNP A1 (A1) down-modulates the post transcriptional activity of Rex protein of HTLV-1. Here, we tested whether the ectopic expression of a dominant negative mutant (NLS-A1-HA) defective in shuttling activity or knockdown of the hnRNPA1 gene using RNA interference could inhibit Rex-mediated export of viral mRNAs in HTLV-1 producing C91PL T-cells. We show that the expression of NLS-A1-HA does not modify the export of Rex-dependent viral mRNAs. Conversely, inhibiting A1 expression in C91PL cells by RNA interference provoked an increase in the Rex-dependent export of unspliced and singly spliced mRNAs. Surprisingly, we also observed a significant increase in proviral transcription and an accumulation of unspliced mRNAs, suggesting that the splicing process was affected. Finally, A1 knockdown in C91PL cells increased viral production by these cells. Thus, hnRNP A1 is implicated in the modulation of the level of HTLV-1 gene expression in T cells transformed by this human retrovirus.
Conclusions
These observations provide an insight into a new cellular control of HTLV-1 replication and suggest that hnRNP A1 is likely part of the regulatory mechanisms of the life cycle of this human retrovirus in T cells.
==== Body
Background
The human T cell leukemia/lymphotropic virus type 1 is the etiologic agent of adult T cell leukemia (ATL), an aggressive and fatal leukemia of CD4+ T lymphocytes [1,2] and is also associated with a neurological demyelinating disease, tropical spastic paraparesis (TSP) or HTLV-I associated myelopathy (HAM)[3]. Infection by HTLV-1 transforms T cells in vitro and in vivo, a process that has been associated with upregulation of specific cellular genes involved in T cell activation and proliferation during the course of viral infection [4-6]. The completion of the replication cycle of HTLV-1 leading to the production of new particles is dependent on two non-structural HTLV-1 encoded regulatory proteins, Tax and Rex, which act at the transcriptional and post-transcriptional levels, respectively [7,8]. The 40-kDa Tax protein trans-activates transcription of the provirus, through its interaction with cellular transcription factors and with Tax response elements present in the 5' long terminal repeat (LTR). The post-transcriptional activity of the 27-kDa Rex protein, an RNA-binding protein, is mediated by its interaction with the Rex response element (XRE) located on the U3/R region of the 3'LTR present on all viral transcripts [9]. When expressed at a critical threshold, Rex is able to direct the cytoplasmic expression of unspliced gag-pol and singly-spliced env mRNAs, at the expense of the multiply-spiced tax/rex mRNA [10,11]. We have recently reported that heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) interferes with the binding of Rex to the XRE, thus leading to a functional impairment of this viral protein [12].
The ubiquitously expressed hnRNP A1 is an abundant nuclear protein that participates in RNA processing, alternative splicing and chromosome maintenance as well as in the nucleocytoplasmic transport of mRNAs [13-18]. This protein contains two RNA-binding domains and a glycine-rich domain implicated in protein-protein interactions. Predominantly located in the nucleus, this cellular protein has the ability to shuttle continuously between the nucleus and the cytoplasm [19-21]. The signal that mediates both nuclear import and export has been identified as a 38-aa sequence, termed M9, located at the C-terminus of hnRNP A1, and is involved in the nucleo-cytoplasmic trafficking of mRNAs [22].
As indicated above, we have provided evidence that hnRNP A1 impairs the post-transcriptional regulation of HTLV-1 gene expression, by interfering with the binding of Rex to the XRE [12]. In the present study, we first demonstrate that the mutation of a putative binding site of hnRNP A1 to the XRE leads to an increase of the post-transcriptional activity of Rex. Next, to further address the effect that hnRNP A1 might exert on viral replication in vivo, we elected to investigate its implication in HTLV-1 producing T cells. Two experimental approaches were implemented: impairment of the functional activity of the endogenous hnRNP A1 by ectopic expression of a dominant negative mutant and knockdown of the hnRNPAl gene expression using RNA interference (siRNA). We report that inhibition of hnRNP A1 expression and functionality were achieved, leading to an increase of viral transcription together with an increase of cytoplasmic expression of viral mRNAs and of viral production. These observations by providing insight into a new cellular control of HTLV-I replication, suggest that hnRNP A1 is likely part of the regulatory mechanisms of the life cycle of this human retrovirus.
Results
A putative hnRNP A1 binding site has been identified, close to the minimal Rex binding site in the stem-loop D of the XRE (Fig 1A). To further evaluate the role of this binding site in the impairment of the functional activity of Rex, two punctual mutations were performed in the CMV/XRE vector containing the indicator luc gene (Fig 1B). These mutations modify the UAGGUA sequence into CCGGUA, and the UACCUA sequence into UACCGG, respectively, thus generating the CMV/mutXRE vector. Either vector (CMV/XRE and CMV/mutXRE), or the control vector (CMV 128, containing only the luc gene) were then transiently transfected in Jurkat cells in the absence or in the presence of a Rex-expressing plasmid. It was observed that, in presence of Rex, luc expression in cells transfected with the CMV/mutXRE vector was more than 3-fold higher than that in cells transfected with the CMV/XRE vector (Fig 1C). These results indicate that the putative hnRNPAl binding site close to the Rex binding site on the SLD sequence in the XRE is directly or indirectly implicated in down-modulating the post-transcriptional activity of Rex. Since the mutations affect a putative binding site for hnRNP A1, these results suggest that hnRNP A1 might be the effector of this down-regulation. To further delineate how this cellular protein perturbs the life cycle of HTLV-1, we elected to investigate its implication in HTLV-1 producing T cells. Two experimental approaches were implemented: impairment of the endogenous hnRNP A1 by ectopic expression of a dominant negative mutant (NLS-A1-HA) defective in shuttling activity and knockdown of the hnRNP A1 gene using RNA interference (RNAi).
Figure 1 Functional characterization of HTLV-1 mutated XRE sequence. (A) Schematic representation of the HTLV-1 XRE. On the left, the XRE corresponds to U3 and R sequences within the HTLV-1 long terminal repeat, and consists of four stem-loops. On the right, the predicted secondary structure of the stem-loopD (SLD) with the minimal Rex binding site and the mutations introduced within the putative hnRNP A1 binding site are indicated. (B) Schematic view of the reporter plasmid CMV/XRE. (C) Effect of mutations within the XRE sequence on the Rex trans-activation capacity. Jurkat cells were transfected with 1 μg of the indicated reporter plasmid in the presence or not of Rex expression plasmid (200 ng) and the constitutive internal control tk-renilla luciferase vector (10 ng). Data are expressed as normalized luciferase activity and the error bars represent the standard deviations from three independent experiments.
A nucleus-localized shuttling-deficient hnRNP A1 mutant does not affect the post-transcriptional activity of Rex
The NLS-A1-HA construct contains the bipartite-basic type NLS of hnRNP K fused in frame with the N-terminus of an HA-tagged hnRNP A1 mutant, which lacked both nuclear import and export activities and inhibits hnRNP A1-dependent mRNA export when microinjected into nuclei of Xenopus laevis oocytes [22,23]. This hnRNP A1 mutant which retains the hnRNP A1 nuclear localization, lacks nuclear export activity [24]. As such, the nucleus-localized NLS-A1-HA has the potential to compete with wild-type hnRNP A1 for binding to mRNAs, and for its nuclear export. A retroviral vector LXSP-NLS-A1-HA was used to ectopically express this dominant negative mutant in the HTLV-1 transformed C91PL T cells. In these cells, Rex governs the cytoplasmic accumulation of unspliced (gag/pol) and singly-spliced (env) mRNAs. After a few days of culture in presence of puromycin, immunostaining of the resistant population revealed that about 30% of the cells were displaying HA labelling (Fig. 2). Dual immunostaining indicated that both endogenous hnRNP A1 (anti-hnRNP A1, red) and ectopically expressed NLS-A1-HA (anti-HA, green) displayed a nuclear diffuse staining excluding the nucleoli.
Figure 2 Expression of a dominant negative mutant of hnRNP A1 in HTLV-1 producing C91PL cells. Confocal microscopy of untransduced (a) or NLS-A1-HA transduced (b)-C91PL cells after dual immunofluorescence staining with anti-HA (green) and anti-hnRNP A1 (red) antibodies; the right panels show the overlay of the green and red staining;
We next investigated whether overexpression of this defective hnRNP A1 mutant was interfering with the expression of viral mRNAs. Quantification of the nuclear and the cytoplasmic levels of unspliced gag/pol, singly spliced env and doubly spliced tax/rex mRNAs was performed by RQ-PCR involving pair of primers specific of each viral mRNA (Fig. 3A). The comparative analysis of the viral mRNAs expression pattern between the control (LXSP) and NLS-A1-HA cells revealed a small increase of unspliced gag/pol and of doubly spliced tax/rex mRNAS in the latter, whereas no modification was observed for the singly spliced env mRNAs (Fig. 3B). The ratio of nuclear to total RNA and that of cytoplasmic to total RNA allowed to calculate a nuclear export rate (NER). Whereas the cytoplasmic expression of tax/rex mRNAs was slightly enhanced in cells expressing the NLS-A1-HA mutant, the NER of the unspliced and singly spliced mRNAs was not affected (Fig 3C). As the cytoplasmic expression of these mRNAs is Rex dependent, these results indicate that the ectopic expression of the NLS-A1-HA mutant in C91 PL cells does not interfere with the functionality of Rex. However and surprisingly, a more than 4-fold increase of the p19gag amount in the supernatant medium of NLS-A1-HA-transduced cells (2786 ± 154 pg/ml) was observed, when compared to the respective control cells (678 ± 104 pg/ml). Taken together, these results indicate that the impairment of the hnRNP A1 functionality might favour the translation of cytoplasmic viral mRNAs.
Figure 3 Effect of ectopic expression of a dominant negative mutant of hnRNP A1 in HTLV-1 producing C91PL cells. (A) Primer location on HTLV-1 mRNA; (B) Analysis of the nucleo-cytoplasmic distribution of viral gene expression in NLS-A1- and LXSP- transduced cells. Four days after transduction, mRNAs were extracted from the nuclear and cytoplasmic compartments of each cell type and levels of unspliced (gag/pol), singly spliced (env) and doubly spliced (tax/rex) mRNAs were reverse transcribed and quantified by real-time quantitative PCR (RQ-PCR), by using specific primers. Results are expressed as the amount of nuclear (grey bar) and cytoplasmic (black bar) indicated mRNA relative to β-actin. (C) Evaluation of the nuclear export rate (NER) of Rex-dependent (gag/pol plus env) mRNA and of Rex-independent (tax/rex) mRNA in NLS-A1- or LXSP- transduced C91PL cells. Numbers are the ratio between cytoplasmic (C) to total (T) RNA and nuclear (N) to total RNA.
Efficient inhibition of hnRNP A1 by retrovirus-delivered siRNAs
We next evaluated whether HTLV-1 replication is modulated by RNA interference with hnRNP A1 gene expression. To that aim, two oligonucleotides encoding siRNA directed against hnRNP A1, one targeting an RNA sequence located on the 5' end (34-nt after the translation start site), and the other an RNA sequence close to the 3'end (548-nt after translation start site) were each inserted in the pRS retroviral vector [25], as indicated in Materials and Methods. Both pRS-siRNA+34 and PRS-siRNA+548 vectors, as well as the pRS empty vector were used to produce recombinant retroviral particles used to transduce Jurkat T cells at a multiplicity of infection (m.o.i.) of 5. After four days of puromycin selection to eliminate nontransduced cells, the siRNA mediated-depletion of hnRNP A1 mRNAs was measured by quantitative RT-PCR. While targeting the 5'end (+34) was found inefficient, targeting the 3'end (+548) reduced the level of hnRNP A1 transcripts to 10% of those detected in untransduced Jurkat cells or in Jurkat cells transduced with empty (pRS) retroviral particles (Fig. 4A). Importantly, the siRNA-mediated reduction in A1 levels did not provoke cell death. Immunoblotting analysis of the PRS-siRNA +548 cells showed a strong reduction of the hnRNP A1 protein level, when compared to that in the pRS-siRNA+34 cells and in control cells (Fig 4B). Furthermore, the levels of the splicing factor ASF/SF2 were not modified in these cells. These data indicate that expression of hnRNP A1 is specifically repressed in the pRS-siRNA+548-transduced Jurkat cells.
Figure 4 RNAi-mediated reduction of hnRNP A1 expression in Jurkat cells. (A) hnRNP A1 mRNA levels in cells transduced with the indicated retroviruses were determined by RQ-PCR. Levels in knockdown cells are given as percent mRNA reduction relative to the level in control cells transduced with empty pRS virus. Standard deviations are from at least three determinations performed in duplicate. (B) Equal amounts of protein from either nontransduced (lane1) or transduced with the indicated virus (lanes 2 to 4) were analyzed by immunoblotting. Actin and ASF/SF2 were used as control. Note that hnRNP A1 was significantly depleted in cells transduced with siRNA+548, whereas ASF/SF2 was not affected.
hnRNP A1 depletion in HTLV-1-producing T lymphocytes altered the transcriptional profile and increased the post-transcriptional activity of Rex
The above described retroviral vector system was used to mediate the in situ synthesis of siRNAs and to suppress specifically hnRNP A1 gene expression in C91PL cells. Retroviruses produced from pRS-siRNA+548 and from the pRS empty vector were used to transduce these cells with a m.o.i. of 5. Four days after transduction, hnRNP A1 depletion was assessed by quantitative PCR analysis of cytoplasmic mRNAs. In siRNA-transduced C91PL cells, that transcript represented 32% of that in control pRS transduced cells (Fig. 5A). Interestingly, a western blot analysis of cell lysates further showed that hnRNP A1 was barely detected in siRNA-transduced C91 PL cells, whereas the levels of Rex, or of hnRNP C1/C2 or of actin were found unchanged (Fig. 5B). Furthermore, a flow cytometry analysis of siRNA-transduced C91PL cells reveals that hnRNP A1 was detected in 6.1% of these cells, whereas it was detected in about 70% of the control cells (Fig. 5C).
Figure 5 Analysis of hnRNP A1 depletion in HTLV-1 producing C91PL cells. (A) Analysis of hnRNP A1 mRNA levels in cells transduced with the indicated retroviruses. Four days after transduction, cytoplasmic RNA were extracted, reverse transcribed with oligo-dT, and levels of hnRNP A1 mRNA were determined by RQ-PCR. (B) Expression of hnRNP A1, Rex and hnRNP C1/C2 was monitored by immunoblotting of total protein extract from C91PL cells transduced with the indicated virus. Equivalent protein loading was confirmed by immunoblotting with an anti-actin antibody. (C) Detection of hnRNP A1 and p19gag expression in C91PL cells transduced with the indicated virus. Dot plots showing both hnRNP A1 and HTLV-1 gag expressions in one representative experiment. The percentage of cells in each quadrant is indicated.
We next investigated whether the decrease in hnRNP A1 expression in C91PL cells was interfering with the expression of viral mRNAs. Real-time quantitative PCR assays were performed to quantify viral mRNAs by using the same primer pairs described above. Results (from two different transduction experiments) assessing the amount of total viral mRNAs (Fig 6A) revealed that suppression of hnRNP A1 in siRNA-transduced C91PL cells was leading to a significant increase of viral transcription (1.7 to 1.8 fold), when compared to PRS control cells. Then, the analysis of the relative nuclear and cytoplasmic levels of unspliced gag/pol, singly spliced env and doubly spliced tax/rex mRNAs indicated that the expression of unspliced gag/pol mRNA was 2 and 3-fold enhanced respectively in the nucleus and cytoplasm of siRNA-transduced C91PL cells, whereas the expression and the distribution of spliced env mRNAs were not significantly altered (Fig. 6B). A slight increase of the doubly-spliced tax/rex mRNAs was observed in both compartments.
Figure 6 Effect of hnRNP A1 depletion on viral gene expression. (A) Quantification of total viral gene expression in siRNA-transduced C91PL cells by quantitative PCR. Nuclear and cytoplasmic mRNAs were extracted from siRNA (black bars)- or control PRS (white bars)- transduced C91PL cells. Equal amounts of mRNA were reverse transcribed with oligo-dT and subjected to RQ- PCR. Results are expressed as the relative levels of total viral mRNA to cellular β-actin. Error bars indicate standard deviations. (B) Analysis of the nucleo-cytoplasmic expression of viral genes. Four days after transduction, mRNAs were extracted and analyzed as in Fig. 3B. Results are expressed as the amount of nuclear (grey bar) and cytoplasmic (black bar) indicated mRNA relative to β-actin. (C) Evaluation of the nuclear export rate (NER) of Rex-dependent (gag/pol plus env) mRNA and of Rex-independent (tax/rex) mRNA in PRS- or siRNA- transduced C91 PL cells.
These results suggest that inhibition of hnRNP A1 in C91PL cells mainly correlates with a defect in the splicing of genomic mRNAs. The NER of the unspliced and singly spliced mRNAs was significantly higher in siRNA-treated cells than in control cells, whereas the cytoplasmic expression of tax/rex mRNAs, which is Rex-independent was not modified (Fig. 6C). As the nucleo-cytoplasmic transport of the former is Rex-dependent, these observations propose that the depletion of hnRNPAl correlates with an increase of Rex activity. Finally, whereas a flow cytometry analysis indicated a similar percentage of p19gag producing cells in siRNA-transduced C91PL cells and in control cells, the quantification of 19gag in the supernatant medium of siRNA-transduced cells revealed a 1.5-fold increase of the p19gag amount (1017 ± 26 pg/ml), compared to that in control cells (678 ± 104 pg/ml).
Collectively, these data support that the hnRNP A1 depletion in HTLV-1-producing T cells increases viral transcription, is correlated with a defect in the splicing process at the level of the gag/pol transcript and increases the post-transcriptional activity of Rex leading to an increase of viral production.
Discussion
The ubiquitously expressed hnRNP A1, an RNA-binding protein, is a nucleocytoplasmic shuttling hnRNP that accompanies eukaryotic mRNAs from the active site of transcription to that of translation. As such, hnRNP A1 is involved in a variety of important cellular functions, including RNA splicing, transport, turnover and translation. We have previously shown that hnRNP A1 decreases the post-transcriptional activity of the Rex protein of HTLV-1, by interfering with the binding of the viral protein on its response element, present on the 3' LTR of all viral RNAs. Here we first report that the mutation of a putative binding site of hnRNP A1 in the XRE enhances the functional activity of Rex. This observation obtained through transient transfection experiments, confirms that A1 proteins could antagonize the post-transcriptional activity of Rex, by a competitive mechanism.
We have next investigated the role of hnRNP A1 in HTLV-1 transformed C91PL cells, which produce HTLV-1 virions. These express the three differentially spliced (the unspliced gag/pol, the singly spliced env and the doubly spliced tax/rex) mRNAs, which encode the structural and regulatory proteins. The gag/pol and env mRNAs are dependent on Rex for their cytoplasmic expression. To determine whether hnRNP A1 interferes with viral replication, we first examined the effect of the ectopic expression of an hnRNP A1 mutant (NLS-A1-HA) defective in nuclear export activity. This mutant was previously used to assess the potential role of hnRNP A1 in nucleocytoplasmic shuttling activity in normal and leukemic myelopoiesis. Interestingly it was found that the ectopic expression of this dominant negative form of hnRNP A1 resulted in the downmodulation of the nucleocytoplasmic trafficking of cellular mRNAs that encode proteins affecting the phenotype of normal and transformed myeloid progenitors [24]. In the present study, we showed that NLS-A1-HA- C91PL cells expressed a higher level of total viral transcripts than that observed in control cells, suggesting that the ectopic expression of this hnRNP A1 mutant correlated with an increased proviral transcription and/or stability of the viral RNA.
Furthermore, no modification of the nuclear export rate was observed in the NLS-A1-HA-transduced C91PL cells, indicating that the activity of Rex was not impaired. Finally, as both endogenous hnRNP A1 and the NLS-A1-HA mutant, which are nucleus-localized and consequently able to access the XRE did not decrease the Rex-dependent nucleo-cytoplasmic expression of the viral mRNAs, we should therefore speculate that the simultaneous presence of both types of A1 forbids them to bind the XRE with maximal efficiency. Interestingly, the increase of p19gag produced by the NLS-A1-HA C91PL cells suggests that the retention of the endogenous hnRNP A1 in the nucleus is favouring an increase in the translation of viral mRNAs
We have then proceeded to the knockdown of hnRNP A1 gene using the retrovirus-mediated RNA interference. This system was first validated in transduction experiments performed in Jurkat T cells. A puromycin-selected population of cells was obtained in which a strong overall specific reduction of hnRNP A1 was observed. Note that this hnRNP A1-depleted Jurkat cells were not affected in their growth even for a long time culture (data not shown). This is consistent with other studies showing that si-RNA-mediated reduction in A1 levels did not affect cell division nor provoke cell death in normal cell lines [26].
We next performed siRNA depletion of hnRNP A1 in C91PL cells and have observed a significant increase in proviral transcription, as demonstrated by the higher level of viral transcripts than that in control cells (Figure 6A). Furthermore, the level of unspliced transcripts was found to be predominant, compared to the singly-and doubly-spliced transcripts, in the hnRNP A1 depleted cells, pleading for a splicing default (Fig. 6B). Finally, the increase of the nuclear export of unspliced and singly spliced mRNAs suggests that the knockdown of hnRNP A1 allows a better accessibility of Rex to the XRE and leads to the enhancement of the post- transcriptional activity of Rex. This is in good correlation with the increase in the production of viral particles, as ascertained by the quantification of the p19gag protein. Since hnRNP A1 has been implicated in nuclear export of cellular mature mRNAs [27] as well as translational and/or posttranslational events of viral mRNAs (our study), it is possible that its depletion could affect the expression of several transcription and/or splicing factors, leading to an effect, for instance, on the splicing process of viral mRNAs.
Of the two experimental approaches used in the present study to apprehend the implication of hnRNP A1 on HTLV-1 replication in in vitro HTLV-1-transformed T-cells, that consisting in the depletion of this cellular protein by RNA interference provides evidence for the role of hnRNP A1 in restraining the viral life cycle at both transcriptional and post-transcriptional levels. We conclude from these findings that down-regulation of hnRNP A1 has an important role on the replicative potential of HTLV-1 in T lymphocytes. Consequently, these data allows us to define hnRNP A1 as a cellular protein endowed with an anti-HTLV-1 activity.
Methods
pRS construct directing the synthesis of siRNA and Plasmids
The vector pRetro-SUPER (pRS) was used to generate biologically active siRNAs from the Pol III H1-RNA gene promoter [25]. Two annealed 64-bp synthetic oligonucleotides were used: 5'-gatccccAGCAAGAGATGGCTAGTGCttcaagagaGCACTAGCCATCTCTTGCTtttttgga aa-3', and 5'-gatccccCAGCTGAGGAAGCTCTTCAttcaagagaTGAAGAGCTTCCTCAGCTGtttttgga aa-3'. The sequence of each oligonucleotide was designed (Oligoengine) to encode two 19-nt (in capital letters) reverse complements homologous to a portion of hnRNP A1 (nucleotides 34–53 for the first construct, and nucleotides 548–567 for the second one) separated by a 9-nt spacer region, and ending by Bgl II and Hind III sites. Each oligonucleotide was then introduced into pRS resulting in either pRS-siRNA+34 or pRS-siRNA+548 retroviral vectors, respectively. Plasmids pgagpol/MLV and EnvVSV-G were kindly provided by F.L. Cosset (U412-Lyon). LXSP-NLS-A1-HA and empty LXSP retroviral vectors were a kind gift of D. Perrotti and has been described previously [23,24].
For reporter gene analyses, the luciferase plasmid (CMV/XRE) was derived from the reporter plasmid pDM138 containing the CAT gene and the XRE sequences [28]. It expresses, under the control of the cytomegalovirus promoter, a two-exon, one-intron precursor RNA in which the luc gene and the XRE are located within the intron (see Fig. 1B). The mutant plasmid (CMV/mutXRE) was generated using a site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions, and with the following primer, 5'-AAAGCCCTGTCAAAACAGGAAATGGCAAGCGCTTCATCCAGCC-3'. This construct was verified by DNA sequencing before use in transfection. The rex-expression plasmid, containing the wild type Rex sequence under the control of the cytomegalovirus promoter, was a gift from B.C. Cullen.
Cell culture and DNA transfection
Jurkat lymphoblastoid T-cells were incubated at 37°C in a 5% CO2 atmosphere, in RPMI-1640 medium (Invitrogen) supplemented with 10% heat-inactivated fetal calf serum (FCS) and 20 IU/ml penicillin, 20 μg/ml streptomycin. The HTLV-1-transformed T-cell line, C91PL [29] was cultured in complete RPMI medium. The human epithelial 293T cells and the human rhabdomyosarcoma TE cellswere cultured in Dulbecco's minimum eagle medium (DMEM, Invitrogen) supplemented with 10% FCS and 20 IU/ml penicillin, 20 μg/ml streptomycin. These cells seeded at 1.2 × 105 cells per well of a 12-well plate were transfected using the calcium phosphate coprecipitation technique [30]. Jurkat cells were transfected by using the X-treme GENE Q2 transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's indications. The amount of plasmid used in each transfection assay is indicated in the figure legends. To assess the efficiency of the transfection assay, 10 ng of the tk-renilla Luciferase plasmid (Promega) were co-transfected in each assay. Cells were harvested 24 h after transfection, resuspended in 100 μl of passive lysis buffer (Promega) and assayed for both firefly and renilla luciferases by using a Dual-Luciferase Reporter assay system (Promega).
Preparation of viral stocks and transduction of T cells
Fresh viral stocks were prepared by transfecting 293T cells (seeded at 5 × 105 cell/well of a 6-well plate) with 2 μg of pRS or pRS-siRNA together with 1 μg of pgag-pol/MLV and 0,45 μg of env/VSV-G with ExGen 500 reagent (Euromedex). Twelve hours later, the cells were washed once with PBS, and newly produced virions were harvested over 24 h in 1,5 ml of fresh medium. Viral supernatants were clarifed by passage through a 0.45-μm syringe filter and aliquots were stored at -80°C. Titers of virus stocks were determined by infecting rhabdomyosarcoma human TE cells (60% confluent) with serially diluted viral stocks. After infection, cells were split and plated in the presence of puromycin (5 μg/ml); puromycin-resistant colonies were scored after 7 days. Virus titers generally ranged from 3 to 5 × 105 transducing units per ml.
Transduction of Jurkat or of C91 PL T cells with retroviral vectors was carried out as followed: briefly, cells (1 × 106) plated in a 24-well plate were infected at a multiplicity of infection (moi) of 5 with viral stocks in a final volume of 1.0 ml containing 4 μg of polybrene/ml, for 18 h and allowed to recover for 24 hr with fresh medium. When necessary, transduced cells were selected with puromycin 4–5 μg/ml for 4 days and maintained in culture for long time period with 1 μg/ml puromycin.
RNA isolation and real time quantitative RT-PCR
Nuclear and cytoplasmic RNAs were extracted from 2 × 106 cells by using an Rneasy RNA-preparation kit (Qiagen) according to the manufacturer's instructions. To reduce the amount of DNA originating from lysis, samples were treated with Rnase-free Dnase (10 U/μl, Boehringer) for 30 min at 20°C and then for 15 min at 65°C. 500 ng of RNA sample were reverse transcribed by using oligo(dT)12–18 and Superscript II (Life Technologies, Inc.). Reverse transcription was performed for 50 min at 42°C. The total cDNA volume of 20 μl was frozen until real-time quantitative PCR was performed. After thawing for PCR experiments, the cDNA was diluted in distilled water and 2 μl of diluted cDNA was used for each PCR reaction. The realtime quantitative PCR (RQ-PCR) was performed in special lightcycler capillaries (Roche) with a lightcycler Instrument (Roche), by using the LightCycler-FastStart reaction Mix SYBR-Green kit (Roche). The following specific primers were used to detect: hnRNP A1, sense 5'-AAGCAATTTTGGAGGTGGTG-3' and antisens, 5'-ATAGCCACCTTGGTTTCGTG-3', gag/polHTLV-1sense, 5'-CCCTCCAGTTACGATTTCCA-3' and antisens, 5'-GGCTTGGGTTTGGATGAGTA-3', envHTLV-1sense, 5'-CTGTGGTGCCTCCTGAACT-3' and antisens, 5'-AAAGTGGCGAGAAACTTACCC-3', pXIII sense, 5'-ATCCCGTGGAGACTCCTCAA-3' and antisens, 5'-CCAAACACGTAGACTGGGTATCC-3'. β-actin sense,5'-TGAGCTGCGTGTGGCTCC-3' and antisens: 5'-GGCATGGGGGAGGGCATACC-3'.
The thermal cycling conditions consisted of 40 cycles at 95°C for 10 sec, 61°C for 5 sec, 72°C for 10 sec. The fluorescence signal increase of SYBR-GREEN was automatically detected during the 72°C phase of the PCR. Omission of reverse transcriptase in the RT-PCR protocol led to a failure of target gene amplification in the positive controls. Light cycler PCR data were analyzed using LightCycler Data software (Idaho Technology). The software first normalizes each sample by background subtraction of initial cycles. A fluorescence threshold is then set at 5% full scale, and the software determines the cycle number at which each sample reached this threshold. The fluorescence threshold cycle number correlates inversely with the log of initial template concentration. β-actin transcript levels were used to normalize the amount of cDNA in each sample. Melting curve profiles were used to confirm amplification of specific transcripts.
Immunoblotting
Cells were washed and harvested in ice-cold PBS containing protease inhibitors (complete mini EDTA-free, Roche Molecular Biochemicals). Cells were lysed in RIPA buffer (150 mM NaCI, 50 mM Tris-HCI pH 8.0, 0.5% deoxycholate, 0.1% SDS, 0.5% Nonidet P-40, protease inhibitors, 80 U/ml endonuclease) and incubated for 30 min at 4°C. After centrifugation at 12,000 rpm for 10 min at 4°C, the supernatant was assayed for protein content by Bradford assay (Bio-Rad). Equal amounts of proteins were separated by SDS/PAGE.
Cells were lysed in Laemmli buffer and equal amounts of proteins were subjected to 12% SDS-PAGE. They were subsequently blotted onto nitrocellulose membrane (BA, Schleicher & Schuell). The membrane was then blocked overnight at 4°C in blocking buffer (PBS and 0.1% Tween-20) supplemented 10% non-fat powdered milk and probed with the appropriate antibody diluted in blocking buffer plus 10% non-fat powdered milk. The following antibodies were used: rabbit anti-actin (Sigma), mouse anti-ASF/SF2 (gift from Dr. J. Stevenin) mouse monoclonal anti-hnRNP A1 and anti-hnRNP C antibodies (4B10 and 4F4, respectively; gifts from G. Dreyfuss), followed with an anti-rabbit (Immunotech, France) or anti-mouse (Dako) Immunoglobulin G-horse radish peroxidase-conjugated antibody. Blots were then developed using an enhanced chemiluminescence detection system (Renaissance, NEN, Life Science Products). Bands were visualized by using Hyperfilm (Amersham Pharmacia Biotech).
Flow cytometric analysis and Immunostaining
Cells (5 × 105) were washed twice with PBS, resuspended in 3% (vol/vol) paraformaldehyde/PBS for 45 min at room temperature, and permeabilized with 0.5% Triton X-100/PBS for 5 min. After washing with PBS, the cells were incubated with specific antibodies (4B10) diluted in 1% BSA/PBS for 1 h. Cells were washed twice with PBS and were then incubated with FITC-conjugated goat anti-mouse, PE-conjugated goat anti-rabbit in 1% BSA/PBS for 40 min. Cells were washed three times with PBS and resuspended in a 2% paraformaldehyde/PBS solution. The fluorescence intensity was measured on a FACScan instrument (Becton Dickinson Labware, Mountain View, Calif;). The integrated fluorescence of the gated population was measured, and data from 10,000 analyzed events were collected.
For immunostaining, C91PL cells were centrifuged on cytoslides using a cytospin (Thermo Shandon, Pittsburgh, PA), fixed on slides with 3.7% paraformaldehyde for 15 min at room temperature, and permeabilized with 0.5% Triton X100 for 5 min in 4°C. The samples were saturated with PBS containing 0.5% gelatin and 0.25% bovine serum albumin for 1 h and stained for 1 h with a 1/100 dilution of a rabbit polyclonal serum directed against HA (Y11 from Santa Cruz Biotechnology) (NLS-A1-HA staining) or 1/1000 dilution of mouse monoclonal antibodies (4B10) (hnRNP A1 staining) in the same saturation solution. The samples were then washed three times with PBS containing 0.25% gelatin and incubated for 1 h with a 1/100 dilution of the following secondary antibodies: goat anti-rabbit immunoglobulin G conjugated to fluorescein isothiocyanate (green color for HA) and goat anti-mouse immunoglobulin G conjugated to lissamine rhodamine sulfchloride (red color for hnRNP A1) (Jackson Immunoresearch). The samples were washed three times in PBS with 0.25% gelatin and mounted for analysis on a Zeiss LSM 510 laser scanning confocal microscope.
ELISA
p19gag was measured in culture medium using the RETROTEK HTLV p19 Antigen ELISA kit (Zeptometrix). Medium of the cell culture was centrifuged at low speed to remove the cell debris, and filtrated through a 0,45-μm filter. The amount of Gag protein was quantified in the resultant supernatant according to the manufacturer procedure. Results are expressed as pg/ml of p19 protein and are the mean of two different experiments, each point tested in quadruplicate.
Competing interests
The author(s) declare that they have no competing interests.
Acknowledgements
HHB is a recipient of a grant of "Fond National de la Recherche Scientifique-Télévie". This study was supported in part by INSERM in the frame of "Coopération franco-beige" INSERM/CFB/FNRS 2003 and by ARC (Association pour la Recherche sur le cancer n°5669 to L.G.).
==== Refs
Poiesz B Ruscetti P Gazdar A Bunn P Minna J Gallo R 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 USA 1980 77 7415 7419 6261256
Johnson J Harrod R Franchini G Molecular biology and pathogenesis of the human T-cell leukaemia/lymphotropic virus Type-1 (HTLV-1) Int J Exp Pathol 2001 82 135 147 11488989 10.1046/j.1365-2613.2001.00191.x
Osame M Pathological mechanisms of human T-cell lymphotropic virus type I-associated myelopathy CHAM/TSP) J Neurovirol 2002 8 359 364 12402162 10.1080/13550280260422668
Yoshida M Multiple targets of HTLV-I for dysregulation of host cells Seminars in Virology 1996 7 349 360 10.1006/smvy.1996.0042
Yoshida M Multiple viral strategies of HTLV-1 for dysregulation of cell growth control Annu Rev Immunol 2001 19 475 496 11244044 10.1146/annurev.immunol.19.1.475
Gatza M Watt J Marriott S Cellular transformation by the HTLV-I Tax protein, a jack-of-all-trades Oncogene 2003 22 5141 5149 12910251 10.1038/sj.onc.1206549
Cullen BR Mechanism of action of regulatory proteins encoded by complex retroviruses Microbiol Rev 1992 56 375 394 1406488
Green PL Chen ISY Levy JA Molecular features of the human T-cell leukemia virus. Mechanisms of transformation and leukemogenicity The retroviridae 1994 3 Plenum Press 277 311
Gröne M Hoffmann E Berchtold S Cullen BR Grassmann R A single stem-loop structure within the HTLV-1 rex response element is sufficient to mediate Rex activity in vivo Virology 1994 204 144 152 8091649 10.1006/viro.1994.1518
Ballaun C Parrington GK Dobrovnik M Rusche J Hauber J Bohnlein E Functional analysis of human T-cell leukemia virus type I rex-response element: direct RNA binding of Rex protein correlates with in vivo activity J Virol 1991 65 4408 4413 2072457
Grassmann R Berchtold S Aepinus C Ballaun C Böhnlein E Fleckenstein B In vitro binding of human T-cell leukemia virus Rex protein to the rex -response element of viral transcripts J Virol 1991 65 3721 3727 1904103
Due Dodon M Hamaia S Martin J Gazzolo L Heterogeneous nuclear ribonucleoprotein A1 interferes with the binding of the human T cell leukemia virus type 1 rex regulatory protein to its response element J Biol Chem 2002 277 18744 18752 11893730 10.1074/jbc.M109087200
Matter N Marx M Weg-Remers S Ponta HHP Konig H Heterogeneous ribonucleoprotein A1 is part of an exon-specific splice-silencing complex controlled by oncogenic signaling pathways J Biol Chem 2000 275 35353 35360 10958793 10.1074/jbc.M004692200
Mayeda A Krainer A Regulation of alternative pre-mRNA splicing by hnRNP A1 and splicing factor SF2 Cell 1992 68 365 375 1531115 10.1016/0092-8674(92)90477-T
Del Gatto-Konczak F Olive M Gesnel M Breathnach R A1 recruited to an exon in vivo can function as an exon splicing silencer Mol Cell Biol 1999 19 251 260 9858549
Ford L Wright W Shay J A model for heterogeneous nuclear ribonucleoproteins in telomere and telomerase regulation Oncogene 2002 21 580 583 11850782 10.1038/sj.onc.1205086
LaBranche H Dupuis S Ben-David Y Bani M Wellinger R Chabot B Telomere elongation by hnRNP A1 and a derivative that interacts with telomeric repeats and telomerase Nat Genet 1998 19 103 104 9620756 10.1038/575
Eperon I Makarova 0 Mayeda A Munroe S Caceres J Hayward D Krainer A 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
Pinol-Roma S Dreyfuss G Shuttling of pre-mRNA binding proteins between nucleus and cytoplasm Nature 1992 355 730 732 1371331 10.1038/355730a0
Pollard V Michael W Nakielny S Siomi M Wang F Dreyfuss G A novel receptor-mediated nuclearprotein import pathway Cell 1996 86 985 994 8808633 10.1016/S0092-8674(00)80173-7
Siomi M Eder P Kataoka N Wan L Liu Q Dreyfuss G Transportin-mediated nuclear import of heterogeneous nuclear RNP proteins J Cell Biol 1997 138 1181 1192 9298975 10.1083/jcb.138.6.1181
Izaurralde E Jarmolowski A Beisel C Mattaj IW Dreyfuss G A role for the M9 transport signal of hnRNP A1 in mRNA nuclear export J Cell Biol 1997 137 27 35 9105034 10.1083/jcb.137.1.27
Michael WM Choi M Dreyfuss G A nuclear export signal in hnRNP A1: a signal-mediated, temperature-dependent nuclear protein export pathway Cell 1995 83 415 422 8521471 10.1016/0092-8674(95)90119-1
Iervolino A Santilli G Trotta R Guerzoni C Cesi V Bergamaschi A Gambacorti-Passerini C Calabretta B Perrotti D hnRNP A1 nucleocytoplasmic shuttling activity is required for normal myelopoiesis and BCR/ABL leukemogenesis Mol Cell Biol 2002 22 2255 2266 11884611 10.1128/MCB.22.7.2255-2266.2002
Brummelkamp T Bernards R Agami R Stable suppression of tumorigenicity by virus-mediated RNA interference Cancer Cell 2002 2 243 247 12242156 10.1016/S1535-6108(02)00122-8
Patry C Bouchard L Labrecque P Gendron D Lemieux B Toutant J Lapointe E Wellinger R Chabot B Small interfering RNA-mediated reduction in heterogeneous nuclear ribonucleoparticule A1/A2 proteins induces apoptosis in human cancer cells but not in normal mortal cell lines Cancer Res 2003 63 7679 7688 14633690
Dreyfuss G Matunis MJ Pinol-Roma S Burd CG hnRNP proteins and the biogenesis of mRNA Annu Rev Biochem 1993 62 289 321 8352591 10.1146/annurev.bi.62.070193.001445
Hope T Bond B McDonald D Klein N Parslow T Effector domains of human immunodeficiency virus type 1 Rev and human T-cell leukemia virus type I Rex are functionally interchangeable and share an essential peptide motif J Virol 1991 65 6001 6007 1920623
Popovic M Lange-Wantzin G Mann D Gallo RC Transformation of human umbilical cord-blood T-cells by human T-cell leukemia/lymphoma virus Proc Natl Acad Scl USA 1983 80 5402 5406
Chen C Okayama H High-efficiency transformation of mammalian cells by plasmid DNA Mol Cell Biol 1987 7 2745 2752 3670292
| 15703079 | PMC551596 | CC BY | 2021-01-04 16:36:39 | no | Retrovirology. 2005 Feb 9; 2:8 | utf-8 | Retrovirology | 2,005 | 10.1186/1742-4690-2-8 | oa_comm |
==== Front
CytojournalCytoJournal1742-6413BioMed Central London 1742-6413-2-41571591010.1186/1742-6413-2-4ResearchThe Anal Pap Smear: Cytomorphology of squamous intraepithelial lesions Arain Shehla [email protected] Ann E [email protected] Premi [email protected] Shikha [email protected] Department of Pathology & Laboratory Medicine, Cedars-Sinai Medical Center, 8700 Beverly Blvd. Los Angeles, CA 90048, USA2005 16 2 2005 2 4 4 22 12 2004 16 2 2005 Copyright © 2005 Arain et al; licensee BioMed Central Ltd.2005Arain 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
Anal smears are increasingly being used as a screening test for anal squamous intraepithelial lesions (ASILs). This study was undertaken to assess the usefulness and limitations of anal smears in screening for ASILs.
Methods
The cytomorphological features of 200 consecutive anal smears collected in liquid medium from 198 patients were studied and findings were correlated with results of surgical biopsies and/or repeat smears that became available for 71 patients within six months.
Results
Adequate cellularity was defined as an average of 6 or more nucleated squamous cells/hpf. A glandular/transitional component was not required for adequacy. Dysplastic cells, atypical parakeratotic cells and bi/multinucleated cells were frequent findings in ASIL while koilocytes were infrequent. Smears from LSIL cases most frequently showed mildly dysplastic and bi/multinucleate squamous cells followed by parakeratotic cells (PK), atypical parakeratotic cells (APK), and koilocytes. HSIL smears contained squamous cells with features of moderate/severe dysplasia and many APKs. Features of LSIL were also found in most HSIL smears.
Conclusions
In this study liquid based anal smears had a high sensitivity (98%) for detection of ASIL but a low specificity (50%) for predicting the severity of the abnormality in subsequent biopsy. Patients with cytologic diagnoses of ASC-US and LSIL had a significant risk (46–56%) of HSIL at biopsy. We suggest that all patients with a diagnosis of ASC-US and above be recommended for high resolution anoscopy with biopsy.
==== Body
Note
For corresponding Editorial, please see Leiman, 2005 [25]
Background
The incidence of anal squamous carcinoma and its precursor lesions has increased in recent years particularly among men having sex with men (MSM) [1]. Prior to the human immunodeficiency virus (HIV) epidemic the incidence of anal cancer in this high risk population was estimated at 36.9 per 100,000 [2], similar to the incidence of cervical cancer prior to adoption of routine cervical cytology screening programs. Among MSM, the incidence of anal cancer in HIV positive individuals has been estimated to be twice that in HIV negative individuals [3,4]. The American Cancer Society projected that about 4,010 new cases of anal cancer would be diagnosed in the United States in 2004, (up from 3,400 cases in 2000) and that about 580 persons would die of the disease during the year [5].
Anal and cervical lesions share many histological and pathological characteristics including the implication of human papilloma virus (HPV) in the pathogenesis of precursor squamous intraepithelial lesions and invasive cancer [6]. Just as routine Pap smear screening has dramatically reduced the incidence of cervical cancer, it is anticipated that screening populations at high risk for anal squamous intraepithelial lesions (ASILs) will reduce the incidence of anal cancer in these individuals. Accordingly we and other laboratories are experiencing a substantial increase in the number of anal smears submitted for cytologic evaluation. This study was performed to assess the usefulness and limitations of anal smears in screening for ASILs.
Materials And Methods
After approval from the IRB, 200 consecutive anal smears submitted from 198 patients were retrieved from the files of the pathology department at Cedars-Sinai Medical Center. The samples had all been collected from the anal canal using the Rovers endocervex brush (Therapak Corp., Irwindale, CA, distributor for Rovers Medical Devices, OSS, The Netherlands), the Digene cervical sampler brush (Digene Corp. Gaithersburg, MD), or the brush from SurePath sample collection kit (TriPath Care Tech, TriPath Imaging, Inc. Burlington, NC) (Figure 1) and submitted in liquid medium (SurePath™, TriPath Imaging™, Burlington, NC). All of the patients were males between the ages of 24 and 67 years (mean: 40.7 yrs., median: 41 yrs). HIV status was available for 79 patients, 37 of whom were HIV positive.
Figure 1 Collection brushes. A. Brush from SurePath sample collection kit (TriPath Care Tech, TriPath Imaging, Inc. Burlington, NC.) B. Rovers endocervex brush (Therapak Corp., Irwindale, CA, distributor for Rovers Medical Devices, OSS, The Netherlands) C. Digene cervical sampler brush (Digene Corp., Gaithersburg, MD)
Retrieved slides were reviewed by three cytopathologists and evaluated for cellularity and presence of anucleated squamous cells, glandular/ transitional cells (G/TZ), parakeratotic cells (PKs), atypical parakeratotic cells (APKs), koilocytes, binucleated and/or multinucleated squamous cells (B/MSCs), and dysplastic cells. The number of cells exhibiting each of these morphologic features was recorded as none, rare (no more than 2 cells/smear), and present (3 or more cells/smear). For this study cellularity was defined as the average number of nucleated squamous cells per 40x high power field (nsc/hpf) calculated by counting 10 hpfs. All of the anal smears had been reported using a modified Bethesda 2001 System terminology recommended for cervical smears [7]. After discrepancies were resolved by re-evaluation, discussion, and concurrence by at least two cytopathologists, the diagnoses were as follows: unsatisfactory due to insufficient cellularity (17 smears), negative for intraepithelial lesion or malignancy (NIL; 58 smears), atypical squamous cells of undetermined significance (ASC-US; 42 smears), low grade squamous intraepithelial lesion (LSIL; 59 smears), atypical squamous cells of undetermined significance cannot exclude high grade squamous intraepithelial lesion (ASC-H; 17 smears), and high grade squamous intraepithelial lesion (HSIL; 7 smears).
Revised cytologic diagnoses were correlated with concurrent and/or follow up tissue biopsies or with repeat anal smears all obtained within six months. Statistical analyses were performed using the Fishers Test. A two sided p value of 0.05 was considered as significant [8].
Results
Cellularity
For purposes of this study we required an average of at least 6 nsc/hpf for cellularity to be considered adequate. This was based on the observation that only smears averaging 6 or more nsc/hpf included abnormal cytologic diagnoses ranging from ASC-US through HSIL whereas smears averaging 5 or fewer nsc/hpf were either NIL or ASC-US. 91% (181) of the 200 smears contained an average of 6 or more nsc/hpf. Of the 19 cases that averaged fewer than 6 nsc/hpf, 17 were designated as unsatisfactory and excluded from the study while two that contained atypical squamous cells were reported as ASC-US and included in the subsequent morphologic analysis. Of note, all of the 7 smears with HSIL and 16/17 smears with ASC-H were cellular with an average of 8 or more nsc/hpf.
Anucleated squamous cells
Anucleated squamous cells were present in smears that were NIL and in smears with diagnoses ranging from ASC-US to HSIL. There were numerous anucleated squamous cells in 7 smears that were reported as unsatisfactory. Among the abnormal smears, neither the presence nor number of anucleated squamous cells correlated with cytologic diagnosis.
Glandular/transitional cells
Three or more groups of G/TZ were present in 56% (103) of the smears while rare G/TZ were seen in an additional 18% (33) smears. Of these smears 68% (93/136) had an abnormal cytologic diagnosis (27 ASC-US, 45 LSIL, 14 ASC-H, 7 HSIL). In comparison 26% (47) of smears contained no G/TZ of which 68% (32) were reported as abnormal (15 ASC-US, 14 LSIL, 3 ASC-H) suggesting that the presence of glandular cells did not facilitate abnormal diagnoses. Again all of the 7 smears with HSIL and 12/17 smears with ASC-H contained 3 or more groups of glandular cells, while two smears with ASC-H contained only rare glandular cells.
Parakeratotic cells
Parakeratotic cells were observed in 71% (130) of the smears in the study. Parakeratotic cells were observed in negative (63%, 37/58) as well as in abnormal (74%, 93/125) cases. In negative cases rare parakeratotic cells were observed more often (in 72%, 27 cases) whereas in smears with epithelial abnormalities presence of rare parakeratotic cells and frequent (>3) parakeratotic cells were about evenly distributed (46%, 43 cases vs 54%, 50 cases).
Atypical parakeratotic cells
APKs were found in 40% (74) of the 183 smears. The number of cases showing APKs increased with the severity of the dysplasia. There were 3 or more APKs in 22% (41) of the smears constituting 7% of ASC-US, 41% of LSIL, 53% of ASC-H and 71% of HSIL cases. Rare APKs were found in 18% (33) of the smears that were interpreted as ASC-US or above. APKs were not found in any of the negative smears and were seen in 72% of the SIL smears.
Koilocytes
Classical koilocytes were infrequent (17%) in all diagnostic categories (Figure 2). Three or more koilocytes were seen in only 10% (6/59) of the LSIL smears and in 6% (1/17) of the ASC-H cases. Rare koilocytes were found in 2% (1/42) of the ASC-US, 20% (12/59) of the LSIL, and 14% (1/7) of the HSIL smears.
Figure 2 Low grade squamous intraepithelial lesion with typical koilocyte. Papanicolaou stain × 40×
Multinucleation
B/MSCs were observed more frequently in abnormal smears – 81% (101/125) as compared to 33% (19/58) of the NIL smears (Figure 3). Among the abnormal cases, 3 or more B/MSCs were present in 59% (74/125 cases) which included 31% (13/42) of ASC-US, 75% (44/59) of LSILs, 76% (13/17) of ASC-Hs, and 43% (4/7) of HSILs. Rare B/MSCs were observed in the remaining 41% cases which included 31% (13/42) of ASC-US, 15% (9/59) of LSILs, 18% (3/17) of ASC-Hs, and 29% (2/7) of HSILs. Although B/MSCs were observed in some smears with cytologic diagnosis of NIL, their presence correlated significantly with an abnormal cytologic diagnosis (p < 0.0001).
Figure 3 Low grade squamous intraepithelial lesion with bi- and multinucleated cells Papanicolaou stain × 40×
Dysplastic squamous cells
84% (50/59) of the smears diagnosed as LSIL contained 3 or more squamous cells with features of mild dysplasia (Figure 4), 6 cases had rare mildly dysplastic cells, and 3 smears contained typical koilocytes but no dysplastic cells. Three or more cells exhibiting moderate/severe dysplasia were present in all smears diagnosed as HSIL (Figure 5). ASC-H cases contained 2 or fewer abnormal cells with features of high grade dysplasia in 5 cases, whereas the remaining showed small cells with dense cytoplasm and atypical nuclei raising the possibility of atypical metaplastic and/or atypical parakeratotic cells. 71% of the smears diagnosed as ASC-H (12/17) and HSIL (5/7) also contained 3 or more mildly dysplastic cells and rare mildly dysplastic cells were found in an additional 4 ASC-H and 2 HSIL smears.
Figure 4 Low grade squamous intraepithelial lesion with mildly dyplastic cells Papanicolaou stain × 40×
Figure 5 A. & B. High grade squamous intraepithelial lesion showing small to medium severely dysplastic cells. Papanicolaou stain × 40×
Table 1 summarizes the frequency of these cytomorphologic features with respect to the cytodiagnostic categories. The most frequent findings in smears diagnosed as LSIL were mildly dysplastic and B/MSCs followed by PKs, APKs, and koilocytes. Smears diagnosed as HSIL contained multiple squamous cells with features of moderate/ severe dysplasia, many APKs, and varying numbers of PKs, B/MSCs, and koilocytes. Each of the smears diagnosed as HSIL also contained some mildly dysplastic cells but classical koilocytes were infrequent.
Table 1 Frequency and distribution of cytologic findings in anal smears
Cytologic diagnosis (n = 183) Cytologic features
Parakeratosis Atypical Parakeratosis Koilocytes Bi/Multi-nucleation Mild dysplasia Moderate-severe dysplasia
NIL (n = 58) + - - + - -
ASC-US (n = 42) + + + + + -
LSIL (n = 59) ++ ++ + +++ +++ -
ASC-H (n = 17) ++ ++ + +++ ++ +
HSIL (n = 7) ++ +++ + ++ ++ +++
+, ++, +++ indicates feature is present in 3 or more cells in 1–33%, 34–74%, or >75% of cases, respectively
n = number of cases
Correlation with follow up diagnosis
Within six months of the index anal smear, follow up consisting of 56 biopsies and 15 smears became available for 39% (71) of the 183 smears constituting 39% (181) of the patients in the study. As shown in Table 2, 86% (57 of 66) smears diagnosed as ASC-US or above were confirmed as abnormal on subsequent biopsy (54) or repeat smear (12). Follow up for 11 smears diagnosed as ASC-US yielded 4 negative, 2 AIN I, 1 AIN II, and 4 AIN III. Five smears diagnosed as LSIL were negative, 11 were AIN I, and 20 were AIN II-III on subsequent follow up. HPV Digene Hybrid Capture II assay was performed on 3 of the 4 ASC-US cases and 3 of the 5 LSIL cases that were negative on follow up. The 3 ASC-US cases tested negative for HPV DNA. The 3 LSIL cases tested positive for both low and high risk HPV DNA and repeat smears at 8 and 10 months respectively showed persistent LSIL in 2 of these cases. Biopsy confirmed 100% of the HSIL diagnoses and 76% (13/17) of the ASC-H diagnoses. Two cases diagnosed as ASC-H on cytology showed AIN I on biopsy; no follow up became available for the remaining 2 cases that had been diagnosed as ASC-H. Only 5 smears diagnosed as NIL had follow up biopsy; 4 were negative and 1 showed AIN II.
Table 2 Follow up diagnoses at 6 months
Cytologic diagnoses Diagnosis at followup‡
Negative AIN I AIN II AIN III
NIL (5) 4 - 1 -
ASC-US (11) 4† 2 1 4
LSIL (36) 5* 11 17 3
ASC-H (15) - 2 5 8
HSIL (4) - - 1 3
Total cases (71) 13 15 25 18
†3 of these cases tested negative for HPV DNA utilizing Digene HCII
*3 of these cases tested positive for high and low risk HPV DNA utilizing Digene HCII and 2 of these cases showed persistent LSIL on follow up at 8 and 10 months respectively
‡Follow up constitutes a composite of 56 biopsies and 15 repeat smears
Discussion
ASIL presents unique challenges in diagnosis and clinical management. By decreasing deaths from opportunistic infections, widespread use of highly active antiretroviral agents and other therapies have done much to improve survival of HIV infected individuals. However, because these therapies do not impact the incidence of HPV infections or malignancies in these individuals, the increased life span of HIV+ individuals probably provides the primary explanation for the rapid and continuing increase in HPV associated AIN that these individuals are experiencing [9-12]. With the help of cytology screening, anal squamous carcinoma may be one of very few preventable malignancies in these individuals.
Anal cytology has been shown to be a cost-effective screening method for detection of ASIL in populations at high risk for anal carcinoma[13]. To date there are few studies that address selected cytomorphologic features and diagnostic limitations associated with anal cytology. Based on the follow up available in our study, a diagnosis of ASC-US and above detected 86% of AINs. If one includes the 5 LSIL smears that were negative on follow up biopsy (all 5 confirmed as LSIL on smear review by three cytopathologists, 3 additionally confirmed by repeat smears testing and/or HPV DNA), then the detection rate increase to 94%. Only one AIN lesion was NIL on cytology. This further confirms that anal smears are a sensitive means for detection of ASIL with a sensitivity of 98%. However, as seen in our study anal cytology was a poor predictor of the severity of AIN lesions and frequently underdiagnosed these lesions. Specificity was calculated at only 50%. Follow up for 5 of 11 (46%) ASC-US smears showed AIN II-III and follow up in 20 of 36 (56%) LSIL smears showed AIN II-III. Conversely, of the 43 cases with AIN II-III on biopsy, only 4 (9%) had been correctly diagnosed as HSIL and only 13 (30%) had been reported as ASC-H while 26 (60%) had been reported as LSIL or below on cytology. The percent cases correctly diagnosed as HSIL may be improved from 9 to 13 (21%) if the 5 ASC-H cases with only 1–2 high grade dysplastic cells in the smear were also reported as HSIL. However, this is difficult in "real life" particularly since ASC-H cases frequently also contain atypical parakeratotic cells. In summary, cytology underdiagnosed 35% (25) of the 71 cases with follow up. There were no high-grade overcalls. In our study, a diagnosis of ASC-H or HSIL accurately predicted the presence of AIN II-III in 90% of cases. However, a cytologic diagnosis of ASC-US or LSIL also held a 46–56% chance that a high-grade AIN would be present on biopsy. This figure is high when compared to cervical cytology where ASC-US and LSIL have been associated with only a 5–17% chance of HSIL on biopsy [14,15].
Prior experience with anal smears as documented in the literature[16,17] reveals that anal smears have low sensitivity and specificity for AIN lesions with poor detection of high grade lesions. Defining abnormal cytology to include ASC-US and ASIL, Palefsky et al [16] reported the sensitivity of anal cytology for detection of biopsy-proven ASIL to be 69% in 407 HIV-positive and 47% in 251 HIV-negative homosexual or bisexual men. The authors also note that the grade of disease on anal cytology did not always correspond to the histologic grade, a finding similar to ours. Anal smears were obtained by dacron swabs in this study. Similarly, Panther et al [18] reported that anal cytology is an inaccurate predictor of the presence of HSIL, regardless of HIV status. The authors analyzed 153 paired specimens of anal cytology and anal biopsies or surgical excisions and obtained a sensitivity of only 47% for detection of a high-grade lesion (ASIL II, III, or invasive squamous cell cancer). Moreover, in their study a cytologic diagnosis of ASC-US (n = 30) was associated with a broad distribution of histologic diagnoses (7 NIL, 11 AIN I, 7 AIN II, or 5 AIN III). Thus, the authors concluded that the presence of any abnormal anal cytologic finding indicates a potential for HSIL on histologic examination. Our study supports this finding. We attribute the higher detection rate for AIN in our study to the collection of specimens in liquid medium using brushes resulting in greater cellularity of our specimens. Liquid-based preparations have also been shown to virtually eliminate poor fixation/air drying artifacts and markedly reduce obscuring fecal contamination thereby providing superior quality material compared to conventional smears [19,20]. A comparable sensitivity level of 92% has been reported by Friedlander et al [17] utilizing thin prep liquid based collection medium (Cytyc, Boxborough, MA).
There is a paucity of literature regarding criteria for adequate anal cytology samples. The 2001 Consensus Conference in Bethesda [7] suggested that 3 – 6 nsc/hpf may be considered adequate for SurePath preparations. An average of 6 or more nsc/hpf detected 123 of the 125 of the abnormal cases in this study (2 undetected ASC-US had lower cellularity). Moreover, although smears with diagnoses of HSIL or ASC-H contained 8 or more nsc/hpf, no statistical association was observed between smear cellularity and undetected HSIL lesions. Thus, for SurePath preparations an average of 6 or more nsc/hpf is recommended as an adequacy guideline.
The presence of G/TZ was not a prerequisite for adequacy in our study. Smears with and without G/TZ detected the same percentage (68%) of abnormal cases. Although most HSIL and ASC-H smears contained 3 or more groups of G/TZs, absence of G/TZ did not correlate statistically with undetected AIN II/III lesions. Thus we do not consider the presence of G/TZ as essential for adequacy, a situation analogous to cervical Pap smears [7,21,22]. Interestingly, we did not encounter any cases of atypical glandular cells of undetermined significance, glandular dysplasia, or adenocarcinoma in our smears. At this time, it is not clear whether individuals at increased risk for ASIL are also at increased risk for anorectal glandular dysplasia and adenocarcinoma.
On review of the morphological features of AIN lesions in cytology smears, we noticed some salient features. Dysplastic cells were the most reliable indicators of ASIL/AIN. Typical koilocytes were infrequent, observed in only 17% of SILs, a finding previously observed by Darragh et al [19] who reported that koilocytes were (a) less frequently observed in anal smears than in cervical smears and (b) absent in some smears that were diagnostic for AIN. APKs, on the other hand, were frequent, present in 72% SILs, and helpful in the diagnosis of ASIL. They were observed most frequently and in greatest numbers in HSIL lesions. Friedlander et al [17], in a review of 70 ThinPrep anal smears for selected cytomorphologic features reported APKs in 62% and koilocytes in 21% of smears. They emphasized the "ubiquitous presence of atypical keratinized squamous cells" and caution against overinterpretation of these cells as indicative of HSIL or squamous carcinoma. B/MSCs were also good indicators of abnormal smears. Although, they may be seen in small numbers in negative smears, when present in large numbers, B/MSCs should trigger a search for ASIL. Parakeratotic cells, although frequently observed were not helpful in the diagnosis of ASIL, a finding supported by Friedlander et al [17] who observed parakeratotic cells in 84% of their study cases. Similar studies in cervical smears have shown that parakerstosis in otherwise negative Pap smears, is not a reliable marker for cervical intraepithelial neoplasia [23,24].
In ASC-H and HSIL, high grade squamous cells are usually small, found as single cells or small sheets admixed with mildly dysplastic cells and atypical parakeratotic cells. Careful scrutiny is required to not miss these high grade lesions.
Our experience with anal cytology also indicates that other infectious agents are rarely diagnosed in anal smears. Candida was present in one case. Herpes or trichomonads were not seen.
Conclusions
To summarize, liquid based anal smears provide a sensitive method for screening populations at increased risk for ASIL but have a low specificity for predicting the severity of the lesion. Patients with cytologic diagnosis of ASC-US and LSIL have a significant risk of having HSIL and should be recommended for high resolution anoscopy with biopsy.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SA participated in the acquisition, analysis and interpretation of data and helped to draft the manuscript.
AEW participated in its design, in the acquisition, analysis and interpretation of data and helped to write the manuscript.
PT participated in its design, in the acquisition, analysis and interpretation of data and helped to write the manuscript.
SB conceived of the study, participated in its design, in the acquisition, analysis and interpretation of data and helped to write the manuscript.
All authors read and approved the final manuscript.
Acknowledgements
Co-editors of CytoJournal Vinod B. Shidham, MD, FRCPath, FIAC and Barbara F. Atkinson, MD thank: the academic editor Gladwyn Leiman MBBCh, FIAC, FRCPath, Director of Cytopathology, Fletcher-Allen Health Care, Professor of Pathology, University of Vermont, 111 Colchester Avenue, Burlington, VT 05401, USA [email protected] for organizing and completing the peer-review process for this manuscript.
==== Refs
Daling JR Madeleine MM Johnson LG Schwartz SM Shera KA Wurscher MA Carter JJ Porter PL Galloway DA McDougall JK Human papillomavirus, smoking, and sexual practices in the etiology of anal cancer Cancer 2004 101 270 280 15241823 10.1002/cncr.20365
Daling JR Weiss NS Klopfenstein LL Cochran LE Chow WH Daifuku R Correlates of homosexual behavior and the incidence of anal cancer Jama 1982 247 1988 1990 7062503 10.1001/jama.247.14.1988
Melbye M Cote TR Kessler L Gail M Biggar RJ High incidence of anal cancer among AIDS patients. The AIDS/Cancer Working Group Lancet 1994 343 636 639 7906812 10.1016/S0140-6736(94)92636-0
Goedert JJ Cote TR Virgo P Scoppa SM Kingma DW Gail MH Jaffe ES Biggar RJ Spectrum of AIDS-associated malignant disorders Lancet 1998 351 1833 1839 9652666 10.1016/S0140-6736(97)09028-4
American Cancer Society I All about anal cancer wwwcancerorg/docroot/CRI/CRI_2xasp?sitearea=&dt=47 2004
Palefsky JM Anal squamous intraepithelial lesions in human immunodeficiency virus-positive men and women Semin Oncol 2000 27 471 479 10950374
Solomon D Davey D Kurman R Moriarty A O'Connor D Prey M Raab S Sherman M Wilbur D Wright TJ Young N The 2001 Bethesda System: terminology for reporting results of cervical cytology Jama 2002 287 2114 2119 11966386 10.1001/jama.287.16.2114
Uitenbroek DG SISA-Binomial <http://homeclaranet/sisa/binomialhtm>(1 Jan 2002) 1997
Ortholan C Francois E Gerard JP [Preneoplastic anal lesions and anal canal carcinoma] Bull Cancer 2003 90 405 411 12850763
Chin-Hong PV Palefsky JM Natural history and clinical management of anal human papillomavirus disease in men and women infected with human immunodeficiency virus Clin Infect Dis 2002 35 1127 1134 12384848 10.1086/344057
Kreuter A Reimann G Esser S Rasokat H Hartmann M Swoboda J Conant MA Tschachler E Arasteh K Altmeyer P Brockmeyer NH [Screening and therapy of anal intraepithelial neoplasia (AIN) and anal carcinoma in patients with HIV-infection] Dtsch Med Wochenschr 2003 128 1957 1962 14502448 10.1055/s-2003-42360
Horster S Thoma-Greber E Siebeck M Bogner JR Is anal carcinoma a HAART-related problem? Eur J Med Res 2003 8 142 146 12765859
Goldie SJ Kuntz KM Weinstein MC Freedberg KA Welton ML Palefsky JM The clinical effectiveness and cost-effectiveness of screening for anal squamous intraepithelial lesions in homosexual and bisexual HIV-positive men Jama 1999 281 1822 1829 10340370 10.1001/jama.281.19.1822
Wright TCJ Cox JT Massad LS Twiggs LB Wilkinson EJ 2001 Consensus Guidelines for the management of women with cervical cytological abnormalities Jama 2002 287 2120 2129 11966387 10.1001/jama.287.16.2120
The ASCUS-LSIL Triage Study (ALTS) Group A randomized trial on the management of low-grade squamous intraepithelial lesion cytology interpretations Am J Obstet Gynecol 2003 188 1393 1400 12824968 10.1067/mob.2003.462
Palefsky JM Holly EA Hogeboom CJ Berry JM Jay N Darragh TM Anal cytology as a screening tool for anal squamous intraepithelial lesions J Acquir Immune Defic Syndr Hum Retrovirol 1997 14 415 422 9170415
Friedlander MA Stier E Lin O Anorectal cytology as a screening tool for anal squamous lesions: cytologic, anoscopic, and histologic correlation Cancer 2004 102 19 26 14968414 10.1002/cncr.11888
Panther LA Wagner K Proper J Fugelso DK Chatis PA Weeden W Nasser IA Doweiko JP Dezube BJ High resolution anoscopy findings for men who have sex with men: inaccuracy of anal cytology as a predictor of histologic high-grade anal intraepithelial neoplasia and the impact of HIV serostatus Clin Infect Dis 2004 38 1490 1492 15156490 10.1086/383574
Darragh TM Jay N Tupkelewicz BA Hogeboom CJ Holly EA Palefsky JM Comparison of conventional cytologic smears and ThinPrep preparations from the anal canal Acta Cytol 1997 41 1167 1170 9250316
Sherman ME Friedman HB Busseniers AE Kelly WF Carner TC Saah AJ Cytologic diagnosis of anal intraepithelial neoplasia using smears and cytyc thin-preps Mod Pathol 1995 8 270 274 7617653
Birdsong GG Pap smear adequacy: Is our understanding satisfactory...or limited? Diagn Cytopathol 2001 24 79 81 11169883 10.1002/1097-0339(200102)24:2<79::AID-DC1014>3.0.CO;2-3
Baer A Kiviat NB Kulasingam S Mao C Kuypers J Koutsky LA Liquid-based Papanicolaou smears without a transformation zone component: should clinicians worry? Obstet Gynecol 2002 99 1053 1059 12052599 10.1016/S0029-7844(02)01998-1
Cecchini S Iossa A Ciatto S Bonardi L Confortini M Cipparrone G Colposcopic survey of Papanicolaou test-negative cases with hyperkeratosis or parakeratosis Obstet Gynecol 1990 76 857 859 2170888
Zahn CM Askew AW Hall KL Barth WHJ The significance of hyperkeratosis/parakeratosis on otherwise normal Papanicolaou smears Am J Obstet Gynecol 2002 187 997 1001 12388995 10.1067/mob.2002.126640
Leiman G Anal sreening cytology Cytojournal 2005 2 5 15715911 10.1186/1742-6413-2-5
| 15715910 | PMC551597 | CC BY | 2021-01-04 16:36:23 | no | Cytojournal. 2005 Feb 16; 2:4 | utf-8 | Cytojournal | 2,005 | 10.1186/1742-6413-2-4 | oa_comm |
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-81570749110.1186/1471-2474-6-8Study ProtocolProlonged conservative treatment or 'early' surgery in sciatica caused by a lumbar disc herniation: rationale and design of a randomized trial [ISRCT 26872154] Peul Wilco C [email protected] Houwelingen Hans C [email protected] der Hout Wilbert B [email protected] Ronald [email protected] Just AH [email protected] Joseph ThJ [email protected] Ralph TWM [email protected] Bart W [email protected] Department of Neurosurgery, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands2 Department of Medical Statistics, Leiden University Medical Center, Leiden, The Netherlands3 Department of Medical Decision Analysis, Leiden University Medical Center, Leiden, The Netherlands4 Department of General Practice, Leiden University Medical Center, Leiden, The Netherlands5 Department of Neurology, Medical Center Haaglanden, The Hague, The Netherlands6 Department of General Practice, University Medical Center Rotterdam (Erasmus MC), PO Box 1736 Rotterdam, The Netherlands2005 11 2 2005 6 8 8 10 1 2005 11 2 2005 Copyright © 2005 Peul et al; licensee BioMed Central Ltd.2005Peul 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 design of a randomized multicenter trial is presented on the effectiveness of a prolonged conservative treatment strategy compared with surgery in patients with persisting intense sciatica (lumbosacral radicular syndrome).
Methods/design
Patients presenting themselves to their general practitioner with disabling sciatica lasting less than twelve weeks are referred to the neurology outpatient department of one of the participating hospitals. After confirmation of the diagnosis and surgical indication MRI scanning is performed. If a distinct disc herniation is discerned which in addition covers the clinically expected site the patient is eligible for randomization. Depending on the outcome of the randomization scheme the patient will either be submitted to prolonged conservative care or surgery. Surgery will be carried out according to the guidelines and between six and twelve weeks after onset of complaints. The experimental therapy consists of a prolonged conservative treatment under supervision of the general practitioner, which may be followed by surgical intervention in case of persisting or progressive disability. The main primary outcome measure is the disease specific disability of daily functioning. Other primary outcome measures are perceived recovery and intensity of legpain. Secondary outcome measures encompass severity of complaints, quality of life, medical consumption, absenteeism, costs and preference. The main research question will be answered at 12 months after randomization. The total follow-up period covers two years.
Discussion
Evidence is lacking concerning the optimal treatment of lumbar disc induced sciatica. This pragmatic randomized trial, focusses on the 'timing' of intervention, and will contribute to the decision of the general practictioner and neurologist, regarding referral of patients for surgery.
==== Body
Background
One of the greatest advantages of publishing the design of a randomized controlled trial (RCT) before results are available is the accessibility to criticism of the methodological quality irrespective of the results. Firstly the scientific reader must be enabled to search for epidemiological shortcomings when the results differ from the expected outcome as compared to results in line with one's expectations. Secondly, it is possible to more extensively elaborate the background and rationale of the research question, the study population, the chosen treatments and outcome measures, as compared to publications describing the trial results. Thirdly, but not less important, publishing the design of a RCT is instrumented in preventing publication bias in subsequent meta-analyses. Studies with non-significant results are less likely to be published than those with significant results [1,2]. It is a considerable loss for data pooling that unpublished trial results are omitted. After pre-publishing the study design even unpublished data can be used in a systematic review, since these can be required from the study group. This article describes the rationale and parallel group design of a RCT in which the optimal timing of disc surgery for sciatica will be investigated.
The lumbosacral radicular syndrome (LSRS or LRS; also called sciatica) is typically characterized by radiating pain in the dermatome of a lumbar or sacral spinal nerve root. Occasionally more than one root is involved. Contained in the syndrome pain may be accompanied with lumbar fixation, reflex abnormalities motor and sensory disturbances. In diagnosis includes stenosis of the spinal and/or root canal, infection, multiple sclerosis, autoimmune or metabolic neuropathy, and tumour. This study will be restricted to herniations at the lowest three lumbar disc levels, since these represent the most common sites. In the vast majority of cases LSRS is the result of a herniated disc. In the Netherlands annually between 60,000 and 75,000 new cases of LSRS are diagnosed by the General Practitioner (GP) [3]. The presumed direct medical costs of treatment of LSRS are € 133 million each year [4]. Most of these costs are attributable to in-hospital treatment; only a small portion is incurred by GP's or physiotherapists (€ 3.2 million). In a study, performed in 1988, more than 11.000 patients were operated in the Netherlands and this frequency did not change in the past years [4,5]. The combined direct and indirect costs are estimated to be € 1,2 billion per year [6]. The indirect costs are considerable due to the high rate of production loss caused by sciatica.
The natural history of LSRS is in general favourable. In 60–80 percent of patients, the leg pain decreased or disappeared within 6–12 weeks after onset [7-9,51]. These patients no longer experienced problems at work or in their private lives after three months. The minority with lasting complaints beyond three months further decreases with time. At one year only a small proportion of herniated discs continues to produce discomfort and disability. At present it is not possible to identify these latter groups of patients in an early stage of their disease by means of intensity of pain, neurological deficit, root irritation signs, or diagnostic imaging. For this reason it is not helpful to perform early diagnostic imaging (CT or MRI), unless a disease entity different from disc herniation is considered. After the indication for surgery has been set diagnostic imaging is helpful in defining the exact site of disc herniation and its anatomical relationship with the nerve root involved.
Since the first publication on lumbar disc surgery by Mixter and Barr [17] many studies have demonstrated the success of surgery for the treatment of LSRS. Unfortunately only a few prospective studies investigated the difference in outcome between surgical and conservative care [7,8,18-22]. The published treatment results vary as much as the frequency of reported complications and the recurrence rate.
The only study, which compared surgery with conservative care directly in a RCT, was performed by Weber more than 20 years ago [7,8]. He found better results for surgery at one-year follow-up. At four and ten years follow-up the results of surgical and conservative care no longer differed. Being the only published RCT comparing surgical and conservative care, this study regrettably carries some important methodological flaws in both design and outcome measures when compared to today's epidemiological standard rules [23]. One of the main shortcomings is the exclusion of patients, who do have an indication for surgery because of "intolerable" pain. Those are the current patients who ask for surgery and are not comparable to the randomized population of Weber. Therefore it is impossible to extrapolate and generalize these results to the treatment policy of today.
Since 1983 a few cohort studies have been published on non-surgical treatment of patients with at least six weeks of leg pain with good short-term results at one-year follow-up [25,22]. These studies also suffer from methodological flaws. The only conclusion that can be drawn from these reports and the study of Weber is that the policy of prolonged conservative care can be effective, as a result of the favourable natural course of LSRS. Epidemiological and clinical studies have shown that most lumbar disc protrusions resolve spontaneously with the elapse of time [15,16]. Another finding is that prolonged conservative care appears safe and without complications if the patient remains active. Recent population based studies however state that the natural history is not favourable at all [50].
Whether particular demographic findings, symptoms, physical signs and/or MRI findings either separately or combined do have prognostic value has not been investigated scientifically yet. It would be of great value if one were able to identify early in the course of the disease those patients who will have an unfavourable outcome without surgery.
In spite of the known favourable natural course the surgical rate in the Netherlands is quite high [10]. We perform six times as many lumbar discectomies compared to Scotland, four times the number in England and two times the number in Sweden. In the latter study comparing 12 Western countries the United States is the only country where more operations are performed for the indication LSRS. There are no substantial differences in the incidence of this disease in the countries mentioned that can explain the difference in surgical rates. There is no indication [6] that the surgical rate has changed under influence of the consensus reports [11,12,11]. Actually change was not likely to occur because the published guidelines were representative for daily practice and normal care before 1996 in the Netherlands. With respect to the indications for and timing of surgery no evidence in the literature is available to either support or contradict these guidelines. These guidelines were produced after agreement between all medical (sub-) disciplines involved in the care for patients with LSRS. Our high surgical rate, as contradictory as it may seem, may reflect good clinical practice.
Because of the observation that most people recover from their complaints in the first 6–8 weeks [9,51] this period of persistent radicular leg pain is considered a good indication for surgery in the Netherlands. Although there is consensus that surgery is only offered in case of persistent pain, the timing of this treatment seems to depend on local production capacity and patient and doctor preferences rather than on evidence-based practice. This lack of evidence for the timing of surgery after the 6–8 week period explains the large variations in daily practice. Exact data on the problems associated with surgery, such as surgical failure, recurrent disc herniation and adverse effects are limited. This is one of the reasons that in some regions surgery will only be carried out after a period of 3–6 months of LSRS. [14].
It is not known whether the relative high rate of disc surgery in the Netherlands is cost-effective or not, compared to other countries [15,16].
In summary, consensus is missing on the preferred timing of disc surgery, due to insufficient evidence that a prolonged conservative care strategy is effective. More insight is needed into the potential short-term effects of a relative early surgery strategy, as compared to an extended wait-and-see period. In particular the effects on the return to work or resumption of previous daily activities as well as the complications of both strategies have not yet been investigated.
The main goal of this comparative study is to investigate whether the completion of a 6–12 weeks period of lasting radicular pain constitutes a solid indication for surgery and is superior to prolonged conservative care. A secondary goal is to identify possible subgroups of patients who will substantially benefit from one of the proposed treatment strategies. The cost-effectiveness results will be a trade-off between a quicker relief of leg pain in the surgery group versus the advantage of lower costs and avoiding the negative effects of surgery in the conservatively treated group. The difference in disease related quality of life depends on the duration of persisting pain and disability after randomization in the prolonged conservative care group.
This study to investigate this scientific gap in our understanding of the effectiveness of surgery for LSRS is in line with a recommendation by the Dutch Health Council in 1999 to the Minister of Health [4] and the current Cochrane Review [15,16].
The results of this trial will lead to a more rational use of the existing guidelines if the hypothesis is rejected. If the latter is accepted and prolongation of the conservative treatment policy is more cost-effective than surgery after 6–12 weeks, the current guidelines for the timing of surgery need correction.
Methods/design
To answer the main research question the investigators propose to conduct a multi-centre comparative randomized clinical trial with parallel group design. The main research question will be answered after a follow-up of six months (Figure 1). The complete follow-up will last two years. The multi-centre design is necessary to collect enough patients in two years. The Medical Ethics Committee of all participating hospitals approved the study protocol.
Figure 1 Flow chart of the Sciatica Trial
Patients
All patients between 18 and 65 years with sciatica of less than 12 weeks duration are eligible for this study. Because of the multi-centre (15 hospitals) design the patients in a large region in the western part of the Netherlands can be included in this trial if they meet the in- and exclusion criteria (Table 1). Because these are the only hospitals, which treat lumbar disc herniations in this area, included patients will reflect a representative population treated in primary and secondary care. Inclusion of patients will be started after a visit to the neurological outpatient clinics. Randomization will start after at least 6 weeks persistent disabling pain in the dermatome of the leg served by the L4, L5 or S1 root. All 1100 GP's involved will be informed about this study and receive information about developments and the results of the trial. They will refer patients within the first 6–12 weeks after onset sciatica.
Table 1 Selection criteria for trial eligibility
Inclusion criteria:
• Age 18–65 yr.
• Persistent radicular pain in the L4, L5 or S1 dermatome with or without mild neurological deficit
• Severe disabling leg pain of 6–12 weeks duration
• Evidence of a unilateral disc herniation confirmed on MRI
• Sufficient knowledge of Dutch language
• Informed consent
Exclusion criteria:
• Cauda equina syndrome or severe paresis (MRC<3)
• Complaints of a lumbosacral radicular syndrome in the same dermatome within the past 12 months
• A history of unilateral disc surgery on the same level
• Spinal canal stenosis
• Degenerative or lytic spondylolisthesis
• Pregnancy
• "Severe life-threatening" or psychiatric illness
• Planned (e)migration to another country in the year after randomization
During the first visit to the neurological outpatient clinic the patient's history will be taken and a standardized neurological examination will be performed. During this visit the neurologist will inform the patient on the cause and course of a lumbosacral radicular syndrome and convey the doubt regarding the timing of surgery for this condition. The study will be explained to the patient and in case of a positive reaction an appointment is made to meet one of the research nurses as soon as possible.
Preferably the study MRI scans will be performed after informed consent during the first visit to the research nurse. Because the patient needs some time to consider participation a second visit will be planned at least two days after the first visit to the outpatient clinic. The research nurse will give all extra information needed to understand the trial and will ask the patient if he/she agrees to be randomized. Informed by the radiologist and surgeon, the research nurse will only randomize the patient during the third visit if the MRI confirms the presence of unilateral disc herniation and the patient is eligible according to the inclusion and exclusion criteria. The patient will not be aware of detailed MRI data. The radiologist and neurosurgeon independently using a standardized Case Record Form (CRF) will register the MRI findings. The MRI will be performed according to a standardized protocol and including Gadolinium series for the intended subgroup analysis.
Treatment allocation
Patients will randomly be allocated to either surgery within 1–2 weeks or prolonged conservative treatment by their GP. Patients, their doctors and research nurses can obviously not be blinded for the allocated treatment. Blinding of the outcome measurements is not possible, due to the fact that mainly self-reported outcomes are used. A randomization list is prepared for every participating hospital. Permuted blocks of random number patients are formed to ensure near-equal distribution of patients over the two randomization arms in the hospitals. Using random number tables generates the random sequence of the permuted blocks. The data manager, who is not involved in the selection and allocation of patients will prepare coded, sealed envelopes containing the treatment allocation. During the second patient visit the research nurse will open the envelope together with the patient and appointments will be made for the allocated treatment, either surgery or referral back to the GP, to ensure that treatment is started as soon as possible after randomization. This will be done after checking all the criteria and especially the persistence of pain with disability in daily functioning. A letter about the allocated treatment arm informs all caregivers. Although the principal investigator will not include and operate upon trial patients he may be biased with a preference for surgery, which could theoretically influence analysis. Therefore the principal investigator is blinded for the allocated treatment. As he is not involved in treatment of the study population blinding during later analysis is only possible after blinding during the randomization and follow-up period.
Interventions
After randomization two groups of patients will exist.
Group A; the surgically treated patients and group B; the conservatively managed patients.
Surgical treatment
(A) will be performed in the conventional manner with microscope or loupe magnification. The investigators prefer the standard surgical approach because the other (minimally invasive) surgical approaches have limited indications, are not more cost-effective, and have a long learning curve. During the transflaval approach care is undertaken to minimize bony removal and on the other hand to prevent overstretching of the compromised nerve root. In addition to removal of herniated disc material as much as possible nuclear material will be removed with pituitary forceps, curettes and rongeurs in order to prevent recurrence. The participating treating doctors are 2 orthopaedic- and 12 neurosurgeons with large experience in the standard approach with loupe magnification or microscope. A standardized CRF will register the findings of the surgeon and the herniated disc material will be investigated histologically for granular infiltration.
Surgery will take place as soon as possible and within a maximum of two weeks after randomization. Hospital admission will be 2–7 days, including the day of surgery. During the immediate post-operative period the patients will be mobilised with the help of a physiotherapist. At home guidance is confirmed by their own physiotherapist. The frequency will be 2 times a week for 8 weeks.
Conservative management
(B) will be conducted by the general practitioner (GP) or neurologist when necessary. The GP will provide ample information about the favourable prognosis of LSRS. The treatment of LSRS is aimed primarily at pain relief and maintenance/restoration of normal day-to-day activities. Unfortunately, the effect of giving information and counselling has not been studied specifically among LSRS patients. However, various studies have evaluated the effect of such support for people suffering from other pain syndromes [24]. Inferences can reasonably be made from the findings of these studies. Hence, it may be assumed that adequate and unambiguous information about what is wrong (the nature of the condition) and what the patient can expect (the prognosis), together with trustworthy counselling can reduce the anxiety and uncertainty felt by the patients and thus ease the pain [12]. The GP's will encourage the patients to continue with normal day-to-day activities in so far as possible. When necessary analgesic medication can be prescribed according to the guidelines. The GP will advise the patients to stay active and if possible return to work and/or their leisure activities.
After the first consultation the GP will make a follow-up schedule. During the next visit the patient and doctor will look at the changes since the first visit to determine whether there is any improvement in the ability to perform normal activities. The doctor will check the efficacy of the prescribed pain medication and may adjust the dose or sort of analgesics according to the NHG guidelines. In these guidelines paracetamol is the first choice. If not effective, NSAID's (ibuprofen, diclofenac or naproxen) are to be prescribed. Only in the event of severe disabling pain morphine may be given for a restricted period of time. By preference all analgesics should be taken at fixed times of the day rather than on a 'if necessary' basis. If the GP and the patient conclude that there is considerable kinesiophobia because of the fear that the radicular or low back pain will increase, the help of a physiotherapist can be recommended. Guided by the GP (and physiotherapist) the patient will upgrade his or her activities according to the agreed time schedule [25,26]. The guide will be time, not the intensity of the pain. The GP will be free in her/his choice of prescription of medication and referral to physiotherapists. The research nurse will register the conservative management strategy after communication with the responsible GP. In case of progressive neurological deficit or worsening intolerable pain the GP can refer the patient back to the research nurse or neurosurgeon. If, six months after randomization, the patient has still not improved or suffers from intermittent LSRS, surgical treatment will be offered. Some patients will ask for surgery earlier because of worsening drug resistant leg pain. In these cases and in the case of a progressive neurological deficit, surgery will be performed in consultation with the patient. If after maximum conservative treatment and counselling the patient is still not able to cope with the functional disability surgery can be requested. If surgery in these cases is not offered by the study-group the patient does have the right to have a second opinion with an undependable neurosurgeon of another university hospital.
Outcome assessment
In the LSRS the most common complaints are pain and disability to perform normal daily activities. We will use below described validated outcome parameters, which will be assessed by means of questionnaires. Patients are not informed about their earlier scores. Follow-up examinations by the research nurse will take place 8, 26 and 52 weeks after randomization and the patients will keep a diary (table 2). In between at 2, 4, 12, 38, and 78 and after 104 weeks the main questionnaire (primary outcome measures) will be filled in at home and send to the data centre.
Table 2 Data collection and outcome measures
Time in weeks ? 0 2,4 8 12 26 38 52 78 104
Likert X X X X X X X X X X
Neurological examination X X X X
Severity of complaints (VAS) X X X X X X X X X X
McGill X
Health Status (SF 36) X X X X X
Functional Status (RDQ) X X X X X X X X X X
EuroQol/VAS Q-of-life X X X X X X X X X X
MRI X X
Costs X X X X X X X X X X
Prolo X X X X
Complications X X X X
Surgery X X X X X
SFBI X X X X X X
Primary outcome measures
1) Roland Disability Questionnaire for Sciatica.' This illness-specific 23-item functional assessment questionnaire is frequently used for low back pain and sciatica [38,39]. Scores range from 0 to 23, reflecting a simple unweighted sums of items endorsed by the respondent. Patients with high scores at baseline do have a severe disabling LSRS. To define recovery a difference of at least 11 points from baseline has to be seen [38,17]. The Roland Questionnaire for Sciatica has a documented high level of internal consistency; construct validity, and responsiveness [38,39]. It is the main primary outcome measure in this trial.
2) Perceived recovery.' This is a seven-point Likert scale measuring the perceived recovery, varying from 'completely recovered' to 'worse than ever'. This outcome scale has been used in previous studies and appears to be valid and responsive to change [27]. Next to this global self-assessment a job and hobby specific Likert will be scored. During the intake of the study the patient will be asked to rank their five most important functional disabilities in daily live (work, hobby), which they can use in their own evaluation overall and in separate items.
3) VAS pain in the leg. This parameter will measure the experienced intensity of pain in the leg during the week before visiting the research nurse. Pain will be assessed on a horizontal 100 mm scale varying from 0 mm, 'no pain in the leg', to 100 mm, 'the worst pain ever'. Patients do not see the results of earlier assessments and will score the pain experienced at the visit. [28-32].
Secondary outcome measures
1) EuroQol classification system and VAS rating personal health. A cost-utility analysis will be performed using QALY's based on the EuroQol questionnaire, which has been validated in many studies and is easy to fill out [41,42,51]. The EuroQol will be measured twice a week during the first four weeks and at all follow-up moments. Patients describe their general health status using the EuroQol classification system, consisting of 5 questions on mobility, self care, usual activities, pain/discomfort, and anxiety/depression [44]. From the EQ-5D classification system, the EQ-5D utility index will be calculated [43]. This utility measure reflects how the general public values the health status described by the patient, which is preferred for economic evaluations from a societal perspective. Patients also rated their personal health using a visual analog scale (VAS) ranging from worst imaginable health to best imaginable health.
2) Short-Form 36 (SF-36). Quality of life was also assessed using the RAND-36 questionnaire. This is a generic health status questionnaire, which can easily be filled out at home. The questionnaire consists of 36 items on physical and social functioning has 8 domains; 1) physical functioning, 2) physical restrictions, 3) emotional restrictions, 4) social functioning, 5) somatic pain, 6) general mental health, 7) vitality, 8) general health perception. This questionnaire has been used frequently and was validated in studies on low back pathology and surgery [33-37]. From the RAND-36, the SF-6D utility index was calculated. Like the EQ-5D, this SF-6D reflects the general public's valuation of the health described by the patient. The SF-6D is a recent instrument that has not been used much yet, but it richer classification system could make it a more sensitive utility measure than the EuroQol measure.
3) Sciatica Frequency and Bothersome Index (SFBI). This is a scale from 0 to 6, which can assess the frequency (0 = not at all to 6 = always) and bothersomeness (0 = not bothersome to 6 = extreme bothersome) of back and leg symptoms. The sum of the results of four symptom questions yields both indexes, ranging from 0 to 24: leg pain; numbness and/or tingling in the leg; weakness in the leg or foot; pain in the back or leg while sitting. [17].
5) PROLO-scale. This scale measures the evaluation of the research nurse of the functional-economic status of the patients. This parameter has been used in studies on the difference in functional outcome between different techniques of lumbar spine fusion [40].
6) VAS pain in the back. This parameter measures the intensity of the pain in the back experienced during the week before visiting the research nurse. Assessment will be based on a horizontal 100 mm scale varying from 0 mm, 'no pain in the back', to 100 mm, 'the worst pain ever'. Patients do not see the results of earlier assessments and will score their pain during the visit. This parameter is included because a lot of patients with LSRS also have back pain in varying intensities, which can change after surgery or conservative treatment.
Other outcome measures
1) Costs. The societal costs during the first year will be estimated in accordance with the recent pharmacoeconomic guideline [47,48]. The costs of hospital admission and surgery will be based on an integral top-down cost analysis in three large regional participating hospitals (aggregated according to the total number of patients per department). From this institutional analysis, the constant costs per admission and the variable costs per admission day will be estimated. From these constant and variable costs, the individual costs of hospital admission and surgery for all patients can be estimated, using the duration of the hospitalization. In the study an MRI is performed in all cases. The costs of this MRI will only be calculated for patients undergoing surgery, because in the normal situation MRI would only be performed when a surgical indication exists.
Patients will register other health care needs in a diary (including physiotherapy, visits to GP's and specialists, nursing care and medication). Each diary covers a period of 3 months and will be discussed with the patient during the follow-up visits to the research nurse. The volume of health care will be assessed using standard prices [48].
In the diary the patient will also register direct non-medical costs (including time costs, travel expenses and domestic help). To estimate productivity costs the patients will also report absenteeism in the diary. At the follow-up visits, the research nurse will register the work situation, work efficiency and gross wages. Absenteeism will be valued according to the friction-cost method.
2) Incidence of (re-) surgery. One of the goals of the policy for group B is to avoid surgery while achieving at least the same effects. The surgical rate is therefore an indication of the success or failure of this policy. The incidence of re-operation at the same disc level in group A will be an indication of the failure rate for surgery.
3) Side-effects or complications that are ascribed to the treatment are recorded by the patients, their treating physicians and the research nurses.
4) MRI findings. The results of the differences between the baseline MRI and the MRI made 52 weeks after randomization are important secondary outcome measures. The difference in size of the disc herniation (in mm), nerve root compression, and amount of scar tissue will be registered. Failures of surgery can be recognized by inadequate disc removal or decompression of the nerve. The data will be gathered, using a standardized CRF, which will be filled out by the local radiologist, orthopaedic- or neurosurgeon and (neuro-) radiologist
Sample size
The result of this study is based on the short-term success of surgical intervention and will be a trade-off between a quicker relief of leg pain versus an advantage in cost-effectiveness for conservatively managed patients. The sample size is calculated on the basis of the Roland Disability Questionnaire for Sciatica averaged during the 12 months follow-up period. The numbers used for this sample-size are drawn from the Maine Lumbar Spine Study 1 year and recently published 5-year results [19,55]. The difference in the Roland score between the surgical- and non-surgical group in this study did not change between 3 and 12 months follow-up as shown in their study [19] and can be averaged over the first year. The main aim of this study is to measure the short-term functional difference at 12 months follow-up. Surgical treatment is considered better when the post treatment change is at least 4 points more when compared to the conservative treatment arm [38] and constant over time. Considering this constant difference and a mean standard deviation =10 over the first year [55] 140 patients per treatment arm are needed to reach a power (1-β) of 0,90 with α = 0.05 (two-sided). To answer the main research question 280 patients are needed for analysis with at least 12 months follow-up. The aim is to enrol 300 (150 per arm) patients in the study, including 8 % loss to follow-up after 1 year. The total number of operated patients each year in all participating hospitals exceeds 1400. With this number of patients also a clinically important difference in median time to recovery of two months can be detected by survival analysis. Although the time to recovery is the main issue, the problem of recurrent complaints is still not solved in the different approaches of survival and proportional hazard analysis.
Statistical and cost analysis
Baseline comparability will be investigated by descriptive statistics to examine if randomisation was successful. Differences in success rates between both groups are calculated, together with 95 per cent confidence intervals. In addition to an analysis of the difference in recovery between the two groups (as explained under the paragraph sample size) analyses of the difference in time to recovery will be carried out. Due to lack of data in the literature we could not base our sample size calculations on these differences. Survival-analysis is used to calculate differences in median time to recovery. Continuous outcomes are evaluated as change scores (differences between baseline measurement and each follow-up measurement). Multivariable analyses are performed to adjust for the eventual differences between the groups at baseline in prognostic indicators. All the analyses are performed according to the intent-to-treat principle. An additional per protocol analysis is performed comparing patients in the wait-and-see group who received surgery with patients in the same group who had not and with patients in the surgery group. To compare the actual treatment sec instead of strategies an explorative analysis will be performed in subgroups off all patients who actually received surgery and who did not receive surgery in both groups. All patients who withdraw from the study are included in the analysis until the time of withdrawal.
The result of this study will be a trade off between the disadvantages of surgery (hospitalisation, reduced quality of life and costs) versus the possible advantages (earlier relief of pain and return to work). For that reason recovery, measured as an 11 point difference in score when compared to baseline (Roland Disability Questionnaire for Sciatica), is the clinically most relevant patient outcome. Quality of Life (SF-36) and perceived recovery are important to compare the reduced quality of life from surgery to the possibly prolonged pain from conservative therapy and also to be able to compare cost-effectiveness with that of other spine interventions. The EuroQol is important to obtain cost-utility ratio's that can be compared with those of a wide range of other interventions. Utilities are obtained from the descriptive classification system of the EuroQol, using the model described by Dolan [43,53]. Conservative treatment may decrease costs compared to surgery but possibly at the expense of delayed effectiveness. In an incremental cost-effectiveness analysis, societal costs during the first year will be compared to the primary outcome measure (Roland Disability Questionnaire for Sciatica, averaged over the first year), Quality of Life (SF-36, during the first year) and perceived recovery (7-points Likert scale). Cost-effectiveness analyses with these effectiveness measures have been conducted before, allowing comparison with other spine interventions.
Finally, to answer the second research question explorative analyses are conducted to investigate whether the treatment effect after two, six and twelve months varies in specific subgroups of patients (Table 3).
Table 3 Selected prognostic variables for subgroup analysis
Demographic Variables
• Age < 39 years versus > 39 years,
• Intellectual versus physical demanding job,
Anamnestic and Neurological Variables
• Acute start LSRS versus slow start,
• History of backpain versus no history,
• Influence of coughing, sneezing on complaints versus no influence,
• Difficulty to put on shoes and/or socks versus no difficulty,
• Straight leg raising ≤ 30 degrees versus > 30 degrees,
• Positive crossed straight leg raising sign versus negative sign,
• VAS-pain > 70 versus < 69 mm,
• Tingling/numbness in pain area versus no tingling (9),
• Pain leg worse by sitting versus no worsening (9),
• McGill affective high score versus low score,
Radiological Variables
• MRI disc sequester versus contained disc herniation,
• MRI circumferential gadolinium enhancement versus no enhancement of disc herniation,
• Mediolateral versus median and lateral disc herniation,
• High versus low height of disc level (height 9 mm),
Miscellaneous Variables
• Preference for surgery versus no preference for surgery.
• Disc Herniation at L5S1 vs. L4L5
Using logistic regression for success rate and linear regression for severity of the disability, each prognostic indicator is checked for interaction with treatment. If the interaction term is significant, a stratified analysis will be performed.
Discussion
In this article the rationale and design of a pragmatic RCT on the cost-effectiveness of timing of disc surgery for LSRS is described. The only randomized trial [7] so far on this subject only included patients where the caregiver was in doubt about the surgical indication. Patients with severe disabling pain were not randomized [8]. The Sciatica Trial is directed to those patients with a clear surgical indication according to current usual care. The study is pragmatic because it acknowledges that sometimes it may not be possible to postpone surgery for every conservative care patient until 6 months after allocation and that some patients will recover before surgery is performed in the surgical group. In these cases we consider it unethical to hold on to the randomized treatment. Because of the Intent-to-Treat analysis these cases will be analysed in their own allocated randomization arm and will not cause methodological problems because it is two healthcare strategies that are compared, as opposed to two treatments. The objective of this trial is to provide evidence on the preferred timing of disc surgery for sciatica. A prolonged conservative treatment strategy is compared to the international guideline advise of surgery after 6–8 weeks LSRS. The intended size of the study population is sufficiently large to detect short and long term differences between both strategies.
Abbreviations
GP = General Practitioner
LSRS = Lumbosacral Radicular Syndrome
RCT = Randomized Controlled Trial
VAS = Visual Analogue Scale
Competing Interests
The author(s) declare that they have no competing interests.
Author's contributions
WP designed the study is responsible for the protocol. HH is responsible for the calculation of the sample size and contributed to the design of analysis. WH is responsible for the design of the cost-effectiveness analysis. RB has contributed in the case record forms and is responsible for the database ProMIse. JE contributed to the involvement of the GP's. JT structured the ideas about the diagnostics of LSRS and intake by neurologists. RT is the neurosurgical supervisor of WP. BK is the epidemiological supervisor of WP. All authors participated in the trial design and coordination. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The Sciatica Trial is funded by ZonMW/NOW. Furthermore we do want to thank the six researchnurses and datamanagers of the Sciatica Research Team for their work in making the start of this trial possible in the participating hospitals.
==== Refs
Egger M Smith GD Bias in location and selection of studies BMJ 1998 316 61 6 9451274
Thornton A Lee P Publication bias in meta-analysis: its causes and consequences J Clin Epidemiol 2000 53 207 16 10729693 10.1016/S0895-4356(99)00161-4
Van de Velden J de Bakker DH Basisrapport: Morbiditeit in de huisartsenpraktijk Utrecht Nivel 1990
Van den Bosch JH Kardaun JW Ziekten van het zenuwstelsel in Nederland Den Haag: SDU/uitgeverij 1993
SIG Zorginformatie Utrecht 1996 Gegevens uit de landelijke registratie van Nederlandse Ziekenhuizen
Van Tulder Koes BW Bouter LM A cost-of-illness study of back pain in the Netherlands Pain 1995 62 233 40 8545149 10.1016/0304-3959(94)00272-G
Weber H Lumbar Disc Herniation. A controlled prospective study with ten years of observation Spine 1983 8 131 140 6857385
Weber H Lumbar Disc Herniation. A prospective study of prognostic factors including a controlled trial J of Oslo City Hospital 1978 28 33 64
Vroomen PC The diagnosis and conservative treatment of Sciatica Thesis Maastricht 1998
Cherkin DC Deyo RA An international comparison of back surgery rates Spine 1994 19 1201 1206 8073310
Smeele IJ van den Hooghen JM Mens JM E.a. NHG-standaard Lumbosacraal Radiculair Syndroom Huisarts Wet 1996 39 78 89
Stam J Consensus over diagnostiek en behandeling van het lumbosacraal radiculair syndroom Ned Tijdschr Geneesk 1996 140 2621 27
CBO Consensus Het lumbosacraal radiculair syndroom 1996
Gezondheidsraad 1999/18 Diagnostiek en behandeling van het lumbosacraal radiculair syndroom
Gibson JN Grant IC Waddell G The Cochrane Review of Surgery for Lumbar Disc Prolapse and Degenerative Lumbar Spondylosis Spine 1999 24 1820 32 10488513 10.1097/00007632-199909010-00012
Gibson JN Grant IC Waddell G Surgery for lumbar disc prolapse (Cochrane Review) Cochrane Library 2000 Oxford: Update Software
Mixter WJ Barr J Rupture of the intervertebral disc with involvement of the spinal canal N Engl J Med 1934 211 210 15
Spanfort EV Lumbar disc herniation: a computer aided analysis of 2504 operations Acta Orthopaedica Scandinavica 1972 142
Atlas SJ Deyo RA The Maine Lumbar Spine Study, part II. 1-Year Outcomes of Surgical and Nonsurgical Management of Sciatica Spine 1996 15 1777 1786 8855462 10.1097/00007632-199608010-00011
Hakelius A Prognosis in Sciatica: A clinical follow-up of surgical and nonsurgical treatment Acta Orthop Scand 1970 1 76 4916637
Malter AD Deyo RA 5-Year Reoperation Rates After Different Types of Lumbar Spine Surgery Spine 1998 23 814 820 9563113 10.1097/00007632-199804010-00015
Saal JA Natural history and non-operative treatmenet of lumbar disc herniation Spine 1996 2S 9S 9112320
Bessette L Liang MH Classics in Spine. Surgery literature revisited Spine 1996 21 259 263 8742199 10.1097/00007632-199602010-00001
Malter AD Deyo RA Cost-Effectiveness of lumbar discectomy for the treatment of herniated intervertebral disc Spine 1996 21 1050 1055 10.1097/00007632-199605010-00011
Saal JA Saal JS The natural history of lumbar intervertebral disc extrusions treated non-operatively Spine 1989 15 683 6 2218716
Gatchel Turk DC Psychological approaches to painmanagement: a practitioner's handbook 1996 The Guilford Press
Main CJ Wood PLR Hollis S The distress and risk assesment method Spine 1992 17 42 52 1531554
Fordyce WE Behavioral methods for chronic pain and illness 1976 St Louis: CV Mosby Company
Deyo RA Outcome measures for low back pain research. A proposal for standardized use Spine 1998 23 2003 2013 9779535 10.1097/00007632-199809150-00018
Scott J Huskisson EC Graphic representation of pain Pain 1976 2 175 184 1026900 10.1016/0304-3959(76)90114-7
Huskisson Measurement of pain Lancet 1974 1127 1131 4139420 10.1016/S0140-6736(74)90884-8
Joyce CRB Zutshi DW Comparison of fixed interval and Visual Analogue Scales for rating Chronic Pain Eur J Clin Pharmacol 1975 8 415 420 1233242 10.1007/BF00562315
Melzack R Katz J Pain measurement in persons in pain Textbook of Pain
Collins SL Moore A The visual analogue pain intensity scale: what is moderate pain in millimetres? Pain 1997 72 95 97 9272792 10.1016/S0304-3959(97)00005-5
Ware JE Sherbourne C The MOS 36-item short-form survey (SF 36): Conceptual framework and item selection Med Care 1992 30 473 483 1593914
Van der Zee K Sanderman R De psychometrische kwaliteiten van de MOS 36-item Short Form Health Survey (SF-36) in een Nederlandse populatie T Soc Gezondheidsz 1993 71 183 191
Brazier JE Harper R Validating the SF-36 health survey questionnaire BMJ 1992 305 160 64 1285753
Aaronson NK Acquadro C International quality of life assesment (IQOLA) project Q Of Life Research 1992 1 349 51 10.1007/BF00434949
Stansfeld SA Roberts R Assessing the validity of the SF-36 General Health Survey Q Of Life Research 1997 6 217 24
Patrick DL Deyo RA Assessing health related quality of life in patients with sciatica Spine 1995 20 1899 909 8560339
Gommans I Koes BW Tulder MW, Koes BW, Bouter LM Validity and responsiveness of the Dutch adaptation of the Roland Disability Questionnaire Low Back Pain 1996 EMGO 57 70
Prolo DJ Oklund SA Toward uniformity in evaluating results of lumbar spine operations Spine 1986 11 601 606 3787326
Dolan PP Modeling valuations for EuroQol health states Med Care 1997 35 1095 1108 9366889 10.1097/00005650-199711000-00002
EuroQol Group A new facility for the measurement of health-related quality of life Health Policy 1990 16 199 208 10109801 10.1016/0168-8510(90)90421-9
Matsubara Y Serial changes on MRI in lumbar disc herniations treated conservatively Neuroradiology 1995 37 378 83 7477838
Bozzao A Galluci M MRI imaging assessment of natural history in patiënts treated without surgery Radiology 1992 185 135 41 1523297
College voor zorgverzekeringen Richtlijnen voor farmaco-economisch onderzoek CvZ, afdeling FO/G&s 1999
Oostenbrink JB Koopmanschap MA Rutten FF Handleiding voor kostenonderzoek, methoden en richtlijnprijzen voor economische evaluaties in de gezondheidszorg CvZ 2000
Jansen SJT Stiggelbout AM Unstable preferences: a shift in valuation or an effect of the elicitation procedure? Med Decis Making 2000 20 62 71 10638538
Tubach F Beaute J Natural History and prognostic indicators of sciatica J of Clin Epidemiology 2004 57 174 179 10.1016/S0895-4356(03)00257-9
Hofstee DJ Gijtenbeek JMM Hoogland PH Houwelingen HC Kloet A Lotters F Tans JTJ Westeinde Sciatica Trial: randomized controlled study of bedrest and physiotherapy for acute sciatica J Neurosurg (Spine 1) 2002 96 45 49
Stiggelbout AM Eijkemans MJC Kiebert GM Kievit J Leer JWH De Haes JCJM The "utility" of the Visual Analog Scale in medical decision-making and technology assessment: is it an alternative to the Time Trade-Off? Int J Technol Assessm Health Care 1996 12 291 8
Dolan P Modelling valuations for health states: the effect of duration Health Policy 1996 38 189 203 10162421 10.1016/0168-8510(96)00853-6
Stolk EA Busschbach JJ Caffa M Meuleman EJ Rutten FF Cost utility analysis of sildenafil compared with papaverine-phentolamine injections BMJ 2000 29 1165 8 10784537 10.1136/bmj.320.7243.1165
Atlas SJ Keller RB Deyo RA Surgical and non-surgical Management of Sciatica Secondary to a Lumbar Disc Herniation. Five-Year Outcomes from the Maine Lumbar Spine Study Spine 2001 15 1179 1187 11413434 10.1097/00007632-200105150-00017
| 15707491 | PMC551598 | CC BY | 2021-01-04 16:32:05 | no | BMC Musculoskelet Disord. 2005 Feb 11; 6:8 | utf-8 | BMC Musculoskelet Disord | 2,005 | 10.1186/1471-2474-6-8 | oa_comm |
==== Front
BMC BiochemBMC Biochemistry1471-2091BioMed Central London 1471-2091-6-11571590410.1186/1471-2091-6-1Research ArticleIncreasing stability of water-soluble PQQ glucose dehydrogenase by increasing hydrophobic interaction at dimeric interface Tanaka Shunsuke [email protected] Satoshi [email protected] Stefano [email protected] Koji [email protected] Department of Biotechnology, Tokyo University of Agriculture and Technology, 2-24-13 Naka-machi, Koganei, Tokyo, 184-8588, Japan2005 16 2 2005 6 1 1 13 8 2004 16 2 2005 Copyright © 2005 Tanaka et al; licensee BioMed Central Ltd.2005Tanaka 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
Water-soluble quinoprotein glucose dehydrogenase (PQQGDH-B) from Acinetobacter calcoaceticus has a great potential for application as a glucose sensor constituent. Because this enzyme shows no activity in its monomeric form, correct quaternary structure is essential for the formation of active enzyme. We have previously reported on the increasing of the stability of PQQGDH-B by preventing the subunit dissociation. Previous studies were based on decreasing the entropy of quaternary structure dissociation but not on increasing the interaction between the two subunits. We therefore attempted to introduce a hydrophobic interaction in the dimeric interface to increase the stability of PQQGDH-B.
Results
Amino acid residues Asn340 and Tyr418 face each other at the dimer interface of PQQGDH-B, however no interaction exists between their side chains. We simultaneously substituted Asn340 to Phe and Tyr418 to Phe or Ile, to create the two mutants Asn340Phe/Tyr418Phe and Asn340Phe/Tyr418Ile. Furthermore, residues Leu280, Val282 and Val342 form a hydrophobic region that faces, on the other subunit, residues Thr416 and Thr417, again without any specific interaction. We simultaneously substituted Thr416 and Thr417 to Val, to create the mutant Thr416Val/Thr417Val. The temperatures resulting in lose of half of the initial activity of the constructed mutants were increased by 3–4°C higher over wild type. All mutants showed 2-fold higher thermal stability at 55°C than the wild-type enzyme, without decreasing their catalytic activities. From the 3D models of all the mutant enzymes, the predicted binding energies were found to be significantly greater that in the wild-type enzyme, consistent with the increases in thermal stabilities.
Conclusions
We have achieved via site-directed mutagenesis the improvement of the thermal stability of PQQGDH-B by increasing the dimer interface interaction. Through rational design based on the quaternary structure of the enzyme, we selected residues located at the dimer interface that do not contribute to the intersubunit interaction. By substituting these residues to hydrophobic ones, the thermal stability of PQQGDH-B was increased without decreasing its catalytic activity.
==== Body
Background
Water-soluble quinoprotein glucose dehydrogenase (PQQGDH-B) from Acinetobacter calcoaceticus has great potential for application as a constituent of an electron mediator-type glucose sensor. The conventionally utilized enzyme for glucose measurement, glucose oxidase (GOD), inherently utilizes oxygen as its electron acceptor during the oxidation of glucose. In contrast, PQQGDH-B is completely independent of oxygen, resulting in improved accuracy and rapidity during glucose measurement. However, because the substrate specificity and stability of PQQGDH-B remain somewhat inferior to those of GOD, we have been engaged in the improvement of these enzymatic properties through protein engineering to further increase the application of PQQGDH-B in glucose monitoring systems [1-17].
The subunit structure of PQQGDH-B was determined to be homodimeric with no activity observed in its monomeric form [17]. Correct quaternary structure is essential for the formation of active enzyme and dissociation of the dimer conformation triggers inactivation of the enzyme. We have previously reported on the increasing of the stability of PQQGDH-B against the dissociation of quaternary structure by chemical cross-linking [12], by constructing tethered enzyme [13], and by introducing Cys residues at the dimer interface to form a novel intersubunit disulfide bond [14]. All of these attempts were based on decreasing the entropy by decreasing the possibility of dimer dissociation, but no attempts had been made to increase the interaction between the two subunits.
In this paper, we report on our rational designing of hydrophobic interaction in the dimer interface to increase the stability of PQQGDH-B. We identified protein regions at the dimer interface where potential novel hydrophobic interactions could be introduced by amino acid substitution, thereby increasing the stability.
Results
Modeling novel dimer hydrophobic core
Among 19 amino acid residues located at the dimer interface, 8 residues were predicted not to be involved in the formation of hydrogen bonds, electrostatic interactions, or hydrophobic interactions (Fig. 1). We focused on the residues Asn340, Tyr418, Thr416, and Thr417, as they are not involved in the formation of the active site cavity and are therefore suitable candidates for amino acid substitution. Although residues Asn340 and Tyr418 face each other on the surface of the dimer interface (Fig. 2), no interaction exists between their side chains. We therefore simultaneously substituted Asn340 to Phe and Tyr418 to Phe or Ile to create the mutants Asn340Phe/Tyr418Phe and Asn340Phe/Tyr418Ile, respectively. Furthermore, the hydrophobic region composed of residues Leu280, Val282, and Val342 faces residues Thr416 and Thr417, again with no specific interaction. We therefore substituted Thr416 to Val and Thr417 to Val to create the double mutant Thr416Val/Thr417Val.
Figure 1 The amino acid residues located at the dimer interface. The two subunits are represented in red and blue, respectively, using the RasMol molecular visualization software 26. The 19 amino acid residues at the interface are shown in space filling format, of which 8 residues (orange and light blue) are predicted not be involve in hydrogen bond formation, electrostatic interaction, or hydrophobic interaction at the interface.
Figure 2 Hydrophobicity of PQQGDH-B dimer interface. Hydrophobic regions are shown in green and hydrophilic regions are shown in blue. Residues that have been substituted in this study are indicated. The structural images were generated using Molecular Operating Environment.
Characterization of mutant enzymes
The activity and stability of each of the three constructed double mutant enzymes were then analyzed. All three mutant enzymes showed slightly higher thermal stability than wild-type PQQGDH-B upon incubation for 10 min at various temperatures (Fig. 3). The temperatures resulting in lose of half of the initial activity were shifted by approximately 3°C higher in the mutants compared to the wild type (Table 1). As the time course of thermal inactivation at 55°C follows first-order kinetics (Fig. 4), half-lives were calculated using logarithmic regression of residual activity. The wild-type enzyme inactivates at 55°C with a half-life of 9.5 ± 1.4 min, while all three double mutants showed greater thermal stability, with half-lives of 5–6°C greater (Table 1).
Figure 3 Thermal stability of wild-type and mutant PQQGDH-Bs. Residual activity was measured at 25°C after 10-min incubations at different temperatures of the following protein samples (0.075 μg/mL): Wild-type ◆, Asn340Phe/Tyr418Phe □, Asn340Phe/Tyr418Ile △, and Thr416Val/Thr417Val ○.
Table 1 Kinetic parameters and thermal stability of PQQGDH-Bs.
Vmax(U/mg) Km(mM) Th(°C)a t1/2 (min)b
Wild type 3030 20 53.9 ± 0.9 9.5 ± 1.4
Asn340Phe/Tyr418Phe 3100 20 57.7 ± 0.4 14.9 ± 1.1
Asn340Phe/Tyr418Ile 2500 20 57.5 ± 0.3 15.5 ± 1.5
Thr416Val/Thr417Val 2800 16 56.5 ± 0.9 14.8 ± 1.4
a Th represents the temperature at which half the initial activity is lost in 10 min.
b t1/2 represents the half-life at 55°C.
Figure 4 Time course of thermal inactivation of wild-type and mutant PQQGDH-Bs. Protein samples (0.075 μg/mL) were incubated at 55°C and aliquots were taken at different times to measure residual activity. The analyzed samples contained the following PQQGDH-Bs: Wild-type ◆, Asn340Phe/Tyr418Phe □, Asn340Phe/Tyr418Ile △, Thr416Val/Thr417Val ○.
Investigation of the kinetic parameters of the wild-type and mutant enzymes shows that the mutations did not significantly affect the overall kinetic properties of PQQGDH-B (Table 1). The specific activities of Asn340Phe/Tyr418Phe (3100 U/mg), Asn340Phe/Tyr418Ile (2500 U/mg), and Thr416Val/Thr417Val (2800 U/mg) were close to that of the wild-type enzyme (3030 U/mg). Except for Thr416Val/Thr417Val, which had a Km value of 16 mM, the mutants had Km values identical to 20 mM Km value of the wild-type enzyme.
Discussion
PQQGDH-B is a 6-blade β-propeller protein, with each blade consisting of a 4-stranded anti-parallel β-sheet (W-motif) [18]. The strands in each W-motif are labeled A-D from the inside to the outside of the molecule [18,19]. All strands are connected by loops, which are named according to the strands they connect. Based on PQQGDH-B structural information, these loop regions have been associated with a number of important functions, such as substrate binding, co-factor binding, and formation of the enzyme active site. As with other β-propeller proteins, the catalytic site and substrate-binding pocket of PQQGDH-B is made up of the cleft formed by loops DA and BC [20]. The enzyme surface composed of loops AB and CD is therefore located opposite the functional region. In the present study, we have introduced mutations at Asn340, Tyr418, Thr416, and Thr417, which are all located in the loop 5CD region. Considering that these residues are located far from the functional region and do not contribute in the structure of functional region, it is not surprising that their substitutions did not significantly alter the enzyme's kinetic parameters, particularly its catalytic activity.
Based on the predicted structure of the mutant enzymes shown in Figure 5, the binding energy of each subunit was calculated, using both the AMBER89 and CHARMM22 force fields, and compared with those of the wild-type enzyme (Table 2). As expected, the increases in hydrophobicity at the subunit interface resulted in increases in their binding energies calculated by both methods. Estimation of the number of hydrophobic interactions based on these same predicted models revealed 4 to 5 novel hydrophobic interactions at the interface of all the mutants, while none were found in the wild-type one. These results based on structural predictions are consistent with the observed improvements in thermal stability.
Table 2 Predicted binding energy of each mutant subunit calculated using AMBER89 and CHARMM22 force fields.
Binding energy (kcal/mol)
AMBER89 CHARMM22
Wild type -249 -118
Asn340Phe/Tyr418Phe -291 -134
Asn340Phe/Tyr418Ile -273 -126
Thr416Val/Thr417Val -278 -131
Figure 5 Hydrophobicity of mutant PQQGDH-B dimer interface. Hydrophobic regions are shown in green while hydrophilic regions are shown in blue. The interfaces shown are those of the Asn340Phe/Tyr418Phe (left), Asn340Phe/Tyr418Ile (middle), and Thr416Val/Thr417Val (right) mutants of PQQGDH-B, with their respective mutation sites circled. The structural images were generated using Molecular Operating Environment.
The addition of hydrophobic interactions in other enzymes has been reported to result in thermostability increases of 2 to 10°C [21-24], comparable to the results of the current study. Recent observations of enzymes from thermophilic organisms indicate that their extraordinary thermal stability is due to hydrophobic interactions. Oligomeric enzyme stability was also reported to be improved through engineering to increase the hydrophobic interaction at the oligomer interface [25]. However, these studies were based on homology analyses between mesophilic and thermophilic bacteria together with random mutagenesis library analyses [21-24]. Although some sequences have been found to be homologous to PQQGDH-B, these are all putative ORFs with no functional information reported. Therefore, there exists no reliable template to help improve the stability of this enzyme by increasing the dimer interface interaction.
Conclusions
We improved PQQGDH-B's thermal stability by increasing the dimer interface interaction through rational design based on its quaternary structure. We demonstrated that this can be achieved by selecting residues located at the dimer interface that do not contribute to the intersubunit interaction and substituting them to hydrophobic ones via site-directed mutagenesis. In each case tested, the enzyme's thermal stability was increased without decreasing its catalytic activity. This rational design approach will provide relevant information for future designs by combining with other mutant PQQGDH-Bs with narrowed substrate specificity and improved catalytic efficiency.
Methods
Chemicals
Glucose, phenazine methosulfate (PMS), 2,6-dichlorophenolindophenol (DCIP), and magnesium chloride were obtained from Kanto Kagaku (Tokyo, Japan), 3-(N-morpholino) propane sulfonate (MOPS) from Dojin (Kumamoto, Japan), and pyrroloquinoline quinone from Mitsubishi Gas Chemical Company (Tokyo, Japan). All other regents were of analytical grade. KpnI was obtained from TOYOBO (Osaka, Japan) and HindIII from New England BioLabs (Beverly, USA).
Site-directed mutagenesis
The structural gene for wild-type PQQGDH-B was previously amplified by polymerase chain reaction (PCR) and inserted into the expression vector pTrc99A (Pharmacia) to create pGB [15]. A 1.2-kbp KpnI-HindIII fragment containing the PQQGDH-B gene was transferred from pGB to pKF18k and mutagenesis was carried out with the Mutan-Express Km kit (Takara) according to the manufacturer's instructions with the oligonucleotides Asn340Phe (5'-GGTGGGACAAAGAATTTACCAGTCC-3'), Tyr418Phe (5'-CGGTACAGCGTCATCAAAAGTAGTGC-3'), Tyr418Ile (5'-CGGTACAGCGTCATCAATAGTAGTGC-3'), and Thr416Val/Thr417Val (5'-CAGCGTCATCATAAACAACGCTATAAGTTGGATC-3'). The mutations (underlined) were confirmed by automated DNA sequencing (ABI PRISM Genetic analyzer 310, Applied Biosystems). The mutated genes were digested with KpnI and HindIII and were replaced into pGB to construct expression vectors containing mutated PQQGDH-B. Numbering of the amino acid positions starts from the first residue of the signal peptide (24 residues).
Enzyme preparation and assay
The PQQGDH-B genes were expressed in Escherichia coli and the enzymes purified as previously reported [15,16]. Following a 30-min preincubation in 10 mM MOPS-NaOH (pH 7.0) containing 1 μM PQQ and 1 mM CaCl2 at room temperature (25°C) to produce the holoenzyme, GDH activity was measured by using 0.6 mM PMS and 0.06 mM DCIP. The enzyme activity was determined by measuring the decrease in absorbance of DCIP at 600 nm.
Analysis of PQQGDH-B stability
The thermal stability of wild-type and mutant PQQGDH-B was determined with 0.075 μg/mL protein, as previously reported [15]. Thermal inactivation experiments were carried out by incubating each holoenzyme in 200 μL of 10 mM MOPS-NaOH, pH 7.0, at 55°C. Aliquots were taken every 5 min and kept at 4°C for 2 min, followed by incubation at room temperature for 30 min. The residual enzyme activity was determined as described above. Since the initial time course for thermal inactivation at 55°C followed first-order kinetics, the thermal stability of each mutant enzyme was expressed as a half-life. The thermal stability of Asn340Phe/Tyr418Phe, Asn340Phe/Tyr418Ile and Thr416Val/Thr417Val were also determined by incubating purified enzyme at various temperatures for 10 min. The residual activities were determined as described above, and were compared with the initial activities.
Prediction of three-dimensional structure and quaternary-dimensional structures
Three-dimensional and quaternary structures were predicted using Molecular Operating Environment (MOE) (Chemical Computing Group Inc., Quebec, Canada). By using the available PDB data of the wild-type PQQGDH-B, 1QBI [18], we made the appropriate substitutions with all possible side-chain orientations to predict the structures of the Asn340Phe/Tyr418Phe, Asn340Phe/Tyr418Ile and Thr416Val/Thr417Val mutants. After addition of hydrogen atoms to the PQQGDH-B structure and optimization of orientation of some hydrogen atoms by MOE, the structures were subjected to energy minimization using the AMBER89 or CHARMM22 force field within the MOE program until the final energy gradient was < 0.01 kcal/mol·Å.
Authors' contributions
ST carried out the site-directed mutagenesis of PQQGDH-B as well as the preparation and characterization of the resulting proteins. SI carried out the 3D modeling and participated in the design of the study. SF participated in interpretation of the results and in drafting the manuscript. KS conceived of the study, participated in its design and coordination, as well as in drafting the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We are grateful to Rie Yamoto for her assistance in preparing the manuscript.
==== Refs
D'Costa EJ Higgins IJ Turner APF Quinoprotein glucose dehydrogenase and its application in an amperometric glucose sensor Biosensors 1986 2 71 87 3454651 10.1016/0265-928X(86)80011-6
Yokoyama K Sode K Tamiya E Karube I Integrated biosensor for glucose and galactose Anal Chim Acta 1989 218 137 142 10.1016/S0003-2670(00)80291-3
Smolander M Livio H-L Rasanen L Mediated amperometric determination of xylose and glucose with an immobilized aldose dehydrogenase electrode Biosensors & Bioelectronics 1992 7 637 643 1337972 10.1016/0956-5663(92)85021-2
Sode K Nakasono S Tanaka M Matsunaga T Subzero temperature operating biosensor utilizing an organic solvent and quinoprotein glucose dehydrogenase Biotechnol Bioeng 1993 42 251 254 10.1002/bit.260420214
Ye L Hammerle M Olsthoorn AJJ Schuhmann W Schmidt H-L Duine JA Heller A High current density "wired" quinoprotein glucose dehydrogenase electrode Anal Chem 1993 65 238 241
Katz E Schlereth DD Schmidt H-L Reconstitution of the quinoprotein glucose dehydrogenase from its apoenzyme on a gold electrode surface modified with a monolayer of pyrroloquinoline quinone Electroanal Chem 1994 368 165 171 10.1016/0022-0728(93)03094-6
Kost GJ Vu H-T Lee JH Bourgeois P Kiechle FL Martin C Miller SS Okorodudu AO Podczasy JJ Webster R Whitlow KJ Multi-center study of oxygen-insensitive handheld glucose point-of-care testing in critical care/hospital/ambulatory patients in the United States and Canada Crit Care Med 1998 26 581 590 9504590 10.1097/00003246-199803000-00036
Laurinavicius V Kurtinaitiene B Liauksminas V Jankauskaite A Simkus R Meskys R Boguslavsky L Skotheim T Tanenbaum S Reagentless biosensor based on PQQ-dependent glucose dehydrogenase and partially hydrolyzed polyarbutin Talanta 2000 52 485 493 10.1016/S0039-9140(00)00396-9
Razumiene J Meskys R Gureviciene V Laurinavicius V Reshetova MD Ryabov AD 4-Ferrocenyllphenol as an electron transfer mediator in PQQ-dependent alcohol and glucose dehydrogenase-catalyzed reactions Electrochemistry Commun 2000 2 307 311 10.1016/S1388-2481(00)00024-2
Schmidt B Oxygen-independent oxidases A new class of enzymes for application in diagnostics Clinica Chim Acta 1997 26 33 37 10.1016/S0009-8981(97)00164-2
Mullen WH Churchhouse J Vadgama P Enzyme electrode for glucose based on the quinoprotein glucose dehydrogenase Analyst 1985 110 925 928 4061851 10.1039/an9851000925
Takahashi Y Igarashi S Nakazawa Y Tsugawa W Sode K Construction and characterization of glucose enzyme sensor employing engineered water soluble PQQ glucose dehydrogenase with improved thermal stability Electrochemistry 2000 68 907 911
Sode K Shirahane M Yoshida H Construction and characterization of a linked-dimeric pyrroloquinoline quinone glucose dehydrogenase Biotechnol Lett 1999 21 707 710 10.1023/A:1005518610946
Igarashi I Sode K Stabilization of quaternary structure of water-soluble quinoprotein glucose dehydrogenase Mol Biotech 2003 24 97 103 10.1385/MB:24:2:97
Sode K Ohtera T Shirahane M Witarto AB Igarashi S Yoshida H Increasing the thermal stability of the water-soluble pyrroloquinoline quinone glucose dehydrogenase by single amino acid replacement Enz Microbial Technol 2000 26 491 496 10.1016/S0141-0229(99)00196-9
Igarashi S Ohtera T Yoshida H Witarto AB Sode K Construction and characterization of mutant water-soluble PQQ glucose dehydrogenases with altered Km value – site-directed mutagenesis studies on the putative active site Biochem Biophys Res Commun 1999 264 820 824 10544015 10.1006/bbrc.1999.1157
Igarashi S Okuda J Ikebukuro K Sode K Molecular engineering of PQQGDH and its applications Arch Biochem Biophys 2004 428 52 63 15234269 10.1016/j.abb.2004.06.001
Oubrie A Rozeboom HJ Kalk KH Duine JA Dijkstra BW The 1.7Å crystal structure of the apo-form of the soluble quinoprotein glucose dehydrogenase from Acinetobacter calcoaceticus reveals a novel internal conserved sequence repeat J Mol Biol 1999 289 319 333 10366508 10.1006/jmbi.1999.2766
Faber HR Groom CR Baker HM Morgan WT Smith A Baker EN 1.8Å crystal structure of the C-terminal domain of rabbit serum haemopexin Structure 1995 3 551 559 8590016 10.1016/S0969-2126(01)00189-7
Oubrie A Rozeboom HJ Kalk KH Olsthoorn AJJ Duine JA Dijkstra BW Structure and mechanism of soluble quinoprotein glucose dehydrogenase EMBO J 1999 18 5187 5159 10508152 10.1093/emboj/18.19.5187
Kallwass HKW Surewicz W Parris W Macfarlane ELA Luyten MA Kay CM Gold M Jones JB Single amino acid substitution can further increase the stability of a thermophilic L-lactate dehydrogenase Protein Eng 1992 5 769 774 1287656
Kirino H Aoki M Aoshima M Hayashi Y Ohba M Yamagishi A Wakagi T Oshima T Hydrophobic interaction at the subunit interface contributes to the thermostability of 3-isopropylmalate dehydrogenase from an extreme thermophile, Thermus thermophilus Eur J Biochem 1994 220 275 281 8119295
Akanuma S Yamagishi A Tanaka N Oshima T Serial increase in the thermal stability of 3-isopropylmalate dehydrogenase from Bacillus subtilis by experimental evolution Protein Sci 1998 7 698 705 9541402
Ohkuri T Yamagishi A Increased thermal stability against irreversible inactivation of 3-isopropylmalate dehydrogenase induced by decreased van der Waals volume at the subunit interface Protein Eng 2003 16 615 621 12968079 10.1093/protein/gzg071
Glaser F Steinberg DM Vakser IA Ben-Tal N Residue frequencies and pairing preferences at protein-protein interfaces Proteins 2001 43 89 102 11276079
Sayle S Milner-White EJ RasMol: Biomolecular graphics for all Trends in Biochemical Sciences TIBS 1995 20 374 7482707
| 15715904 | PMC551599 | CC BY | 2021-01-04 16:36:49 | no | BMC Biochem. 2005 Feb 16; 6:1 | utf-8 | BMC Biochem | 2,005 | 10.1186/1471-2091-6-1 | oa_comm |
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-321571793010.1186/1471-2105-6-32Methodology ArticleA note on generalized Genome Scan Meta-Analysis statistics Koziol James A [email protected] Anne C [email protected] Department of Molecular and Experimental Medicine, The Scripps Research Institute, 10550 North Torrey Pines Road, MEM216, La Jolla, CA 92037, USA2005 17 2 2005 6 32 32 24 11 2004 17 2 2005 Copyright © 2005 Koziol and Feng; 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
Wise et al. introduced a rank-based statistical technique for meta-analysis of genome scans, the Genome Scan Meta-Analysis (GSMA) method. Levinson et al. recently described two generalizations of the GSMA statistic: (i) a weighted version of the GSMA statistic, so that different studies could be ascribed different weights for analysis; and (ii) an order statistic approach, reflecting the fact that a GSMA statistic can be computed for each chromosomal region or bin width across the various genome scan studies.
Results
We provide an Edgeworth approximation to the null distribution of the weighted GSMA statistic, and, we examine the limiting distribution of the GSMA statistics under the order statistic formulation, and quantify the relevance of the pairwise correlations of the GSMA statistics across different bins on this limiting distribution. We also remark on aggregate criteria and multiple testing for determining significance of GSMA results.
Conclusion
Theoretical considerations detailed herein can lead to clarification and simplification of testing criteria for generalizations of the GSMA statistic.
==== Body
Background
Wise, Lanchbury and Lewis [1] introduced a rank-based statistical technique for meta-analysis of genome scans, the Genome Scan Meta-Analysis (GSMA) method, and derived its exact null distribution using a clever inclusion/exclusion argument. Koziol and Feng [2] provided an alternative derivation of the null distribution of the GSMA statistic via a combinatoric approach involving probability generating functions, and suggested an Edgeworth series approximation to its exact null distribution that improves upon the Wise [1] normal approximation.
Levinson [3] described two generalizations to the GSMA statistic: (i) a weighted version of the GSMA statistic, so that different studies could be ascribed different weights for analysis; and (ii) an order statistic approach, reflecting the fact that a GSMA statistic can be computed for each chromosomal region or bin across the various genome scan studies. Wise [1] had suggested that each chromosomal region (bin) be about 30 cM, leading to a total of about n = 120 bins spanning the entire genome, and correspondingly 120 GSMA statistics. Wise [1] and Koziol and Feng [2] had investigated the marginal distribution of any of these (exchangeable) GSMA statistics, whereas under the order statistic formulation of Levinson [3], the joint distribution of the entire set of GSMA statistics is taken into account. In this note, we consider both generalizations in turn. In particular, (i) we provide an Edgeworth approximation to the null distribution of the weighted GSMA statistic, analogous to that in Koziol and Feng [2]; and (ii) we examine the limiting distribution of the GSMA statistics under the order statistic formulation, and quantify the relevance of the pairwise correlations of the GSMA statistics across different bins on this limiting distribution. We conclude with remarks concerning the Levinson [3] aggregate criteria and multiple testing for determining significance of GSMA results.
Results
The GSMA statistics
We first introduce some notation. Let Xij, i = 1, ..., m, j = 1, ..., n, denote the rank of any particular linkage test statistic (e.g., LOD score) in the jth chromosomal region (bin) from the ith study, with each study being ranked separately. Levinson [3] rank the bins from 1 = "best" to n = "worst" on the basis of, say, maximum LOD score or lowest p value observed within each bin, but the reverse ranking from 1 = "worst" to n = "best" is also feasible. In practice, m can be as few as 4 (e.g., [4,5]); and, following Wise [1], n is generally about 120. The GSMA statistics are then S1, ..., Sn, where . The exact (marginal) null distribution of each Sj was derived in Wise [1]; in the notation of Levinson [3], PAvgRnk, the "pointwise probability" of any Sj, is obtained from its marginal null distribution. The normal approximation to the exact distribution of the Sj is straightforward: the Sj are identically distributed, and each Sj has an approximate normal distribution with mean and variance under the null hypothesis that ranks are randomly assigned within each study. Koziol and Feng [2] provided an Edgeworth correction to this approximation, and recommended that the correction be used, at least for m ≤ 12.
The weighted GSMA statistic
Levinson [3] proposed a weighted version of the GSMA statistic, namely, , with the weight wi ascribed to the ith study reflecting the relative linkage information from that study. (We are temporarily omitting the j subscript for clarity.) The normal approximation to the marginal null distribution of Sw is straightforward, and depends on the two parameters and . The combinatorial argument utilized by Koziol and Feng [2] to derive the exact distribution of the unweighted GSMA statistic (which relies on probability generating functions) is generally no longer applicable in the weighted setting. Nevertheless, as in Koziol and Feng [2], we here provide an Edgeworth correction that may be applied to the weighted GSMA statistic. To this end, we equivalently consider the linear transform , where . We then have E[Rw] = 0, , and
(We have used Koziol and Feng [2], eqn. 11, for .)
The Edgeworth Type A series approximation to the density of up to 4th order terms, is f (z) = φ (z)[1 + c4H4 (z)], where φ(·) is the standard normal density function, , H4 (·) is the 4th degree Hermite-Chebyshev polynomial, H4 (z) = z4 - 6z2 + 3, and the constant c4 given by (Stuart and Ord [6], eqn. 6.42). Furthermore, the cumulative distribution function of the Edgeworth series is given simply by
where Φ(·) denotes the cumulative distribution function of the standard normal distribution, c4 and φ(·) are as above, and H3(·) is the 3rd degree Hermite-Chebyshev polynomial, H3 (z) = z3 - 3z (Stuart and Ord [6], eqn. 6.43).
In practice, we would expect the Edgeworth series approximation to provide an adequate representation of the exact distribution of the weighted GSMA statistic, in a manner analogous to the unweighted case [2]. Here, we briefly investigate the adequacy of the Edgeworth approximation, using an example from Lewis et al. [7]. They had applied the GSMA methodology to data from m = 20 schizophrenia genome scans, and found strong evidence for linkage on chromosome 2q, as well as suggestive evidence for linkage at several other chromosomal locations. The rank data for each scan are available online at D.F. Levinson's website (accessed July 14, 2004) [8], and we use these data to reconstruct first the unweighted GSMA statistics Sj, j = 1, 2, ..., 120, corresponding to the 120 bins spanning the entire genome, then their preferred weighted versions. Lewis [7] had recommended weights for each individual study proportional to the square root of the number of affected cases for that study. From Levinson's website [8], the individual weights wi, i = 1, 2, ..., 20, are 2.32, 1.77, 1.20, 1.17, 1.17, 1.16, 1.15, 1.08, 1.03, 1.01, 0.95, 0.88, 0.80, 0.80, 0.68, 0.67, 0.59, 0.54, 0.53, 0.51, greater than a four-fold range.
We simulated the null distribution of the weighted GSMA statistic , where with m = 20, n = 120, by drawing each Xi as an independent random integer from 1 to 120 (that is, a uniform distribution of the integers from 1 to 120), then forming Rw with the Levinson [8] weights. We used the random number generator in R [9], to produce 10,000 values for Rw. We then formed , and compared the empirical distribution of the 10,000 Z values with the Edgeworth approximation as described above; for comparative purposes, we also computed the normal approximation, which is based on matching the first two moments rather than the first four moments with Edgeworth. Figure 1 shows the resulting quantile-quantile plot of the empirical distribution of the weighted GSMA Z values with both a normal approximation, panel A, and the Edgeworth approximation, panel B. Note that, even in this setting of the weighted combination of m = 20 individual GSMA statistics, the normal approximation is particularly ill-fitting in the tails. Agreement in the tails would be of particular relevance in practical applications, as these represent the areas of potentially significant findings (p-values). The noticeable disagreement in the tails between the weighted GSMA statistic and its normal approximation is largely ameliorated with the Edgeworth approximation. With attendant computational savings, the Edgeworth approximation provides a practical means of determining significance of weighted GSMA results compared to simulation; tail probabilities derived from the normal approximation should only be used with extreme caution.
The GSMA order statistics
We turn next to order statistic considerations (and reintroduce the subscript j). The Levinson [3] order statistic approach to inference relating to the GSMA statistics takes into account the inherent ordering of the Sj: their Pord refers to the probability of any observed Sk given the kth bin's place in the ordering of all of the Sj. We here derive approximations to this distribution. Let , j = 1, ..., n, and T(1) <T(2) < ... T(n) denote the order statistics. We note first that the Tj have (approximately) a singular symmetric multivariate normal distribution, with means 0, variances 1, and correlations . That the joint distribution is singular follows from the observation that for all i, hence, is identically 0. If we dismiss the correlations as negligible (of absolute magnitude < 0.01 for n > 100), then the Tj are (approximately) independent, identically distributed N(0,1) (standard normal) random variates, and the cumulative distribution function (cdf) Fk of the kth order statistic T(k) is given by
with Φ(·) as above (David [10], eqn. 2.1.3).
We briefly examine whether correlations can be ignored when determining the distributions of the T(j). Numerical computation of the distributions of the order statistics from a symmetric multivariate normal distribution is feasible in a number of cases; we here examine perhaps the most relevant case, concerning the extreme T(n). Note that
Prob (T(n) ≤ x) = Prob (T1 ≤ x,T2 ≤ x, ...,Tn ≤ x); (2)
this latter probability may be calculated in R using the mvtnorm package [9], based on methodology by Genz [11,12]. With n = 120, we depict in Figure 2 a Q-Q plot of the (approximate) distribution of T(n) under independence, eqn. (1), compared to the distribution from eqn. (2) with pairwise correlations . The independence model tends to agree quite closely to the correlation model in this particular case, especially in the critical right tail, and has the virtue of numerical simplicity. We remark that one might improve slightly on the normal independence model by incorporating the Edgeworth correction into the individual cumulative distribution functions in equation (1).
Aggregate criteria and multiple testing
Levinson [3] had proposed an aggregate criterion for detecting linkage based on both the marginal distributions and the order statistic distributions of the GSMA statistics. In particular, they argued that bins that have achieved both PAvgRnk < 0.05 and Pord < 0.05 "are the most likely to contain linked loci or to be adjacent to linked bins". Note that their criterion entails both the marginal distribution of the Tj, through PAvgRnk, and the (joint) order statistic distribution of the Tj, through Pord. We remark that there is some redundancy to the aggregate criterion {PAvgRnk < 0.05 and Pord < 0.05}, as can be seen through consideration of critical values relating to their aggregate criterion. With the normal approximation to the distribution of each normalized GSMA statistic Tj, the criterion {PAvgRnk < 0.05} is equivalent to the criterion {Tj > 1.645}. The criterion {Pord < 0.05} may be computed from eqn. (1), and depends on the ordering of the individual Tj. With n = 120, then for the ten largest order statistics T(120), T(119), ..., T(111), the criterion {PAvgRnk < 0.05 and Pord < 0.05} reduces to {Pord < 0.05}, since their 95th percentiles under their joint order statistic distribution exceed 1.645 [implying that, if {Pord < 0.05} obtains, then {PAvgRnk < 0.05} will automatically be satisfied]; and, for the remaining order statistics T(110), T(109), ..., T(1), the criterion {PAvgRnk < 0.05 and Pord < 0.05} reduces to {PAvgRnk < 0.05}, equivalently, {T(j) > 1.645}, as their 95th percentiles under the order distribution, eqn. (1), are less than 1.645 [implying that, if {PAvgRnk < 0.05} obtains, then {Pord < 0.05} will automatically be satisfied].
We conclude with a remark concerning multiple testing. Levinson [3] suggested a simple Bonferroni correction for multiple testing when determining the significance of GSMA results. In particular, they used the criterion {PAvgRnk < 0.000417} (0.05 corrected for 120 tests) for evidence that a bin is likely to contain a linked locus or loci. One can improve on this procedure by using Holm's [13] paradigm for multiple testing rather than Bonferroni. We illustrate Holm's [13] procedure by returning to the Lewis [7] study with m = 20 schizophrenia genome scans. As noted above, we used the online data to reconstruct the normalized unweighted GSMA statistics Tj,j = 1, 2, ..., 120, corresponding to the 120 bins spanning the entire genome. With m = 20 studies, we shall invoke the normal approximation to the distributions of the individual Tj.
Lewis [7] had extensively investigated various criteria for linkage from the 20 schizophrenia genome scans, and we shall not reproduce their analyses. Rather, we here illustrate a graphical procedure for the simultaneous evaluation of p-values from tests on the same data; this procedure is immediately applicable to the simultaneous consideration of the 120 GSMA statistics. The procedure, originally suggested by Schweder and Spjøtvoll [14], consists of a probability plot of the p-values versus the uniform distribution. Koziol [15] subsequently suggested that Holm's [13] paradigm for multiple testing be applied to Schweder and Spjøtvoll's [14] scenario, for a graphical determination of the number of true hypotheses.
Let us briefly review the Holm [13] method, which is an extension of the Bonferroni method for multiple comparisons. Suppose we compare the smallest p-value P(1) among n p-values with α/n and we find that the p-value is less than α/n. Then our multiple testing problem has been reduced by one test, and we should compare the next smallest p-value P(2) to . In general, we would compare P(i) with . Holm's [13] step-down test begins with i = 1, comparing P(i) with , and stops as soon as P(i) exceeds , rejecting at overall level α all prior tests with smaller p-values. The Holm [13] method, like Bonferroni, makes no assumption on the dependence of tests, but beyond P(1) is less conservative than Bonferroni.
In Figure 3A we present a probability plot of the 120 p-values corresponding to the 120 individual Tj statistics, which we have recomputed from the online Levinson dataset [8]. On this plot, the points corresponding to the "true" hypotheses of no linkage in individual bins should roughly fall along a straight line passing through the origin. We have also superimposed the Bonferroni and Holm boundaries for overall alpha level 0.05 and n = 120 p-values; but, the two boundaries are virtually indistinguishable. There is little indication of large departures from the global null hypothesis of no linkage.
In Figure 3B we rescale the y-axis, and focus solely on the Bonferroni and Holm boundaries. Differences are most readily apparent for the largest ordered p-values. On the other hand, with a large number of hypotheses (here 120), the improvement of Holm over Bonferroni at the smallest ordered p-values is marginal at best. As a reviewer has presciently remarked, the Holm procedure generally is most helpful (advantageous) relative to Bonferroni with only a small number of hypotheses.
In Figure 3C we zoom in on the part of the probability plot nearest the origin; we here have superimposed the Holm [13] boundary. In accord with Lewis [7], we find that only one GSMA statistic achieves statistical significance at overall alpha level 0.05, namely, the statistic corresponding to bin 2.5. [Recall that the Holm and Bonferroni boundaries coincide at the smallest p-value, P(1).] That is, in this particular instance, the unweighted GSMA statistics with either Bonferroni or Holm [13] correction for multiple testing identify statistically evidence for linkage on chromosme 2q.
Conclusion
For practitioners utilizing GSMA statistics, the question arises as to whether the methods proposed here as well as in Koziol and Feng [2] are merely of theoretical interest, or have practical import. If one utilizes solely the unweighted GSMA statistic, and chooses to consider its marginal distribution (corresponding to the PAvgRnk formulation of Levinson [2]), then the exact null distribution of the GSMA statistic is available from Wise [1] or Koziol and Feng [2], and should be preferred over any approximate methods. If the exact null distribution is computationally intractable for practitioners, then the Edgeworth approximation of Koziol and Feng [2] provides a simple and accurate means of assessing significance; we would argue that the Edgeworth approximation is preferable to a normal approximation in this instance. When weights are introduced into the GSMA statistic, then the combinatoric arguments of Wise [1] and Koziol and Feng [2] will typically be insufficient to derive the exact null distribution [though we remark that a moment generating function approach patterned after the probability generating function formulation of Koziol and Feng [2] can be brought to bear on this problem.] One can either simulate the null distribution or derive an Edgeworth approximation: we do not believe either method enjoys global advantages over the other. We caution against simple reliance on a normal approximation: in the situation investigated here, Figure 1, the weighted combination of m = 20 individual GSMS statistics, the normal approximation is particularly ill-fitting in the tails. [Agreement in the tails is of particular relevance to practitioners, as these represent the areas of potentially significant findings (p-values).] As for the order statistic formulation and the aggregate criteria of Levinson [3], we believe that the theoretical considerations given in this paper can lead to clarification and simplification of testing criteria.
Authors' contributions
JAK conceived, designed and drafted the manuscript. ACF performed the statistical simulation.
Both authors read and approved the final manuscript.
Acknowledgements
This research was supported in part by NIH grant RR00833 to the Scripps General Clinical Research Center. We thank the reviewers for their insightful comments and suggestions.
Figures and Tables
Figure 1 Quantile-quantile plots of the weighted GSMA statistic vs. a normal approximation and an Edgeworth approximation. A. Quantile-quantile plot of the null distribution of the weighted GSMA statistic, x-axis, versus the Edgeworth approximation, y-axis. B. Quantile-quantile plot of the empirical null distribution of the weighted GSMA statistic, x-axis, versus the Edgeworth approximation, y-axis. The empirical null distribution of the weighted GSMA statistic was obtained from 10,000 simulations, with m = 20 scans, n = 120 bins per scan, and weights for the 20 scans taken from Lewis [7]. The normal approximation shares the first two moments as the weighted GSMA statistic; the Edgeworth approximation shares the first four moments as the weighted GSMA statistic. Quantiles are depicted from the .0001 percentage point to the .9999 percentage point.
Figure 2 Quantile-quantile Plot of T(n) Distribution. Quantile-quantile plot of the approximate normal distribution of T(n), the largest (order) GSMA statistic, under the correlation model (eqn. 2, with pairwise correlations ), x-axis, versus the independence model (eqn.1), y-axis. Following Wise [1], we chose n = 120. Quantiles are depicted from the .001 percentage point to the .999 percentage point.
Figure 3 Probability plot of GSMA schizophrenia statistics. A. Probability plot of the 120 p-values corresponding to the 120 GSMA statistics T1 , ..., T120 from the 20 schizophrenia genome scans reported in Lewis [7], versus the (expected) uniform distribution. Also depicted are the Bonferroni (solid line) and Holm (dotted line) boundaries at overall alpha level 0.05. B. Inset of Figure 3A, in which we display solely the Bonferroni and Holm boundaries. We have rescaled the y-axis so as to emphasize the differences in the boundaries, and have relabeled the x-axis to correspond to the fact that we here have n = 120 ordered p-values. C. Inset of Figure 3A, illustrating the 12 smallest ordered p-values (circles), along with a Holm [13] boundary (solid line) at overall alpha level 0.05. [We are using the integer ordering of the x-axis as in Figure 3B.] Only the first p-value, corresponding to bin 2.5, falls outside this boundary. The bins depicted here, from left to right, are: 2.5, 3.2, 11.5, 5.5, 20.2, 8.2, 6.1, 2.6, 22.1, 1.6, 1.7, and 5.3.
==== Refs
Wise LH Lanchbury JS Lewis CM Meta-analysis of genome searches Ann Hum Genet 1999 63 263 72 10738538 10.1046/j.1469-1809.1999.6330263.x
Koziol JA Feng AC A note on the genome scan meta-analysis statistic Ann Hum Genet 2004 68 376 380 15225163 10.1046/j.1529-8817.2004.00103.x
Levinson DF Levinson MD Segurado R Lewis CM Genome scan meta-analysis of schizophrenia and bipolar disorder, part I: Methods and power analysis Am J Hum Genet 2003 73 17 33 12802787 10.1086/376548
Chiodini BD Lewis CM Meta-analysis of 4 coronary heart disease genome-wide linkage studies confirms a susceptibility locus on chromosome 3q Arterioscler Thromb Vasc Biol 2003 23 1863 1868 12947017 10.1161/01.ATV.0000093281.10213.DB
Demenais F Kanninen T Lindgren CM Wiltshire S Gaget S Dandrieux C Almgren P Sjogren M Hattersley A Dina C Tuomi T McCarthy MI Froguel P Groop LC A meta-analysis of four European genome screens (GIFT consortium) shows evidence for a novel region on chromosome 17p11.2-q22 linked to type 2 diabetes Hum Mol Genet 2003 12 1865 1873 12874106 10.1093/hmg/ddg195
Stuart A Ord JK Kendall's Advanced Theory of Statistics 1987 5 New York: Oxford University Press [Distribution Theory, Vol 1.]
Lewis CM Levinson DF Wise LH DeLisi LE Straub RE Hovatta I Williams NM Schwab SG Pulver AE Faraone SV Brzustowicz LM Kaufmann CA Garver DL Gurling HM Lindholm E Coon H Moises HW Byerley W Shaw SH Mesen A Sherrington R O'Neill FA Walsh D Kendler KS Ekelund J Paunio T Lonnqvist J Peltonen L O'Donovan MC Owen MJ Wildenauer DB Maier W Nestadt G Blouin JL Antonarakis SE Mowry BJ Silverman JM Crowe RR Cloninger CR Tsuang MT Malaspina D Harkavy-Friedman JM Svrakic DM Bassett AS Holcomb J Kalsi G McQuillin A Brynjolfson J Sigmundsson T Petursson H Jazin E Zoega T Helgason T Genome scan meta-analysis of schizophrenia and bipolar disorder, part II: Schizophrenia Am J Hum Genet 2003 73 34 48 12802786 10.1086/376549
Levinson Research Page
The R Project for Statistical Computing
David HA Order Statistics 1970 New York: John Wiley
Genz A Numerical computation of multivariate normal probabilities Journal of Computational and Graphical Statistics 1992 1 141 150
Genz A Comparison of methods for the computation of multivariate normal probabilities Computing Science and Statistics 1993 25 400 405
Holm S A simple sequentially rejective multiple test procedure Scand J Statist 1979 6 65 70
Schweder T Spjøtvoll E Plots of p-values to evaluate many tests simultaneously Biometrika 1982 69 493 502
Koziol JA A note on plots of p-values to evaluate many tests simultaneously Biom J 1989 8 969 972
| 15717930 | PMC551600 | CC BY | 2021-01-04 16:02:52 | no | BMC Bioinformatics. 2005 Feb 17; 6:32 | utf-8 | BMC Bioinformatics | 2,005 | 10.1186/1471-2105-6-32 | oa_comm |
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-351572535710.1186/1471-2105-6-35Methodology ArticleTwo-part permutation tests for DNA methylation and microarray data Neuhäuser Markus [email protected] Tanja [email protected]öckel Karl-Heinz [email protected] Institute for Medical Informatics, Biometry and Epidemiology, University of Duisburg-Essen, Hufelandstr. 55, D-45122 Essen, Germany2005 22 2 2005 6 35 35 10 12 2004 22 2 2005 Copyright © 2005 Neuhäuser 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
One important application of microarray experiments is to identify differentially expressed genes. Often, small and negative expression levels were clipped-off to be equal to an arbitrarily chosen cutoff value before a statistical test is carried out. Then, there are two types of data: truncated values and original observations. The truncated values are not just another point on the continuum of possible values and, therefore, it is appropriate to combine two statistical tests in a two-part model rather than using standard statistical methods. A similar situation occurs when DNA methylation data are investigated. In that case, there are null values (undetectable methylation) and observed positive values. For these data, we propose a two-part permutation test.
Results
The proposed permutation test leads to smaller p-values in comparison to the original two-part test. We found this for both DNA methylation data and microarray data. With a simulation study we confirmed this result and could show that the two-part permutation test is, on average, more powerful. The new test also reduces, without any loss of power, to a standard test when there are no null or truncated values.
Conclusion
The two-part permutation test can be used in routine analyses since it reduces to a standard test when there are positive values only. Further advantages of the new test are that it opens the possibility to use other test statistics to construct the two-part test and that it avoids the use of any asymptotic distribution. The latter advantage is particularly important for the analysis of microarrays since sample sizes are usually small.
==== Body
Background
The addition of a methyl group at the carbon-5 position of cytosine is a modification of DNA called DNA methylation. In mammalian cells, DNA methylation is essential for proper development [1]. The methylation patterns of tumor cells are altered compared to those of normal cells, moreover, there are also differences between different types of cancer as shown for subtypes of leukemia [2] and lung cancer [3]. Thus, DNA methylation analysis promises to become a powerful tool in cancer diagnosis [4].
DNA methylation data can be obtained using the MethyLight technology [5]. When the tested region is not or only partially methylated the result is negative (undetectable methylation, null values). In contrast, samples that show methylation will have a value greater than 0 [4]. Thus, DNA methylation data obtained with MethyLight have a clump of zero observations and a continuous nonzero part. For such a data structure, two-part models as proposed by Lachenbruch [6-8] are applicable. In that approach, the test statistic is the sum of two squared statistics, one comparing the proportions of zeros and one comparing the positive values. For example, one can use the binomial test and the Wilcoxon rank sum test. The asymptotic null distribution of the sum of the squares of the two test statistics is χ2 with two degrees of freedom (df = 2).
In microarray data it is relatively common that small and negative expression levels were clipped-off to be equal to an arbitrarily chosen cutoff value. Recent examples used different cutoff values: 1, 20, and 50, respectively [9-11]. Aside from the fact that negative values make no biological sense, there are, with regard to oligonucleotide arrays from Affymetrix, two primary reasons for truncating the values [12]. First, spots at the low intensity range are generally more vulnerable to noise, thus, it is thought that the technology produces a poor discrimination at low levels of expression [13]. Second, the focus is on expression of identified genes and expressed sequence tags. Differences at negative or low values may result from differences in binding to the mismatch probes. Since it is generally not known what binds to the mismatch probes, the differences at negative or low values cannot be attributed to target genes.
Due to the truncation there are two different types of data: truncated values and original observations. Since the truncated values are not just another point on the continuum of possible values, it would be inappropriate to use a standard statistical method that would treat all values equally [14]. The two different types of data should be analyzed separately. Therefore, the two-part model, comparing the proportions of truncated values and the distribution of positive values, is applicable. Note that negative expression levels are not possible when the Affymetrix Microarray Suite (MAS) 5.0 software is used. However, small values are possible and may be truncated.
Since, in microarray experiments, the sample sizes, i.e. the numbers of replications, are usually very small [15,16], the use of the asymptotic distribution of Lachenbruch's two-part test statistic may be questionable. Thus, we carried out permutation tests with the two-part statistic. For a permutation test all possible permutations under the null hypothesis are generated. In our situation we permute the group labels for the whole sample, i.e. for truncated values (or the null values in case of methylation data) and original observations. Then, the test statistic is calculated for each permutation. The null hypothesis can then be accepted or rejected using the permutation distribution of the test statistic, the p-value being the probability of the permutations giving a value of the test statistic as supportive or more supportive of the alternative than the observed value [17,18]. Thus, inference is based upon how extreme the observed test statistic is relative to other values that could have been obtained under the null hypothesis.
We found that, in the case of a two-part model, the permutation test is not only a way to avoid the use of an asymptotic distribution, but also is a more powerful test, i.e. a test that produces, on average, smaller p-values. In addition, the permutation test reduces, without any loss of power, to a single test if no truncated (or null) values were present. Thus, the proposed test is applicable in routine use whether or not truncated (or null) values occur. After the definition of the tests in the following section, we present our findings for DNA methylation data and microarray data. We then confirm the results using simulations.
Two-part tests
As briefly mentioned above a two-part test statistic is the sum of two squared statistics, one comparing the proportions of truncated values and one comparing the positive values. Let n1 and n2 be the numbers of independent observations regarding one gene (or one region in case of methylation data, respectively), for two groups to be compared. The observed numbers of truncated values (or null values in case of methylation data) in the two groups are denoted by m1 and m2. To compare these numbers m1 and m2 Lachenbruch [6] used the statistic
where , , and . Under the null hypothesis the proportions of truncated values are not different between the two groups, and B2 is asymptotically χ2-distributed with df = 1. B2 is always well defined, unless there are only truncated values in both groups or no truncated values at all. For these two extreme cases we set B2 = 0.
For the second part in Lachenbruch's two-part model one can use different tests, Lachenbruch [6-8] considered the Wilcoxon rank sum test, Student's t test, and the Kolmogorov-Smirnov test. The use of the latter test in a two-part model, however, was too liberal, the type I error rate was close to 0.065 (for a significance level of α = 0.05 [7,8]). Both other tests work well. Here, we apply the Wilcoxon test for two reasons. This test was used in the original analyses of the data we use below [3,11]. Nonparametric tests based on ranks are more appropriate for non-normally distributed data such as microarray data [19].
The standardized rank sum statistic based on the non-truncated (or positive values in case of methylation data) values is defined as
where RS is the rank sum, i.e. the sum of the ranks in group 1. When there are ties within the non-truncated values the denominator slightly changes [[20], p. 109]. For the extreme case that there are only truncated values in at least one group we set W = 0.
The test statistic for the two-part test is X2 = B2 + W2. Under the null hypothesis of no difference between the groups, X2 is asymptotically χ2-distributed with df = 2 [6,7]. Alternatively, a permutation test can be performed with the statistic X2. This test, called two-part permutation test here, is a permutation test based on the sum statistic X2. It is carried out by permuting the group labels for the whole sample. Thus, all observations, truncated and non-truncated values (or null and positive values in case of methylation data) are reallocated to the groups. When performing this two-part permutation test the exact permutation distribution of X2 is determined. This distribution, computed by generating all possible permutations or, for the p-values given below in Table 1, using a simple random sample of 20,000 permutations, is used to compute the p-value. Since it is a permutation test based on the sum X2, it is neither necessary to determine the permutation distributions of the summands B2 and W2 nor to calculate the p-values of the univariate tests related to B2 and W2.
Application to actual methylation data
We use DNA methylation data from 7 regions and 87 lung cancer cell lines, 41 lines are from small cell lung cancer and 46 lines from non-small cell lung cancer [3,4]. The proportion of positive values for the different regions ranges from 39 to 100% for the small cell lung cancer and from 65 to 98% for the non-small cell lung cancer. The data are available at . Siegmund et al. [4] transformed the data by standardizing the positive values on the natural-log scale. However, for the tests applied here, this transformation has no influence.
Table 1 presents the p-values of the two tests. For most regions the two-part permutation test gives a smaller p-value than the original two-part test. The only exception is the region APC. However, the original two-part test's p-value for this region is 0.1684 and, for a p-value of this size, a small change in the value is usually of no importance.
Application to actual microarray data
Tschentscher et al. [11] performed an experiment with HG-U95Av2 oligonucleotide arrays in order to compare patients with uveal melanomas with and without monosomy 3. Expression values were calculated by use of the MAS 4.0 software. The data are available at . The sample size in this microarray experiment is 10 per group. As mentioned above, expression levels below 50 were set to 50. This data truncation occurred in 2,215 (28%) out of 7,902 genes. First, we consider these 2,215 genes. Figure 1 displays the number of genes for the different number of truncated values per gene.
Table 2 shows the frequencies of different size groups of the p-values. Often, the two-part permutation test gives a smaller p-value than the original two-part test. For instance, for 19 genes the p-value is ≤ 0.001 when the latter test is applied. The permutation test gives a p-value ≤ 0.001 for these 19 and for 39 additional genes. As usual, the majority of genes do not show any indication of being differentially expressed. For these genes a slight change in the p-value is of no importance. Thus, in Table 3, we consider the genes for which the p-value of the original two-part test is ≤ 0.1. Out of the 2,215 genes 514 remain. As shown in Table 3 the p-values of the two-part permutation test are, on average, distinctly smaller than those of the original two-part test.
For a large proportion of genes (72% in this data set) there are no truncated values. In that case, the two-part statistic reduces to the sum 0 plus the squared standardized rank sum W2. Of course, one has to define a priori whether the two-part test or the Wilcoxon test will be used to analyze the data. If the original two-sided (asymptotic) Wilcoxon test were chosen and applied, one could compare W2 with critical values from a χ2 distribution with one degree of freedom (df). If the two-part were chosen there is df = 2 and, in case of no truncated values, power is lost compared to the original Wilcoxon test. For instance, the 95% percentile of the χ2 distribution with df = 1 is 3.84, but it is 5.99 for df = 2. The permutation test using the sum statistic X2 does not suffer from this power loss: When there are no truncated values there is, of course, no difference in the proportions of truncated values, and the test statistic X2 is, for every permutation, the sum 0 + W2. Thus, the two-part permutation test reduces to the exact two-sided Wilcoxon rank sum test when there are no truncated values. Consequently, this permutation test does not only give smaller p-values, but it is also applicable in routine use whether or not truncated values are present.
Simulation study
The two different tests, the original two-part test and the two-part permutation test, were compared in a Monte Carlo simulation study performed using SAS version 8.2, 5,000 simulation runs were generated for each configuration. The sample size of 10 per group was chosen as in the microarray experiment presented above. In some configurations some randomly chosen values were set to 0 according to binomial distributions with the probabilities p1 and p2, and the remaining observations were generated according to a lognormal distribution (with median 1 and σ = 1). Then, the values of one group were shifted if applicable. In some other configurations all observations were generated according to the lognormal distribution and, in one group, shifted. Then, values smaller than a cutoff value were truncated.
The type I error rates of the two tests are very similar. With e.g. p1 = p2 = 0.3 and a significance level of α = 0.05 the simulated type I error rates were 0.049 for both the original two-part test and the two-part permutation test.
Table 4 displays results for situations with a difference between the two groups. As above, only those comparisons were regarded for which there is some indication of a difference, i.e. a p-value ≤ 0.1 of the original two-part test. In all considered configurations the median of the difference between the p-values of the original two-part test and the two-part permutation test is positive. The finding that the p-values of the permutation test are smaller corresponds to a higher power of this test. The power is given in Table 5, as shown the power of the two-part permutation test is at least as high as that of the original two-part test. There is only one exception, the latter test is slightly more powerful in one situation, i.e. when the proportion of zeros is higher in group 1 and the positive values are larger in group 1.
Discussion
Previous research demonstrated that a two-part test is appropriate and powerful in the presence of a clump of zero observations (i.e. truncated or null values). In this paper we propose a permutation test for such situations with two types of data. Usually, nonparametric tests can be performed based on an asymptotic distribution or based on a permutation null distribution. The two approaches often give similar results, especially when the sample sizes are large. However, in the case of a two-part test one cannot simply replace the asymptotic distributions of B2 and W2 by the exact permutation distributions. If so, one would compute two exact p-values although the aim of a two-part test is to receive one p-value that combines information from both parts. Therefore, the two-part permutation test uses the exact permutation distribution of the sum statistic X2. That this permutation distribution of the sum is generated rather than to simply replace the asymptotic distributions of the summands by their exact permutation distributions may be the reason why the permutation test is more powerful.
A disadvantage of a permutation test is that it can be computer-intensive. However, this issue is less relevant now due to faster algorithms [[18], chap. 13] and the advent of high-speed PCs. Furthermore, one can carry out a permutation test based on a random sample out of the possible permutations, as we did for the DNA methylation data (see Table 1).
In microarray experiments it is common to investigate thousands of genes simultaneously. The approach presented here for the identification of differentially expressed genes is to consider a univariate testing problem for each gene. A correction for the multiplicity of genes is a subsequent step, that is, like the previous step of normalizing the data, outside the scope of this paper. A common approach to the multiplicity problem is to consider a procedure for testing the genes simultaneously for differential expression with the test on an individual gene being implied in the simultaneous test. For such a procedure different proposals have been made recently. For instance, there are methods based on the p-values of the tests from individual genes [21-23]. In a similar manner, the multiplicity of regions can be managed in DNA methylation data.
Conclusion
Aside from the shown improvement in power, the proposed two-part permutation test has three important advantages. First, it avoids the use of any asymptotic distribution and, therefore, can safely be applied in case of small sample sizes that are common in microarray experiments. Second, it reduces without any loss of power to the exact Wilcoxon test if there were no truncated (or zero) values. Thus, it can be used in routine analyses. Third, the permutation test opens the possibility to use other tests to construct the two-part test. Thus, tests with unknown or non-standard null distributions can be used. For instance, one could replace the Wilcoxon test by the Baumgartner-Weiß-Schindler test [24] that was recently recommended for the analysis of gene expression data [19].
Authors' contributions
MN performed the statistical analyses and drafted the manuscript. TB prepared the microarray data. TB and KHJ participated in the design of the simulation study and helped to draft the manuscript. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Number of genes with the given number of truncations per gene (data from the microarray experiment of Tschentscher et al. [11], only genes with at least one truncation)
Table 1 p-values of the original two-part test and the two-part permutation test for seven regions in lung cancer cell lines; DNA methylation data from Siegmund et al. [4]
Region Original two-part test Two-part permutation test1
CALCA p = 0.0005 p = 0.0004
PTGS2 p = 0.0002 p = 0.0001
MTHFR p = 0.0007 p = 0.0002
MGMT-M1 p = 0.0860 p = 0.0857
APC p = 0.1684 p = 0.1712
MYOD1 p = 0.0132 p = 0.0111
ESR1 p = 0.0040 p = 0.0030
1performed based on simple random samples of 20,000 permutations (when using 20,000 permutations a 95%-confidence interval for the p-value is the observed p-value ± 0.007 when p = 0.5, or ± 0.003 when p = 0.05).
Table 2 Frequencies of different size groups of the p-values of the original two-part test and the two-part permutation test for genes with at least one truncation; data from the microarray experiment of Tschentscher et al. [11]
p-value of the original two-part test p-value of the two-part permutation test
≤ 0.001 > 0.001 and ≤ 0.01 > 0.01 and ≤ 0.05 > 0.05 and ≤ 0.1 > 0.1
≤ 0.001 19 0 0 0 0
> 0.001 and ≤ 0.01 39 56 0 0 0
> 0.01 and ≤ 0.05 0 50 164 8 0
> 0.05 and ≤ 0.1 0 0 49 112 17
> 0.1 0 0 0 53 1648
Table 3 Differences between the p-values of the original two-part test and the two-part permutation test for genes with at least one truncation and small p-values (i.e. the p-value of the original test must be as small as mentioned under condition), a positive difference means that the two-part permutation test has a smaller p-value than the original test; data from the microarray experiment of Tschentscher et al. [11]
Condition Number of remaining genes Mean difference (± SD) Median difference Quartiles of the difference
p ≤ 0.1 514 0.0065 (± 0.0126) 0.0047 0.0012, 0.0124
p ≤ 0.05 336 0.0057 (± 0.0064) 0.0040 0.0018, 0.0095
p ≤ 0.01 114 0.0025 (± 0.0018) 0.0019 0.0010, 0.0035
p ≤ 0.001 19 0.0006 (± 0.0003) 0.0006 0.0003, 0.0009
Table 4 Results of the simulation study: Differences between the p-values of the original two-part test and the two-part permutation test for data sets with small p-values (i.e. p-value of the original test ≤ 0.1), a positive difference means that the two-part permutation test has a smaller p-value than the original test; p1 and p2 are the probabilities for zero values and the positive values in group 1 are shifted by μ; 5,000 data sets with n1 = n2 = 10 were generated for each configuration
Configuration Number of data sets Mean difference (± SD) Median difference Quartiles of the difference
p1 = p2 = 0, μ = 2.5 4538 0.0118 (± 0.0145) 0.0057 0.0013, 0.0161
p1 = p2 = 0.3, μ = 2.5 4038 0.0008 (± 0.0069) 0.0020 0.0008, 0.0033
p1 = 0.4, p2 = 0.2, μ = 2.5 4346 0.0010 (± 0.0059) 0.0019 0.0008, 0.0033
p1 = 0.2, p2 = 0.4, μ = 2.5 4270 0.0011 (± 0.0056) 0.0018 0.0006, 0.0030
p1 = 0, p2 = 0.3, μ = 2.5 4883 0.0039 (± 0.0055) 0.0020 0.0005, 0.0051
p1 = 0.3, p2 = 0, μ = 2.5 4893 0.0040 (± 0.0053) 0.0022 0.0007, 0.0049
p1 = 0, p2 = 0.4, μ = 0 3427 -0.0033 (± 0.0119) 0.0006 -0.0092, 0.0029
cutoff value = 0.5, μ = 2.5 4621 0.0059 (± 0.0081) 0.0034 0.0009, 0.0072
cutoff value = 1, μ = 2.5 4860 0.0017 (± 0.0052) 0.0013 0.0003, 0.0032
Table 5 Results of the simulation study: Powers of the original two-part test and the two-part permutation test; 5,000 data sets with n1 = n2 = 10 were generated for each configuration (notation as in Table 4, significance level α = 0.05)
Configuration Original two-part test Two-part permutation test
p1 = p2 = 0, μ = 2.5 0.82 0.93
p1 = p2 = 0.3, μ = 2.5 0.67 0.67
p1 = 0.4, p2 = 0.2, μ = 2.5 0.76 0.75
p1 = 0.2, p2 = 0.4, μ = 2.5 0.74 0.74
p1 = 0, p2 = 0.3, μ = 2.5 0.93 0.96
p1 = 0.3, p2 = 0, μ = 2.5 0.95 0.97
p1 = 0, p2 = 0.4, μ = 0 0.45 0.47
cutoff value = 0.5, μ = 2.5 0.85 0.90
cutoff value = 1, μ = 2.5 0.92 0.92
==== Refs
Tsou JA Hagen JA Carpenter CL Laird-Offringa IA DNA methylation analysis: a powerful new tool for lung cancer diagnosis Oncogene 2002 21 5450 5461 12154407 10.1038/sj.onc.1205605
Model F Adorjan P Olek A Piepenbrock C Feature selection for DNA methylation based cancer classification Bioinformatics 2001 17 S157 S164 11473005
Virmani AK Tsou JA Siegmund KD Shen LYC Long TI Laird PW Gazdar AF Laird-Offringa IA Hierarchical clustering of lungcancer cell lines using DNA methylation markers Cancer Epidemiology, Biomarkers & Prevention 2002 11 291 297 11895880
Siegmund KD Laird PW Laird-Offringa IA A comparison of cluster analysis methods using DNA methylation data Bioinformatics 2004 20 1896 1904 15044245 10.1093/bioinformatics/bth176
Eads CA Danenberg KD Kawakami K Saltz LB Blake C Shibata D Danenberg PV Laird PW MethyLight: a high-throughput assay to measure DNA methylation Nucleic Acids Research 2000 28 E32 10734209 10.1093/nar/28.8.e32
Lachenbruch PA Analysis of data with clumping at zero Biometrische Zeitschrift 1976 18 351 356
Lachenbruch PA Comparison of two-part models with competitors Statistics in Medicine 2001 20 1215 1234 11304737 10.1002/sim.790
Lachenbruch PA Analysis of data with excess zeros Statistical Methods in Medical Research 2002 11 297 302 12197297 10.1191/0962280202sm289ra
Jelinek DF Tschumper RC Stolovitzky GA Iturria SJ Tu Y Lepre J Shah N Kay NE Identification of a global gene expression signature of B-chronic lymphocytic leukemia Molecular Cancer Research 2003 1 346 361 12651908
Küppers R Klein U Schwering I Distler V Bräuninger A Cattoretti G Tu Y Stolovitzky GA Califano A Hansmann M-L Dalla-Favera R Identification of Hodgkin and Reed-Sternberg cell-specific genes by gene expression profiling Journal of Clinical Investigation 2003 111 529 537 12588891 10.1172/JCI200316624
Tschentscher F Hüsing J Hölter T Kruse E Dresen IG Jöckel K-H Anastassiou G Schilling H Bornfeld N Horsthemke B Lohmann DR Zeschnigk M Tumor classification based on gene expression profiling shows that uveal melanomas with and without monosomy 3 represent two distinct entities Cancer Research 2003 63 2578 2584 12750282
Ibrahim JG Chen M-H Gray RJ Bayesian models for gene expression with DNA microarray data J Am Stat Assoc 2002 97 88 99 10.1198/016214502753479257
Tamayo P Slonim D Mesirov J Zhu Q Kitareewan S Dmitrovsky E Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation Proc Nat Acad Sci USA 1999 96 2907 2912 10077610 10.1073/pnas.96.6.2907
Delucchi KL Bostrom A Methods for analysis of skewed data distributions in psychiatric clinical studies: working with many zero values American Journal of Psychiatry 2004 161 1159 1168 15229044 10.1176/appi.ajp.161.7.1159
Gadbury GL Page GP Heo M Mountz JD Allison DB Randomization tests for small samples: an application for genetic expression data Applied Statistics 2003 52 365 376
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
Manly BFJ Randomization, Bootstrap and Monte Carlo Methods in Biology 1997 2 Chapman and Hall, London, U.K
Good PI Permutation Tests 2000 2 Springer, New York, NY
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
Hollander M Wolfe DA Nonparametric statistical methods 1999 2 Wiley, New York, NY
Zaykin DV Zhivotovsky LA Westfall PH Weir BS Truncated product method for combining P-values Genetic Epidemiology 2002 22 170 185 11788962 10.1002/gepi.0042
Dudbridge F Koeleman BPC Rank truncated product of P-values, with application to genomewide association scans Genetic Epidemiology 2003 25 360 366 14639705 10.1002/gepi.10264
Storey JD Tibshirani R Statistical significance for genomewide studies Proceedings of the National Academy of Sciences USA 2003 100 9440 9445 10.1073/pnas.1530509100
Baumgartner W Weiß P Schindler H A nonparametric test for the general two-sample problem Biometrics 1998 54 1129 1135
| 15725357 | PMC551601 | CC BY | 2021-01-04 16:02:52 | no | BMC Bioinformatics. 2005 Feb 22; 6:35 | utf-8 | BMC Bioinformatics | 2,005 | 10.1186/1471-2105-6-35 | oa_comm |
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-141570750110.1186/1471-2148-5-14Research ArticleEvolution of the relaxin-like peptide family Wilkinson Tracey N [email protected] Terence P [email protected] Geoffrey W [email protected] Ross AD [email protected] Howard Florey Institute of Experimental Physiology and Medicine, University of Melbourne, Australia2 Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia2005 12 2 2005 5 14 14 14 10 2004 12 2 2005 Copyright © 2005 Wilkinson et al; licensee BioMed Central Ltd.2005Wilkinson 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 relaxin-like peptide family belongs in the insulin superfamily and consists of 7 peptides of high structural but low sequence similarity; relaxin-1, 2 and 3, and the insulin-like (INSL) peptides, INSL3, INSL4, INSL5 and INSL6. The functions of relaxin-3, INSL4, INSL5, INSL6 remain uncharacterised. The evolution of this family has been contentious; high sequence variability is seen between closely related species, while distantly related species show high similarity; an invertebrate relaxin sequence has been reported, while a relaxin gene has not been found in the avian and ruminant lineages.
Results
Sequence similarity searches of genomic and EST data identified homologs of relaxin-like peptides in mammals, and non-mammalian vertebrates such as fish. Phylogenetic analysis was used to resolve the evolution of the family. Searches were unable to identify an invertebrate relaxin-like peptide. The published relaxin cDNA sequence in the tunicate, Ciona intestinalis was not present in the completed C. intestinalis genome. The newly discovered relaxin-3 is likely to be the ancestral relaxin. Multiple relaxin-3-like sequences are present in fugu fish (Takifugu rubripes) and zebrafish (Danio rerio), but these appear to be specific to the fish lineage. Possible relaxin-1 and INSL5 homologs were also identified in fish and frog species, placing their emergence prior to mammalia, earlier than previously believed. Furthermore, estimates of synonymous and nonsynonymous substitution rates (dN/dS) suggest that the emergence of relaxin-1, INSL4 and INSL6 during mammalia was driven by positive Darwinian selection, hence these peptides are likely to have novel and in the case of relaxin-1, which is still under positive selection in humans and the great apes, possibly still evolving functions. In contrast, relaxin-3 is constrained by strong purifying selection, demonstrating it must have a highly conserved function, supporting its hypothesized important neuropeptide role.
Conclusions
We present a phylogeny describing the evolutionary history of the relaxin-like peptide family and show that positive selection has driven the evolution of the most recent members of the family.
==== Body
Background
The relaxin-like peptide family includes: relaxin-1, relaxin-2, relaxin-3, and the insulin-like (INSL) peptides, INSL3, INSL4, INSL5 and INSL6. All share high structural similarity with insulin due to the presence of six cysteine residues, which confer two inter-chain and one intra-chain disulfide bonds. Thus, it was postulated that relaxin and insulin had derived from a common ancestral gene and were therefore grouped as the insulin superfamily, which later included insulin-like growth factors I and II (IGF-1 and -2) (reviewed in [1]).
Despite less than 50% predicted sequence similarity across members of the insulin superfamily, primary structural determinants are retained, resulting in the similar tertiary structures of relaxin and insulin (reviewed in [1]). The structures of insulin, relaxin [1], relaxin-3 [2] and INSL3 [3] are formed by the cleavage of the pro-hormone peptide into three chains (A, B and C), removal of the C chain and the formation of three disulfide bridges between six invariant cysteine residues found on the A and B chains, to produce an active protein. Based on primary sequence similarity the native structures of human relaxin-1 (H1 relaxin), INSL4, 5 and 6 should be similar, but to date this has not been confirmed.
Various studies have highlighted the importance of a [Arg-X-X-X-Arg-X-X-Ile] motif in the relaxin B chain for interaction with the relaxin receptor and biological activity, which is dependent upon the presence of this motif [4]. Coupled with the insulin-like cysteine bond pattern, the presence of this motif is used to distinguish relaxin sequences. Hence, three relaxin genes are present in humans: relaxin-1 is found only in humans and the great apes, its expression is limited to the decidua, placenta and prostate [5]; relaxin-2 is the major circulating form of relaxin in the human [6] and the functional equivalent to the relaxin-1 in all non-primates; while relaxin-3 was only recently discovered and shows brain specific expression [2]. Throughout this paper relaxin will be used to refer to relaxin-2 in humans and great apes, and the equivalent relaxin-1 in all other mammals (table 1).
Table 1 The relaxin-2 gene in humans and great apes is the equivalent of the relaxin-1 gene in non-primate species and will be referred to as relaxin throughout this paper. Non-primate species only have RLN1 and RLN3 genes.
Humans and Great Apes All other mammals e.g. mouse Peptide abbreviation Gene name
Relaxin-1 Relaxin# RLX1 RLN1
Relaxin-2# RLX2 RLN2
Relaxin-3 Relaxin-3 RLX3 RLN3
#indicates functional equivalence
Relaxin has been well characterized in a reproductive context.. It is a product of the ovary and/or placenta in most species studied, and has various roles in pregnancy and parturition, which differ between species [7]. Recent advances have revealed relaxin to be a multifunctional hormone, with numerous non-reproductive roles (reviewed in [1]). Of the other relaxin-like peptides, INSL3, or relaxin-like factor (RLF), is closely related to relaxin and is critical for testis descent [8,9]. Early placenta insulin-like peptide (EPIL), placentin or insulin-like 4 (INSL4) is primate specific and likely to have diverged from a common relaxin ancestor before the duplication that led to the two relaxin genes seen in humans and great apes [10]. Insulin-like 5 and 6 (INSL5 and INSL6) were identified by searching EST databases with the cysteine motif conserved in all insulin-like peptides [11,12]. Both INSL5 and 6 peptides show higher sequence similarity to relaxin rather than insulin and are functionally uncharacterised.
Until recently, receptors for all peptides of the relaxin family were unknown. Given the high degree of structural similarity between relaxin and insulin, it had been believed that the relaxin receptor would also be a tyrosine kinase receptor, similar to the insulin receptor. However, it was finally demonstrated in 2002, that relaxin activated two previously orphan leucine rich repeat containing heterotrimeric guanine nucleotide binding protein-coupled receptors (GPCR), LGR7 and LGR8 [13]. Further studies have shown that LGR7 is the relaxin receptor [13], although it also specifically interacts with relaxin-3 [14], and that LGR8 is the INSL3 receptor [15,16]. Even more recently, another two GPCRs which are activated by relaxin-3, the somatostatin- and angiotensin- like peptide receptor (SALPR or GPCR135) and GPCR142 have been identified [17,18]. More recently GPCR142 has been shown to be the receptor for INSL5, [19] while GPCR135 appears to be the specific receptor for relaxin-3. Surprisingly, while LGR7, LGR8, GPCR135 and GPCR142 are all Type I GPCRs, they are from different branches within this family and are only very distantly related [17,18,20]. No other relaxin-like peptides have been shown to interact with these receptors [13,14,16-18].
Relaxin and INSL3 have primarily been of interest as hormones of pregnancy and reproduction; therefore it was assumed that the largely uncharacterised INSL4, 5 and 6 would share similar functions. However the discovery of the brain specific relaxin-3, and the widespread expression of INSL5 and its receptor GPCR142 have resulted in a re-evaluation of these assumptions and raised the need for a new approach to investigating these peptides.
Relaxin evolution has confounded researchers for decades. High sequence variability in relaxins across closely related species is a well-known feature of this peptide, however startling similarities have been observed between very distant species such as pigs and whales [21]. Other studies have reported the existence of an invertebrate relaxin, a hormone with "relaxin-like" properties has been described in protozoa (T. pyriformis) [22], ascidians (H. momus) [23] and tunicates (C. intestinalis) [23]. A cDNA and peptide sequence with almost 100% similarity to porcine relaxin was isolated from C. intestinalis [24]. Contrastingly, numerous efforts to identify and sequence a relaxin ortholog in bovine and other ruminants have been unsuccessful. A likely non-functional single copy relaxin-like gene has been found in the ovine [25]. Despite this controversy very few phylogenetic studies of relaxin, or the other relaxin-like peptides, have been reported. An in-depth analysis analysing only relaxin-1, -2 and INSL3 in several primates has been performed [26] and a more recent paper has discussed the evolution of the family without including a detailed phylogenetic analysis [27].
The increasing availability of genomic data has provided an opportunity to clarify the area of relaxin evolution using phylogenetic analysis. The relaxin-like peptide family phylogeny shows relaxin-3 is likely to be the ancestral relaxin, emerging prior to the divergence of fish, presumably with a function in the brain. Multiple relaxin-3 sequences are present in the fish and frogs, while the phylogeny also suggests that a relaxin with reproductive functions is present in these lineages. However, relaxin has been lost in the chicken and its genome instead contains two relaxin-3-like sequences. Evolutionary rate analysis shows positive Darwinian selection to be driving the most recent members of the family, INSL6, INSL4 and relaxin-1.
Results
Sequence similarity searches and multiple alignment
Table 2 outlines the relaxin-like peptide sequences used in these phylogenetic analyses, from human, mouse, rat, dog, chimpanzee, pig, chicken, wallaby, R. esculenta, X. tropicalis, X. laevis, fugu fish, zebrafish, rhesus monkey and rainbow trout. New sequences identified during similarity searches have been highlighted. The source of each newly identified sequence is annotated as a footnote (i.e. genomic, ESTs). It should be noted that more mammalian relaxin and INSL3 sequences have been identified than were included in these analyses. Inclusion of these sequences did not improve the accuracy of the phylogeny, and as the aim of this study was to determine the evolutionary history of the entire family, they were omitted. All available non-mammalian relaxin-like sequences with accompanying nucleotide sequences were included.
Table 2 GenBank accession numbers for relaxin-like sequences. Accession numbers for all sequences included in the phylogeny are listed. New sequences identified in this study are highlighted in bold and accession numbers shown in brackets underneath. Phylogenetic analyses showed TrRLX3f, DrRLX3c, XtRLX3a and XlRLX3 sequences (shown in italics) to be relaxin rather than relaxin-3 homologs and are therefore listed in the relaxin (RLX2) column.
Species RLX1 RLX2 RLX3 INSL3 INSL4 INSL5 INSL6
Human P04808 P04090 Q8WXF3 P51460 Q14641 Q9Y5Q6 Q9Y581
Chimpanzee S42783 P51455 BK005156b BK005155b BK005152b BK005153b BK005154b
Mouse - CAA81611 Q8CHK2 O09107 - Q9WUG6 Q9QY05
Rat - J00780 Q8BFS3 AAD33663 - - Q9WV41
Dog - Q9TRM8 AAEX01023146 AAEX01022997 AAEX01053723
Wallaby AAM22209
R. esculenta CAC16108
Fugufish RLX3f
BK005388b RLX3a-c RLX3d,e
Zebrafish RLX3c
BK005255b RLX3a
BK005228 RLX3b
BK005252a,b
RLX3d (b,c)
Chicken RLX3a
BK005146a RLX3b
BK005533b
Pig P01348 Q8HY17 P51461
X. tropicalis RLX3a
BK005229a RLX3b(b,d)
RLX3c(b,d)
X. laevis RLX3
BK005230a
O. mykiss RLX3
BK005147a
M. mulatta INSL4
BK005251a
aSequences determined from multiple overlapping ESTs.
bSequences determined from genomic data.
cSequence from the Ensembl zebrafish genome assembly [68](GENSCAN00000013529 on supercontig NA8544).
dSequence from the Xenopus genome v2.0 [58].
Searches of the completed human, mouse and rat genomes did not identify any novel relaxin-like peptides, however, likely non-functional INSL5 genes were revealed in the rat and dog genomes. A sequence with high similarity to INSL5 was found (Genbank Accession No: NW_047717.1), with a frameshift mutation in the first exon of the gene, which introduces a stop codon, resulting in a protein truncated early in the B chain (data not shown). The recently completed dog genome also contains a sequence highly homologous to the human INSL5 peptide (Genbank Accession No: NW_ AAEX01024390.1). However this sequence does not encode an open reading frame, with two stop codons present in the C peptide sequence and one upstream of a homologous B chain sequence. Furthermore, there was no recognizable signal peptide or Methionine start codon upstream of this B chain sequence and it was missing the critical Cys-29 in the B chain.
Orthologs of all relaxin-like peptides were identified in the completed chimpanzee genome (table 2). Two relaxin-like sequences were identified in the completed chicken genome (GgRLX3a, b in table 2 and figure 1), both with high similarity to human relaxin-3. Comparing the putative B and A domains of GgRLX3a and b with those of human relaxin-3 shows them to be 81% and 75% similar respectively (data not shown). Three sequences with a high similarity to human relaxin-3 (73%, 92% and 90% respectively; data not shown) were also discovered in the X. tropicalis genome (XtRLX3a-c, in table 2 and figure 1).
Figure 1 Multiple sequence alignment of the relaxin-like peptide family. Amino acid sequences of the B and A domains only were aligned using ClustalW, then edited by hand using Seaview to remove gaps. This alignment was then used for all phylogenetic analyses. Newly identified sequences are highlighted in bold and italics. Invariant cysteine residues are indicated by asterisks (*) and the relaxin specific B-chain motif [RxxxRxxI/V] is shown. Sequences are clustered into subfamilies (A and B) based on primary sequence similarity and phylogenetic analysis. Hsa = Homo sapiens, Pt = Pan troglodytes, Mmul = Maca mulatta, Mm = Mus musculus, Rn = Rattus norvegicus, Cf = Canis familiaris, Ss= Sus scrofa, Re = Rana esculenta, Me = Macropus eugenii, Xl = Xenopus laevis, Xt = Xenopus tropicalis, Dr = Danio rerio, Tr = Takifugu rubripes, Gg = Gallus gallus, Om = Oncorhynchus mykiss.
Sequences with high similarity to relaxin-3 have previously been reported in the fugu fish, TrRLX3a-e [27] and zebrafish, DrRLX3a [27,28] (see table 2 and figure 1). These searches identified a sixth relaxin-like sequence in the fugu fish, TrRLX3f, and another three in the zebrafish, DrRLX3b-d (table 2 and figure 1). Unlike the sequences previously identified in the fugu fish [27], TrRLX3f is more similar to human relaxin-2 (60%) than human relaxin-3 (54%) (Data not shown). Of the zebrafish sequences DrRLX3b and d both show 77% similarity to human relaxin-3 in their B and A domains (data not shown). DrRLX3c is only 60% similar to human relaxin-3 and 54% similar to human relaxin-2 (data not shown).
Searches of partially completed genomes identified relaxin-like sequences in X. laevis (XlRLX3), Oncorhynchus mykiss (rainbow trout) (OmRLX3) and an INSL4 ortholog in the rhesus monkey (table 2 and figure 1). While OmRLX3 shows high similarity to human relaxin-3 (76%, data not shown), XlRLX3 is less similar (69%, data not shown).
The presence of a relaxin gene in ruminants could not be determined due to the incomplete bovine genomic data currently available. No bovine ESTs with a similarity to relaxin were found and the presence of a bovine equivalent to the ovine likely non-functional genomic relaxin sequence could not be confirmed. However, a bovine EST was identified (BI682322) with high similarity to human relaxin-3 (79% identity to the translated EST product) starting from the end of the B chain (45F in the human pro-hormone sequence).
Searches of invertebrate genomic and EST databases failed to identify a relaxin-like gene in any invertebrate or prokaryote. Although a C. intestinalis relaxin-like cDNA sequence has previously been reported [24], our searches failed to confirm this finding. The published relaxin sequence could not be found in the completed C. intestinalis genome.
The B and A chains from all relaxin-like peptides identified in the following species: human, chimpanzee, rhesus monkey, pig, mouse, rat, wallaby, chicken, fugu fish, zebrafish, rainbow trout,R. esculenta, X. laevis and X. tropicalis were aligned using ClustalW and edited to remove all gaps, which are problematic in phylogenetic analysis (figure 1). Only the six cysteine residues responsible for conferring structure are conserved across all the relaxin-like peptides. Striking identity is seen amongst relaxin-3 sequences, especially in the B chain. Much lower similarity is seen amongst relaxin sequences, apart from the cysteine motif, only the relaxin-specific B chain motif is conserved. C peptide sequences show only negligible similarity between even closely related species making them impossible to align accurately. The C peptide is cleaved from the mature form of relaxin-1, 3 and INSL3, is believed to be cleaved from the mature form of all other relaxin-like peptides and was therefore excluded from all sequences.
Phylogeny of the relaxin-like peptide family
The alignment of B and A domains described above (figure 1) was used to construct phylogenetic trees with the maximum parsimony (MP), neighbour-joining (NJ) and maximum likelihood (ML) methods. MP and NJ methods produced conflicting trees, both with low bootstrap support for most of the major branches, and the ML tree failed to resolve many relationships (data not shown). Based on the relationships that could be determined with a degree of confidence, the sequences were divided into two clusters to be analysed separately (as shown in figure 1). Cluster A contained relaxin-3 and INSL5 sequences, while cluster B contained relaxin-1, relaxin-2, INSL3, INSL4 and INSL6 sequences. Several fish and frog sequences with lower sequence similarity to relaxin-3 (DrRLX3c, TrRLX3f, XlRLX3 and XtRLX3, figure 1) were grouped with cluster B based on preliminary phylogenetic analysis (data not shown). These sequences have, therefore, been listed as relaxin homologs rather than relaxin 3 (table 2). Also placed in cluster B are the sequences previously isolated from the tammar wallaby, MeRLX, [29] and edible frog, ReRLX [30]. Despite sequence similarity to relaxin-3, previous functional characterization and expression profiles of MeRLX and ReRLX indicates they are relaxin, rather than relaxin-3 homologs.
None of the tree construction methods employed was able to completely resolve the phylogeny of either cluster. Bootstrap values in the MP and NJ generated trees were very low (below 50%, data not shown) and Tree-Puzzle failed to resolve the position of several sequences. This is primarily because of the short sequences used, as the C peptide can not be used to increase sequence length and thus improve the output from the tree generation methods, an inferred tree was produced instead. The Tree-Puzzle tree was resolved using topologies conserved between the MP and NJ trees and then reconciled against a species tree using GeneTree. The inferred gene trees were then edited to minimize the incongruence (the number of losses and inferred duplications) with the species tree. The inferred cluster A Additional file 1 and the cluster B trees Additional file 2 were combined to produce the phylogenetic tree of the complete relaxin-like peptide family (figure 2). Branch confidence levels are indicated on figure 2; branches without notation are inferred only. This gene tree was then reconciled with the species tree (figure 3).
Figure 2 Evolutionary relationships among relaxin-like peptides. Topology shown is a consensus tree based on MP (maximum parsimony), ML (maximum likelihood) and NJ (neighbour-joining) analysis of the amino acid alignment shown in figure 1. Consensus tree was produced and edited using TreeView to correlate topology with known genomic information about the family. Human insulin used as an outgroup. Where possible, confidence values are shown at branches: * >50%, ** >75%, all other branches are inferred. Hsa = Homo sapiens, Pt = Pan troglodytes, Mmul = Maca mulatta, Mm = Mus musculus, Rn = Rattus norvegicus, Cf = Canis familiaris, Ss = Sus scrofa, Re = Rana esculenta, Me = Macropus eugenii, Xl = Xenopus laevis, Xt = Xenopus tropicalis, Dr = Danio rerio, Tr = Takifugu rubripes, Gg = Gallus gallus, Om = Oncorhynchus mykiss.
Figure 3 Reconciled tree for the relaxin-like peptide family. The consensus tree of relaxin-like peptides (figure 2) from human, chimpanzee, mouse, dog, rat, pig, wallaby, chicken, fugu fish, zebrafish, rainbow trout, R. esculenta, X. laevis and X. tropicalis was reconciled using GeneTree with a species tree complied from a phylogeny of model organisms [65]. Squares indicate duplication events, red dotted lines indicate absent genes, either lost from those species (in grey), or not yet sequenced. While used to construct the gene tree as an outgroup, insulin has been removed from the reconciled tree. Hsa = Homo sapiens, Pt = Pan troglodytes, Mmul = Maca mulatta, Mm = Mus musculus, Rn = Rattus norvegicus, Cf = Canis familiaris, Ss = Sus scrofa, Re = Rana esculenta, Me = Macropus eugenii, Xl = Xenopus laevis, Xt = Xenopus tropicalis, Dr = Danio rerio, Tr = Takifugu rubripes, Gg = Gallus gallus, Om = Oncorhynchus mykiss.
Analysis of the reconciled tree shows a major duplication event occurred early in the vertebrate lineage, giving rise to two subfamilies (clusters A and B respectively). Another duplication in subfamily A, prior to the emergence of fish, resulted in two lineages, which evolved into relaxin-3 and INSL5 in mammals. Interestingly, several non-mammalian relaxin-3-like sequences grouped with INSL5, implicating them as possible INSL5 homologs (GgRLX3b, OmRLX3, DrRLX3b, DrRLX3d, TrRLX3d and TrRLX3e). The reconciled tree also shows two additional fish-specific duplications in subfamily A. In the fugu fish genome a third duplication has occurred, resulting in three putative relaxin-3 (TrRLX3a, b, c) and two INSL5 homologs (TrRLX3d, e). In subfamily B there were four duplications, all were after the divergence of birds and reptiles and likely to have occurred during mammalian evolution. These events resulted in INSL3, INSL6, relaxin-1, relaxin-2 and INSL4.
Synonymous (dS) and Nonsynonymous (dN) substitution rate estimates
Results show the relaxin-like peptides are under varying selection pressures (table 3). Pairwise comparisons of human and chimpanzee orthologs provide the only way to compare all members of the family between two species. RLN1, RLN2 and INSL6 have high dN/dS rate estimates, with results for RLN1 and INSL6 suggesting positive Darwinian selection. The extremely high estimate for INSL6 (99) is caused by having a dS of 0 (i.e. no synonymous substitutions), resulting in a division by 0 for the rate estimate, which is represented as 99 rather than infinity. All other human and chimpanzee sequences compared were identical and thus produced dN/dS estimates of 0.
Table 3 Synonymous (dS) and nonsynonymous (dN) substitution rate estimates for all relaxin-like genes. Substitution rates were estimated using the Yang and Neilsen, 2000 method as implemented in yn00 in the PAML suite. Estimations were made using pairwise alignments of the nucleotide sequences of the B and A domains of human and chimpanzee or human and mouse genes. A dN/dS of 99 represents infinity and indicates that all substitutions detected are nonsynonymous, while na indicates that a dN/dS measurement is not possible as the sequences being compared are identical, or have only synonymous substitutions.
human- chimpanzee human-mouse
dN dS dN/dS dN dS dN/dS
RLN1 0.055 0.039 1.4 _a
RLN2 0.017 0.023 0.7 0.56 1.06 0.5
RLN3 0 0 na 0.04 1.83 0.02
INSL3 0 0 na 0.17 1.31 0.1
INSL4 0 0.027 na _a
INSL5 0 0.023 na 0.16 1.24 0.1
INSL6 0.006 0 99 0.33 0.65 0.5
INS 0 0.13 na 0.033 1.44 0.02
a These genes are not present in the mouse.
Human and mouse orthologs were used to estimate rates for the other members of the family. RLN2 and INSL6 show the highest estimates again, although are much lower than comparisons with chimpanzee sequences, the INSL6 estimate suggesting weak purifying selection instead of positive selection. The very low substitution rate observed for RLN3 (0.02) shows this peptide to be under strong purifying selection, at a similar rate to insulin (INS) (table 3). Rates vary among the other members of the family from ~0.1 for INSL3 and INSL5 to ~0.5 for RLN2 and INSL6. As INSL4 is not present in mice and the human and chimpanzee sequences were identical, the INSL4 sequence from the rhesus monkey was used instead (data not shown). This comparison yielded a dN/dS estimate of 0.5, indicating weak purifying selection.
Substitution rate estimates for the individual B and A domains were determined in a similar fashion, using human and chimpanzee comparisons for RLN1, RLN2 and INSL6; human, rhesus monkey comparisons were used for INSL4 and human, mouse comparisons were used for RLN3, INSL3 and INSL5 (figure 4). These comparisons show the B domains of relaxin-1 and INSL6 to be under positive selection (both estimates were 99). The B domains of INSL4 and relaxin-2 also have high substitution rates (1.0 and 0.7 respectively), but are not high enough to suggest positive rather than neutral or weak purifying selection. All A domains are under the effects of fairly strong purifying selection, except for that of relaxin-1, which is under only very weak selection pressures (0.8). Interestingly, while the B domains of relaxin-2, INSL6 and INSL4 all have very high dN/dS estimates, the A domains of these genes have very low estimates. This is in contrast with the other members of the family, relaxin-3, INSL3 and INSL5, which all have higher dN/dS estimates in the A domain than the B domain.
Figure 4 Synonymous (dS) and nonsynonymous (dN) substitution rate estimates for individual B and A domains of each relaxin-like gene. Substitution rates (dN/dS) were estimated using the Yang and Neilsen, 2000 method as implemented in yn00 in the PAML suite. Human and chimpanzee comparisons were used for RLN1 and INSL6; human and rhesus monkey comparisons were used for INSL4 and human, mouse comparisons were used for RLN2, RLN3, INSL3 and INSL5.
Positive selection tests
To confirm the pairwise dN/dS (or ω) estimates, more sophisticated codon-based substitution models (reviewed in [31]) were used. As pairwise comparisons have already shown positive selection to be acting on RLN1 and INSL6, INSL4 was analysed further. The phylogenetic tree of all sequences used is shown in figure 5. Branch-specific likelihood analysis of the data, which assumes a constant ω ratio across all sites in a sequence, was used to test whether the INSL4 branch (branch A, figure 5) has a different ω ratio than all other branches. While the two-ratios model indicates a ω ratio of 1.1 for branch A (table 4), the LRT comparing this with the one-ratio model shows this to be statistically insignificant (P = 0.6, d.f. = 1, table 5).
Figure 5 Phylogeny of mammalian RLN2 and INSL4 genes. Tree generated using Tree-Puzzle using a gamma distribution, the Dayhoff model of substitution and 10 000 puzzling steps. Confidence values are shown as percentages on each branch. The INSL4 branch (labeled A on the tree) was tested for positive selection. Hsa = Homo sapiens, Mm = Mus musculus, Rn = Rattus norvegicus, Ss = Sus scrofa, Me = Macropus eugenii, Mmul = Maca mulatta, Cd = Camelus dromedaries, Gc = Galago crassicaudatus, Pt = Pan troglodytes, Ggo = Gorilla gorilla, Ph = Papio hamadryas, Fc = Felis catus, Cf = Canis familiaris, Oc = Oryctolagus cuniculus, Ma = Mesocricetus auratus, Ec = Equus caballus.
Table 4 Parameter estimates for INSL4 under different branch models, site models and branch-site models. Models implemented in Codeml from the PAML suite. Parameters in boldface indicate positive selection. Sites potentially under positive selection are numbered using the human INSL4 sequence in figure 1 as the reference.
Model ρ ℓ Parameter est. Positively selected sites
1 ratio (R0) 1 -1460.1 ω = 0.7
Branch specific
2 ratios (R2) 2 -1459.9 ω0 = 0.7(background),
ω1 = 1.1 (branch A)
Site specific
Neutral (M1) 1 -1424.4 ρ0 = 0.1, ρ1 = 0.8 not allowed
Selection (M2) 3 -1416.6 ρ0 = 0.1, ρ1 = 0.6,
ρ2 = 0.2, ω2 = 2.9 15L (P >0.99)
14H, 27H, 28R, 36V (P >0.95)
Discrete (M3) (k = 2) 3 -1421.4 ρ0 = 0.2, ω0 = 0.03,
ρ1 = 0.8, ω1 = 1.0 37 sitesa (P >0.99)
Discrete (M3) (k = 3) 5 -1413.8 ρ0 = 0.1, ω0 = 0.01,
ρ1 = 0.5, ω1 = 0.6,
ρ2 = 0.3, ω2 = 2.0 15L, 27H, 28R (P >0.99)
14H, 25G, 26R, 30D, 36V
(P > 0.90)
Beta (M7) 2 -1419.6 ρ0 = 0.2,q = 0.2 not allowed
Beta&ω (M8) 4 -1414.6 ρ0 = 0.8,p = 0.3,
q = 0.06,
ρ1 = 0.2, ω1 = 2.6 15L, 28R (P > 0.95)
14H, 27H, 36V (P > 0.90)
Branch-Site
Model A 3 -1418.4 ρ0 = 0.1, ρ1 = 0.6,
ρ2 = 0.2, ω2 = 3.0 In the foreground lineage:
13K, 37I (P > 0.95)
Model B 5 -1416.4 ρ0 = 0.1, ω0 = 0.01,
ρ1 = 0.6, ω1 = 0.6,
ρ2 = 0.3, ω2 = 3.2 In the foreground lineage:
13K, 37I (P > 0.95)
In the background lineages:
no significant sites
a1A, 2A, 3E, 5R, 6G, 10R, 11F, 12G, 14H, 15L, 16L, 17S, 18Y, 20P, 25G, 26R, 27H, 28R, 29F, 30D, 31P, 32F, 35E, 36V, 37I, 39D, 40D, 41G, 42T, 43S, 44V, 45K, 47L. Note that these sites should be treated with caution as ω under this model is not significantly higher than 1.
Table 5 Likelihood ratio test statistics (2δ) for the INSL4 data set.
2δ d.f. P-value
LRT of ω at branch A
1 ratio (R0) vs. 2 ratios (R2) 0.3 1 0.6
LRTs of variable ω's among sites
M0 vs. M3 (k = 3) 92.5 2 <0.0001
M1 vs. M2 15.7 2 0.0004
M7 vs. M8 10.1 2 0.006
LRT's of variable ω's along branch A
M1 vs. Model A 12.0 2 0.002
M3 (k = 2) vs. Model B 10.0 2 0.007
Site-specific models, which allow the ω ratios to vary between sites in a sequence, were also applied to the data. The ω ratio was found to vary considerably among amino acid sites. The discrete model with K = 3 site classes was the best fit to the data with a log likelihood value of -1413.8 (table 4). This model suggests that 31% of sites are under positive selection (ω2 = 2.00), while half (54%) are under weak purifying selection (ω1 = 0.6) and the other 15% constrained under extreme purifying selection (ω0 = 0.01) (listed in table 4). Eight amino acids are identified as under positive selection at the 90% cut off (14H, 15L, 25G, 26R, 27H, 28R, 30D, 36V, see table 4). All but two of these are within the A chain of INSL4. The LRT of M3 with its null model (M0) shows these results to be significant (P < 0.0001, d.f. = 2, table 5). Similar results are seen with Model M8.
Lastly, branch-site models A and B were applied to the data. These models extend the site and branch specific models by allowing the ω ratios to vary among lineages and sites and were used to test for specific sites under positive selection on the INSL4 branch. Model A, which fits the data significantly better than its null model M0 (P = 0.02, d.f. = 2, table 5) identifies 2 sites (13K and 37I) under positive selection in the INSL4 branch at the 95% cut off (table 4). Model B, which allows the ω ratio to vary both in the foreground lineage (the INSL4 branch) and in the background branches also fits the data significantly better than its null model, M3 with k = 2 (P = 0.007, d.f. = 2, table 5), and identifies the same two positively selected sites as Model A (13K and 37I) in the INSL4 branch (table 4). Model B also confirms the results produced by the discrete model (M3 with k = 3 site classes), showing 27% of sites of under positive selection, 58% under weak purifying selection (ω2 = 0.58) and 14% under very strong purifying selection (table 4).
Discussion
While relaxin evolution has been the centre of much controversy (relaxin is often cited as a gene that conflicts with the Darwinian theory of evolution [24,32-34]), this report is the first attempt to describe the evolutionary history of the whole relaxin-like peptide family from a phylogenetic perspective. Previous studies have concentrated on the primate relaxins and relaxin-like factors [26], or not included detailed phylogenetic analyses [27]. We have sought to expand upon these by incorporating sequences identified in all the available completed genomes with a subset of cloned relaxin-like sequences, particularly those from non-mammalian species.
None of the phylogenetic tree construction programs used was able to completely resolve the evolution of the relaxin-like peptide family. This is likely due to variable divergence across the family and the short sequence length [35]. Incorporating results from the MP and NJ methods suggested positions for several branches that were unresolved after ML analysis. Minimizing incongruence between the gene and species trees by reducing the number of assumed duplications in the reconciled tree also provided a method to infer the evolutionary history of this family.
Similarly to previously published results, searches of available genomic and EST data failed to identify any novel members of the relaxin-like peptide family [28]. Given the stringent and well-described insulin family signature that revolves around the invariant cysteine residues that confer the insulin-like structure seen across the superfamily, we find it improbable that any novel relaxin or insulin-like sequences will be identified.
The presence of an invertebrate relaxin has been of speculation since 1983 when relaxin-like activity was first detected in the protozoa, T. pyriformis [22]. Similar activity was reported in H. momus [23] and in C. intestinalis, where a cDNA sequence almost identical to pig relaxin was found [24]. However, our searches of all completed invertebrate genomes (including C. intestinalis) failed to identify any relaxin-like sequences, including the published sequence. Multiple insulin-like peptides have been found in several invertebrates, including: Bombyxi mori (silkworm) [36], D. melanogaster [37] and C. elegans [38]. As these sequences lack the relaxin-specific motif, and show no homology to other relaxin family peptides, they are not considered part of the relaxin subfamily and therefore have not been included in these analyses. Much of the controversy surrounding relaxin evolution concerns the identification of an invertebrate relaxin sequence (a cDNA sequence from Ciona intestinalis) almost identical to pig relaxin (Georges and Schwabe, 1999). Completion of the C. intestinalis and other invertebrate genomes has allowed us to conclude that there is not a relaxin-like sequence in any invertebrate sequenced to date. If an invertebrate relaxin does exist, it does not contain the relaxin-specific motif characterized in vertebrates.
A hallmark of relaxin sequences is their high variability, even amongst closely related species. Relaxin-like peptide sequences isolated from two whales are almost identical to porcine relaxin [21], however as these sequences were derived from amino acid sequencing, without nucleotide or and genomic sequence available, they have not been able to be included in these phylogenetic analyses.
The presence of a functional relaxin in the ruminant lineage has yet to be confirmed [25]. More genomic data is required to confirm the presence of a non-functional relaxin gene sequence in the bovine, similar to that observed in the ovine [25]. Searches of the preliminary bovine genome assembly have failed to find a relaxin gene. Interestingly, a relaxin sequence has been identified in the camel [39] and relaxin expression found in the closely related llama and alpaca [40]. While classified as a ruminant, Camelidae have a unique reproductive anatomy and physiology [41]. A bovine EST (BI682322) with high similarity to exon 2 of human relaxin-3 was identified. Confirmation of the presence of relaxin and relaxin-3 orthologs in ruminants awaits further sequencing of the bovine and ovine genomes.
The presence of an avian relaxin has also been of speculation. While relaxin-like activity has been reported in the chicken [42], an avian relaxin-like peptide or gene has not been identified. While two relaxin-3-like genes were identified in the nearly completed chicken genome, no avian relaxin gene was found. As no other relaxin-like genes were found, the reported relaxin activity may be due to one of the relaxin-3-like genes.
The phylogeny of the relaxin-like peptide family indicates relaxin-3 is the ancestral relaxin, appearing prior to the divergence of teleosts. The finding of multiple relaxin-3-like genes in the fugu fish and zebrafish suggests multiple lineage-specific duplications of a single relaxin-3-like ancestor have occurred in fish [27]. However, the possibility the other mammal specific relaxin-like peptides emerged earlier before being lost in the teleost can not be excluded [27]. We find it more likely that these duplications, and the resulting multiple relaxin-3-like genes, are fish specific and due to genome wide duplications hypothesized to have occurred during fish evolution [43]. Phylogenetic analyses show multiple fish homologs of both the mammalian relaxin-3 and INSL5 genes, meaning that two relaxin-3-like genes existed prior to the genome duplication event proposed to have occurred in the teleost ancestor. The putative fish relaxin homolog was either, present in the teleost ancestor, duplicated and the second copy lost or emerged shortly after or, as a result of, the genome-wide duplication event.
While termed relaxin-3-like based on sequence similarity, phylogenetic analysis indicates that several non-mammalian sequences (OmRLX3, DrRLX3b, DrRLX3d, TrRLX3d, TrRLX3e and GgRLX3b) could be INSL5 homologs. None of the sequences found in the complete X. tropicalis genome were placed in this group, while there are members present in the more ancient fish lineage and the younger avian lineage. It is possible that this gene has either been lost, or remains unidentified, in the X. tropicalis genome. A sequence with similarity only to the B chain of relaxin-3 was also found, but a corresponding A chain match was not, however, there is a gap in the genome assembly upstream which might contain the missing domain. Future assemblies of the Xenopus genome should resolve this issue. These results suggest that INSL5 could have emerged during teleost evolution, far earlier than previously believed. Unlike the mammal-specific relaxin-like genes, which are clustered together (on chromosome 9 in the human and chromosome 19 in the mouse), INSL5 is localized independently (chromosome 1 in the human and chromosome 4 in the mouse). These findings are of particular interest in the analysis of INSL5, which is still functionally uncharacterised.
All the potential non-mammalian INSL5 homologs retain the relaxin-specific B chain [RxxxRxxI/V] motif, hence would be capable of interacting with the relaxin receptor, LGR7, and thus functionally classified as a relaxin. Recent studies have shown INSL5 is a high affinity ligand for GPCR142 but not GPCR135, LGR7 or LGR8 [19]. As the residues required for interaction with GPCR135 and GPCR142 are not known, it is unknown whether the non-mammalian INSL5 homologs would interact with GPCR142, GPCR135 and/or LGR7.
Phylogenetic results from this study suggest the presence of a relaxin homolog in fish and frogs, although not in the chicken. Relaxin sequences have previously been isolated and peptide sequenced from either the ovaries or testes of the edible frog [30], little skate (Raja erinacea) [44], spiny dogfish (Squalus acanthias) [45], Atlantic stingray (Dasyatis sabina) [46] and the sand tiger shark (Odontaspis taurus) [47]. While having high similarity with relaxin-3, these sequences are not relaxin-3 homologs (as the B chain of the stingray sequence is lacking the relaxin-specific motif, it is not a functional relaxin [46] and has not been considered further). Based on the expression of all these genes in reproductive organs such as the testes and ovaries, and the failure to find the R. esculenta gene expressed in the brain using northern blot analysis [30], we believe these to be among the first relaxin peptides with a reproductive function. Based on the similarity with relaxin-3 observed in these sequences, the ancestral relaxin homolog, and its new reproductive function, is likely to have emerged prior to the divergence of teleosts. A complete picture of relaxin-like peptides present in non-mammalian genomes will be invaluable in understanding the evolution of relaxin from neuropeptide to reproductive hormone.
The ancestral RLN3 gene is under very strong purifying selection, highlighting the importance of its highly conserved function, likely to be in the brain [2]. As high divergence is a hallmark of relaxin sequences, it is somewhat unsurprising that RLN2 is under only weak purifying selection. We suggest that this lack of selective pressure has contributed to the high sequence divergence seen between many relaxins (e.g. human and mouse) and the differences in relaxin's functions observed across mammals.
Information about the selective constraints placed upon these peptides, can provide valuable insight into the nature of interactions with their receptors. Based on selection pressures we can conclude that the interactions between relaxin-3 and GPCR135, INSL5 and GPCR142 are very specific, while the binding of relaxin to LGR7 is much looser. In this context the cross-reactivity seen between LGR7 and INSL3 or H1 relaxin, which are both similar to relaxin in sequence but especially in structure, is understandable, as is the lack of binding between GPCR135 and GPCR142 with any other relaxin-like peptide. Unexpectedly, synonymous and nonsynonymous substitution rate estimates for RLN1 and INSL6 show these to be under positive selection. Positive selection is often difficult to observe using pairwise comparisons that average over the whole length of a sequence, making these results even more striking. While pairwise comparisons failed to confirm positive selection was acting on INSL4, further statistical tests suggested that positive Darwinian selection acted on several sites in the INSL4 sequence after its emergence. Further analysis will be required to confirm these sites as important in the acquisition of a new receptor and a new function by INSL4, particularly in light of recent studies that question the reliability of ML methods to accurately detect positive selection acting on single sites [48-50]. We are encouraged that both branch-specific and site-specific ML models find positive selection to be acting on INSL4.
When the B and A domains of each gene were analyzed separately, further differences in selection pressures became apparent. The interaction between relaxin and its receptor has been thought to be primarily mediated through the B chain of the peptide [4], so the finding that selection pressures are stronger on the A chain of relaxin-1, INSL4 and INSL6 was unexpected. We also find it noteworthy that INSL4, INSL6 and relaxin-1 are the most recent members of the family to emerge and all appear to be under the effects of positive Darwinian selection. INSL6 emerged during mammalian development, INSL4 and RLN1 during primate evolution, they remain functionally uncharacterized and INSL4 and INSL6 are without known receptors. The low selection pressure on the B domain and the strong constraints placed on the A domain of INSL4 and INSL6 suggests that, unlike the B chain mediated interaction of relaxin and INSL3 with their receptors, the interaction of these peptides with their receptors could be dependant on the A chain instead. The low dN/dS rate observed for INSL5 indicates this peptide to be evolutionary stable and of functional importance. In particular the constraints placed on both A and B chains of INSL5 suggest a well-defined receptor interaction system, while the total absence of these constraints on either chain within relaxin-1 suggests the opposite, that perhaps this peptide is still evolving its function.
Conclusions
We present here a phylogeny for the relaxin-like peptide family. Relaxin has long been used as an example of a gene that conflicts with the Darwinian theory of evolution [24,32-34]. However, we have shown that these can issues can be resolved when studied in the context of the rest of the relaxin-like peptide family, in particular the new, but likely ancestral relaxin, relaxin-3.
We have demonstrated that positive selection has been a driving force in the recent expansion of the relaxin-like peptide family during mammalian evolution. While strong purifying selection has maintained the structural core of these peptides by constraining the insulin superfamily cysteine motif, outside these residues, positive selection has acted after at least three gene duplication events (which generated INSL6, INSL4 and relaxin-1) to allow these new genes to acquire a new receptor and novel functions. Given the known roles of relaxin and INSL3 in reproduction (and the likely similar roles of INSL4 and INSL6 given the specificity of their expression in reproductive tissues) these findings correlate with a general trend towards rapid evolution in several reproduction associated genes [51-54]. We anticipate that further analysis of the coevolution of the relaxin-like peptides with their receptors will contribute much towards our understanding of the pleiotropic actions of this family as well as mechanisms involved in the evolution of peptide hormone systems.
Methods
Sequences and sequence similarity searches
Amino acid and nucleotide sequences of cloned relaxin-like peptide family members from the following species were obtained from GenBank [55]: human (Homo sapiens) H1 relaxin, H2 relaxin, H3 relaxin, INSL3, INSL4, INSL5, INSL6; mouse (Mus musculus) relaxin, relaxin-3, INSL3, INSL5, INSL6; rat (Rattus norvegicus) relaxin, relaxin-3, INSL3, INSL6; dog (Canis familiaris) relaxin; pig (Sus scrofa) relaxin, relaxin-3, INSL3; edible frog (Rana esculenta) relaxin and tammar wallaby (Macropus eugenii) relaxin (see table 2 for accession numbers). Five published relaxin-like sequences previously identified in the fugu fish (TrRLX3a-e) [27] and the zebrafish (DrRLX3a) [28] were also used. There are several partial relaxin-like peptide sequences available, however only sequences with corresponding nucleotide sequence data were utilized in this study.
Sequence similarity searches using TBLASTN [56] were conducted using the B and A chain sequences of each family member to identify other mammalian, vertebrate and invertebrate relaxin-like peptides. The following databases were searched: human, mouse, rat, dog, chimpanzee (Pan troglodytes), fugu fish, zebrafish, fruit fly (Drosophila melanogaster), mosquito (Anopheles gambiae), Caenorhabditis elegans, all yeast, all plant and all bacterial genomes at NCBI [57], X. tropicalis [58] and C. intestinalis [59], Expressed Sequence Tags (EST), Genome Survey Sequences (GSS), and High-Throughput Genomic Sequences (HTGS) databases [55]. While the classical cysteine motif of the insulin superfamily was used to distinguish sequences as members of this family, relaxin homologs were distinguished by the additional presence of the specific relaxin motif [RXXXRXXI/V] in the B chain of the derived peptide sequence.
Multiple sequence alignment and phylogenetic analysis
Amino acid sequences were aligned using ClustalW [60] with default parameters. The alignments were edited to delete the C and signal peptide sequences, leaving only the B and A domains, which were further edited to minimize gaps and then concatenated. Human insulin was included as an outgroup.
Phylogenetic trees were constructed using maximum parsimony (MP): implemented in PHYLIP [61] using ProtPars, Neighbour-joining (NJ): implemented in PHYLIP using ProtDist and Neighbour and maximum likelihood (ML): implemented in Tree-Puzzle [62]. Data analyzed in PHYLIP was bootstrapped 1000 times using SeqBoot and consensus trees derived using Consense. Tree-Puzzle was run with a two-rate model of heterogeneity, the Dayhoff model of substitution and 50 000 puzzling steps. Trees were edited using TreeView [63].
Reconciliation of gene and species trees
Gene trees of relaxin-like peptides were reconciled with a species tree using GeneTree [64]. Reconciled trees are an attempt to resolve incongruence between gene and species trees by predicting gene duplications and losses [64]. The species tree was based on a phylogeny of model organisms [65]. The reconciled tree was edited to minimize incongruence, primarily by reducing inferred duplications.
Estimation of synonymous and nonsynonymous substitution rates
Pairwise nucleotide sequence alignments of human and chimpanzee, human and rhesus monkey (Maca mulatta) and human and mouse orthologs were constructed using ClustalW [60] and edited to limit alignments to the B and A domains only, which were then concatenated. Synonymous (dS) and nonsynonymous (dN) substitution rates were estimated using the methods of Yang and Nielsen [66] as implemented in yn00 in the PAML suite [67].
Testing for positive selection
The following relaxin-1 and INSL4 nucleotide sequences were aligned using ClustalW: human H2 relaxin (X00948), INSL4 (L34838), chimpanzee(Pan troglodytes) relaxin-2 (Z27245), INSL4 (BK005152); gorilla (Gorilla gorilla) relaxin-2 (Z27228, Z27237); rhesus monkey relaxin (A34936), INSL4 (BK005251); bush baby (Galago crassicaudatus) relaxin (AF317625); baboon (Papio hamadryas) relaxin (Z27246, Z27224); camel (Camelus dromedarius) relaxin (AF254739); cat (Felis catus) relaxin (AF233688); dog (Canis familiaris) relaxin (AF233687); guinea pig (Cavia porcellus) relaxin (S85964); rabbit (Oryctolagus cuniculus) relaxin (S45940); hamster (Mesocricetus auratus) relaxin (S79879) and horse (Equus caballus) relaxin (AB000201). The alignment was edited as described previously. A ML tree was constructed from this alignment using TreePuzzle [62] and the method of Yang and co-workers [31] was used to test for positive selection in the INSL4 branch. Using Codeml from the PAML suite, several models were fitted to the data. The branch specific models, One-ratio (R1) and Two-ratios (R2) were used to detect lineage-specific changes in selective pressure. The site specific models, Neutral (M1), Selection (M2), Discrete (M3) with 2 and 3 site classes, Beta (M7) and Beta&ω (M8), were also used to test for individual residues under positive selection. The branch-site models A and B were used to detect positive selection in a subset of sites in a specified branch. Likelihood ratio tests (LRT) were used to assess their goodness of fit, by comparing a model that does allow for dN/dS > 1 against a model that does not (i.e. a null model). Therefore, the branch specific LRT was R2 vs. R1. The site specific LRTs were M3, M2 and M8 against their respective null models, M0, M1 and M7. The branch-site models A and B were tested against M1 and M3 with k = 2 site classes respectively. Positively selected sites with a posterior probability of P(ω>1) >0.90 were listed.
Authors' contributions
TW performed all sequence and phylogenetic analysis and drafted the manuscript, TS participated in phylogenetic analysis, design and coordination of study, GWT participated in the design of the study, and RADB participated in phylogenetic analysis, conceived of the study and participated in its design and coordination.
Supplementary Material
Additional File 1
Phylogeny of cluster A- relaxin-3 and INSL5. Phylogeny of Cluster A constructed from a ClustalW alignment of the B and A domain amino acid sequences from relaxin 3 and INSL5 peptides. Consensus tree generated from MP (Protpars in PHYLIP), ML (TreePuzzle) and NJ (Neighbour in PHYLIP) methods and edited in Treeview to minimize species tree incongruence. Human insulin was used as an outgroup. Where possible, confidence values are shown at branches: * >50%, ** >75%, all other branches are inferred. Hsa = Homo sapiens, Pt = Pan troglodytes, Mm = Mus musculus, Rn = Rattus norvegicus, Cf = Canis familiaris, Ss = Sus scrofa, Xt = Xenopus tropicalis, Dr = Danio rerio, Tr = Takifugu rubripes, Gg = Gallus gallus, Om = Oncorhynchus mykiss.
Click here for file
Additional File 2
Phylogeny of cluster B- relaxin-1, 2, INSL3, INSL4 and INSL6. Consensus phylogeny of Cluster B constructed from a ClustalW alignment of the B and A domain amino acid sequences from relaxin 1, 2, INSL3, INSL4, INSL6 peptides. Consensus tree generated from MP (Protpars in PHYLIP), ML (TreePuzzle) and NJ (Neighbour in PHYLIP) methods and edited in Treeview to minimize species tree incongruence. Human insulin was used as an outgroup. Where possible, confidence values are shown at branches: * >50%, ** >75%, all other branches are inferred. Hsa = Homo sapiens, Pt = Pan troglodytes, Mmul = Maca mulatta, Mm = Mus musculus, Rn = Rattus norvegicus, Cf = Canis familiaris, Ss = Sus scrofa, Re = Rana esculenta, Me = Macropus eugenii, Xl = Xenopus laevis, Xt = Xenopus tropicalis, Dr = Danio rerio, Tr = Takifugu rubripes.
Click here for file
Acknowledgements
The authors wish to thank Toby Sargeant (WEHI) for his assistance with phylogenetic analysis and the recommendations of two anonymous reviewers. This work was supported by an Institute Transitional Block Grant from the National Health and Medical Research Council (NHMRC) of Australia (983001). RADB is a recipient of a NHMRC RD Wright Fellowship.
==== Refs
Sherwood OD Relaxin's physiological roles and other diverse actions Endocrine reviews 2004 25 205 234 15082520 10.1210/er.2003-0013
Bathgate RA Samuel CS Burazin TC Layfield S Claasz AA Reytomas IG Dawson NF Zhao C Bond C Summers RJ Parry LJ Wade JD Tregear GW Human relaxin gene 3 (H3) and the equivalent mouse relaxin (M3) gene. Novel members of the relaxin peptide family. Journal of Biological Chemistry 2002 277 1148 1157 11689565 10.1074/jbc.M107882200
Bullesbach EE Schwabe C The primary structure and the disulfide links of the bovine relaxin-like factor (RLF). Biochemistry 2002 41 274 281 11772026 10.1021/bi0117302
Bullesbach EE Schwabe C The relaxin receptor-binding site geometry suggests a novel gripping mode of interaction. Journal of Biological Chemistry 2000 275 35276 35280 10956652 10.1074/jbc.M005728200
Hansell DJ Bryant-Greenwood GD Greenwood FC Expression of the human relaxin H1 gene in the decidua, trophoblast, and prostate Journal of Clinical Endocrinology and Metabolism 1991 72 899 904 2005217
Winslow JW Shih A Bourell JH Weiss G Reed B Stults JT Goldsmith LT Human seminal relaxin is a product of the same gene as human luteal relaxin Endocrinology 1992 130 2660 2668 1572287 10.1210/en.130.5.2660
Sherwood OD E. Knobil JDN Relaxin The Physiology of Reproduction 1994 New York, Raven 861 1009
Zimmermann S Steding G Emmen JMA Brinkmann AO Nayernia K Holstein AF Engel W Adham IM Targeted Disruption of the Insl3 Gene Causes Bilateral Cryptorchidism Molecular Endocrinology 1999 13 681 691 10319319 10.1210/me.13.5.681
Nef S Parada LF Cryptorchidism in mice mutant for Insl3. Nature Genetics 1999 22 295 299 10391220 10.1038/10364
Bieche I Laurent A Laurendeau I Duret L Giovangrandi Y Frendo JL Olivi M Fausser JL Evain-Brion D Vidaud M Placenta-specific INSL4 expression is mediated by a human endogenous retrovirus element Biology of Reproduction 2003 68 1422 1429 12606452 10.1095/biolreprod.102.010322
Conklin D Lofton-Day CE Haldeman BA Ching A Whitmore TE Lok S Jaspers S Identification of INSL5, a new member of the insulin superfamily. Genomics 1999 60 50 56 10458910 10.1006/geno.1999.5899
Lok S Johnston DS Conklin D Lofton-Day CE Adams RL Jelmberg AC Whitmore TE Schrader S Griswold MD Jaspers SR Identification of INSL6, a new member of the insulin family that is expressed in the testis of the human and rat Biology of Reproduction 2000 62 1593 1599 10819760
Hsu SY Nakabayashi K Nishi S Kumagi J Kudo M Sherwood OD Hseuh AJW Activation of Orphan Receptors by the Hormone Relaxin Science 2002 295 671 674 11809971 10.1126/science.1065654
Sudo S Kumagai J Nishi S Layfield S Ferraro T Bathgate RA Hsueh AJ H3 relaxin is a specific ligand for LGR7 and activates the receptor by interacting with both the ectodomain and the exoloop 2 Journal of Biological Chemistry 2003 278 7855 7862 12506116 10.1074/jbc.M212457200
Kumagai J Hsu SY Matsumi H Roh JS Fu P Wade JD Bathgate RA Hsueh AJ INSL3/Leydig insulin-like peptide activates the LGR8 receptor important in testis descent Journal of Biological Chemistry 2002 277 31283 31286 12114498 10.1074/jbc.C200398200
Bogatcheva NV Truong A Feng S Engel W Adham IM Agoulnik AI GREAT/LGR8 is the only receptor for insulin-like 3 peptide Molecular Endocrinology 2003 17 2639 2646 12933905 10.1210/me.2003-0096
Liu C Eriste E Sutton S Chen J Roland B Kuei C Farmer N Jornvall H Sillard R Lovenberg TW Identification of relaxin-3/INSL7 as an endogenous ligand for the orphan G-protein coupled receptor GPCR135 J Biol Chem 2003 278 50754 50764 14522968 10.1074/jbc.M308995200
Liu C Chen J Sutton S Roland B Kuei C Farmer N Sillard R Lovenberg TW Identification of relaxin-3/INSL7 as a ligand for GPCR142 J Biol Chem 2003 278 50765 50770 14522967 10.1074/jbc.M308996200
Liu C Kuei C Sutton S Chen J Bonaventure P Wu J Nepomuceno D Wilkinson T Bathgate RAD Eriste E Sillard R Lovenberg TW INSL5 is a high affinity specific agonist for GPCR142 (GPR100) J Biol Chem 2005 292 300 15525639
Hsu SY Kudo M Chen T Nakabayashi K Bhalla A van der Spek PJ van Duin M Hsueh AJW The Three Subfamilies of Leucine-Rich Repeat-Containing G Protein-Coupled Receptors (LGR): Identification of LGR6 and LGR7 and the Signaling Mechanism for LGR7 Molecular Endocrinology 2000 14 1257 1271 10935549 10.1210/me.14.8.1257
Schwabe C Bullesbach EE Heyn H Yoshioka M Cetacean relaxin. Isolation and sequence of relaxins from Balaenoptera acutorostrata and Balaenoptera edeni. Journal of Biological Chemistry 1989 264 940 943 2910872
Schwabe C LeRoith D Thompson RP Shiloach J Roth J Relaxin extracted from protozoa (Tetrahymena pyriformis). Molecular and immunologic properties. Journal of Biological Chemistry 1983 258 2778 2781 6402504
Georges D Viguier-Martinez MC Poirier JC Relaxin-like peptide in ascidians II: bioassay and immunolocalization with anti-porcine relaxin in three species. General and Comparative Endocrinology 1990 79 429 438 2272464 10.1016/0016-6480(90)90073-U
Georges D Schwabe C Porcine relaxin, a 500 million-year-old hormone? the tunicate Ciona intestinalis has porcine relaxin. FASEB Journal 1999 13 1269 1275 10385617
Roche PJ Crawford R Tregear G A single-copy relaxin-like gene sequence is present in sheep Molecular and Cellular Endocrinology 1993 91 21 28 7682520 10.1016/0303-7207(93)90250-N
Klonisch T Froehlich C Tetens F Fischer B Hombach-Klonisch S Molecular Remodeling of Members of the Relaxin Family During Primate Evolution Mol Biol Evol 2001 18 393 403 11230540
Hsu SY New insights into the evolution of the relaxin-LGR signaling system Trends in Endocrinology and Metabolism 2003 14 303 309 12946872 10.1016/S1043-2760(03)00106-1
Bathgate RA Scott D Chung S Ellyard D Garreffa A Tregear G Searching the human genome database for novel relaxin-like and insulin-like peptides Letter in Peptide Science 2002 8 129 132 10.1023/A:1016210004087
Bathgate RA Siebel AL Tovote P Claasz AA Macris M Tregear GW Parry LJ Purification and characterization of relaxin from the tammar wallaby (Macropus eugenii): bioactivity and expression in the corpus luteum. Biology of Reproduction 2002 67 293 300 12080031
de Rienzo G Aniello F Branno M Minucci S Isolation and characterization of a novel member of the relaxin/insulin family from the testis of the frog Rana esculenta Endocrinology 2001 142 3231 3238 11416046 10.1210/en.142.7.3231
Bielawski JP Yang Z Maximum likelihood methods for detecting adaptive evolution after gene duplication Journal of Structural and Functional Genomics 2003 3 201 212 12836699 10.1023/A:1022642807731
Schwabe C Bullesbach EE Relaxin: structures, functions, promises, and nonevolution FASEB J 1994 8 1152 1160 7958621
Schwabe C Gowan LK Reinig JW Evolution, relaxin and insulin: a new perspective Annals of the New York Academy of Sciences 1982 380 6 12 7044236
Schwabe C Warr GW A polyphyletic view of evolution: the genetic potential hypothesis Perspectives in Biology and Medicine 1984 27 465 485 6374609
Dores RM Rubin DA Quinn TW Is it possible to construct phylogenetic trees using polypeptide hormone sequences? General and Comparative Endocrinology 1996 103 1 12 8812320 10.1006/gcen.1996.0088
Nagasawa H Kataoka H Isogai A Tamura S Suzuki A Mizoguchi A Fujiwara Y Suzuki A Takahashi SY Ishizaki H Amino acid sequence of a prothoracicotropic hormone of the silkworm Bombyx mori Proc Nat Acad Sci USA 1986 83 5840 5843 16593744
Brogiolo W Stocker H Ikeya T Rintelen F Fernandez R Hafen E An evolutionarily conserved function of the Drosophila insulin receptor and insulin-like peptides in growth control Current Biology 2001 11 213 221 11250149 10.1016/S0960-9822(01)00068-9
Pierce SB Costa M Wisotzkey R Devadhar S Homburger SA Buchman AR Ferguson KC Heller J Platt DM Pasquinelli AA Liu LX Doberstein SK Ruvkan G Regulation of DAF-2 receptor signaling by human insulin and ins-1, a member of the unusually large and diverse C. elegans insulin gene family Genes and Development 2001 15 672 686 11274053 10.1101/gad.867301
Hombach-Klonisch S Abd-Elnaeim MMM Skidmore JA Leiser R Fischer B Klonisch T Ruminant relaxin in the pregnant one-humped camel (Camelus dromedarius) Biology of Reproduction 2000 62 839 846 10727251
Bravo PW Stewart DR Lasley BL Fowler ME Hormonal indicators of pregnancy in the llamas and alpacas Journal of the American Veterinary Medical Association 1996 208 2027 2030 8707678
Abd-Elnaeim MMM Saber A Hassan A Abou-Elmagd A Klisch K Jones CJP Leiser R Development of the Areola in the Early Placenta of the One-humped Camel (Camelus dromedarius): A Light, Scanning and Transmission Electron Microscopical Studydromedarius) during the second half of pregnancy Anatomia, Histologia, Embryologia 2003 32 326 334 14651479 10.1111/j.1439-0264.2003.00465.x
Brackett KH Fields PA Dubois W Chang ST Relaxin: An Ovarian Hormone in an Avian Species (Gallus domesticus) General and Comparative Endocrinology 1997 105 155 163 9038247 10.1006/gcen.1996.6819
Van de Peer Y Taylor JS Meyer A Are all fishes ancient polyploids? J Struct Funct Genomics 2003 2 65 73 12836686 10.1023/A:1022652814749
Bullesbach EE Schwabe C Callard IP Relaxin from an oviparous species, the skate (Raja erinacea) Biochemical and Biophysical research communications 1987 143 273 280 3827922
Steinetz BG Schwabe C Callard IP Goldsmith LT Dogfish shark (Squalus acanthias) testes contain a relaxin. Journal of Andrology 1998 19 110 115 9537298
Bullesbach EE Schwabe C Lacy ER Identification of a glycosylated relaxin-like molecule from the male atlantic stingray, Dasyatis sabina Biochemistry 1997 36 10735 10741 9271504 10.1021/bi970393n
Reinig JW Daniel LN Schwabe C Gowan LK Steinetz BG O'Byrne E Isolation and characterization of relaxin from the sand tiger shark (Odontaspis taurus) Endocrinology 1981 109 537 543 7250055
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
Zhang J Frequent False Detection of Positive Selection by the Likelihood Method with Branch-Site Models Mol Biol Evol 2004 21 1332 1339 15014150 10.1093/molbev/msh117
Wyckoff GJ Wang W Wu C Rapid evolution of male reproductive genes in the descent of man Nature 2000 403 304 309 10659848 10.1038/35002070
Ting C Tsaur S Wu M Wu C A rapidly evolving homeobox at the site of a hybrid sterility gene Science 1998 282 1501 1504 9822383 10.1126/science.282.5393.1501
Swanson WJ Clark AG Waldrip-Dail HM Wolfner MF Aquadro CF Evolutionary EST analysis identifies rapidly evolving males reproductive proteins in Drosophila Proc Natl Acad Sci USA 2001 98 7375 7379 11404480 10.1073/pnas.131568198
Rooney AP Zhang J Rapid evolution of a primate sperm protein: relaxation of functional constraint or positive darwinian selection? Molecular Biology and Evolution 1999 16 706 710 10335662
National Centre for Biotechnology Information
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. Nucleic Acids Research 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Genomic Biology
JGI Xenopus tropicalis v2.0 Home
JGI Ciona v1.0
Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Research 1994 22 4673 4680 7984417
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
Page RD TREEVIEW: An application to display phylogenetic trees on personal computers. Computer Applications in the BioSciences 1996 12 357 358 8902363
Page RD GeneTree: comparing gene and species phylogenies using reconciled trees Bioinformatics 1998 14 819 820 9918954 10.1093/bioinformatics/14.9.819
Hedges SB The origin and evolution of model organisms Nature Reviews Genetics 2002 3 838 849 12415314 10.1038/nrg929
Yang Z Nielsen R Estimating Synonymous and Nonsynonymous Substitution Rates Under Realistic Evolutionary Models Molecular Biology and Evolution 2000 17 32 43 10666704
Yang Z PAML: a program package for phylogenetic analysis by maximum likelihood 1997 13 555 556 9367129
Ensembl zebrafish genome browser Comput Appl Biosci
| 15707501 | PMC551602 | CC BY | 2021-01-04 16:37:16 | no | BMC Evol Biol. 2005 Feb 12; 5:14 | utf-8 | BMC Evol Biol | 2,005 | 10.1186/1471-2148-5-14 | oa_comm |
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-71571322810.1186/1471-2156-6-7SoftwareComparative linkage analysis and visualization of high-density oligonucleotide SNP array data Leykin Igor [email protected] Ke [email protected] Junsheng [email protected] Nicole [email protected] Martin R [email protected] Richard JH [email protected] Wing Hung [email protected] Carsten [email protected] Cheng [email protected] Department of Biostatistics, Harvard School of Public Health, Boston, MA 02115, USA2 Department of Biostatistical Science, Dana-Farber Cancer Institute, 44 Binney Street, Boston, MA 02115, USA3 Department of Computer Science, University of Illinois at Chicago, Chicago, IL 60607, USA4 Molecular Otolaryngology Research Labs, University of Iowa, Iowa City, IA 52242, USA5 Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA 02115, USA6 Department of Statistics, Stanford University, Stanford, CA, 94305, USA7 Affymetrix Inc., 3380 Central Expressway, Santa Clara, CA 95051, USA2005 15 2 2005 6 7 7 6 7 2004 15 2 2005 Copyright © 2005 Leykin et al; licensee BioMed Central Ltd.2005Leykin 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 identification of disease-associated genes using single nucleotide polymorphisms (SNPs) has been increasingly reported. In particular, the Affymetrix Mapping 10 K SNP microarray platform uses one PCR primer to amplify the DNA samples and determine the genotype of more than 10,000 SNPs in the human genome. This provides the opportunity for large scale, rapid and cost-effective genotyping assays for linkage analysis. However, the analysis of such datasets is nontrivial because of the large number of markers, and visualizing the linkage scores in the context of genome maps remains less automated using the current linkage analysis software packages. For example, the haplotyping results are commonly represented in the text format.
Results
Here we report the development of a novel software tool called CompareLinkage for automated formatting of the Affymetrix Mapping 10 K genotype data into the "Linkage" format and the subsequent analysis with multi-point linkage software programs such as Merlin and Allegro. The new software has the ability to visualize the results for all these programs in dChip in the context of genome annotations and cytoband information. In addition we implemented a variant of the Lander-Green algorithm in the dChipLinkage module of dChip software (V1.3) to perform parametric linkage analysis and haplotyping of SNP array data. These functions are integrated with the existing modules of dChip to visualize SNP genotype data together with LOD score curves. We have analyzed three families with recessive and dominant diseases using the new software programs and the comparison results are presented and discussed.
Conclusions
The CompareLinkage and dChipLinkage software packages are freely available. They provide the visualization tools for high-density oligonucleotide SNP array data, as well as the automated functions for formatting SNP array data for the linkage analysis programs Merlin and Allegro and calling these programs for linkage analysis. The results can be visualized in dChip in the context of genes and cytobands. In addition, a variant of the Lander-Green algorithm is provided that allows parametric linkage analysis and haplotyping.
==== Body
Background
The oligonucleotide Mapping 10 K arrays [1] have been used for linkage analysis [2-4] and their advantages in genome coverage and information content compared to microsatellite-based assays has been demonstrated. The array contains 11,550 SNPs with an average heterozygosity rate of 0.32 and an average marker distance of 0.31 cM. However, the commonly used multi-point linkage analysis software packages such as GeneHunter [5,6] and Merlin [7] are command-line programs and it is not straightforward to find genes in the regions of high linkage scores. In addition, the haplotyping results are represented commonly in a text format without any gene context.
Here we report the development of a new software tool called CompareLinkage that can be used for automated conversion of Mapping 10 K genotype data into the "Linkage" format for linkage analysis in Merlin, GeneHunter and Allegro [8]. In addition the program can convert the pedigree information and SNP marker information into the "Linkage" format. After performing the linkage analysis using one or more of these programs, the CompareLinkage software can export the linkage score information into the dChip software [9-11] to visualize the results within a chromosome window. In addition, we implemented a variant of the Lander-Green [5,12] algorithm into the dChipLinkage module to analyze pedigrees with up to 18 bits (bits = 2n-f ; with n = number of non-founders and f = number of founders) using the parametric linkage analysis method. We are currently testing and validating the implementation of the algorithm which will be described in detail elsewhere. The linkage score curves, genotypes and haplotypes are graphically displayed in a dChip chromosome window which has the genes, cytoband and SNP marker information included. Together the CompareLinkage and dChip software programs provide for the first time a graphical user interface (GUI) and an automated procedure for comparative linkage analysis utilizing three commonly used linkage software programs.
Implementation
The CompareLinkage software for comparative linkage analysis using Merlin and Allegro
To analyze large pedigrees rapidly and to compare the linkage analysis results of different software packages, we developed a software tool called CompareLinkage to automate the following processes: (1) Converting of Affymetrix Mapping 10 K genotype data, pedigree files and marker information into the "Linkage" format [13], and detecting and fixing incompatibilities in pedigree genotypes. The input genotype text file for CompareLinkage can be a single text file containing genotypes for each sample or a combined text file as exported by the Affymetrix GDAS 3.0 software. (2) Automatically calling the software packages Merlin and Allegro for linkage analysis and converting the analysis results (LOD or non-parametric linkage (NPL) scores) into the input files for dChip to visualize the results in the context of genes and cytobands. (3) The SNP genotype data in the "Linkage" format can be converted into the dChip input files (genotype, pedigree and marker information files) to perform parametric linkage analysis by dChipLinkage. All steps are discussed in detail at the CompareLinkage software manual provided on the software website. All these functionalities are useful for cross-validation of linkage results and to identify concordance and discordances between different linkage analysis programs as well as between parametric and non-parametric linkage results.
A graphical user interface (GUI) for Windows was also implemented in Java. In this GUI users are allowed to set their own working directory and the location of the Perl interpreter through the "Setting" menu. CompareLinkage's functions of converting file formats and getting dChip input files are incorporated through the "Convert" and "GetCurve" menu (Figure 1). Since computing is usually time-consuming, the code of calling the Perl program is executed in separate thread to provide better interaction. The output of the Perl program can be viewed in the message window (Figure 2).
Figure 1 The CompareLinkage GUI dialog for choosing pedigree, genome information and genotype call files.
Figure 2 The intermediate output of the CompareLinkage GUI.
The dChipLinkage software module
The Affymetrix Mapping 10 K array CEL files and genotype TXT files can be imported into dChip and visualized along cytobands and genes as previously reported [9,11]. The information of the SNPs such as their genetic and physical distance and allele frequencies from three ethnic groups (Asian, African American and Caucasian) is obtained from the Affymetrix website [14] and converted into the genome information files for dChip. The information of the reference genes and cytobands is obtained from the UCSC genome bioinformatics database [15] for the matching human genome assembly (hg12 or hg15) of the SNP information, and is organized into the refGene and cytoband file provided with dChip.
We implemented a variant of the Lander-Green [5,6,16] algorithm in the dChipLinkage module of dChip to perform multipoint parametric linkage analysis and compute a LOD score at each SNP position. Disease allele frequencies, penetrance information and phenocopy information for dominant and recessive disease models can be selected by the user through a dialog (Figure 3). The Mendelian genotype errors inconsistent with parental genotypes are detected and set to missing genotypes. To handle other genotyping errors or wrongly mapped SNP markers, we assume a conservative genotyping error rate of 0.01 [1] (user adjustable) and regard observed genotypes as phenotypes in the likelihood computation [17]. As a result, the computation of the probability of the observed genotype data at one marker given an inheritance vector v involves the summation over all the possible real genotypes (or equivalently the founder allele configurations):
Figure 3 The dChipLinkage dialog for specifying linkage analysis parameters.
where Fi represents the ith of all the possible founder allele configurations and is independent of v. P(real genotypes i | v, Fi) is 1 since an inheritance vector and founder allele configuration uniquely determines the real genotypes, and P(observed genotypes | real genotypes i) involves comparing the real genotype and observed genotype for all the individuals and multiplying the probability by the error rate of 0.01 (default value) for each disagreement and 0.99 for each agreement. We also use the matrix-vector multiplication algorithm and bit reduction due to founder phase symmetry described in [16], and the founder allele factoring technique reported in [6,17] to speed up the computation of single-locus and accumulative likelihood vectors as well as the likelihood vector of disease phenotypes.
We use the forward-backward computation in the Lander-Green algorithm to obtain the marginal probability distribution of inheritance vector at each SNP marker position given the data of all the markers on a chromosome. In addition the most likely inheritance vector at each marker given the genotype data of all the markers on this chromosome is calculated [6]. Conditioned on the most likely inheritance vector at a marker and the observed genotype data, we can find the most likely founder allele configurations. When there are competing inheritance vectors with the same largest marginal probabilities at a marker, we select the one with fewer crossover events from the last marker since the distance between adjacent markers are small (average 300 kb) and it is therefore less likely to have multiple crossover events between two markers in a pedigree [7]. Together these procedures give the haplotyping results of the pedigree data. dChipLinkage visualizes the haplotyping result in either the haplotype view or the ordered genotype view.
Results
The comparative linkage analysis using Merlin, GeneHunter, Allegro and dChipLinkage
CompareLinkage can format Affymetrix Mapping 10 K SNP genotype output files and genotype files into the "Linkage" format and convert genome information and pedigree files into the formats suitable for Merlin (Version 0.10.2), GeneHunter (Version 2.1) and Allegro (Version 1.2). CompareLinkage removes all non-informative markers and calls the PedCheck software [18] to detect genotype incompatibilities using the pedigree information. A Mendelian genotype inconsistency at a SNP is handled by setting the genotype of this SNP in all the individuals to missing. For the analysis in GeneHunter, overlapping segments of large chromosomes are prepared, with each segment containing 150 or fewer markers with 75 markers in common between adjacent segments. Linkage scores are computed as the mean of two scores for the same marker from the two overlapping fragments. We ran genome-wide linkage analysis using all the three software packages and dChipLinkage for the 10 K SNP genotype data of three families: 5026.10 (Figure 4; autosomal recessive non-syndromic deafness disease, 13 bits, Asian), CR (Figure 5; recessive, 17 bits, Asian) and ER (Figure 6; dominant, 17 bits, Caucasian). For the parametric analysis, we use a disease frequency of 0.001, a penetrance value of 0.99 and a phenocopy of 0.01 for all the families and all the software packages. For GeneHunter and Allegro we ran both nonparametric and parametric analysis. For Merlin, the NPL_all statistic is computed. The allele frequencies are calculated based on the actual genotype data in each family. The LOD score or NPL score are computed at the position of the SNP makers. After running the analysis for all chromosomes, the two chromosomes with the largest LOD scores were selected from each pedigree and compared below.
Figure 4 The pedigree structure of family 5026.10. The PED 4.2 software is used to draw the pedigrees.
Figure 5 The pedigree structure of family CR
Figure 6 The pedigree structure of family ER.
Figures 7, 8, 9, 10, 11, 12 show the comparative LOD score and nonparametric score plots in dChip for these chromosomes analyzed with GeneHunter, Merlin, Allegro and dChipLinkage. The vertical red line in the figures indicates the significance threshold and is set to 3 for parametric analysis (LOD scores) and to 3.7 for non-parametric analysis (NPL score) based on statistical significance recommended by Lander and Kruglyak [19]. The linkage scores largely agree with each other in the regions with significant LOD/NPL scores. GeneHunter, Merlin and Allegro detect the peaks in the chromosome 1 and 3 of the consanguineous family 5026.10 but compute lower LOD scores than dChipLinkage (indicated by arrows in Figure 7 and 8). For another consanguineous family CR with a recessive disease, all software packages detect similar peak regions in the two chromosomes denoted as A and B (Figure 9 and 10). For the family ER with a dominant disease, dChipLinkage computes similar overall patterns but reports a possible sporadic and non-significant peak (LOD < 1.6) in each chromosome (indicated by arrows in Figure 11 and 12).
Figure 7 The comparative linkage results of the chromosome 1 of the family 5026.10 using CompareLinkage and dChipLinkage. The genotype calls are displayed on the left in yellow (AB), red (AA) and blue (BB), with SNPs on rows and samples on columns. The sample names and the disease status (1 = Unaffected and 2 = Affected) are displayed on the top. The linkage scores of different software are displayed on the right in the shaded box. The lower and upper limits of the shaded box (such as [-10, 6]) are in the brackets on the bottom of the curve. The red vertical line indicates the threshold of 3.0 for LOD scores and 3.7 for NPL scores. This line is user adjustable.
Figure 8 The comparative linkage results of the chromosome 3 from the family 5026.10. The figure format is the same as Figure 7.
Figure 9 The comparative linkage results of the chromosome A of the family CR.
Figure 10 The comparative linkage results of the chromosome B of the family CR.
Figure 11 The comparative linkage results of the chromosome A of the family ER.
Figure 12 The comparative linkage results of the chromosome B of the family ER.
Linkage analysis and visualization using dChipLinkage
To do parametric linkage analysis in dChipLinkage, a pedigree file is needed (Figure 13C). The file is similar to the standard pedigree file format but has an additional "Array" column matching each individual in the pedigree file to the corresponding genotype information in the genotype file through array names (the header line in Figure 13A). The data importing and analysis steps are:
Figure 13 The genotype file (A), genome information file (B) and pedigree file (C) used by dChipLinkage for analysis.
1. Open dChip.
2. Select the Analysis menu and the Get External Data function to read in the genotype file in the text format (Figure 13A).
3. Select the genome information file downloaded from the dChip website (Figure 13B). This file is provided in three versions, each containing the SNP information like TSC SNP ID and genetic map locations but having different allele frequencies for each of the three ethnic groups (Asian, Caucasian and African Americans).
4. Select the Analysis menu and the Chromosome function to display the genotype calls, genes and cytobands along the chromosome
5. After the program has displayed the genotype data, select the Chromosome menu and the Linkage function to start the dChipLinkage module (Figure 3). Specify the pedigree file (Figure 13C) and other linkage parameters. Depending on whether the dChip "Chromosome View" displays one or all chromosomes, the linkage analysis will be performed for one or all chromosomes accordingly. For the analysis of the 5026.10 family, the recessive disease model is assumed, and a penetrance of 0.99, phenocopy of 0.01, disease allele frequency of 0.001 and a SNP marker error rate of 0.01 are used. The SNP allele frequencies in the genome information file are used and truncated to values between 0.001 and 0.999. This family has 13 bits and it takes about 20 minutes for the whole genome linkage analysis.
Using dChipLinkage to analyze the 5026.10 family, we were able to identify a region on the chromosome 1 (Cytogenetic region: 1p36.32 – 1p36.22) with LOD scores of greater than 2.3 (Figure 7, indicated by arrow). The most interesting gene in this region is ESPIN, which has previously been shown to be involved in deafness in mice [20] and two frameshift mutations in the gene have just recently been associated with deafness in two consanguineous families [21]. Sequence analysis of the locus revealed that the parents (the individual 1 and 2 in Figure 4) and the unaffected child (the individual 6) are heterozygous for the insertion mutation and the affected children are homozygous (data not shown). In addition a novel locus with a maximum LOD score of 2.77 was identified on the chromosome 3 (Figure 8, indicated by arrow). The peak region on the chromosome 3 is about 2 Mb wide (Figure 14C). Using the GeneHunter software, we compute a maximum expected LOD score of 2.78 for this family under the specified parameters. Therefore we extract the most linkage information based on the dense SNP markers in this region. Figure 14 shows the LOD score curve together with genotype calls, inferred haplotypes and ordered genotypes based on haplotyping. In Figure 15 the results are presented in the context of cytobands and genes. The Chromosome/Export SNP data function can also export the text information of the SNPs, genes and cytobands in the region with linkage scores exceeding the threshold.
Figure 14 (A) In the genotype view, the red, blue, yellow and white colors represent genotype call AA, BB, AB and No Call. (B) The inferred haplotypes indicating ancestor origins are displayed in correspondence to the genotype view. The different colors represent distinct founder chromosomes. For each individual (column), the father allele haplotype is displayed on the left and mother allele haplotype on the right. (C) In the ordered genotype view, the red and blue colors represent the A and B genotype of father allele (left) and mother allele (right) in each individual (column). The LOD score curve is displayed in the shaded box on the right. The left boundary and right boundary of the box represent value of -2 and 3, and the red vertical line represents 2.
Figure 15 (A) The peak LOD score region is enlarged and displayed proportionally to real chromosomal distance in the context of genes and cytobands. LOD score peaks are shown at the q-arm of chromosome 3 (114.18 -117.00 Mb, maximal LOD = 2.77). The shaded curve region has the same range as Figure 14. (B) A enlarged view of the peak region with more details of the individual SNPs and genes. The transcription starting site of the genes are used to display their positions.
After the linkage computation is finished, the inferred haplotype information can be visualized. In the haplotype view (Figure 14 and 16), one can view the inference on how the founder chromosomes are crossed over and inherited by the descendants. The different colors represent distinct founder chromosomes, and for each individual, the father allele haplotype is displayed on the left and mother on the right. Since a pedigree contains no phase information of the founders [6], in the linkage computation we can assume that one child of each founder always inherits the whole grandfather-descent chromosome. This assumption does not affect the LOD score computation but reduces the number of bits in the Lander-Green algorithm by the number of founders and consequently reduces the analysis time. This is the reason that in Figure 14B the individual 1 has both father and mother haplotypes in pure color and individual 2 has only the father haplotype in pure color. By inspection of the observed genotype and the inferred haplotypes (Figure 16), one can see that only in the peak LOD score region all the affected children (individual 3, 4, 5 and 7) are homozygous and that the unaffected child (individual 6) is heterozygous. All the affected individuals share two copies of the identical chromosome segment (the pink color between the two arrows) presumably containing the disease locus. By two very close crossover events respectively in individual 6 (indicated by the black arrow) and individual 7 (indicated by the white arrow), the LOD score implicates the possible disease gene in a 2 Mb region and one can easily search the physical map for candidate disease genes in this region in the dChip chromosome view (Figure 15).
Figure 16 The genotypes (A) and inferred haplotypes (B) from family 5026.10 on the peak score region of chromosome 3 are shown (for more details see the legend in Figure 14). In the peak LOD score region all the affected children (3, 4, 5 and 7) inherited the same ancestral allele in the consanguineous family and the unaffected child (6) inherited two different ancestral alleles.
Discussion and conclusions
We have developed the CompareLinkage software for easy comparison and analysis of genotype datasets with common multi-point linkage analysis software programs. It provides functions such as automated data formatting and the calling of linkage analysis software programs to facilitate comparative linkage analysis. The results can be visualized in a chromosome window in the context of genes, cytobands and SNPs in dChip's user friendly graphical interface. The linkage scores of other linkage software packages can be saved into the dChip score file format through CompareLinkage and viewed in the dChip chromosome viewer. This provides the interface to view other computed statistics such as linkage disequilibrium scores along the chromosomes. We have also implemented a variant of the Lander-Green algorithm as the dChipLinkage module for parametric linkage analysis of small pedigrees. It can analyze all chromosomes for families with up to 18 bits within one hour on a PC with one gigabyte memory. This is useful for recessive and consanguineous families whose bits are often small.
The comparison analysis of three Mapping 10 K array data sets show similar results in regions with significant LOD scores across all the four software packages. The regions with concordant LOD/NPL scores should provide more confidence in the candidate disease loci. However, there are clear differences in isolated regions. This emphasizes the challenge of a comparative analysis using different linkage algorithm implementations. We hypothesize that the differences between the software programs in peak locations are attributable to:
1. The specific algorithm implementation in each program.
2. The difference between parametric – and non-parametric analysis.
3. The existence of undetected genotype errors in the data sets which could falsely deflate LOD scores [17,22]. dChipLinkage uses an error model to automatically handle genotype errors and avoid sporadic LOD score peaks due to undetected non-Mendelian errors, and results in a smoother LOD curve as seen in Figure 7, 8, 9, 10, 11, 12. However, this error handling algorithm involves more iterations and increases the computation time. There are further techniques to reduce the memory and time requirement of the Lander-Green algorithm [7,8,23,24]
In light of the discordance between the results from common linkage software packages and from dChipLinkage, we will validate dChipLinkage implementation using additional datasets and the CompareLinkage software.
In summary, the CompareLinkage and dChipLinkage software automate the comparative linkage analysis and visualization using multiple software packages. With these tools users will be able to increase their confidence in candidate regions and can use the visualization tools to explore the disease associated genome regions.
Availability and requirements
Project name: The CompareLinkage software and the dChipLinkage software module
Project home page:
Operating system(s): Windows (dChipLinkge); Windows (CompareLinkage and its graphical interface), Unix (CompareLinkage command line version)
Programming language: Visual C++ 6.0 (dChipLinkge); Perl and Java (CompareLinkage software)
Other requirements: None
License: None.
Any restrictions to use by non-academics: No restrictions
Authors' contributions
CR, CL and WHW conceived of the study, and participated in its design and coordination. NM and RJHS generated the 5026.10 family data, and MP generated the CR and ER family data. IL implemented the CompareLinkage software and performed the comparative analysis using multiple linkage analysis software packages. JC implemented its graphical user interface (GUI). CL implemented the dChipLinkage module. KH participated in the design and analysis of the study. IL, CR and CL drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We thank Hajime Matsuzaki, Patricia Dahia, Robert Sean Hill, Steven Boyden for helpful discussions. This work is supported by NIH grant 1R01HG02341 and P20-CA96470 (IL, KH and WHW), RO1-DC02842 (Richard J.H. Smith), NIH DK54931 (Martin R. Pollak), and grants from Friends of Dana-Farber Cancer Institute (CL) and Claudia Adams Barr Program in Cancer Research (CL).
==== Refs
Kennedy GC Matsuzaki H Dong S Liu WM Huang J Liu G Su X Cao M Chen W Zhang J Liu W Yang G Di X Ryder T He Z Surti U Phillips MS Boyce-Jacino MT Fodor SP Jones KW Large-scale genotyping of complex DNA Nat Biotechnol 2003 21 1233 1237 12960966 10.1038/nbt869
Matsuzaki H Loi H Dong S Tsai YY Fang J Law J Di X Liu WM Yang G Liu G Huang J Kennedy GC Ryder TB Marcus GA Walsh PS Shriver MD Puck JM Jones KW Mei R Parallel genotyping of over 10,000 SNPs using a one-primer assay on a high-density oligonucleotide array Genome Res 2004 14 414 425 14993208 10.1101/gr.2014904
Middleton FA Pato MT Gentile KL Morley CP Zhao X Eisener AF Brown A Petryshen TL Kirby AN Medeiros H Carvalho C Macedo A Dourado A Coelho I Valente J Soares MJ Ferreira CP Lei M Azevedo MH Kennedy JL Daly MJ Sklar P Pato CN Genomewide linkage analysis of bipolar disorder by use of a high-density single-nucleotide-polymorphism (SNP) genotyping assay: a comparison with microsatellite marker assays and finding of significant linkage to chromosome 6q22 Am J Hum Genet 2004 74 886 897 15060841 10.1086/420775
John S Shephard N Liu G Zeggini E Cao M Chen W Vasavda N Mills T Barton A Hinks A Eyre S Jones KW Ollier W Silman A Gibson N Worthington J Kennedy GC Whole-genome scan, in a complex disease, using 11,245 single-nucleotide polymorphisms: comparison with microsatellites Am J Hum Genet 2004 75 54 64 15154113 10.1086/422195
Lander ES Green P Construction of multilocus genetic linkage maps in humans Proc Natl Acad Sci U S A 1987 84 2363 2367 3470801
Kruglyak L Daly MJ Reeve-Daly MP Lander ES Parametric and nonparametric linkage analysis: a unified multipoint approach Am J Hum Genet 1996 58 1347 1363 8651312
Abecasis GR Cherny SS Cookson WO Cardon LR Merlin--rapid analysis of dense genetic maps using sparse gene flow trees Nat Genet 2002 30 97 101 11731797 10.1038/ng786
Gudbjartsson DF Jonasson K Frigge ML Kong A Allegro, a new computer program for multipoint linkage analysis Nat Genet 2000 25 12 13 10802644 10.1038/75514
Lin M Wei LJ Sellers WR Lieberfarb M Wong WH Li C dChipSNP: significance curve and clustering of SNP-array-based loss-of-heterozygosity data Bioinformatics 2004 20 1233 1240 14871870 10.1093/bioinformatics/bth069
Li C Wong WH Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection Proc Natl Acad Sci U S A 2001 98 31 36 11134512 10.1073/pnas.011404098
Li C Wong WH Parmigiani G, Garrett ES, Irizarry R and Zeger SL DNA-Chip Analyzer (dChip) The analysis of gene expression data: methods and software 2003 New York, Springer 120 141
Lange K Mathematical and statistical methods for genetic analysis 2002 2 New York, Springer-Verlag
Lathrop M Ott J Linkage User's Guide
Affymetrix Affymetrix Mapping 10K Array - Support Materials [http://www.affymetrix.com/support/technical/byproduct.affx?product=10k]
UCSC UCSC Genome Bioinformatics [http://genome.ucsc.edu/]
Kruglyak L Daly MJ Lander ES Rapid multipoint linkage analysis of recessive traits in nuclear families, including homozygosity mapping Am J Hum Genet 1995 56 519 527 7847388
Sobel E Papp JC Lange K Detection and integration of genotyping errors in statistical genetics Am J Hum Genet 2002 70 496 508 11791215 10.1086/338920
O'Connell JR Weeks DE PedCheck: a program for identification of genotype incompatibilities in linkage analysis Am J Hum Genet 1998 63 259 266 9634505 10.1086/301904
Lander E Kruglyak L Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results Nat Genet 1995 11 241 247 7581446 10.1038/ng1195-241
Zheng L Sekerkova G Vranich K Tilney LG Mugnaini E Bartles JR The deaf jerker mouse has a mutation in the gene encoding the espin actin-bundling proteins of hair cell stereocilia and lacks espins Cell 2000 102 377 385 10975527 10.1016/S0092-8674(00)00042-8
Naz S Griffith AJ Riazuddin S Hampton LL Battey JFJ Khan SN Wilcox ER Friedman TB Mutations of ESPN cause autosomal recessive deafness and vestibular dysfunction J Med Genet 2004 41 591 595 15286153 10.1136/jmg.2004.018523
Douglas JA Boehnke M Lange K A multipoint method for detecting genotyping errors and mutations in sibling-pair linkage data Am J Hum Genet 2000 66 1287 1297 10739757 10.1086/302861
Markianos K Daly MJ Kruglyak L Efficient multipoint linkage analysis through reduction of inheritance space Am J Hum Genet 2001 68 963 977 11254453 10.1086/319507
Kruglyak L Lander ES Faster multipoint linkage analysis using Fourier transforms J Comput Biol 1998 5 1 7 9541867
| 15713228 | PMC551603 | CC BY | 2021-01-04 16:38:18 | no | BMC Genet. 2005 Feb 15; 6:7 | utf-8 | BMC Genet | 2,005 | 10.1186/1471-2156-6-7 | oa_comm |
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-211572071110.1186/1471-2164-6-21Research ArticleIdentification of a new family of putative PD-(D/E)XK nucleases with unusual phylogenomic distribution and a new type of the active site Feder Marcin [email protected] Janusz M [email protected] Laboratory of Bioinformatics and Protein Engineering, International Institute of Molecular and Cell Biology, Trojdena 4, 02-109 Warsaw, Poland2005 18 2 2005 6 21 21 11 11 2004 18 2 2005 Copyright © 2005 Feder and Bujnicki; licensee BioMed Central Ltd.2005Feder and Bujnicki; 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
Prediction of structure and function for uncharacterized protein families by identification of evolutionary links to characterized families and known structures is one of the cornerstones of genomics. Theoretical assignment of three-dimensional folds and prediction of protein function even at a very general level can facilitate the experimental determination of the molecular mechanism of action and the role that members of a given protein family fulfill in the cell. Here, we predict the three-dimensional fold and study the phylogenomic distribution of members of a large family of uncharacterized proteins classified in the Clusters of Orthologous Groups database as COG4636.
Results
Using protein fold-recognition we found that members of COG4636 are remotely related to Holliday junction resolvases and other nucleases from the PD-(D/E)XK superfamily. Structure modeling and sequence analyses suggest that most members of COG4636 exhibit a new, unusual variant of the putative active site, in which the catalytic Lys residue migrated in the sequence, but retained similar spatial position with respect to other functionally important residues. Sequence analyses revealed that members of COG4636 and their homologs are found mainly in Cyanobacteria, but also in other bacterial phyla. They undergo horizontal transfer and extensive proliferation in the colonized genomes; for instance in Gloeobacter violaceus PCC 7421 they comprise over 2% of all protein-encoding genes. Thus, members of COG4636 appear to be a new type of selfish genetic elements, which may fulfill an important role in the genome dynamics of Cyanobacteria and other species they invaded. Our analyses provide a platform for experimental determination of the molecular and cellular function of members of this large protein family.
Conclusion
After submission of this manuscript, a crystal structure of one of the COG4636 members was released in the Protein Data Bank (code 1wdj; Idaka, M., Wada, T., Murayama, K., Terada, T., Kuramitsu, S., Shirouzu, M., Yokoyama, S.: Crystal structure of Tt1808 from Thermus thermophilus Hb8, to be published). Our analysis of the Tt1808 structure reveals that we correctly predicted all functionally important features of the COG4636 family, including the membership in the PD-(D/E)xK superfamily of nucleases, the three-dimensional fold, the putative catalytic residues, and the unusual configuration of the active site.
==== Body
Background
The PD-(D/E)XK domain is ubiquitously found in enzymes involved in metabolism of nucleic acids, mostly in nucleases with diverse biological functions. The first structurally characterized members of the PD-(D/E)XK superfamily were restriction enzymes (REases) (reviews: [1,2]). Crystallographic studies revealed that this superfamily groups together many nucleases with different cellular functions, including: phage λ exonuclease [3], bacterial enzymes exerting ssDNA nicking in the context of methyl-directed and very-short-patch DNA repair: MutH [4] and Vsr [5], Tn7 transposase TnsA [6], a family of archaeal Holliday junction resolvases (Hjc and Hje) from different species of Archaea [7-9], a Holliday junction resolvase (endonuclease I) from phage T7 [10], and an archaeal XPF/Rad1/Mus81 family nuclease that cleaves branched structures generated during DNA repair, replication, and recombination [11].
All members of the PD-(D/E)XK superfamily share a common structural core, comprising a mixed β-sheet of 4 or 5 strands flanked on both sides by α-helices [1,2,12]. These secondary structures are often embedded in very different peripheral elements, which sometimes constitute the majority of the protein. The common β-sheet serves as a scaffold for a weakly conserved active site, typically comprising two or three acidic residues (Asp or Glu) and one Lys residue, which together form the hallmark bipartite catalytic motif (P)D...Xn...(D/E)XK (where X is any amino acid). The Lys residue serves to position a water molecule for an in-line attack on the scissile phosphodiester bond, while the carboxylate residues coordinate a Mg2+ ion, which acts as a cofactor. Despite the wealth of structural and biochemical data, obtained mainly for REases (summarized in a collection of reviews: [13]), there is still controversy over the exact catalytic mechanism and the number of metal ions required (1, 2, or 3) by PD-(D/E)XK nucleases [14,15]. Moreover, it was found that some members of the PD-(D/E)XK superfamily developed different variants of the active site. In Vsr and its homologs, the (D/E)XK half-motif was replaced by "FxH" and an additional, unique catalytic His residue appeared in another part of the common three-dimensional fold [5]. In some REases, the acidic residue from the (D/E)XK half-motif was found to have "migrated" to another region of the polypeptide in a way that the position of the carboxylate group in the active site is generally maintained as in the "orthodox" members of the PD-(D/E)XK superfamily, despite the side chain is attached to another place in the backbone [16-19]. In a few enzymes, the conserved Lys was found to be replaced by a Glu, Gln, or Asn residue [20-22].
Crystallographic analyses have also revealed the PD-(D/E)XK fold in proteins that do not function as deoxyribonucleases at all and exhibit no conservation of the active site with the above-mentioned enzymes. The structure of the C-terminal catalytic domain of tRNA splicing endoribonuclease (RNase) EndA is identical to the minimal core of the PD-(D/E)XK fold [23], yet this protein lacks the Mg2+ binding site common to its cousins that cleave phosphodiester bonds in DNA. Remarkably, on the opposite side of the common fold, EndA developed a different active site, whose geometric configuration is very similar to that of a His-Tyr-Lys triad in structurally unrelated RNase A [24]. Finally, the N-terminal domain (NTD) of the RPB5 subunit of RNA polymerase from Saccharomyces cerevisiae exhibits perfect conservation of the restriction enzyme-like structure, but lacks any catalytic residues – it is postulated that it functions as a nucleic acid binding domain devoid of any catalytic activity [25].
The divergence exhibited by the members of the PD-(D/E)XK superfamily is remarkable. Even enzymes with very similar biological functions, such as REases that recognize and cleave the same substrate, can exhibit little or no significant sequence similarity. Thus, most of the afore-mentioned enzymes were considered unrelated until the corresponding crystal structures were solved. Only in a few cases the membership in the PD-(D/E)XK superfamily was successfully predicted using bioinformatics (in some cases backed up by mutagenesis of hypothetical catalytic residues) before the actual structures were determined [26-29]. The catalogue of members of the PD-(D/E)XK superfamily is therefore far from being complete and it is expected that new lineages will be discovered as new sequences appear in the databases. Here, we predict that a large uncharacterized protein family with an unusual phylogenetic distribution is likely to represent a new branch of PD-(D/E)XK nucleases.
Results
Sequence analysis of COG4636 reveals remote similarity to PD-(D/E)XK nucleases
In the course of analyses of proteins with unknown structures, we came across a family of sequences grouped together in the Clusters of Orthologous Groups (COG) database [30] as COG4636 and annotated as "uncharacterized protein conserved in Cyanobacteria". Analyses of cross-references to other databases revealed no functional information about any member of this family. Nonetheless, preliminary analysis of sequence conservation combined with secondary structure prediction revealed a characteristic pattern of α-helices and β-strands associated with conserved carboxylate residues (review: [31]), which suggested that members of COG4636 may belong to the PD-(D/E)XK superfamily (Figure 1). The multiple sequence alignment revealed nearly perfect conservation of a "PD" half-motif, but only partial conservation of the "(D/E)XK" half-motif. Specifically, instead of the Lys residue most members of COG4636 possessed a hydrophobic amino-acid, such as Leu or Val. This suggested that the apparent similarity to the pattern of catalytic residues typical for the PD-(D/E)XK superfamily may be either spurious or indicate a new family of enzymes with an active site devoid of the otherwise conserved residue. We searched for homologs of the analyzed family, beyond sequences from complete genomes grouped together in COG4636, by carrying PSI-BLAST searches of the nr database. Altogether, we collected 435 sequences with significant similarity to COG4636, which will be hereafter referred to as "COG4636+". No statistically significant sequence similarity was detected to any protein with an experimentally determined function.
Figure 1 Multiple sequence alignment of selected representatives of the extended COG4636+ family. The selection of representative sequences includes the modeled protein from Nostoc (motif H-PD-EXX-K, members from G. violaceus with different order of putative catalytic residues (Gv1: H-PD-EXK; Gv2: S-PD-EXD-K; Gv3: H-PD-EXD; Gv4: H-PD-EXX-N; Gv5: Q-PD-EXX-K), and members of mono-phyletic clusters from D. hafniense, C. aurantiacus, S. coelicolor, T. thermophilus, and G. violaceus). The positions of putative catalytic residues are labeled with "*". The variable termini, which could not be confidently aligned, are not shown; the number of omitted residues is indicated. A complete alignment of full-length sequences is available for download from . Amino acids are colored according to the physico-chemical properties of their side-chains (negatively charged: red, positively charged: blue, polar: magenta, hydrophobic: green). Conserved residues are highlighted. Elements of predicted secondary structure (helices and strands) are indicated by tubes and arrows, respectively.
In order to test the hypothesis of the evolutionary connection between COG4636 and the PD-(D/E)XK superfamily we carried out the fold-recognition analysis, which allows to predict the three-dimensional fold of the target protein by matching its sequence with the available protein structures and assessing the sequence-structure compatibility using a combination of criteria, such as sequence similarity, match of secondary structure elements, compatibility of residue-residue contacts, etc. (review: [32]). Sequences of individual members of COG4636 were therefore submitted to the GeneSilico protein fold-recognition metaserver [33]. Disappointingly, no methods reported statistically significant matches between these sequences and proteins with known structures. Only a few threading methods that explicitly use the structural information from the templates (FUGUE, INBGU, mGenTHREADER, SAM-T02, and 3DPSSM) reported, in some cases, matches to structures of PD-(D/E)XK nucleases, but never at the first position of the ranking. However, in the course of CASP-5 protein structure prediction contest we found that the fold-recognition operation for strongly diverged proteins can be greatly improved by limiting the analysis to the conserved core, i.e. omission of strongly diverged regions and non-conserved insertions, as well as using a refined multiple sequence alignment rather than allowing the servers to build their own sequence profiles from unrefined PSI-BLAST results [34]. Thus, we modified the multiple sequence alignment of the COG4636+ family by removing strongly diverged termini that could not be reliably aligned, and submitted to the meta-server only the core section, comprising ca. 110 aa. This time, as expected, fold-recognition analysis of a well-defined protein core gave unambiguous results: mGenTHREADER, SPARKS, and FUGUE reported structures of Holliday junction resolvases Hjc and Hje, members of the PD-(D/E)XK fold [7-9], at the first positions of their rankings, with significant scores (0.45, -2.08, and 3.46, respectively). Results obtained from the primary servers have been supported by the consensus server Pcons [35], which reported the Hjc and Hje enzymes at the first four position of its ranking, with scores 1.38-1.20, compared to the insignificant score 0.61 for the subsequent fold in the ranking.
Modeling and model-based identification of a putative active site
In order to identify the putative active site of newly predicted members of the PD-(D/E)XK superfamily, we modeled the structure of one of the COG4636+ members, whose sequence was close to the consensus calculated for the whole family (hypothetical protein all3650 from Nostoc sp. PCC 7120, GI: 17231142) and used it as a platform to study the three-dimensional arrangement of conserved residues. A homology model of all3650 was constructed using the "FRankenstein's Monster" approach (see Methods and ref. [34]), starting with the unrefined alignments between the consensus sequence and the structures of Hjc and Hje enzymes (1gef, 1hh1, and 1ob8) reported by threading methods. Initially, the model of the protein core was constructed by iterating the homology modeling procedure, evaluation of the sequence-structure fit by VERIFY3D, merging of fragments with best scores, and local realignment in poorly scored regions. Local realignments were constrained to maintain the overlap between the secondary structure elements found in the template structures, and those predicted for the target. This procedure was stopped when the regions in the protein core (helices and strands) obtained acceptable VERIFY3D score (>0.3) or their score could not be improved by any manipulations, while the average VERIFY3D score for the whole model could not be improved. The final alignment between all3650 and the three structures used as templates is shown in Figure 2. The final model of the core, comprising residues 39–188, obtained a poor average VERIFY3D score of 0.13 due to low scores in the variable loops that could not be modeled with confidence. However, the secondary structure elements (with the exception of the C-terminal helix), obtained an acceptable average score of 0.37. It is important to note that all catalytic residues of the PD-(D/E)XK fold are found in the stable regions of regular secondary structure rather than in loops [36]. The variable N-terminus, which could not be modeled because of the strong divergence and the lack of appropriate template structures, was added "de novo" using the fragment insertion method ROSETTA [37]. The coordinates of the final, full-length model (Figure 3) are available as supplementary material [see Additional file 1] and on-line at
Figure 2 Fold-recognition alignment between all3650 and structures of Hjc and Hje. Amino acids are colored according to the physico-chemical properties of their side-chains. Conserved residues are highlighted. Secondary structure elements experimentally identified in Hjc and Hje and predicted for all3650 are shown between the target and the template sequences. Known and predicted catalytic residues are indicated by "*" (above the alignment for the target, below the alignment for the templates).
Figure 3 Homology model of all3650. Helices and strands are shown in green and yellow, respectively. The predicted catalytic residues are shown in the wireframe representation and labeled. The termini are indicated.
The model of all3650 reveals a typical PD-(D/E)XK nuclease-like spatial arrangement of one Lys ε-amino group (from the residue K127) and two carboxylate groups (from residues D83 and E111) (Figure 4). The modeled structure suggest also an additional highly conserved His residue (H43) that could be a part of the metal ion-binding site or be involved in substrate-binding. Strikingly, in all3650 as well as in the great majority of sequences from the COG4636+ family, the conserved Lys (K127 in all3650) is found not in the common position in the same β-strand as the conserved Glu residue (E111 in all3650), but in a spatially adjacent α-helix. Thus, the predicted active site is formed by a "PD-EXX-K" sequence motif. This "migration" of the presumptive catalytic Lys residue and retention of the original position of the spatially adjacent carboxylate in COG4636+ members resembles the situation reported for a number of restriction enzymes such that as Cfr10I, NgoMIV, Ecl18kI, SsoII, and PspGI [16,18,19,38]. In the latter enzymes, however, it is the carboxylate that is relocated and the original position of the Lys residue is retained, such the active site is formed by a "PD-XXK-E" sequence motif (Figure 4).
Figure 4 Spatial conservation of the PD-(D/E)XK active site in all3650, Hjc, and NgoMVI. A) The predicted structure of all360 is shown in the same orientation as the crystal structures of the bona fide PD-(D/E)XK nucleases: B) Holliday junction resolvase Hje (1ob8 in PDB [9]) and C) REase NgoMIV (1fiu in PDB [78] to illustrate the spatial conservation of side-chains in the active site (the carboxylate residues in red and the Lys residue in blue), despite the lack of their conservation in the PD-EXX-K, PD-DXK, and PD-XXK-E variants of the sequence motif. Only the common core is shown, terminal regions and insertions have been omitted for clarity of the presentation.
Inspection of the multiple sequence alignment reveals that only two carboxylates (corresponding to D83 and E111 in all3650) are practically invariant in the COG4636+ family, while all the others undergo various substitutions (Figure 1). In a small group of sequences (represented by a hypothetical protein gll0909 from G. violaceus, GI: 37520478) the Lys residue is present both at the "classical" and alternative position, thereby forming a "PD-EXK-K" variant of the active site. This arrangement resembles a putative evolutionary intermediate between the "classical" active site and the newly discovered rearranged variant. In another lineage of the COG4636+ family, an Asp residue appears in the position normally occupied by Lys in the C-terminal half-motif. Some of the members of this lineage (exemplified by glr2344 from G. violaceus, GI: 37521913) exhibit therefore the "PD-EXD-K" motif, but the majority (exemplified by hypothetical protein glr1284 from G. violaceus, GI: 37520853) lack the Lys residue and exhibit only the "PD-EXD" variant. In another lineage (represented by gll1896 from G. violaceus, GI: 37521465) the Lys residue is replaced by Asn to form the "PD-EXX-N" variant of the predicted active site. The conserved His residue (H43 in all3650) is present in most members of the COG4636+ family, with the exception of a small lineage of closely related proteins (represented by gll1896 from G. violaceus, GI: 37520579) in which it is substituted by Gln, and a larger group of more diversified sequences, in which it is substituted by Thr or Ser. Most members of the latter group possess a Lys or Arg residue in the "catalytic" position and hence exhibit "PD-EXK-K" (see above) or "PD-EXR-K" variants of the active site. It will be very interesting to determine experimentally, which of those residues in different configurations are involved in catalysis, and which are only auxiliary. In particular, it would be interesting to find if both or either of the Lys residues present in the potential "intermediate" versions of the active site are required for catalysis.
Phylogenomic analysis of the COG4636+ family
Sequence searches of the nr database at the NCBI revealed that the great majority of members of the COG4636+ family (382 of total 435) originate from Cyanobacteria; of these, 84% were found in just 6 genomes (G. violaceus PCC 7421, Nostoc punctiforme PCC 73102, Crocosphaera watsonii WH 8501, Nostoc sp. PCC 7120, Anabaena variabilis ATCC 29413, Synechocystis sp. PCC 6803). It is astonishing that members of COG4636+ represent over 2% of all protein-encoding genes of G. violaceus PCC 7421 (95 of 4430 total [39]), other completely sequenced genomes of Cyanobacteria are completely devoid of them or encode only 1 or 2 sequences from this family. We were not able to identify any members of the COG4636+ family in the sequences derived from seawater samples collected from the Sargasso Sea [40] and deposited in the "environmental samples" database at the NCBI. Since the prevalent Cyanobacteria found in the Sargasso Sea are Synechococcus and Prochlorococcus, the lack of COG4636+ members in the environmental samples is in good agreement with the paucity of these genes in the fully sequenced genomes of these species.
In order to reconstruct the evolutionary history of the COG4636+ family, we calculated the phylogenetic tree, based on the same reliable section of the multiple sequence alignment that was used for protein structure prediction (see Methods). Unfortunately, in all trees obtained with different methods and parameters, the majority of deep branches received very low bootstrap support (data not shown), hence the relationships within the whole family must be regarded as unresolved. We were able, however, to identify a number of branches with bootstrap support >90%. Many of such branches comprise members from one species only. This situation is characteristic for sequences found in a few non-Cyanobacterial species; for instance 8 sequences from D. hafniense DCB-2 (Firmicutes), 7 sequences from C. aurantiacus (Chloroflexales), and 6 from S. coelicolor (Actinobacteria) each form a separate species branch on the phylogenetic tree, while 14 sequences from T. thermophilus HB27 (Deinococcus-Thermus lineage) form three separate branches. Several monophyletic groups of closely related sequences are also observed in G. violaceus (e.g. a sub-family comprising 7 sequences with GI numbers: 37522824, 37520777, 37522646, 37521452, 37522233, 37520151, 37522558). There is also one branch comprising 6 closely related sequences in C. watsonii, GI numbers: 45527153, 45527776, 45524526, 45527777, 45527775, 45527774). Other statistically significant branches, however, comprise members from different species, suggesting that they were either formed prior to speciation or that their members were transmitted horizontally between different genomes of already existing species.
To identify if members of COG4636+ are encoded by any known mobile genetics elements or if they are preferentially associated with any other proteins, we analyzed the genomic neighborhood of all members of the family. Although we carefully examined annotations of predicted open reading frames (ORFs) in the range of 3000 bp upstream and downstream, we weren't able to identify any recurrent type of proteins, either with respect to the molecular or cellular function or the predicted three-dimensional fold (data not shown). Also no preference for occurrence of COG4636+ family members within or near any apparent mobile genetic elements (putative prophages etc.) was observed. Thus, insertion of the genes encoding putative COG4636+ nucleases seems virtually random. The only notable exception is a neighborhood of another member of COG4636+, suggesting tandem duplication. We identified one instance of 4 consecutively arranged genes in the genome of C. watsonii WH8501, all from the above-mentioned branch of 6 closely related sequences (the other two relatives are located elsewhere on the chromosome). We also found a few tandem duplications: 9 in C. watsonii WH8501 and 5 in G. violaceus PCC7421, 5 in Nostoc sp. PCC6803, 2 in N. punctiforme PCC73102, 2 in A. variabilis ATCC 29413, 2 in Synechocystis sp. PCC6803, 2 in T. thermophilus HB27, 1 in T. erythraeum IMS101 and 1 in M. magnetotatcticum MS-1. In general, however, tandem duplications are rare and the distribution of COG4636+ family members along the chromosomes of Cyanobacteria with completed genomes seems completely erratic (Figure 5).
Figure 5 Localization of COG4636+ family members in the chromosomes of Cyanobacteria with completed genomes. Circular chromosome maps of genomes with at least three genes encoding COG4636+ members (indicated by dots). Genes shown in dark blue are transcribed clockwise (positive reading frame) and those in red are transcribed anticlockwise (negative reading frame). Dots plotted inside the circle indicate that more than one gene is localized in the same region of the map (1/360 of the genome length).
Discussion
Our results suggest that functionally uncharacterized proteins grouped together in COG4636 are a branch of the PD-(D/E)XK superfamily, which has not been identified to date due to a presence of an unusual variant of the active site, which lacks the conserved Lys residue at the typical position in the primary sequence. That the catalytic Lys can migrate in the framework of the active site of PD-(D/E)XK nucleases has been suggested earlier, based on the sequence analysis of another nuclease domain found in site-specific, non-long terminal repeat retrotransposable elements [2], but to date no molecular model was offered to suggest the alternative point for the attachment of the side chain to the protein backbone. Our sequence analysis of the COG4636+ family and the structural model of one of its members explain the problems with identification of the PD-(D/E)XK motif on the sequence level and provide a platform for further studies. Specifically, our analysis points at the most interesting members of the family, which display previously not observed variants of the PD-(D/E)XK active site. Experimental analyses of these proteins and determination of the role of individual amino acids in the evolutionary context may help to better understand the plasticity of the PD-(D/E)XK active site and may settle down the controversy in the field of nucleases regarding the mechanism(s) of the reaction.
Phylogenomic analyses show that putative nucleases grouped in the COG4636+ family are exceptionally abundant in genomes of certain Cyanobacteria, but absent in others. They are typically abundant in the sequenced genomes of freshwater species, but scarce in the genomes of marine species, with the exception of C. watsonii WH 8501, which was isolated from tropical waters of the Western Atlantic and Pacific oceans. It is remarkable that members of COG4636+ are almost absent from the genomes of Synechococcus and Prochlorococcus species thriving in the Sargasso sea, as well as in the environmental samples isolated from that region. On the other hand, in G. violaceus PCC 7421 they comprise over 2% of all protein-encoding genes. This phylogenetic distribution resembles that of mobile genetic elements such as introns or insertion sequences (reviews: [41,42]) and suggests that the contemporary COG4636+ family originated from a few predecessors that underwent extensive horizontal gene transfer and massive proliferation in certain genomes. Monophyly of COG4636+ sequences in non-Cyanobacterial species strongly suggests that proliferation occurred in each of these species independently, following a single event of colonization by horizontal transfer from a Cyanobacterium (or in the case of T. thermophilus – three independent successful colonizations).
We hypothesize that the mechanism by which these putative nucleases induce their proliferation in a genome is similar to that displayed by homing nucleases and restriction enzymes [43], namely to incise the DNA by introducing nicks or double-strand breaks, which stimulates recombination and may lead to tandem duplications and a variety of genomic rearrangements [44-47]. Frequent cleavage of the genomic DNA would be lethal for the cell, therefore if members of COG4636+ are indeed active as nucleases, then they should target rare sequences (in a manner similar to homing endonucleases; review: [48]) or unusual structures in the DNA (similarly to the structure-specific Holliday junction resolvases), or their activity would have to be somehow regulated (inhibited) by interactions with other proteins or cellular processes (for instance by DNA modification). There are known examples of Holliday junction resolvases carried on defective lambdoid prophages [49]. Unfortunately, analysis of the genomic neighborhood shows no preferred association of COG4636+ members with any mobile genetic elements or particular gene families that could give us hints about the cellular processes they could be part of or suggest how their predicted nuclease activity could be inhibited or regulated. Especially, we found no correlation with the presence of known or putative methyltransferases. This suggests that despite sharing the common PD-(D/E)XK fold with REases, COG4636+ members are unlikely to serve as parts of restriction-modification systems, which are known to be abundant in Cyanobacteria [50,51]. It must be noted, however, that multiple solitary DNA methyltransferases were reported in Anabaena PCC 7120 [51], and these enzymes could potentially provide protection against the cleavage of the chromosomal DNA by at least some of the COG4636+ members found in this organism.
One possibility is that COG4636+ members serve as a part of the restriction barrier, similarly to the unrelated NucA family of extracellular nucleases found in Cyanobacteria, e.g. Anabaena sp. PCC 7120 [52] and Microcystis sp. [53]. They could also fulfill a role in maintenance of the identity of the species by controlling the flow of incoming DNA, as recently suggested for restriction-modification systems [54]. From the genomic analyses it appears, however, that the primary function of COG4636+ members is to spread and multiply, and their cellular roles may be merely side-effects of this selfish expansion. It is very likely that their nuclease activity is recombinogenic and may increase the frequency of genomic rearrangements. Moreover, the multiplication of closely related COG4636+ members in certain genomes leads to an abundance of dispersed related DNA sequences, which by themselves may increase the frequency of genome rearrangements by homologous recombination. It was suggested that in the marine Cyanobacteria the factors that increase the genome plasticity might not be promoted by natural selection due to the homeostatic environment of the open ocean [55]. Conversely, the unstable environment of fresh waters might promote the spreading of factors that destabilize the genome by increasing the frequency of recombination and thereby increase the diversity of the population. This is in good agreement with our finding of prevalence of COG4636+ members in Cyanobacteria that thrive in fresh waters and their paucity in marine species (with the exception of C. watsonii WH 8501). Summarizing, it is plausible that members of COG4636+ fulfill an important role in the genome dynamics of Cyanobacteria and other species they colonize. We hope that our predictive study will facilitate experimental determination of the molecular and cellular function of members of this intriguing protein family.
Methods
Sequence analysis
Searches of the non-redundant (nr) database were carried out at the NCBI using PSI-BLAST [56] with default parameters, using different sequences from COG4636 as queries. Significantly similar sequences were retrieved from all searches and pooled together. Identical sequences from the same organism were removed. A multiple sequence alignment was generated using MUSCLE [57] with default parameters and subsequently adjusted manually, based on the analysis results of secondary structure prediction (see below), to ensure that no unwarranted gaps are introduced within α-helices and β-strands. Phylogenetic inference was carried out using the reliable central section of the multiple sequence alignment. The matrix of pairwise distances was calculated from sequences according to the JTT model [58] with gaps treated as missing data. The neighbor-joining (NJ) tree was inferred according to the method of Saitou and Nei [59].
Phylogenomic analysis
The Eutils module from the Biopython package was used as an interface to access remotely the NCBI databases [60]. The Gene Identification numbers of proteins included in the final multiple alignment sequences were used to identify the corresponding GenPept entries, which were downloaded into a local Barkeley database using an in-house developed parser based on the SAX package . The "coded_by" field from each GenPept file was used to identify the corresponding DNA sequence, which were also downloaded into the database. The sequence in the range of 3000 bp upstream or downstream from the region encoding a COG4636+ member were scanned for the presence of annotated Open Reading Frames (ORFs). Initially, the functional categorization of these ORFs was carried out based on the automatic assignment into the PFAM and COG families. In the absence of any recurrent function, the annotations of all ORFs were carefully re-analyzed visually and in uncertain cases, additional searches against the CDD database were carried out [61]. The distribution of COG4636+ members on the chromosome maps was visualized using a program developed in-house specifically for that purpose.
Protein structure prediction
Secondary structure prediction and tertiary fold-recognition was carried out via the GeneSilico meta-server gateway at [33]. Secondary structure was predicted using PSIPRED [62], PROFsec [63], PROF [64], SABLE [65], JNET [66], JUFO [67], and SAM-T02 [68]. Solvent accessibility for the individual residues was predicted with SABLE [65] and JPRED [66]. The fold-recognition analysis (attempt to match the query sequence to known protein structures) was carried out using FFAS03 [69], SAM-T02 [68], 3DPSSM [70], BIOINBGU [71], FUGUE [72], mGENTHREADER [73], and SPARKS [74]. Fold-recognition alignments reported by these methods were compared, evaluated, and ranked by the Pcons server [35].
Homology modeling
Fold-recognition alignments to the structures of selected templates were used as a starting point for homology modeling using the "FRankenstein's Monster" approach [34], comprising cycles of model building, evaluation, realignment in poorly scored regions and merging of best scoring fragments. The positions of predicted catalytic residues and secondary structure elements were used as spatial restraints. Briefly, preliminary models were generated based on the alignments to various template structures returned by the FR servers. The sequence-structure fit in these models was assessed using VERIFY3D [75] and visualized using the COLORADO3D server [76]. The most common and best-scoring fragments were merged to produce a hybrid model, in which the sequence-structure was re-evaluated. In the poorly scoring fragments the alignment was locally modified by shifting the sequences within the limits of predicted secondary structures and a next generation of models corresponding to different alignments was generated. The cycles of evaluation of models, generation of hybrids and local re-alignment in problematic regions continued until the global VERIFY3D score could not be improved. Regions, which could not be modeled because of the lack of the appropriate template structure, were added "de novo" using the fragment insertion method ROSETTA [37].
Note added in Proof
After submission of this manuscript, a crystal structure of one of the COG4636+ members was released in the Protein Data Bank (code 1wdj; Idaka, M., Wada, T., Murayama, K., Terada, T., Kuramitsu, S., Shirouzu, M., Yokoyama, S.: Crystal Structure of Tt1808 from Thermus thermophilus Hb8 To be Published). Our analysis of the Tt1808 structure and its comparison with the model of all3650 confirms our predictions. Tt1808 does indeed exhibit the PD-(D/E)xK fold: the DALI [77] search of the the Protein Data Bank (PDB) database with 1wdj revealed that its 8 closest structural matches with Z-scores in a range of 5.3-3.7 are members of the PD-(D/E)xK superfamily, including the Holliday junction resolvases we used as templates to model the all365 protein. Analysis of the Tt1808 structure (Figure 6) reveals that we correctly predicted the topology of the catalytic domain in all365. We only mispredicted an α-helix in the C-terminus of all365; in Tt1808 this element is replaced by a β-hairpin. We have also successfully modeled the structure of the N-terminal subdomain but failed to predict the interaction between this part and two loops of the catalytic domain (compare Figure 3 and Figure 6). It is important to note that these errors concern regions that do not influence any of our functional interpretations based on the all3650 model. Most importantly, the identity of presumed catalytic residues of all365 was predicted correctly, including the postulated unusual position of the Lys residue (in our model of all365 the side chain of K127 has a different orientation than K130 in Tt1808, but such details are irrelevant to our functional interpretations). It is interesting to note that Tt1808 has the S-PD-EXR-K variant of the active site, and that the side chain of the R118 residue, which replaced the "classical" catalytic Lys, points away from other catalytic residues, on the opposite side of the loop between the "EXR" and "K" elements. Summarizing, we correctly predicted all functionally important features of the COG4636+ family, including the membership in the PD-(D/E)xK superfamily of nucleases, the three-dimensional fold, the putative catalytic residues, and the unusual configuration of the active site.
Figure 6 The crystal structure of Tt1808 (1wdj in PDB). Tt1808 is shown in the same orientation and is colored and labeled in the same way as the homology model of all3650 on Figure 3. Two regions of differences between Tt1808 and the model of all3650 are indicated: the N-terminal subdomain has a similar fold, but different orientation (magenta line) and the C-terminal region folds as a β-harpin (cyan line) rather than as an α-helix.
List of abbreviations
aa, amino acid(s); bp, base pair(s); nt, nucleotide; e, expectation; REase, restriction endonuclease; ORF, product of an open reading frame,
Authors' contributions
MF carried out all sequence analyses and structure predictions using fold-recognition methods and ROSETTA. JMB built the homology model, analyzed spatial vs. sequential conservation of the putative active site, and wrote the manuscript. Both authors have read and accepted the final version of the manuscript.
Table 1 Distribution of COG4636+ family members among different bacteria.
organism / genome phylum habitat data source COG4636+ members
total disrupted
Gloeobacter violaceus PCC 7421 Cyanobacteria calcareous rock C 95 1
Nostoc punctiforme PCC 73102 Cyanobacteria cycad (endosymbiont) WGS 71 7
Crocosphaera watsonii WH 8501 Cyanobacteria marine water WGS 62 1
Nostoc sp. PCC 7120 Cyanobacteria fresh water C 58 1
Anabaena variabilis ATCC 29413 Cyanobacteria fresh water WGS 45 5
Synechocystis sp. PCC 73102 Cyanobacteria fresh water C 36 1
Thermus thermophilus HB27 Deinococcus-Thermus thermal environment C 14 -
Trichodesmium erythraeum IMS101 Cyanobacteria marine water WGS 10 3
Desulfitobacterium hafniense DCB-2 Firmicutes sewage sludge WGS 8 -
Chloroflexus aurantiacus Chloroflexi fresh water (hot springs) WGS 7 -
Streptomyces coelicolor A3(2) Actinobacteria soil C 6 -
Rhodopirellula baltica SH 1 Planctomycetes marine water C 5 1
Moorella thermoacetica ATCC 29413 Firmicutes fresh water (ponds) WGS 3 -
Deinococcus radiodurans R1 Deinococcus-Thermus unknown C 3 -
Magnetospirillum magnetotacticum MS-1 Proteobacteria fresh water (ponds) WGS 2 1
Synechococcus elongatus PCC 73102 Cyanobacteria fresh water WGS 2 -
Aquifex aeolicus VF5 Aquificae fresh water (hot springs) C 2 -
Kineococcus radiotolerans SRS30216 Actinobacteria unknown (isolated from radioactive work area) WGS 2 -
Caulobacter crescentus CB15 Proteobacteria fresh water C 1 -
Thermosynechococcus elongatus BP-1 Cyanobacteria fresh water (hot springs) C 1 -
Synechococcus sp. PCC 73102 Cyanobacteria brackish (euryhaline) and/or marine water UGS 1 -
Microcystis aeruginosa Cyanobacteria fresh water (lakes, ponds and rivers) NR 1 -
Prochlorococcus marinus str. MIT9313 Cyanobacteria marine water C - -
Prochlorococcus marinus subsp. marinus CCMP1375 Cyanobacteria marine water C - -
Prochlorococcus marinus subsp. pastoris CCMP1986 Cyanobacteria marine water C - -
Synechococcus sp. WH 8102 Cyanobacteria marine water C - -
C – Completed genomic sequence, WGS – Whole Genome Shotgun, UGS – Unfinished Genomic Sequence, NR – non-redundant database (NCBI). ORFs were regarded as "disrupted" if they bear frameshift mutations or stop codons.
Supplementary Material
Additional File 1
The additional data file all3650.pdb contains the coordinates of the original all3650 model (obtained before the Tt1808 structure was published) in the PDB format.
Click here for file
Acknowledgements
This analysis was funded by KBN (grant 3P04A01124 to JMB). JMB was also supported by the EMBO/HHMI Young Investigator Award and by a Fellowship from the Foundation for Polish Science. The work of MF was supported by the NIH (Fogarty International Center grant R03 TW007163-01).
==== Refs
Aggarwal AK Structure and function of restriction endonucleases CurrOpinStructBiol 1995 5 11 19
Bujnicki JM Pingoud A Molecular phylogenetics of restriction endonucleases Restriction Endonucleases Nucleic Acids and Molecular Biology Gross HJ 2004 14 Berlin, Springer-Verlag 63 87
Kovall RA Matthews BW Structural, functional, and evolutionary relationships between lambda-exonuclease and the type II restriction endonucleases Proc Natl Acad Sci U S A 1998 95 7893 7897 9653111 10.1073/pnas.95.14.7893
Ban C Yang W Structural basis for MutH activation in E.coli mismatch repair and relationship of MutH to restriction endonucleases Embo J 1998 17 1526 1534 9482749 10.1093/emboj/17.5.1526
Tsutakawa SE Muto T Kawate T Jingami H Kunishima N Ariyoshi M Kohda D Nakagawa M Morikawa K Crystallographic and functional studies of very short patch repair endonuclease Mol Cell 1999 3 621 628 10360178 10.1016/S1097-2765(00)80355-X
Hickman AB Li Y Mathew SV May EW Craig NL Dyda F Unexpected structural diversity in DNA recombination: the restriction endonuclease connection Mol Cell 2000 5 1025 1034 10911996 10.1016/S1097-2765(00)80267-1
Nishino T Komori K Tsuchiya D Ishino Y Morikawa K Crystal structure of the archaeal holliday junction resolvase Hjc and implications for DNA recognition Structure (Camb) 2001 9 197 204 11286886 10.1016/S0969-2126(01)00576-7
Bond CS Kvaratskhelia M Richard D White MF Hunter WN Structure of Hjc, a Holliday junction resolvase, from Sulfolobus solfataricus Proc Natl Acad Sci U S A 2001 98 5509 5514 11331763 10.1073/pnas.091613398
Middleton CL Parker JL Richard DJ White MF Bond CS Substrate recognition and catalysis by the Holliday junction resolving enzyme Hje Nucleic Acids Res 2004 32 5442 5451 15479781 10.1093/nar/gkh869
Hadden JM Convery MA Declais AC Lilley DM Phillips SE Crystal structure of the Holliday junction resolving enzyme T7 endonuclease I Nat Struct Biol 2001 8 62 67 11135673 10.1038/83067
Nishino T Komori K Ishino Y Morikawa K X-ray and biochemical anatomy of an archaeal XPF/Rad1/Mus81 family nuclease: similarity between its endonuclease domain and restriction enzymes Structure (Camb) 2003 11 445 457 12679022 10.1016/S0969-2126(03)00046-7
Venclovas C Timinskas A Siksnys V Five-stranded beta-sheet sandwiched with two alpha-helices: a structural link between restriction endonucleases EcoRI and EcoRV Proteins 1994 20 279 282 7892176
Pingoud A Gross HJ Restriction endonucleases Nucleic Acids and Molecular Biology 2004 14 Berlin, Heidelberg, Springer-Verlag 442
Kovall RA Matthews BW Type II restriction endonucleases: structural, functional and evolutionary relationships Curr Opin Chem Biol 1999 3 578 583 10508668 10.1016/S1367-5931(99)00012-5
Horton JR Blumenthal RM Cheng X Pingoud AM Restriction endonucleases: Structure of the conserved catalytic core and the role of metal ions in the DNA cleavage Restriction endonucleases Nucleic Acids and Molecular Biology Gross H J 2004 14 Berlin, Springer-Verlag 361 392
Skirgaila R Grazulis S Bozic D Huber R Siksnys V Structure-based redesign of the catalytic/metal binding site of Cfr10I restriction endonuclease reveals importance of spatial rather than sequence conservation of active centre residues J Mol Biol 1998 279 473 481 9642051 10.1006/jmbi.1998.1803
Bujnicki JM Rychlewski L Identification of a PD-(D/E)XK-like domain with a novel configuration of the endonuclease active site in the methyl-directed restriction enzyme Mrr and its homologs Gene 2001 267 183 191 11313145 10.1016/S0378-1119(01)00405-X
Pingoud V Kubareva E Stengel G Friedhoff P Bujnicki JM Urbanke C Sudina A Pingoud A Evolutionary relationship between different subgroups of restriction endonucleases J Biol Chem 2002 277 14306 14314 11827971 10.1074/jbc.M111625200
Tamulaitis G Solonin AS Siksnys V Alternative arrangements of catalytic residues at the active sites of restriction enzymes FEBS Lett 2002 518 17 22 11997010 10.1016/S0014-5793(02)02621-2
Newman M Strzelecka T Dorner LF Schildkraut I Aggarwal AK Structure of restriction endonuclease BamHI and its relationship to EcoRI Nature 1994 368 660 664 8145855 10.1038/368660a0
Lukacs CM Kucera R Schildkraut I Aggarwal AK Understanding the immutability of restriction enzymes: crystal structure of BglII and its DNA substrate at 1.5 A resolution Nat Struct Biol 2000 7 134 140 10655616 10.1038/72405
Pingoud V Sudina A Geyer H Bujnicki JM Lurz R Luder G Morgan R Kubareva E Pingoud A Specificity changes in the evolution of Type II restriction endonucleases: a biochemical and bioinformatic analysis of restriction enzymes that recognize unrelated sequences J Biol Chem 2004
Bujnicki JM Rychlewski L Unusual evolutionary history of the tRNA splicing endonuclease EndA: relationship to the LAGLIDADG and PD-(D/E)XK deoxyribonucleases Protein Sci 2001 10 656 660 11344334 10.1110/ps.37101
Li H Trotta CR Abelson J Crystal structure and evolution of a transfer RNA splicing enzyme Science 1998 280 279 284 9535656 10.1126/science.280.5361.279
Todone F Weinzierl RO Brick P Onesti S Crystal structure of RPB5, a universal eukaryotic RNA polymerase subunit and transcription factor interaction target Proc Natl Acad Sci U S A 2000 97 6306 6310 10841537 10.1073/pnas.97.12.6306
Daiyasu H Komori K Sakae S Ishino Y Toh H Hjc resolvase is a distantly related member of the type II restriction endonuclease family Nucleic Acids Res 2000 28 4540 4543 11071943 10.1093/nar/28.22.4540
Kvaratskhelia M Wardleworth BN Norman DG White MF A conserved nuclease domain in the archaeal Holliday junction resolving enzyme Hjc J Biol Chem 2000 275 25540 25546 10940317 10.1074/jbc.M003420200
Aravind L Makarova KS Koonin EV SURVEY AND SUMMARY: holliday junction resolvases and related nucleases: identification of new families, phyletic distribution and evolutionary trajectories Nucleic Acids Res 2000 28 3417 3432 10982859 10.1093/nar/28.18.3417
Bujnicki JM Rychlewski L Grouping together highly diverged PD-(D/E)XK nucleases and identification of novel superfamily members using structure-guided alignment of sequence profiles J Mol Microbiol Biotechnol 2001 3 69 72 11200231
Tatusov RL Fedorova ND Jackson JD Jacobs AR Kiryutin B Koonin EV Krylov DM Mazumder R Mekhedov SL Nikolskaya AN Rao BS Smirnov S Sverdlov AV Vasudevan S Wolf YI Yin JJ Natale DA The COG database: an updated version includes eukaryotes BMC Bioinformatics 2003 4 41 12969510 10.1186/1471-2105-4-41
Bujnicki JM Crystallographic and bioinformatic studies on restriction endonucleases: inference of evolutionary relationships in the "midnight zone" of homology Curr Protein Pept Sci 2003 4 327 337 14529527
Godzik A Fold recognition methods Methods Biochem Anal 2003 44 525 546 12647403
Kurowski MA Bujnicki JM GeneSilico protein structure prediction meta-server Nucleic Acids Res 2003 31 3305 3307 12824313 10.1093/nar/gkg557
Kosinski J Cymerman IA Feder M Kurowski MA Sasin JM Bujnicki JM A "FRankenstein's monster" approach to comparative modeling: merging the finest fragments of Fold-Recognition models and iterative model refinement aided by 3D structure evaluation Proteins 2003 53 Suppl 6 369 379 14579325 10.1002/prot.10545
Lundstrom J Rychlewski L Bujnicki JM Elofsson A Pcons: a neural-network-based consensus predictor that improves fold recognition Protein Sci 2001 10 2354 2362 11604541 10.1110/ps.08501
Fuxreiter M Simon I Protein stability indicates divergent evolution of PD-(D/E)XK type II restriction endonucleases Protein Sci 2002 11 1978 1983 12142452 10.1110/ps.4980102
Simons KT Kooperberg C Huang E Baker D Assembly of protein tertiary structures from fragments with similar local sequences using simulated annealing and Bayesian scoring functions J Mol Biol 1997 268 209 225 9149153 10.1006/jmbi.1997.0959
Pingoud V Conzelmann C Kinzebach S Sudina A Metelev V Kubareva E Bujnicki JM Lurz R Luder G Xu SY Pingoud A PspGI, a type II restriction endonuclease from the extreme thermophile Pyrococcus sp.: structural and functional studies to investigate an evolutionary relationship with several mesophilic restriction enzymes J Mol Biol 2003 329 913 929 12798682 10.1016/S0022-2836(03)00523-0
Nakamura Y Kaneko T Sato S Mimuro M Miyashita H Tsuchiya T Sasamoto S Watanabe A Kawashima K Kishida Y Kiyokawa C Kohara M Matsumoto M Matsuno A Nakazaki N Shimpo S Takeuchi C Yamada M Tabata S Complete genome structure of Gloeobacter violaceus PCC 7421, a cyanobacterium that lacks thylakoids DNA Res 2003 10 137 145 14621292
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
Lambowitz AM Belfort M Introns as mobile genetic elements Annu Rev Biochem 1993 62 587 622 8352597 10.1146/annurev.bi.62.070193.003103
Mahillon J Leonard C Chandler M IS elements as constituents of bacterial genomes Res Microbiol 1999 150 675 687 10673006 10.1016/S0923-2508(99)00124-2
Sadykov M Asami Y Niki H Handa N Itaya M Tanokura M Kobayashi I Multiplication of a restriction-modification gene complex Mol Microbiol 2003 48 417 427 12675801 10.1046/j.1365-2958.2003.03464.x
Gimble FS Invasion of a multitude of genetic niches by mobile endonuclease genes FEMS Microbiol Lett 2000 185 99 107 10754232 10.1016/S0378-1097(00)00069-0
Chinen A Uchiyama I Kobayashi I Comparison between Pyrococcus horikoshii and Pyrococcus abyssi genome sequences reveals linkage of restriction-modification genes with large genome polymorphisms Gene 2000 259 109 121 11163968 10.1016/S0378-1119(00)00459-5
Kobayashi I Behavior of restriction-modification systems as selfish mobile elements and their impact on genome evolution Nucleic Acids Res 2001 29 3742 3756 11557807 10.1093/nar/29.18.3742
Handa N Nakayama Y Sadykov M Kobayashi I Experimental genome evolution: large-scale genome rearrangements associated with resistance to replacement of a chromosomal restriction-modification gene complex Mol Microbiol 2001 40 932 940 11401700 10.1046/j.1365-2958.2001.02436.x
Jurica MS Stoddard BL Homing endonucleases: structure, function and evolution Cell Mol Life Sci 1999 55 1304 1326 10487208 10.1007/s000180050372
Mahdi AA Sharples GJ Mandal TN Lloyd RG Holliday junction resolvases encoded by homologous rusA genes in Escherichia coli K-12 and phage 82 J Mol Biol 1996 257 561 573 8648624 10.1006/jmbi.1996.0185
Lyra C Halme T Torsti AM Tenkanen T Sivonen K Site-specific restriction endonucleases in cyanobacteria J Appl Microbiol 2000 89 979 991 11123471 10.1046/j.1365-2672.2000.01206.x
Matveyev AV Young KT Meng A Elhai J DNA methyltransferases of the cyanobacterium Anabaena PCC 7120 Nucleic Acids Res 2001 29 1491 1506 11266551 10.1093/nar/29.7.1491
Muro-Pastor AM Flores E Herrero A Wolk CP Identification, genetic analysis and characterization of a sugar-non-specific nuclease from the cyanobacterium Anabaena sp. PCC 7120 Mol Microbiol 1992 6 3021 3030 1343821
Takahashi I Hayano D Asayama M Masahiro F Watahiki M Shirai M Restriction barrier composed of an extracellular nuclease and restriction endonuclease in the unicellular cyanobacterium Microcystis sp FEMS Microbiol Lett 1996 145 107 111 8931334 10.1016/0378-1097(96)00395-3
Jeltsch A Maintenance of species identity and controlling speciation of bacteria: a new function for restriction/modification systems? Gene 2003 317 13 16 14604787 10.1016/S0378-1119(03)00652-8
Carr NG Mann NH Bryant DA The oceanic cyanobacterial picoplankton The molecular biology of Cyanobacteria 1994 Dordrecht, Kluwer Academic Publishers
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 NucleicAcidsRes 1997 25 3389 3402
Edgar RC MUSCLE: multiple sequence alignment with high accuracy and high throughput Nucleic Acids Res 2004 32 1792 1797 15034147 10.1093/nar/gkh340
Jones DT Taylor WR Thornton JM The rapid generation of mutation data matrices from protein sequences Comput Appl Biosci 1992 8 275 282 1633570
Saitou N Nei M The neighbor-joining method: a new method for reconstructing phylogenetic trees Mol Biol Evol 1987 4 406 425 3447015
Chapman B Chang J Biopython: python tools for computational biology. ACM SIGBIO Newslett 2000 20 15 19
Marchler-Bauer A Anderson JB DeWeese-Scott C Fedorova ND Geer LY He S Hurwitz DI Jackson JD Jacobs AR Lanczycki CJ Liebert CA Liu C Madej T Marchler GH Mazumder R Nikolskaya AN Panchenko AR Rao BS Shoemaker BA Simonyan V Song JS Thiessen PA Vasudevan S Wang Y Yamashita RA Yin JJ Bryant SH CDD: a curated Entrez database of conserved domain alignments Nucleic Acids Res 2003 31 383 387 12520028 10.1093/nar/gkg087
McGuffin LJ Bryson K Jones DT The PSIPRED protein structure prediction server Bioinformatics 2000 16 404 405 10869041 10.1093/bioinformatics/16.4.404
Rost B Yachdav G Liu J The PredictProtein server Nucleic Acids Res 2004 32 W321 6 15215403
Ouali M King RD Cascaded multiple classifiers for secondary structure prediction Protein Sci 2000 9 1162 1176 10892809
Adamczak R Porollo A Meller J Accurate prediction of solvent accessibility using neural networks-based regression Proteins 2004 56 753 767 15281128 10.1002/prot.20176
Cuff JA Barton GJ Application of multiple sequence alignment profiles to improve protein secondary structure prediction Proteins 2000 40 502 511 10861942 10.1002/1097-0134(20000815)40:3<502::AID-PROT170>3.0.CO;2-Q
Meiler J Baker D Coupled prediction of protein secondary and tertiary structure Proc Natl Acad Sci U S A 2003 100 12105 12110 14528006 10.1073/pnas.1831973100
Karplus K Karchin R Draper J Casper J Mandel-Gutfreund Y Diekhans M Hughey R Combining local-structure, fold-recognition, and new fold methods for protein structure prediction Proteins 2003 53 Suppl 6 491 496 14579338 10.1002/prot.10540
Rychlewski L Jaroszewski L Li W Godzik A Comparison of sequence profiles. Strategies for structural predictions using sequence information Protein Sci 2000 9 232 241 10716175
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
Fischer D Hybrid fold recognition: combining sequence derived properties with evolutionary information Pacific Symp Biocomp 2000 119 130
Shi J Blundell TL Mizuguchi K FUGUE: sequence-structure homology recognition using environment-specific substitution tables and structure-dependent gap penalties J Mol Biol 2001 310 243 257 11419950 10.1006/jmbi.2001.4762
Jones DT GenTHREADER: an efficient and reliable protein fold recognition method for genomic sequences J Mol Biol 1999 287 797 815 10191147 10.1006/jmbi.1999.2583
Zhou H Zhou Y Single-body residue-level knowledge-based energy score combined with sequence-profile and secondary structure information for fold recognition Proteins 2004 55 1005 1013 15146497 10.1002/prot.20007
Luthy R Bowie JU Eisenberg D Assessment of protein models with three-dimensional profiles Nature 1992 356 83 85 1538787 10.1038/356083a0
Sasin JM Bujnicki JM COLORADO3D, a web server for the visual analysis of protein structures Nucleic Acids Res 2004 32 W586 9 15215456 10.1093/nar/gkh032
Holm L Sander C Protein structure comparison by alignment of distance matrices J Mol Biol 1993 233 123 138 8377180 10.1006/jmbi.1993.1489
Deibert M Grazulis S Sasnauskas G Siksnys V Huber R Structure of the tetrameric restriction endonuclease NgoMIV in complex with cleaved DNA Nat Struct Biol 2000 7 792 799 10966652 10.1038/79032
| 15720711 | PMC551604 | CC BY | 2021-01-04 16:39:32 | no | BMC Genomics. 2005 Feb 18; 6:21 | utf-8 | BMC Genomics | 2,005 | 10.1186/1471-2164-6-21 | oa_comm |
==== Front
BMC Mol BiolBMC Molecular Biology1471-2199BioMed Central London 1471-2199-6-41572070810.1186/1471-2199-6-4Research ArticleSelection of reference genes for gene expression studies in human neutrophils by real-time PCR Zhang Xiaozhu [email protected] Lily [email protected] Andrew J [email protected] The James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research, St. Paul's Hospital, University of British Columbia, Vancouver, Canada2005 18 2 2005 6 4 4 5 7 2004 18 2 2005 Copyright © 2005 Zhang et al; licensee BioMed Central Ltd.2005Zhang 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
Reference genes, which are often referred to housekeeping genes, are frequently used to normalize mRNA levels between different samples. However the expression level of these genes may vary among tissues or cells, and may change under certain circumstances. Thus the selection of reference gene(s) is critical for gene expression studies. For this purpose, 10 commonly used housekeeping genes were investigated in isolated human neutrophils.
Results
Initial screening of the expression pattern demonstrated that 3 of the 10 genes were expressed at very low levels in neutrophils and were excluded from further analysis. The range of expression stability of the other 7 genes was (from most stable to least stable): GNB2L1 (Guanine nucleotide binding protein, beta polypeptide 2-like 1), HPRT1 (Hypoxanthine phosphoribosyl transferase 1), RPL32 (ribosomal protein L32), ACTB (beta-actin), B2M (beta-2-microglobulin), GAPD (glyceraldehyde-3-phosphate dehydrogenase) and TBP (TATA-binding protein). Relative expression levels of the genes (from high to low) were: B2M, ACTB, GAPD, RPL32, GNB2L1, TBP, and HPRT1.
Conclusion
Our data suggest that GNB2L1, HPRT1, RPL32, ACTB, and B2M may be suitable reference genes in gene expression studies of neutrophils.
==== Body
Background
Neutrophils are the most numerous granulocytes in blood and are responsible for the first line of host defence. However, neutrophils have frequently been implicated in the pathogenesis of many diseases because they can produce various cytokines, chemokines and other proinflammatory mediators [1,2]. Numerous studies have been performed on the mechanisms that regulate the bioactivity of neutrophils. Understanding patterns of expressed genes may provide insight into complex regulatory networks and help to identify genes implicated in diseases. Quantitative real time PCR is one of the most powerful quantification methods for gene expression analysis. Similar to other methods used in expression studies, data from samples are usually required to be normalized against a set of data or references to correct for the difference in the amount of starting materials. The genes used as references are often referred to as housekeeping genes, assuming that those genes are constitutively expressed in certain tissues and under certain circumstances. However, the literature shows that the expression levels of the so called "housekeeping genes" may vary in different tissues, different cell types, and different disease stages [3-6]. Therefore, the selection of the reference genes is critical for the interpretation of the expression data.
In this study, we investigated 10 commonly used housekeeping genes (Table 1), and found 5 genes could be preferential reference genes for gene expression studies in human neutrophils.
Results
RNA quality and quantity
RNA analysis by an Agilent 2100 Bioanalyzer provided the size profiles and the concentration of the samples. All the RNA samples used in this study were of good quality despite the long neutrophil isolation procedure. Intact rRNA subunits of 28S and 18S were observed on both the gel electrophoresis and electrophotogram, indicating that the degradation of the RNA was minimal (Figure 1).
Expression patterns of the candidate genes in neutrophils
Initial screening for the gene expression pattern suggested that the 10 candidate housekeeping genes were differentially expressed in neutrophils (data not shown). Based on the band intensity of the PCR products, the two lowest expressed genes, two medium expressed genes and the three highest expressed genes were chosen for real-time PCR analysis. ABL1, PBGD and TUBB were excluded from further evaluation due to their extremely low expression level.
Standard curve and real-time PCR
Standard curves were generated by using copy number vs. the threshold cycle (Ct). The linear correlation coefficient (R2) of all the seven genes ranged from 0.976 to 0.999. Based on the slopes of the standard curves, the amplification efficiencies of the standards were from 91%~100%, which were derived from the formula E = 10 1/-slope -1. The Ct values of all the 7 genes in all the unknown samples were within 15.9 to 33.5 cycles, covered by the range of the standard curves. Electrophoresis analysis of all the amplified products from real-time PCR showed a single band with the expected sizes, and no primer dimer was observed. The dissociation plots provided by the ABI Prism 7900HT also indicated a single peak in all the reactions.
The stability and expression level of reference genes in the neutrophils
The gene expression levels were measured by real-time PCR, and the expression stabilities were evaluated by the M value of GeNorm. The ranking of the expression stability in these genes was (from the most stable to the least stable): GNB2L1, HPRT1, RPL32, ACTB, B2M, GAPD and TBP (Figure 2). The M values of GNB2L1, HPRT1, RPL32, ACTB, and B2M were lower than 0.5, and therefore these genes were concluded to be stably expressed housekeeping genes in neutrophils.
A normalization factor (NF) was calculated based on the geometric mean of the copy numbers of these 5 selected reference genes in each sample. After normalization against the NF, the ranking of the relative expression levels was (from high to low): B2M, ACTB, GAPD, RPL32, GNB2L1, TBP, and HPRT1 (Figure 3).
Based on both the expression stability and expression level, our data suggested that B2M and ACTB can be used as a reference gene for high abundance gene transcripts, RPL32 and GNB2L1 for medium abundance transcripts, and HPRT1 for low abundance transcripts in gene expression studies.
Discussion
Real-time PCR is one of the most sensitive and flexible quantification methods for gene expression analysis. It provides simultaneous measurement of gene expression in many different samples for a number of genes. However, many factors in real-time PCR may affect the results, including the selection of the reference genes. An ideal reference gene should be expressed at a constant level among different tissues of an organism, at all stages of development, and should be unaffected by the experimental treatment. However, no one single gene is expressed at such a constant level in all these situations [4,7]. For example, ACTB, GAPD, 18S and 28S rRNA are the most commonly used reference genes, but a number of studies have provided solid evidence that their transcription levels vary significantly between different individuals, different cell types, different developmental stages, and different experimental conditions [3-6]. Therefore, thorough validation of candidate reference genes is critical for accurate analysis of gene expression.
It is also well known that RNA quality and quantity are critical for successful gene expression analysis. Degraded and inaccurately quantified RNA would give misleading results. In this study, the total RNA was extracted from isolated human neutrophils, and usually it takes 2–3 hours from drawing the blood to obtaining the pure neutrophils. RNA degradation is frequently observed. For this reason we performed careful RNA analysis by using an Agilent 2100 Bioanalyzer (Agilent Technologies) before the gene expression study. The results indicated our RNA samples were of good quality. Other quantification methods which need a microgram-level of RNA were not practical for our study because the amount of RNA extracted from the neutrophils from 10 ml blood was very limited (around 3–5 μg).
DNA contamination is another important factor that affects the accuracy of gene expression analysis. In this study, the following steps were taken to prevent and monitor DNA contamination: (1) RNase-free DNase I treatment on all the RNA samples; (2) The primers were designed to be able to distinguish the PCR product derived from mRNA or genomic DNA (Table 2); (3) Dissociation analysis by ABI Prism 7900HT; (4) Gel electrophoresis of all the amplified PCR products. With all these precautions in place we were confident that there was no detectable DNA contamination. The signal from SYBR I was specifically from the desired amplicons, not from artefacts (primer dimers or genomic DNA contamination).
For the reasons discussed above, we have confidence that our gene expression results were accurate and reliable, and we further analyzed the expression stability and expression level. The principle that the expression ratio of two ideal reference genes should be identical in all samples is well established. Based on this principle we found GNB2L1, HPRT1, RPL32, ACTB, and B2M were stably expressed in the neutrophils, and they were used for the calculation of a normalization factor (NF). After normalization we found B2M was the most highly expressed, followed by ACTB, RPL32, GNB2L1, and HPRT1 was the lowest expressed. As the expression level of the reference genes may be an additional factor for consideration in the process of reference gene selection, this ranking of the relative expression level of the candidate reference genes may be informative for future gene expression studies in neutrophils.
Conclusion
To our knowledge, this is the first detailed study of the stability and level of reference gene expression in neutrophils. We found GNB2L1, HPRT1, RPL32, ACTB, and B2M are good choices for reference gene(s) selection. B2M and ACTB can be used for high-abundance mRNA, RPL32 and GNB2L1 for medium-abundance mRNA, and HPRT1 for low-abundance mRNA in expression studies of neutrophils. For more accurate normalization, as suggested by other authors [8], we recommend a combination of the stably expressed genes GNB2L1, HPRT1, RPL32, ACTB, and B2M as a panel of reference genes for the normalization.
Methods
Candidate genes for expression studies
Ten housekeeping genes were selected from commonly used reference genes (ABL1, ACTB, B2M, GAPD, GNB2L1, HRPT1, PBGD, RPL32, TBP, and TUBB). Gene symbols and their full names, gene accession numbers as well as functions are listed in Table 1. These genes were chosen because they have different functions in order to avoid genes belonging to the same biological pathways that may be co-regulated. In selecting the genes to be analyzed, preference was given to pseudogene-free genes in the NCBI linked database (Table 1). All the primers were designed by the software, Primer 3, . Hairpin structure and primer dimerization were analyzed by NetPrimer. Primers spanning at least one intron were chosen to minimize inaccuracies due to genomic DNA contamination. The length of the primers was from 18-mer to 22-mer, GC content was from 45% to 60%, and the expected PCR products range from 114 bp to 318 bp. If the genes have pseudogenes, primers were chosen according to the alignment results between the genes and the pseudogenes, so that the primers were unique to the genes and different from the pseudogenes (Table 2).
Subjects and sample preparation
A total of 15 volunteers were recruited (Table 3). All participants signed an informed consent document. 20 ml of peripheral blood was taken into heparinized tubes. Neutrophil isolation was performed by a Dextran-Ficoll sedimentation and centrifugation method [9]. Briefly, 20 ml blood was mixed with 5% Dextran (100,000–20,000 k Da; Sigma) in RPMI (9:2). After 40 min sedimentation, the white blood cell rich plasma was transferred onto the top of 10 ml Ficoll (Pharmacia), and centrifuged at 2500 rpm for 15 min. The cell pellet contained the neutrophils. The contaminating erythrocytes were removed by hypotonic lysis. The isolated neutrophils were subject to Kimra staining and microscopic cell differential count. The purity of the neutrophils was calculated. Samples with more than 2% eosinophils were excluded from the study. Half of the isolated neutrophils were used for RNA isolation.
RNA extraction and RT-PCR
Total RNA was isolated using RNeasy Mini Kit (Qiagen) as described by the manufacturer. Genomic DNA was eliminated by RNase-free DNase I digestion (Qiagen) during the isolation procedure. Isolated total RNA was analyzed on an Agilent 2100 Bioanalyzer using the RNA 6000 pico labchip Kit (Agilent Technologies). First strand cDNA synthesis was carried out with SuperScript RNase H- Reverse Transcriptase (Invitrogen) and random primers (Invitrogen) in a total volume of 20 μl. Reverse transcription was performed at 37°C for 1 hour followed by 72°C for 15 min.
Amplification of gene transcripts
To screen the basal expression patterns of the candidate genes in neutrophils, three randomly selected samples were tested by PCR with the ten primer pairs (Table 2).
The expression study was performed using a 384 well plate on an ABI Prism 7900HT Sequence Detection System (Applied Biosystems) with QuantiTect SYBR Green PCR Kit (Qiagen). The reactions were performed according to the manufacturer's instructions with minor modifications. The PCR program was initiated at 95°C for 10 min to activate Taq DNA polymerase, followed by 45 thermal cycles of 15 seconds at 94°C, 30 seconds at 58°C and 30 seconds at 72°C. Size analysis of the PCR products (dissociation analysis or meting curve analysis) was performed immediately after the real-time PCR. The temperature range used for the melting curve generation was from 60°C to 95°C. Each sample was analyzed in triplicate wells. In addition, all the reactions were further subject to electrophoresis on 2.5% agarose gels stained with ethidium bromide to confirm the expected PCR products.
Standard curves
The amplified fragments from each primer pair were purified with QIAquick PCR purification Kit (Qiagen), and confirmed by DNA sequencing (University of British Columbia, NAPS Unit). The concentrations of the PCR products were quantified by a spectrophotometer (Perkin-Elmer Lambda 2 UV/VIS Spectrometer), which were further transformed to copy numbers based on the length and base composition of the PCR products. A ten-fold series dilution was made and 10 to 1,000,000 copies were used for generating standard curves in the real-time PCR, plotted as Ct values (cycle numbers of threshold or crossing points) versus logarithms of the given concentrations of the DNA templates.
Determination of Gene stability and expression levels in human neutrophils
Gene stability was also evaluated using the geNorm software program [8]. Briefly, this approach relies on the principle that the expression ratio of two perfect reference genes would be identical in all samples in all experimental conditions or cell types. Variation in the expression ratios between different samples reflects the fact that one or both of the genes are not stably expressed. Therefore, increasing variation in this ratio corresponds to decreasing expression stability. The geNorm program can be used to calculate the gene expression stability measure (M), which is the mean pair-wise variation for a gene compared with all other tested control genes. Genes with higher M values have greater variation in expression. The stepwise exclusion of the gene with the highest M value allows the ranking of the tested genes according to their expression stability. The proposed threshold for eliminating a gene as unstable was M ≥ 0.5. In the final analysis, genes with M value lower than 0.5 were considered as stably expressed genes, and were used for normalization factor (NF) calculation. Using the NF we calculated and ranked the expression level of all the seven genes in our samples.
Abbreviations
ACTB, beta-actin; ALB1, Abelson murine leukemia viral oncogene homolog 1; B2M, beta-2-microglobulin; cDNA, complementary DNA; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GNB2L1 (Guanine nucleotide binding protein, beta polypeptide 2-like 1; HPRT1, Hypoxanthine phosphoribosyltransferase 1; NCBI, National Center for Biotechnology Information; PCR, polymerase chain reactions; PBGD, porphobilinogen deaminase; RPL32, ribosomal protein L32; RT-PCR, reverse transcription-polymerase chain reactions; TBP, TATA-binding protein; TUBB, beta-tubulin
Authors' contributions
XZ performed all the experimental procedures and was the primary author of the manuscript. LD participated in the study design and data analysis. AS conceived of the study, participated in the study design and coordination. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by grants from the British Columbia Lung Association and the American Thoracic Society. AJS is the recipient of a Canada Research Chair in genetics. The authors would like to thank Drs. Peter Paré and James Hogg for their expert reviews of the manuscript.
Figures and Tables
Figure 1 The results of RNA analysis by Agilent bioanalyzer. The first peak is a 20 bp molecular marker. The second and the third peaks are 18S and 28S rRNA.
Figure 2 Gene expression stability of seven candidate reference genes in the neutrophil analyzed by the geNorm program. The threshold for eliminating a gene as unstable was M ≥ 0.5.
Figure 3 The relative expression level normalized against Normalization Factors from the 5 most stable genes (HPRT1, GNB2L1, RPL32, ACTB, and B2M) provided by geNorm. HPRT1 was the lowest expressed gene, and B2M was the highest among the candidate genes in neutrophils.
Table 1 10 selected candidate housekeeping genes
Gene symbol Gene Name Accession Number Function Gene synonyms
mRNA genomic DNA
ABL1 Abelson murine leukemia viral oncogene homolog NM_007313 NT_035014 Cytoplasmic and nuclear protein tyrosine kinase ABL, JTK7, p150, c-ABL, v-abl
ACTB Beta-actin NM_001101 NT_007819 Cytoskeletal structural protein
B2M Bata-2-microglobulin NM_004048 NT_030828 Cytoskeletal protein involved in cell locomotion
GAPD Glyceraldehyde-3-phosphate dehydrogenase NM_002046 NT_009759 Glycolytic enzyme G3PD, GAPDH
GNB2L1 Guanine nucleotide binding protein, β-peptide 2-like 1 NM_006098 NT_077451 Involved in binding and anchorage of protein kinase C H12.3, RACK1, Gnb2-rs1
HPRT1 Hypoxanthine phosphoribosyltransferase 1 NM_000194 NT_011786 Constitutively expressed at low levels, involved in the metabolic salvage of purines in mammals. HPRT, HGPRT
PBGD Porphobilinogen deaminase NM_000190 NT_033899 Deficiency of porphobilinogen deaminase results in acute intermittent porphyria HMBS, AIP, UPS
RPL32 Ribosomal protein L32 NM_000994 NT_005927 Member of the 80 different ribosome proteins
TBP TATA-binding protein NM_003194 NT_007583 Involved in the activation of basal transcription from class II promoter GTF2D, SCA17, TFIID, GTF2D1
TUBB Beta-tubulin NM_001069 NT_034880 Member of the tubulin family of structural proteins
Table 2 Primers for Real-Time PCR
Gene symbol Length Position in cDNA Sequence (5'-3')
ABL1 20 Exon 7 1217-1236 TGACAGGGGACACCTACACA
20 Exon 9 1535-1516 TCAAAGGCTTGGTGGATTTC
ACTB 20 Exon 2 210-229 CATCGAGCACGGCATCGTCA
21 Exon 3 420-400 TAGCACAGCCTGGATAGCAAC
B2M 19 Exon 2 268-287 ACTGAATTCACCCCCACTGA
20 Exon 4 381-362 CCTCCATGATGCTGCTTACA
GAPD 20 Exon 7 728-747 TGGACCTGACCTGCCGTCTA
22 Exon 8 970-948 CCCTGTTGCTGTAGCCAAATTC
GNB2L1 20 Exon 3 327-346 GAGTGTGGCCTTCTCCTCTG
20 Exon 5 550-531 GCTTGCAGTTAGCCAGGTTC
HRPT1 20 Exon 4 322-341 GACCAGTCAACAGGGGACAT
22 Exon 7 516-495 AACACTTCGTGGGGTCCTTTTC
PBGD 18 Exon 11-12 764-781 AGGATGGGCAACTGTACC
20 Exon 13 995-976 GTTTTGGCTCCTTTGCTCAG
RPL32 19 Exon 1/2 33-51 CATCTCCTTCTCGGCATCA
20 Exon 3 185-166 AACCCTGTTGTCAATGCCTC
TBP 20 Exon 4 623-642 GAACCACGGCACTGATTTTC
20 Exon 5 780-761 CCCCACCATGTTCTGAATCT
TUBB 20 Exon 3 240-259 CTTCGGCCAGATCTTCAGAC
20 Exon 4 416-397 AGAGAGTGGGTCAGCTGGAA
Table 3 Characteristics of the study subjects
No. Gender (F/M) Age Diagnosis
Allergy Asthma Healthy
Asian 6 4/2 33 ± 3 3 - 3
Caucasian 9 3/6 33 ± 9 3 2 4
==== Refs
Cassatella MA The production of cytokines by polymorphonuclear neutrophils Immunol Today 1995 16 21 26 7880385 10.1016/0167-5699(95)80066-2
Cassatella MA Neutrophil-derived proteins: selling cytokines by the pound Adv Immunol 1999 73 369 509 10399011
Warrington JA Nair A Mahadevappa M Tsyganskaya M Comparison of human adult and fetal expression and identification of 535 housekeeping/maintenance genes Physiol Genomics 2000 2 143 147 11015593
Thellin O Zorzi W Lakaye B De Borman B Coumans B Hennen G Grisar T Igout A Heinen E Housekeeping genes as internal standards: use and limits J Biotechnol 1999 75 291 295 10617337 10.1016/S0168-1656(99)00163-7
Bustin SA Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays J Mol Endocrinol 2000 25 169 193 11013345 10.1677/jme.0.0250169
Suzuki T Higgins PJ Crawford DR Control selection for RNA quantitation Biotechniques 2000 29 332 337 10948434
Haberhausen G Pinsl J Kuhn CC Markert-Hahn C Comparative study of different standardization concepts in quantitative competitive reverse transcription-PCR assays J Clin Microbiol 1998 36 628 633 9508285
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
Le Cabec V Maridonneau-Parini I Annexin 3 is associated with cytoplasmic granules in neutrophils and monocytes and translocates to the plasma membrane in activated cells Biochem J 1994 303 ( Pt 2) 481 487 7526843
| 15720708 | PMC551605 | CC BY | 2021-01-04 16:22:25 | no | BMC Mol Biol. 2005 Feb 18; 6:4 | utf-8 | BMC Mol Biol | 2,005 | 10.1186/1471-2199-6-4 | oa_comm |
==== Front
BMC Fam PractBMC Family Practice1471-2296BioMed Central London 1471-2296-6-71571591410.1186/1471-2296-6-7Study ProtocolEvaluation of an education and activation programme to prevent chronic shoulder complaints: design of an RCT [ISRCTN71777817] Bruijn Camiel De [email protected] Bie Rob [email protected] Jacques [email protected] Marielle [email protected]öke Albère [email protected] den Heuvel Wim [email protected] der Heijden Geert [email protected] Geert-Jan [email protected] Institute for Rehabilitation Research, Hoensbroek, The Netherlands2 Department of General Practice and Care and Public Health Research Institute, Maastricht University, The Netherlands3 Department of Epidemiology, Maastricht University, The Netherlands4 Department of Medical, Clinical and Experimental Psychology, Maastricht University, The Netherlands5 Hoensbroek Rehabilitation Centre, Hoensbroek, The Netherlands6 Julius Center for Health Sciences and Primary Care, University Medical Centre, Utrecht, The Netherlands7 Pain Management and Research Centre, University Hospital Maastricht, The Netherlands2005 16 2 2005 6 7 7 7 12 2004 16 2 2005 Copyright © 2005 Bruijn et al; licensee BioMed Central Ltd.2005Bruijn 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
About half of all newly presented episodes of shoulder complaints (SC) in general practice are reported to last for at least six months. Early interventions aimed at the psychological and social determinants of SC are not common in general practice, although such interventions might prevent the development of chronic SC.
The Education and Activation Programme (EAP) consists of an educational part and a time-contingent activation part. The aim of the EAP is to provide patients with the proper cognitions by means of education, and to stimulate adequate behaviour through advice on activities of daily living.
Design
The article describes the design of a randomised clinical trial (RCT) to evaluate the effectiveness and cost-effectiveness of an EAP in addition to usual care, compared to usual care only, in the prevention of chronic SC after six months. It also describes the analysis of the cost and effect balance. Patients suffering from SC for less than three months are recruited in general practice and through open recruitment. A trained general practitioner or a trained therapist administers the EAP.
Primary outcome measures are patient-perceived recovery, measured by self-assessment on a seven-point scale, and functional limitations in activities of daily living. Questionnaires are used to study baseline measures, prognostic measures, process measures and outcome measures.
Discussion
The inclusion of patients in the study lasted until December 31st 2003. Data collection is to end in June 2004.
==== Body
Background
Shoulder complaints
Shoulder complaints (SC) have been defined by Sobel & Winters [1] as pain localised in the region of the deltoid muscle, the acromioclavicular joint, the superior part of the trapezoid muscle and the scapula. Radiation of the pain to the arm as well as limitation of the motion of the upper arm and/or the shoulder girdle may be present [1].
SC are characterised by pain in the area between the base of the neck and the elbow, at rest or when elicited by movement of the upper arm (Fig. 1).
Figure 1 Area between the base of the neck and the elbow
Musculoskeletal disorders, of which SC constitute the second largest group after low back disorders, account for the second largest share in healthcare costs and represent the largest group of work-related diseases in the Netherlands [2].
Shoulder complaints in general practice
The point prevalence of SC in the general population in the Netherlands has recently been estimated at 21% [3]. In a British study a lower point prevalence of 14% has been found [4]. The annual incidence of SC as seen by general practitioners (GPs) in the Netherlands lies between 15 and 25 patients per 1000 registered general practice patients[1]. About half of all newly presented episodes in general practice are reported to last for at least six months, while 40 percent of the newly presented episodes result in disability in terms of activities of daily living after one year [5].
The International Association for the Study of Pain (IASP) regards persistent or recurring pain lasting less than three months as acute pain, whereas more than three months of persistent or recurring pain is considered to be chronic pain [6]. The study by van der Windt [5] showed that, according to this cut-off criterion, 51% of patients with a newly presented episode of SC in general practice develop chronic SC, that is, complaints lasting more than three months.
The Dutch College of General Practitioners provides clinical guidelines for the treatment of SC [7]. These guidelines, however, do not include treatment aimed at psychosocial factors such as maladaptive behaviour and inadequate cognitions, known to play a role in the development and persistence of chronic musculoskeletal diseases [8-10]. Treatments addressing such factors are mentioned in the guidelines, but only as a last resort, when the biomedical approach has proved ineffective in reducing the pain.
To date, early interventions aimed at the psychological and social determinants of SC are not common in general practice, although such interventions in the early stages of the SC might prevent the development of chronic complaints [11].
Early intervention in general practice
We hypothesised that an intervention concentrating on psychological and social determinants in the early stages of SC would prevent the development of inadequate cognitions and maladaptive behaviour, ensuring that such inadequate cognitions and behaviour do not play a role in the development of chronic SC later on.
The term cognitions refers to the way patients think about their pain and what the pain means to them, in terms of thoughts, beliefs, attitudes and self-efficacy expectations [12], whereas behaviour refers to the patients' observable actions [13].
Patients' cognitions and behaviour should thus be influenced so as to become adequate cognitions and adaptive behaviour. A relatively brief treatment in the early stages of SC, administered by a trained therapist, may be expected to be effective in preventing the development of chronic SC. The Education and Activation Programme (EAP) that formed the subject of the present study is such an early intervention.
Aim of the study
This paper describes the design of a study to evaluate the clinical effectiveness of an early EAP aimed at using psychological and behavioural factors to prevent chronic SC. In addition, the study is to evaluate the balance between costs and effects. EAP in patients with acute SC is to be compared with treatment according to the Dutch College of General Practitioners guidelines. The Medical Ethics Committee of the Institute for Rehabilitations Research in association with Rehabilitation Foundation Limburg has approved the design of the study presented here. Funding was obtained from the Netherlands Organisation for Scientific Research.
This paper describes also the rationale and content of the EAP.
Education and activation programme
Previous studies have indicated that cognitive behavioural therapies aimed at bio-psychosocial factors are promising instruments for the prevention of chronic musculoskeletal pain [14]. The EAP focuses on the same elements as cognitive behavioural therapies, but is applied at an earlier stage than such therapies [15]. Whereas the latter focus mainly on the elimination of inadequate cognitions and maladaptive behaviour after they have already developed in the course of the SC, the EAP focuses on guiding the patient towards adequate pain behaviour and reinforcing this behaviour at an early stage of the SC.
The aim of the EAP is to prevent the development of inadequate cognitions and maladaptive behaviour in patients with acute SC.
Education is used to maintain or induce adequate cognitions by providing information that is tailored to questions that patients have about their SC. Health care educators frequently assume that giving information equals comprehension, which should automatically translate into changed behaviours as the knowledge is applied [16]. According to Hussey, however, simply receiving a message hardly correlates with understanding it [17]. Effective learning also requires the active participation of the patient[18].
Since inadequate cognitions and maladaptive behaviours are not yet fully developed in patients with acute SC, the focus of the EAP is not on restructuring inadequate cognitions or modifying maladaptive behaviour, but on maintaining or inducing the proper cognitions by education and on maintaining or inducing adequate behaviour by giving advice on activities of daily living. Adequate behaviour is considered to be behaviour in which the patient remains active. A comprehensive description of the EAP is given in the Design section.
Design
Patients
Patients are recruited by GPs and in the open population by advertising in local newspapers.
Patients are eligible for inclusion in the Randomised Clinical Trial (RCT) if they consult their own GP or respond to adverts in a local newspaper with a new episode of SC that has lasted no longer than three months, at rest or when elicited by movement in the shoulder area.
Patients are included if they are 18 years or older and living in the south of the Netherlands. Only newly presented episodes of SC are considered, that is, patients who have not consulted their GP and have not been treated for their SC in the preceding three months. Additional exclusion criteria are given in table 1.
Table 1 Exclusion criteria
• other episodes of SC in the 12 months preceding the consultation with the GP
• prior fractures and/or surgery of the shoulder
• (suspected) referred pain from internal organs
• SC with a confirmed extrinsic cause
• inability to complete a questionnaire independently
• presence of dementia or other severe psychiatric abnormalities
Randomised Clinical Trial
A Randomised Clinical Trial (RCT) with a six-month follow-up is used to evaluate the effectiveness and cost-effectiveness of an EAP to prevent chronicity in patients with acute SC, compared to usual care.
A computer-generated random sequence table is used to randomise the patients to EAP or usual care. Neither the patient nor the GP, nor the trained therapist, can be blinded for the allocated treatment. The trained therapist is also the researcher coordinating the RCT and conducting the data analysis, but is blinded for treatment allocation during the data analysis. The allocation code will be revealed only after the data analysis has been completed.
Treatments
Usual care (UC) is applied according to the Dutch College of General Practitioners guidelines for SC (version 1999)[19]. Management during the first two weeks consists of a wait-and-see policy with information and advice about shoulder complaints, possibly supplemented with analgesics or nonsteroidal anti-inflammatory drugs. If this approach has little or no effect, up to three corticosteroid injections can be given. Physiotherapy is considered for complaints persisting after six weeks or more. If the SC persist, referral to a hospital-based specialist may be considered.
The focus of the EAP is to maintain or induce the proper cognitions by education and to stimulate adequate behaviour by means of advice on activities of daily living. Table 2 shows the components of the EAP. The EAP is administered by specially trained GPs or an ambulant therapist (CDB) trained to provide the EAP. The ambulant therapist administers the EAP when no trained GP is available in the living area of the patient.
Table 2 Elements of the education and activation programme
Education • Information on the origin, nature and prognosis of the SC
• Information on possible interventions and their effects (tailored to the patient's questions and needs)
• Information on the effect of cognitions and behaviour on the perpetuation of the SC
Activation When no alterations in activities have occurred due to the SC
• Positive reinforcement
• Instruction to be aware of possible changes
When alterations in activities have occurred due to the SC
• Identification of up to three altered frequent activities of daily living
• Determination of the desired level of activity and the size of the steps needed to reach this level
The EAP consists of a minimum of two sessions and a maximum of six follow-up sessions over a period of six weeks. Each session may last up to 20 minutes. The first and second sessions are organised in the general practice setting by the trained GP, or at the patient's home by the ambulant EAP therapist. The other sessions are provided by telephone.
Education
The first part of the EAP has an educational purpose, and focuses on information about the origin, nature and prognosis of the SC, possible interventions and their effects, the impact on activities of daily living and its consequences and the patient's own possibility to contribute to recovery. This information is tailored to the patients' questions and needs and is based on the information available in the Dutch College of General Practitioners guidelines for SC.
In addition, the effect of cognitions and behaviour on the perpetuation of the SC is clarified to the patient by an example. If possible, this example refers to a condition or circumstance the patient has experienced, such as a broken bone or back pain. The patient is helped by the trained GP or the trained therapist to explore whether his or her thoughts about the SC are justified. Negative patterns of thinking are modified into adequate and accurate thoughts.
Activation
The second part of the EAP consists of a time contingent activation programme, based on the principles of operant learning. It focuses on gradually increasing activities of daily living, despite the pain.
Potential avoidance of activities is countered by reinforcement of continuation or resumption of usual activities.
Positive reinforcement is used to stimulate patients with a normal activity pattern, in spite of their SC, to continue their activities. This positive reinforcement may be enough to achieve continuation of the desired activities [13]. These patients are also instructed to be aware of possible changes in their activities that could lead to undesirable behaviour such as reduced use of the affected shoulder.
Patients who have reduced their normal activities are helped to identify up to three frequent activities of daily living that they have reduced as a result of the SC. These activities are stepwise gradually increased to the desired level of activity in a time-contingent manner. The desired level of activity and the magnitude of the increases are determined and agreed upon by the EAP therapist and the patient.
The patient and the EAP therapist also plan a progress evaluation, which is used to positively reinforce the patient's behaviour if the gradual increase has been correctly implemented or to adjust the magnitude of the increases if the original objectives prove too optimistic.
Measurements
The first outcome measure is the perceived recovery of the patient. Patients are considered to be recovered when they report to be much improved or fully recovered, on an 7-point ordinal scale, after six months.
The second outcome measure is that of functional limitations in activities of daily living. This variable is assessed by a 16-item questionnaire, the shoulder disability questionnaire (SDQ)[20], with a scoring range of 0 to 16. A reduction of the score on this questionnaire implies a reduction in functional limitations. The outcome measures are recorded at 6, 12 and 26 weeks after randomisation. The SDQ is also measured at baseline. A cost diary [21] is used to assess health care utilisation, direct non-medical costs and indirect costs. A complete overview of baseline measures, prognostic measures, process measures and outcome measures is given in table 3.
Table 3 Variables
Baseline measures T = 0 Demographic variables
• Age
• Gender
• Employment status
Specific disease characteristics
• Affected side
• Possible cause of shoulder complaints
• Duration of complaints
• History of shoulder complaints
Co-morbidity
Physical activity
Workload
Treatment credibility and preference
Prognostic measures T = 0 Mobility of glenohumeral joint
• HIB (hand in back), HIN (hand in neck), passive exorotation
• Active and passive abduction
Mobility of cervicothoracal spine
Severity of main complaint
Psychosocial variables
• Anxiety1
• Depression1
• Somatisation1
• Distress1
Job content
Outcome measures T = 1,2,3 Perceived recovery of complaints Functional limitations to daily activities2
Process measures T = 0,1,2,3 Psychosocial variables
• Kinesiophobia3
• Fear avoidance and beliefs4
• Catastrophising5
• Coping with pain5
• Internal locus of control5
• External locus of control5
Global assessment
Shoulder pain6
General health7
Cost T = 0–26 weeks Health care utilisation8
Direct non-medical costs
Indirect costs
1 four-dimensional complaint list [25]
2 Shoulder Disability Questionnaire [20]
3 Tampa Scale for Kinesiophobia – Dutch version (partly) [26]
4 Fear Avoidance and Beliefs Questionnaire – Dutch version (partly) [27]
5 Pain Coping and Cognition List [28]
6 Shoulder Pain Score [29]
7 Generic Health Related Quality of Life [30]
8 Cost Diary [21]
Data analysis
The statistical analysis will be carried out according to the 'intention-to-treat' principle. Differences between groups, with 95% confidence intervals, will be calculated for each outcome measure. The study groups will be compared by an independent samples t-test for changes since baseline for continuous outcome variables and the chi-square test for categorical outcome variables. In addition, the corresponding baseline value for each continuous outcome will be used as a covariate.
The analysis will be repeated taking any loss-to-follow-up into account by applying a sensitivity analysis in which all patients who are lost to follow-up are first considered to show the largest observed improvement and then the largest observed deterioration in outcome measures.
The analyses of the difference in change for the outcomes at three and six months will account for the repeated measures character of the data. Baseline characteristics that are a priori considered to be possible prognostic factors for outcome variables, as well as post-randomisation differences between the groups, will be handled as potential confounders. Their influence will be evaluated by means of multivariable regression analyses. In the case of confounding, adjusted effect estimates will also be reported.
Sample size
About half of all newly presented episodes of SC in general practice are reported to last for at least six months. A number needed to treat of 4.5 after six months is considered clinically relevant. This implies an absolute reduction of 22% of the proportion of patients with SC after six months. With a two-sided alpha of 0.05 and a statistical power (1-β) of 0.80, 70 patients per treatment group are needed to detect a difference in favour of the EAP compared to usual care after six months.
Embedding in the Dutch Shoulder Disability Study
This RCT is part of the Dutch Shoulder Disability Study, a comprehensive prognostic cohort study on SC, with randomised controlled interventions in subcohorts. The Dutch Shoulder Disability Study is funded by the Netherlands Organisation for Scientific Research (NWO, grant number 904-65-901).
Discussion
Reasons for publishing a study design
There are several reasons to publish a study design before the results are available. The main reason is that it provides an opportunity to counteract publication bias, that is, the phenomenon whereby a study producing positive results is more likely to be published than a study showing no difference between the study groups [22], [23]. Hence, if the design is published but not the results, the study can still be included in a systematic review because data can be retrieved from the researcher [24].
Another reason is that it gives researchers the opportunity to reflect upon the study design independently of the results. When results run counter to the researchers' expectations, methodological flaws are usually examined. But when the results are in line with expectations, methodological flaws are more likely to be overlooked. [22]
The third reason arises from the tendency among randomised controlled studies to deviate from their original designs, mainly because of practical problems. Such deviations from the study design may affect the study results. Publishing the study design forces researchers to test its implementation and to answer for any deviations from the design.
Finally, this article offers us an opportunity to describe the rationale and content of the intervention in greater detail than the methods section of an article reporting the results of the RCT would do [24].
Applicability in general practice
The EAP is a brief intervention that can easily be administered by GPs in addition to the usual care according to the guidelines. This might give GPs an instrument to prevent the development of chronic SC in the early stages of the complaints by focusing on psychological and social determinants.
Time schedule
The inclusion of patients in the study lasted until December 31st 2003. Data collection will be completed in June 2004. Currently, 108 patients have been included and are being followed up.
Abbreviations
SC: Shoulder Complaints
GP: General Practitioner
IASP: International Association for the Study of Pain
EAP: Education and Activation Programme
RCT: Randomised Clinical Trial
UC: Usual Care
CDB: Camiel De Bruijn
SDQ: Shoulder Disability Questionnaire
NWO: Netherlands Organisation for Scientific Research
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CDB participated in the design of the study, will provide the EAP as an ambulant therapist, coordinate the data collection, will perform the statistical analysis and publish the results. RDB participated in the design of the study and will participate in the statistical analysis. JG, MG, WVDH and G-JD also participated in the design of the study. AK participated in the development of the education and activation programme and will conduct the training of the general practitioners that will give the education and activation programme. GVDH conceived of the study, and participated in its design. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Sobel JS Winters JC Shoulder complaints in general practice 1996 , Rijksuniversiteit Groningen
Picavet HSJ Gils HWV Schouten JSAG Klachten Van Het Bewegingsapparaat In De Nederlandse Bevolking 2000 Bilthoven, Rijksinstituut voor Volksgezondheid en Milieu
Picavet HSJ Schouten JSAG Musculoskeletal Pain in the Netherlands: Prevalences, Consequences and Risk Groups, the DMC(3)-study Pain 2003 102 167 178 12620608 10.1016/s0304-3959(02)00372-x
Bongers PM The Cost of Shoulder Pain at Work British Medical Journal 2001 322 64 65 11154606
Windt DA Koes BW Boeke AJ Deville W Jong BAD Bouter LM Shoulder disorders in general practice: prognostic indicators of outcome Br J Gen Pract 1996 46 519 523 8917870
Merskey H Bogduk N Merskey H and Bogduk N IASP Pain Terminology Classification of Chronic Pain 1994 Seattle, IASP Press 209 214
Winters JC Jongh ACD D. A. W. M. van der Windt Jonquiere M Winter AFD G. J. M. G. van der Heijden Sobel JS Goudswaard AN NHG-Standaard Schouderklachten Huisarts en Wetenschap 1999 42 222 231
Linton S A Systematic Review of Psychological Risk Factors for Back and Neck Pain Spine 2000 25 1148 1156 10788861 10.1097/00007632-200005010-00017
Turk DC Jensen TS, Turner JA and Z. WH The Role of Demographic and Psychosocial Factors in Transition from Acute to Chronic Pain 1997 8 IASP Press 185 213
Weiser S Cedraschi C Psychosocial Issues In The Prevention Of Chronic Low Back Pain - A Literature Review Bailliere's Clinical Rheumatology 1992 6 657 684 1477896
Linton SJ Gatchel RJ and Turk DC Prevention With Special Reference To Chronic Musculoskeletal Disorders Psychosocial Factors In Pain 1999 New York, Guilford Publications 374 389
Wit R Dam F Litjens M H. HAS Assessment of Pain Cognitions in Cancer Patients with Chronic Pain Journal of Pain and Symptom Management 2001 22 911 924 11728794 10.1016/S0885-3924(01)00354-2
Fordyce WE Behavioral Methods For Chronic Pain And Illness 1976 Saint Louis, C. V. Mosby Company
Turner JA Keefe FJ Max M Cognitive-Behavioral Therapy For Chronic Pain Pain; an updated review 1999 523 533
Davis P Busch AJ Lowe JC Taniguchi J Djkowich B Evaluation of a Rheumatoid Arthritis Patient Education Program: Impact on Knowledge and Self-Efficacy Patient Education And Counseling 1994 24 55 61 7862595 10.1016/0738-3991(94)90025-6
Hussey LC Strategies for Effective Patient Education Material Design Journal of Cardiovascular Nursing 1997 11 37 46 8982880
Padberg RM Padberg LF Strengthening the Effectiveness of Patient Education: Applying Principles of Adult Education Oncol Nurs Forum 1990 17 65 69 2300506
G. J. M. G. van der Heijden Leffers P Bouter LM Development And Responsiveness Of The Shoulder Disability Questionnaire J Clin Epidemiol 1999
Goossens MEJB M. P. M. H. Rutten - van Molken Vlaeyen JWS S. M. J. P. van der Linden The Cost Diary: A Method To Measure Direct And Indirect Costs In Cost-Effectiveness Research Journal of Clinical Epidemiology 2000 53 688 695 10941945 10.1016/S0895-4356(99)00177-8
Dickersin K The existence of publication bias and risk factors for its occurrence JAMA 1990 263 1385 1389 2406472 10.1001/jama.263.10.1385
Eastbrook PJ Berlin JA Gopalan R Matthews DR Publication Bias in Clinical Research Lancet 1991 337 867 872 1672966 10.1016/0140-6736(91)90201-Y
Ostelo RWJG Koke AJA Beurskens AJHM Vet HCW Kerckhoffs MR Vlaeyen JWS Wolters PMJC Berfelo WM Brandt PA Behavioral-graded Activity Compared with Usual Care after First-time Disk Surgery: Considerations of The Design of a Randomized Clinical Trial J Manipulative Physiol Ther 2000 23 312 319 10863250
| 15715914 | PMC551606 | CC BY | 2021-01-04 16:29:12 | no | BMC Fam Pract. 2005 Feb 16; 6:7 | utf-8 | BMC Fam Pract | 2,005 | 10.1186/1471-2296-6-7 | oa_comm |
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-81570750410.1186/1471-2334-5-8Research ArticleDetection of virulence genes in Malaysian Shigella species by multiplex PCR assay Thong Kwai Lin [email protected] Susan Ling Ling [email protected] SD [email protected] Yasin Rohani [email protected] Institute of Biological Sciences, Faculty of Science, University of Malaya, 50603 Kuala Lumpur, Malaysia2 Department of Medical Microbiology, Faculty of Medicine, University of Malaya, 50603 Kuala Lumpur, Malaysia3 Institute for Medical Research, Jalan Pahang, 50588 Kuala Lumpur, Malaysia2005 14 2 2005 5 8 8 18 8 2004 14 2 2005 Copyright © 2005 Thong et al; licensee BioMed Central Ltd.2005Thong 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 Malaysia, Shigella spp. was reported to be the third commonest bacterial agent responsible for childhood diarrhoea. Currently, isolation of the bacterium and confirmation of the disease by microbiological and biochemical methods remain as the "gold standard". This study aimed to detect the prevalence of four Shigella virulence genes present concurrently, in randomly selected Malaysian strains via a rapid multiplex PCR (mPCR) assay.
Methods
A mPCR assay was designed for the simultaneous detection of chromosomal- and plasmid-encoded virulence genes (set1A, set1B, ial and ipaH) in Shigella spp. One hundred and ten Malaysian strains (1997–2000) isolated from patients from various government hospitals were used. Reproducibility and sensitivity of the assay were also evaluated. Applicability of the mPCR in clinical settings was tested with spiked faeces following preincubation in brain heart infusion (BHI) broth.
Results
The ipaH sequence was present in all the strains, while each of the set1A, set1B and ial gene was present in 40% of the strains tested. Reproducibility of the mPCR assay was 100% and none of the non-Shigella pathogens tested in this study were amplified. The mPCR could detect 100 colony-forming units (cfu) of shigellae per reaction mixture in spiked faeces following preincubation.
Conclusions
The mPCR system is reproducible, sensitive and is able to identify pathogenic strains of shigellae irrespective of the locality of the virulence genes. It can be easily performed with a high throughput to give a presumptive identification of the causal pathogen.
==== Body
Background
Members of the genus Shigella, namely S. flexneri, S. dysenteriae, S. sonnei and S. boydii have caused and continue to be responsible for mortality and/or morbidity in high risk populations such as children under five years of age, senior citizens, toddlers in day-care centres, patients in custodial institutions, homosexual men and, war- and famine-engulfed people. Yearly episodes of shigellosis globally have been estimated to be 164.7 million and of these, 163.2 million were in developing countries and the remaining in industrialized nations. The mortality rate was approximately 0.7% [1]. A recent study by Lee & Puthucheary [2] on bacterial enteropathogens in childhood diarrhoea in a Malaysian urban hospital showed that Shigella spp. was the third most common bacteria isolated. S. flexneri and S. dysenteriae type 1 infections are usually characterized by frequent passage of small amounts of stool and mucus or blood. At times, watery stool followed by typical dysenteric stool maybe present with S. dysenteriae type 1 infection. S. sonnei and S. boydii infections are less severe with watery faeces but little mucus or blood.
Shigellosis is usually a self-limiting infection, however when it subsides, the intestinal ulcers heal with scar tissue formation. Uncomplicated recovery is usual and the organisms rarely cause other types of infections. Adversely, in 3 to 50% of cases, depending on the virulence of the strain, the nutritional and immune status of the host, the initial infection maybe followed by neurological complications or kidney failure. Serious complications do occur at greatest frequencies in malnourished infants, toddlers, older adults and immunocompromised individuals [3,4].
Virulence genes responsible for the pathogenesis of shigellosis may be located in the chromosome or on the inv plasmid borne by the organism. They are often multifactorial and coordinately regulated, and the genes tend to be clustered in the genome. Previously reported PCR-based detection methods concentrated mainly on the ipaH gene alone [5,6] or on ipaH and ial genes in two separate PCR assays [7,8]. As ial is found on the large inv plasmid which is prone to loss or deletions, this gene-based detection may give false negative results. ipaH, on the other hand, is present on both the Shigella chromosome and on a large plasmid and hence, it is a more stable gene to detect. However, the sole presence of ipaH is not an absolute indicator of virulence as loss or deletion of the plasmid renders the bacterium noninvasive and therefore, avirulent. set1A and set1B are chromosomal genes encoding Shigella enterotoxin 1 (ShET1), which cause the watery phase of diarrhoea in shigellosis [9,10]. ial and ipaH are responsible for directing epithelial cell penetration by the bacterium and for the modification of host response to infection, respectively [11-13].
Here, we describe the application of a multiplex PCR (mPCR) design for simultaneous detection of four virulence genes (set1A, set1B, ial and ipaH) in Shigella spp. and to determine the prevalence of these virulence genes in a random selection of Malaysian Shigella strains.
Methods
Bacterial strains and growth conditions
A total of 110 Shigella strains of S. flexneri (n = 84), S. sonnei (n = 15), S. dysenteriae (n = 10) and S. boydii (n = 1) were used in this study. These strains were isolated from patients with diarrhoea in Peninsular Malaysia from 1997–2000, and were provided by the Institute for Medical Research (IMR), Malaysia. Serotyping of the strains (Shigella antisera from Mast Diagnostics, UK) was carried out by the Bacteriological Unit, IMR. All the strains were checked on Salmonella-Shigella (SS) agar before being transferred to Luria Bertani (LB) agar plate, incubated overnight at 37°C for subsequent screening of virulence-associated genes. All strains were stored at -20°C in LB broth containing 15% glycerol.
Development of mPCR
Boiled suspension of bacterial cells was used as DNA template. Previously described primers, obtained from Integrated DNA Techs, USA, for detection of the four virulence genes were applied to the template [8,14,15] (Table 1). Prior to combining all the four primer sets in an mPCR, each pair of primers was optimized singly in separate PCR assays. A typical 25-μl PCR reaction mixture for every primer set consisted of 1x PCR buffer B (Promega, USA), 4 mM MgCl2, 130 μM of each deoxynucleotide (dNTP), 0.5 μM of each primer, 1 U of Taq DNA polymerase (Promega, USA) and 2 μl of DNA template. Amplifications were carried out using a Robocycler Gradient 40 Temperature Cycler (Strategene Cloning Systems, USA). The cycling conditions were an initial denaturation at 95°C for 5 min, template denaturation at 95°C for 50 s, annealing at 55°C for 1.5 min, and extension at 72°C for 2 min for a total of 30 cycles, with a final extension at 72°C for 7 min.
Table 1 Primers used to identify various virulence-associated genes of Shigella spp.
Primer Virulence gene Nucleotide sequences (5' → 3') Size of amplicon (bp) Reference
ShET1A set1A TCA CGC TAC CAT CAA
AGA
TAT CCC CCT TTG GTG
GTA 309 14
ShET1B set1B GTG AAC CTG CTG CCG
ATA TC
ATT TGT GGA TAA AAA
TGA CG 147 14
ial ial CTG GAT GGT ATG GTG
AGG
GGA GGC CAA CAA TTA
TTT CC 320 15
Shig1 ipaH TGG AAA AAC TCA GTG
CCT CT 423 8
Shig2 CCA GTC CGT AAA TTC
ATT CT
Based on the results of individual priming, an mPCR was designed. Various parameters such as concentrations of primers (0.5–0.8 μM), MgCl2 (2 to 4 μM), Taq DNA polymerase (0.6 to 4 U) and dNTPs (100–150 μM) and buffer strength (1.4X to 2.4X) were tested. The simultaneous gene amplifications were performed in a reaction volume of 25 μl consisting of 1.8X PCR buffer B (Promega, USA), 4 mM MgCl2, 130 μM of each dNTP, 0.3 μM of each ShET1B primer, Shig1 and Shig2 primers, 0.5 μM of each ShET1A and ial primers, 1 U of Taq DNA polymerase (Promega, USA) and 2 μl of DNA template. All the reaction mixtures were overlaid with 20 μl of sterile mineral oil. Amplifications were similarly carried out as above.
After initial screening, strain TH13/00 (S. flexneri 2a) was chosen as a positive control for PCR assays. A negative control using sterile distilled water as template was included in every PCR assay. The DNA fragments were separated in 2% agarose gel.
Reproducibility test
The mPCR assay was repeated at least twice with 28 strains to determine the reproducibility of the results, whereby the DNA template of a particular strain was freshly prepared for each repeat.
Specificity test
The specificity of the mPCR assay was tested with 12 other non-Shigella pathogens: Enterobacter cloacae, Salmonella Paratyphi A (ATCC 9281), S. Paratyphi C, S. Typhimurium, S. Enteritidis, S. Typhi (ATCC 7251), Listeria monocytogenes, Pseudomonas aeruginosa, Klebsiella pneumoniae, Citrobacter freundii, Escherichia coli O157:H7 and E. coli O78:H11.
Faecal spiking and sensitivity test
This was based on a modification of that described by Chiu and Ou [16]. Approximately 0.2 g of faeces from a healthy individual was suspended in 1 ml of brain heart infusion (BHI) (Oxoid Ltd., UK) and diluted 10-fold. Then, 1 ml of the diluted faecal suspension was inoculated into 4 ml of BHI and vortexed to obtain a homogenous mixture of broth-faecal suspension. Meanwhile, an overnight culture of S. flexneri 2a was harvested and serially diluted 10-fold with BHI. Then, 250 μl of each dilution of cell culture was mixed with 250 μl of the broth-faecal suspension and 500 μl of BHI in a new eppendorf tube. The tubes were vortexed and preincubated at 37°C for 4 h without shaking. Simultaneously, 100 μl of each diluted culture was plated on LB agar (Oxoid Ltd., UK) to determine the number of viable bacteria in each dilution. After preincubation, mPCR assay was performed on the boiled lysates of each diluted culture. A pure culture of S. flexneri 2a (TH13/00) and an unspiked faecal sample served as positive and negative controls.. The test was repeated with a spiked faecal sample of another healthy individual, and the average detection limit was reported.
Screening of clinical specimens
0.2 g of each faecal sample from 10 patients suffering from diarrhoea in a local tertiary University Hospital was suspended in 1 ml of BHI and diluted 10-fold. A volume of 250 μl of broth-faecal suspension was inoculated into 5 ml of BHI and preincubated at 37°C for 4 h without shaking. Concurrently, 100 μl of the suspension was plated onto MacConkey and SS agar plates and incubated overnight at 37°C. mPCR assay was performed on the boiled lysate of the broth-faecal suspension after preincubation. A pure culture of strain TH13/00 and a Shigella-spiked faecal sample served as a positive control, whilst a PCR reaction mixture without bacterial DNA template and an unspiked faecal sample from a healthy individual acted as a negative control.
Results
Optimization strategies
A monoplex PCR for each primer set was initially carried out based on a published report [17]. Although the concentrations of MgCl2 (3 mM), dNTP (400 μM each) and primers (1 μM each) were used as recommended, unspecific bands were present together with intense primer-dimers. In order to reduce the background noise and primer-dimers, concentrations of 0.5 μM of each primer and 200 μM of each dNTP were used. Further optimizations of MgCl2 concentrations (2 to 4 μM) and dNTP (100,130 and 150 μM each) gave intense amplicons with a clean background in each monoplex amplification (Fig 1, lanes 1–4).
Figure 1 Ethidium bromide-stained agarose gel showing PCR products. Lane M, 100-bp DNA ladder (Promega); lane 1, set1B gene product; lane 2, set1A gene product; lane 3, ial gene product; lane 4, ipaH gene product; lane 5, mPCR product.
Initial attempts to amplify equally all the four genes in a single reaction using the reaction condition in monoplex PCR were not successful. A common practice in mPCRs involving any non-amplification of a required gene ('weak locus') is to increase the amount of primers of the gene at same time with a decrease of the amount of primers for all the loci that can be amplified, especially those with strong amplifications. Hence, the concentrations of primers for both ipaH (Shig) and set1B (ShET1B) were reduced to 0.3 μM each and the primers for both set1A (ShET1A) and ial (ial) genes were maintained at 0.5 μM each. Following optimization of the concentrations of Taq DNA polymerase (0.6 to 4U/25 μL), buffer strength (1.4 X to 2.4 X), dNTPs (140 to 220 μM) and annealing temperatures (49 to 59°C) (at a constant MgCl2 concentration of 4 mM), a more uniform amplification of all the genes with no background noise was obtained (Fig. 1 lane 5) at a final buffer concentration of 1.8X, 1 U Taq DNA polymerase, 130 μM dNTP each and annealing temperature of 55°C.
Prevalence of virulence genes in the Malaysian strains
All the 110 strains of Shigella spp. tested showed the presence of ipaH (Table 2). Conversely, only 41% of the strains had both set1A and set1B genes, and ial gene. Almost all the Shigella strains tested positive for the tandem genes (87%) belonged to S. flexneri 2a serotype. Among the predominant strains of Shigella flexneri in Malaysia, ial was found in serotypes 4a, 6, 3a, 2a, 1a and 3c. All the four genes were present only in S. flexneri 2a and 3a.
Table 2 Prevalence of the four virulence-associated genes in Malaysian Shigella spp.
Serotype Total strains set1B (%) set1A (%) ial (%) ipaH (%)
S. flexneri 1a 3 0 (0) 0 (0) 1 (33) 3 (100)
1b 3 0 (0) 0 (0) 0 (0) 3 (100)
2a 47 41 (87) 41 (87) 19 (40) 47 (100)
3a 18 3 (17) 3 (17) 12 (67) 18 (100)
3c 10 0 (0) 0 (0) 1 (10) 10 (100)
4a 1 1 (100) 1 (100) 1 (100) 1 (100)
6 1 0 (0) 0 (0) 1 (100) 1 (100)
y 1 0 (0) 0 (0) 0 (0) 1 (100)
S. sonnei 15 0 (0) 0 (0) 2 (13) 15 (100)
S. dysenteriae 2 10 0 (0) 0 (0) 8 (80) 10 (100)
S. boydii 6 1 0 (0) 0 (0) 0 (0) 1 (100)
Total 110 45 45 45 110
Reproducibility
Reproducibility for the detection ofset1A, set1B, ial and ipaH genes assayed in the mPCR was 100%. None of the non-Shigella strains tested gave any amplification (data not shown).
Sensitivity
The mPCR assay was tested on 10-fold dilutions of an overnight culture of S. flexneri 2a. All the four virulence-associated genes were detected until 10-3 dilutions (data not shown). This was equivalent to 2.45 × 105 lysate or a minimum of 490 cfu of shigellae per 25 -μLmPCR reaction.
Faecal spiking and sensitivity
An initial experiment using undiluted spiked faecal sample failed to give any PCR amplification (data not shown). When the faecal suspension was diluted and preincubated in BHI for 4 h, the mPCR assay was successful in detecting the presence of the four virulence genes at an average concentration of 5.0 × 104 colony-forming units (cfu) shigellae ml-1 or approximately 100 cfu per reaction mixture (Fig. 2 lane 8).
Figure 2 Faecal-spiking and sensitivity result of mPCR. Lane M, 100-bp DNA ladder (Promega, USA); lane 1, TH13/00 (positive control); lane 2, unspiked faeces (negative control); lane 3, undiluted spiked faeces; lane 4, 10-1 dilution; lane 5, 10-2 dilution; lane 6, 10-3 dilution; lane 7, 10-4 dilution; lane 8, 10-5 dilution; lane 9, 10-6 dilution; lane 10, 10-7 dilution; lane 11, 10-8 dilution; lane 12, 10-9 dilution; lane 13, 10-10 dilution; lane 14, "water blank".
Clinical specimens
A preliminary study on the efficacy of the mPCR assay in the direct detection of the aforementioned Shigella virulence genes on faecal samples was tested on ten diarrhoeal patients. No mPCR product was detected although both the positive controls had amplifications. By conventional culture method, there was no growth of Shigella on the LB, MacConkey and SS agar plates.
Discussion
Numerous studies had been performed to detect virulence genes in Shigella by monoplex PCRs [8,17,18]. Studies involving the combination of chromosomal- and plasmid-encoded virulence genes in a single assay for Shigella detection, on the other hand, are scarce. Although the optimization of mPCR is more tedious and difficult to achieve than monoplexes, the ease of screening a large number of specimens, once the system is optimized, far outweighs the initial problems. The present mPCR system encompasses the presence of virulence genes found in the Shigella chromosome and on the large inv plasmid. Hence, it can determine if the pathogenesis of a particular strain is attributable to its chromosome or the plasmid, or if the strain is still invasive or otherwise, in a single reaction.
Initially, the monoplex PCRs were carried out following reaction conditions as proposed by a previous report [17]. However, we could not reproduce their results and hence had to modify the PCR conditions. Our failure to reproduce identical results despite using similar reagent concentrations and amplification conditions maybe attributed to the different makes of PCR reagents and primers used. Broude et al. [19] had compared amplification efficiencies of two commercial Taq DNA polymerases and found that they displayed different specificity in PCR.
Preferential amplification of one target sequence over another is a known phenomenon in mPCRs and it is usually overcome by increasing the amount of primers for the weaker amplification simultaneously with a decrease of primer concentrations for the stronger amplification. Buffer concentration may also affect mPCR amplifications despite it being seldom considered during monoplex optimization works [20]. Upon adjustment of primers and buffer concentrations, specific and consistent amplification of all the genes in the multiplex combination was achieved.
Although other studies have demonstrated the presence of ial and ipaH in strains of enteroinvasive Escherichia coli (EIEC) [11,13,21], we had not applied the mPCR assay to EIEC strains. It is unfortunate that these strains were not available for our study as EIEC gives rise to similar illness as Shigellosis.
Our study supported the observations of Noriega et al. [9] and Vargas et al. [17] in local Shigella strains. Their studies showed that both set1A and set1B were present exclusively in S. flexneri 2a. The complete correlation between the presence of both set1A and set1B showed that both genes are indeed found in tandem in the Shigella genome. In this study, almost all the Shigella strains positive for the presence of set1A and set1B (41/45 strains) belonged to S. flexneri 2a, thus confirming previous works that both genes are highly conserved in this particular serotype [14].
Both the prevalence of ial and ipaH were independent of the four different species of Shigella tested. Though both ial and ipaH are responsible for invasion-related processes and are found on the inv plasmid, the ial gene cluster resides near a region of the plasmid, which is a hot spot for spontaneous deletions [22]. This probably explains the lower prevalence of ial (45/110 strains) than ipaH (110/110 strains) in the Malaysian Shigella strains. Since invasiveness is a prerequisite for virulence in shigellae and since most of these virulence genes are located on the large plasmid, these strains would have possessed the plasmid when first isolated from patients. Due to storage/subculturing, the plasmid might have been lost together with the virulence-associated genes. By virtue of multiple copies being present on both the chromosome and the inv plasmid [23], ipaH seemed to be less compromised by plasmid loss and/or deletions. As the sole presence of ipaH is not indicative of the invasive phenotype, our mPCR design, which incorporated three other virulence genes, could determine the invasiveness of Shigella strains in epidemiological studies.
Dilution of the faecal sample with BHI was performed to lower the levels of PCR inhibitors such as bilirubin, bile salts and heme in the faeces [16]. An additional step of preincubating the spiked faecal samples also helped to eliminate the natural inhibitors [24]. The short 4-h enrichment step would increase the total number of target sequences caused by more bacterial growth and the overall detection sensitivity of the assay. Although PCR cannot differentiate between dead and viable bacteria, enrichment helped to dilute the concentrations of dead bacteria, thus reducing the probability of detecting them by the subsequent mPCR assay. The sensitivity level achieved in the study was found to be comparable to other studies. Houng et al. [25] detected up to 7.4 × 104 cfu shigellae ml-1 by amplifying the IS 630 sequences in shigella spp.. Yavzori et al. [24] reported a detection level of 104 cfu shigellae per gram of faeces with the use of virF primers. Although it has been reported that ingestion as low as 100 shigellae resulted in clinical disease [26], the highest percentage of volunteers having diarrhoea were administered doses of at least 104 viable organisms. Thus, the average detection limit of mPCR described in this study (5.0 × 104 cfu/ml) is within the common infectious dose for shigellae.
Results from the preliminary clinical screening were promising. Nevertheless, the consideration of other diarrhoeal pathogens being present in the clinical samples cannot be negated. More patient samples are warranted to thoroughly vet the robustness and applicability of the developed mPCR in clinical environments.
One limitation of the present mPCR system is its inability to differentiate Shigella spp., unlike the multiplex reactions based on specific rfc genes developed by Houng et al. [25]. For future research, either set1A or set1B may be omitted from the multiplex system as both genes are shown to exist tandemly. rfc primers of different Shigella origins maybe incorporated to enable the discrimination of Shigella spp. as well as the identification of virulent strains in one assay.
Conclusions
We conclude that the mPCR system is able to identify pathogenic strains of shigellae irrespective of the locality of the virulence genes. The described assay is reproducible, sensitive, can be easily performed and is able to give a presumptive identification of the causal pathogen, which could be confirmed by culture techniques using selective media. An added advantage would be that EIEC, which gives a similar illness, might also be detected by this method, as EIEC also harbours ial and ipaH genes.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SLLH carried out the experiments, data analysis and wrote the manuscript. RMY provided the bacterial strains. SDP contributed to the writing of the manuscript. TKL conceived and co-designed the study, provided input for writing and supervision of the study. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
IRPA grant (06-02-03-1007) from the Ministry of Science, Technology and Environment, Malaysia and Vote F (F0144/2002B) from the University of Malaya, Malaysia, supported this work. We are grateful to Mr. Koh Yin Tee (IMR) for serotyping the bacterial strains.
==== Refs
Kotloff KL Winickoff JP Ivanoff B Clemens JD Swerdlow DL Sansonetti PJ Adak GK Levine MM Global burden of Shigella infections: implications for vaccine development and implementation of control strategies Bull World Health Organ 1999 77 651 666 10516787
Lee WS Puthucheary SD Bacterial enteropathogens isolated in childhood diarrhoea in Kuala Lumpur – the changing trend Med J Malaysia 2002 57 24 30 14569714
Salyers AA Whitt DD Bacterial pathogenesis: a molecular approach 1994 Washington, DC, ASM Press
Khan AM Rabbani GH Faruque ASG Fuchs GJ WHO-ORS in treatment of shigellosis J Diarrhoeal Dis Res 1999 17 88 89 10897893
Gaudio PA Sethabutr O Echeverria P Hoge CW Utility of a polymerase chain reaction diagnostic system in a study of the epidemiology of shigellosis among dysentery patients, family contacts and well controls living in a shigellosis-endemic area J Infect Dis 1997 176 1013 1018 9333160
Dutta S Chatterjee A Dutta P Rajendran K Roy S Pramanik KC Bhattacharya SK Sensitivity and performance characteristics of a direct PCR with stool samples in comparison to conventional techniques for diagnosis of Shigella and enteroinvasive Escherichia coli infection in children with acute diarrhoea in Calcutta, India J Med Microbiol 2001 50 667 674 11478669
Sethabutr O Venkatesan M Murphy GS Eampokalap B Hoge CW Echeverria P Detection of shigellae and enteroinvasive Escherichia coli by amplification of the invasive plasmid antigen H DNA sequence in patients with dysentery J Infect Dis 1993 167 458 461 8421181
Lüscher D Altwegg M Detection of shigellae, enteroinvasive and enterotoxigenic Escherichia coli using the polymerase chain reaction (PCR) in patients returning from tropical countries Mol Cell Probes 1994 8 285 290 7870070 10.1006/mcpr.1994.1040
Noriega FR Liao FM Formal SB Fasano A Levine MM Prevalence of Shigella enterotoxin 1 among Shigella clinical isolates of diverse serotypes J Infect Dis 1995 172 1408 1410 7594690
Rhee SJ Wilson KT Gobert AP Nataro JP Fasano A The enterotoxic activity of Shigella enterotoxin 1 (ShET1) is mediated by inducible nitric oxide synthase activity J Pediatr Gastroenterol Nutr 2001 33 400 416 10.1097/00005176-200109000-00038
Frankel G Riley L Giron JA Valmassoi J Friedmann A Strockbine N Falkow S Schoolnik GK Detection of Shigella in feces using DNA amplification J Infect Dis 1990 161 1252 1256 2189008
Ménard R Prévost MC Gounon P Sansonetti P Dehio C The secreted Ipa complex of Shigella flexneri promotes entry into mammalian cells Proc Natl Acad Sci USA 1996 93 1254 1258 8577750 10.1073/pnas.93.3.1254
Hale TL Genetic basis of virulence in Shigella species Microbiol Rev 1991 55 206 224 1886518
Fasano A Noriega FR Maneval DR JrChanasongcram S Russell R Guandalini S Levine MM Shigella enterotoxin 1: an enterotoxin of Shigella flexneri 2a active in rabbit small intestine in vivo and in vitro J Clin Invest 1995 95 2853 2861 7769126
Frankel G Giron JA Valmassoi J Schoolnik GK Multi-gene amplification: simultaneous detection of three virulence genes in diarrhoeal stool Mol Microbiol 1989 3 1729 1734 2695745
Chiu CH Ou JT Rapid identification of Salmonella serovars in feces by specific detection of virulence genes, invA and spvC, by an enrichment broth cultivation-multiplex PCR combination assay J Clin Microbiol 1996 34 2619 2622 8880536
Vargas M Gascon J De Anta MTJ Vila J Prevalence of Shigella enterotoxins 1 and 2 among Shigella strains isolated from patients with traveler's diarrhea J Clin Microbiol 1999 37 3608 3611 10523561
Gaudio PA Sethabutr O Echeverria P Hoge CW Utility of a polymerase chain reaction diagnostic system in study of the epidemic of shigellosis among dysentery patients, family contacts and well controls living in a shigellosis-endemic area J Infect Dis 1997 176 1013 1018 9333160
Broude NE Zhang L Woodward K Englert D Cantor CR Multiplex allele-specific target amplification based on PCR suppression Proc Natl Acad Sci USA 2001 98 206 211 11136256 10.1073/pnas.98.1.206
Henegariu O Heerema NA Diouhy SR Vance GH Vogt PH Multiplex PCR: critical parameters and step-by-step protocol BioTechniques 1997 23 504 511 9298224
Nataro JP Seriwatana J Fasano A Maneval DR Guers LD Noriega F Dubovsky R Levine MM Morris JG Jr Identification and cloning of a novel plasmid-encoded enterotoxin of enteroinvasive Escherichia coli and Shigella strains Infect Immun 1995 63 4721 4728 7591128
Sasakawa C Kamata K Sakai T Murayama SY Makino S Yoshikawa M Molecular alteration of the 140-megadalton plasmid associated with loss of virulence and Congo red binding activity in Shigella flexneri Infect Immun 1986 51 470 475 3002985
Venkatesan M Buysse JM Kopecko DJ Use of Shigella flexneri ipaC and ipaH gene sequences for the general identification of Shigella spp. and enteroinvasive Escherichia coli J Clin Microbiol 1989 27 2687 2691 2687318
Yavzori M Cohen D Wasserlauf R Ambar R Rechavi G Ashkenazi S Identification of Shigella species in stool specimens by DNA amplification of different loci of the Shigella virulence plasmid Eur J Clin Microbiol Infect Dis 1994 13 232 237 8050436
Houng HSH Sethabutr O Echeverria P A simple polymerase chain reaction technique to detect and differentiate Shigella and enteroinvasive Escherichia coli in human feces Diagn Microbiol Infect Dis 1997 28 19 25 9218914 10.1016/S0732-8893(97)89154-7
DuPont HL Levine MM Hornick RB Formal SB Inoculum size in shigellosis and implications for expected mode of transmission J Infect Dis 1989 159 1126 1128 2656880
| 15707504 | PMC551607 | CC BY | 2021-01-04 16:28:14 | no | BMC Infect Dis. 2005 Feb 14; 5:8 | utf-8 | BMC Infect Dis | 2,005 | 10.1186/1471-2334-5-8 | oa_comm |
==== Front
BMC NeurolBMC Neurology1471-2377BioMed Central London 1471-2377-5-41572535810.1186/1471-2377-5-4Research Articleparkin mutation dosage and the phenomenon of anticipation: a molecular genetic study of familial parkinsonism Poorkaj Parvoneh [email protected] Lina [email protected] Jennifer S [email protected] John G [email protected] Gerard D [email protected] Haydeh [email protected] Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle, USA2 Genomics Institute, Wadsworth Center, New York State Department of Health, Albany, NY, USA3 Department of Neurology, Oregon Health & Science University, Portland, OR, USA4 Departments of Neurology and Pharmacology, University of Washington, and Geriatric Research Education Clinical Center, Veterans Affairs Puget Sound Health Care System, Seattle, WA, USA2005 22 2 2005 5 4 4 6 12 2004 22 2 2005 Copyright © 2005 Poorkaj et al; licensee BioMed Central Ltd.2005Poorkaj 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
parkin mutations are a common cause of parkinsonism. Possessing two parkin mutations leads to early-onset parkinsonism, while having one mutation may predispose to late-onset disease. This dosage pattern suggests that some parkin families should exhibit intergenerational variation in age at onset resembling anticipation. A subset of familial PD exhibits anticipation, the cause of which is unknown. The aim of this study was to determine if anticipation was due to parkin mutation dosage.
Methods
We studied 19 kindreds that had early-onset parkinsonism in the offspring generation, late-onset parkinsonism in the parent generation, and ≥ 20 years of anticipation. We also studied 28 early-onset parkinsonism cases without anticipation. Patients were diagnosed by neurologists at a movement disorder clinic. parkin analysis included sequencing and dosage analysis of all 12 exons.
Results
Only one of 19 cases had compound parkin mutations, but contrary to our postulate, the affected relative with late-onset parkinsonism did not have a parkin mutation. In effect, none of the anticipation cases could be attributed to parkin. In contrast, 21% of early-onset parkinsonism patients without anticipation had parkin mutations.
Conclusion
Anticipation is not linked to parkin, and may signify a distinct disease entity.
==== Body
Background
Mutations in the parkin gene are a common cause of parkinsonism. parkin was originally discovered as the cause of autosomal recessive juvenile parkinsonism [1]. However, recent reports suggest that not all parkin mutations are recessive [2-5], nor is age at onset always early [6-9]. Several studies have found heterozygous mutations in patients with late onset parkinsonism, suggesting that a single parkin mutation predisposes to later disease onset [8-10]. Collectively, these reports imply that parkin may exert a dosage effect in which possession of two mutations (homozygous or compound heterozygous) leads to early-onset parkinsonism, while possession of one normal and one mutant parkin (heterozygous) increases the risk for late-onset parkinsonism.
Several groups have reported what appears to be genetic anticipation in parkinsonism [11-13]. In the kindreds studied by them, the parent generation had typical late-onset parkinsonism and the individuals in the offspring generation developed the disease at much earlier ages. Barring the possibility that this pattern is an artifact of ascertainment bias (which cannot be ruled out until a biological mechanism is found), the intergenerational difference in onset age may be indicative of one of several mechanisms. The most common cause of anticipation is triplet repeat expansions. In a few parkinsonism families, disease has been shown to segregate with pathogenic expansions in SCA loci, but the search for other expansion loci in familial parkinsonism has been unsuccessful [14-16]. Other mechanisms that could result in anticipation include: change in mitochondrial heteroplasmy, which can affect disease severity and age at onset; a modifier gene that segregates independently of the disease gene; and parent/child exposure to a toxin. Gene dosage for dominant mutations can also mimic anticipation. For example, Familial Hypercholestrolemia is an autosomal dominant disorder in which heterozygotes develop the disease after the 4th decade of life, whereas homozygotes show symptoms at much younger ages, sometimes at birth (OMIM *143890).
Is anticipation in parkinsonism related to parkin mutation dosage? We hypothesized that some parkin mutations are dominant: heterozygotes have incomplete penetrance and may develop late-onset parkinsonism, whereas homozygotes and compound heterozygotes have accelerated disease leading to early-onset parkinsonism. This unifying hypothesis was attractive because if true, it could explain the reported variations in mode of inheritance of parkin (recessive in some families, dominant in others), the range in age at onset from juvenile to late onset, and the significantly earlier onset in some of the children of affected parents. To test this hypothesis, we studied parkin in 19 kindreds with early-onset parkinsonism in the index generation, late-onset parkinsonism in the parent generation, and exhibited ≥ 20 years of anticipation.
Methods
Families were recruited from a movement disorder clinic. The probands and affected family members had the clinical diagnosis of idiopathic Parkinson's disease (here referred to as parkinsonism due to lack of autopsy confirmation). Diagnosis was made by a neurologist according to the British Parkinson's Disease Brain Bank criteria except that family history was not an exclusion criterion [17]. These families were identified through an index case (a clinic patient) who reported a family history of parkinsonism. Positive family history was defined as patient reporting a first or second degree relative with parkinsonism, although we did inquire about more distant relatives as well. Clinic patients were enrolled sequentially and their affected relatives were subsequently identified and enrolled. For affected relatives we obtained medical records or personally examined them when possible. In some families the index case had early onset and the relative had late onset, in others, the index case had late onset and the relative with early onset was subsequently identified. The study was approved by the Institutional Review Boards at the participating institutions.
DNA was extracted from blood using standard protocol. Genotyping was blind to phenotype. To identify point mutations, we sequenced both DNA strands of all 12 exons. Exons and 50–100 bp of flanking intronic sequences were PCR-amplified [1], agarose gel-purified (Gene-clean III, Bio101), and directly sequenced by dye-terminator cycle sequencing (ABI, Big-Dye) using an ABI377 sequencer. To identify exon deletions and duplications, we analyzed gene dosage using real-time fluorescence-based PCR (ABI 7700 Sequence Detector). Amplification of subject genomic DNA was performed using fluorescently labeled probes (5' FAM or VIC, 3' TAMRA) and Taqman Universal PCR Mix (ABI) [4,18]. parkin exon amplifications were multiplexed under standard conditions with an 84-bp fragment of a single-copy human β-actin gene (Genbank accession number XM_004814) as an internal control. A standard curve was generated for each parkin exon and for β-actin using 0, 5, 15, 55 and 220 ng of control human genomic DNA. The number of PCR cycles required before the ABI 7700 detects each parkin exon product (CT value) was plotted against the corresponding exon standard curve, thus calculating the relative parkin copy number. The copy number for each exon was normalized to the single-copy actin gene within each multiplexed reaction and to a normal control reference individual, allowing an estimate of the number of copies of parkin. Optimal threshold levels for each primer set were maintained between plate analyses. All samples were analyzed in triplicate.
Results
Nineteen kindreds were chosen for parkin analysis based on having parkinsonism in two consecutive generations, late-onset parkinsonism in the parent generation (onset ages 59 to 89 years, mean 71.0 ± 8.5), early-onset parkinsonism in the offspring generation (onset ages 8 to 47 years, mean 37.2 ± 9.8), and ≥ 20 years of anticipation as measured by the difference in mean ages at onset in two generations. In the younger generation, 2 probands had onset before age 20, 1 was in his twenties, and 16 were over 30 years old. It was hypothesized that the affected individuals with early-onset parkinsonism are homozygous or compound heterozygous, and the parents and other relatives with late-onset parkinsonism are heterozygous. For parkin analysis, we began with the family member from the generation with the earlier onset, so as to enrich for parkin mutations. We expected the majority of the index cases to be compound heterozygous. At the minimum, 16–49% of index cases should have had mutations, since this is the range reported for early-onset sporadic and familial parkinsonism [7,9,19]. However, only one case had parkin mutations. This patient had compound deletions in exon 3 with onset at age 8. Her sister, onset at age 15, also had the compound mutation. The relative with late-onset was an uncle with onset age of 64. Contrary to our hypothesis, the uncle did not have a parkin mutation. The cause of late-onset parkinsonism in the uncle was different from the nieces. In effect, we did not find any cases where we could attribute the intergenerational difference in age at onset to parkin dosage.
We also analyzed parkin in 28 additional patients with early-onset parkinsonism (onset age 14 to 40 years, mean 32.8 ± 7.0), either with (n = 4) or without (n = 24) family history, but with no evidence of anticipation. In this group, 3 probands had onset at or before age 20, 5 were in their twenties, and 20 were over 30 years old. Nine mutations were found in six individuals. Two subjects were compound heterozygous (onset ages 31 and 37) and four were heterozygous (onset ages 14, 25, 37, 37). The frequency of parkin carriers in early-onset parkinsonism without anticipation was 21% which is in the range 16% – 49% reported in the literature (the lower range represents population and clinic based studies similar to ours, while the higher frequencies were found in highly selected autosomal recessive families). It was of interest to determine if any of the parents of the three early-onset parkinsonism patients with compound mutations had developed late-onset parkinsonism. All six parents were heterozygous (5 were confirmed by genotyping, one was inferred), and all had remained free of parkinsonism to the ages of 53, 60, 63, 74, 76 and 78 yrs. These individuals may still develop parkinsonism.
Discussion
Familial parkinsonism with anticipation may be more common than classical dominant and recessive subtypes combined. In our clinic population, among 487 patients studied, 145 had a first or second degree relative with PD; that is 30% which is in line with the published figures for other referral clinics. Among the 145 familial cases, 110 had parkinsonism in consecutive generations, which is compatible with autosomal dominant inheritance, and 35 had an affected sibling or cousin, which is suggestive of recessive inheritance. However, among the 110, only 26 were compatible with a classical dominant pattern (i.e., <10 years intergenerational variation in onset); while 63 exhibited 10–68 years of anticipation, and 7 had 10–17 years of reverse anticipation (14/110 had unknown onset ages). Our clinic is a referral center, which explains the relatively high proportions of early-onset and familial cases. The interesting finding was the relative proportions of autosomal dominant, autosomal recessive and anticipation cases within the familial subtype. Despite its relatively high prevalence, at least in our clinic, familial parkinsonism with anticipation has been largely overlooked in genetic research. A few parkinsonism families have been attributed to expansions in known SCA loci, but in the majority of the kindreds, the cause remains unknown. The more obvious possibilities are mitochondrial inheritance, modifier genes, parent-child exposure to environmental triggers, as yet unidentified triplet repeats or dominant genes with dosage effect, and in some cases, artifactual appearance of anticipation due to ascertainment bias. None of these are mutually exclusive; more than one may be true and operative in different families.
The current parkin literature suggest that possessing two parkin mutations is fully penetrant and leads to early-onset parkinsonism, whereas having only one mutation may be incompletely penetrant and lead to later disease onset. While the causative link to early-onset parkinsonism is widely accepted, the association of parkin with late-onset parkinsonism remains controversial [9,10,20,21]. We postulated that, if parkin heterozygotes are at risk for late-onset disease, then some parkin families should exhibit intergenerational variation in age at onset resembling anticipation, where heterozygous parents develop late-onset parkinsonism and children who inherit two mutations develop early-onset parkinsonism. We hoped to explain the appearance of anticipation in relation to parkin dosage, but the findings do not support this postulate. While the results rule out a link between parkin and anticipation in these families, they do not negate the association of parkin with late-onset parkinsonism.
Conclusion
The phenomenon of anticipation is not due to parkin mutation dosage. The underlying mechanism for anticipation may be genetic or environmental. Identification and a-priori classification of pedigrees that exhibit significant intergenerational age at onset variation, as being distinct from families that display classical dominant pattern, may facilitate gene mapping studies by reducing heterogeneity. Anticipation in parkinsonism merits investigation in its own right, not only because it is a common phenomenon and may account for a large subset of familial parkinsonism, but it may also uncover a novel mechanism in parkinsonism.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PP carried out the molecular genetic studies. LM interviewed the subjects and gathered family histories and medical records. JN performed the neurological examinations. GS participated in study design and supervised molecular genetic studies. JM performed molecular analysis of one large pedigree. HP conceived the study, participated in its design and coordination and drafted the manuscript. All authors read and approved the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors thank the patients, families, and volunteers who participated in the study. Financial support was provided by grants from the National Institutes of Health (RO1 NS36960) and the Veterans' Administration (PADRECC).
==== Refs
Kitada T Asakawa S Hattori N Matsumine H Yamamura Y Minoshima S Yokochi M Mizuno Y shimizu N Mutations in the parkin gene cause autosomal recessive juvenile parkinsonism Nature 1998 392 605 608 9560156 10.1038/33416
Kobayashi T Matsumine H Zhang J Imamichi Y Mizuno Y Hattori N Pseudo-autosomal dominant inheritance of PARK2: two families with parkin gene mutations J Neurol Sci 2003 207 11 17 12614925 10.1016/S0022-510X(02)00358-1
Lucking CB Bonifati V Periquet M Vanacore N Brice A Meco G Pseudo-dominant inheritance and exon 2 triplication in a family with parkin gene mutations Neurology 2001 57 924 927 11552035
Maruyama M Ikeuchi T Saito M Ishikawa A Yuasa T Tanaka H Hayashi S Wakabayashi K Takahashi H Tsuji S Novel mutations, pseudo-dominant inheritance, and possible familial affects in patients with autosomal recessive juvenile parkinsonism. Ann Neurol 2000 48 245 250 10939576
Klein C Pramstaller PP Kis B Page CC Kann M Leung J Woodward H Castellan CC Scherer M Vieregge P Breakefield XO Kramer PL Ozelius LJ Parkin deletions in a family with adult-onset, tremor-dominant parkinsonism: expanding the phenotype. Ann Neurol 2000 48 65 71 10894217
Farrer M Chan P Chen R Tan L Lincoln S Hernandez D Forno L Gwinn-Hardy K Petrucelli L Hussey J Singleton A Tanner C Hardy J Langston JW Lewy bodies and parkinsonism in families with parkin mutations Ann Neurol 2001 50 293 300 11558785 10.1002/ana.1132
Hedrich K Marder K Harris J Kann M Lynch T Meija-Santana H Pramstaller PP Schwinger E Bressman SB Fahn S Klein C Evaluation of 50 probands with early-onset Parkinson's disease for Parkin mutations Neurology 2002 58 1239 1246 11971093
West A Periquet M Lincoln S Lucking CB Nicholl D Bonifati V Rawal N Gasser T Lohmann E Deleuze JF Maraganore D Levey A Wood N Durr A Hardy J Brice A Farrer M Complex relationship between Parkin mutations and Parkinson disease Am J Med Genet 2002 114 584 591 12116199 10.1002/ajmg.10525
Oliveira SA Scott WK Martin ER Nance MA Watts RL Hubble JP Koller WC Pahwa R Stern MB Hiner BC Ondo WG Allen FHJ Scott BL Goetz CG Small GW Mastaglia F Stajich JM Zhang F Booze MW Winn MP Middleton LT Haines JL Pericak-Vance MA Vance JM Parkin mutations and susceptibility alleles in late-onset Parkinson's disease Ann Neurol 2003 53 624 629 12730996 10.1002/ana.10524
Foroud T Uniacke SK Liu L Pankratz N Rudolph A Halter C Shults C Marder K Conneally PM Nichols WC Heterozygosity for a mutation in the parkin gene leads to later onset Parkinson disease Neurology 2003 60 796 801 12629236
Bonifati V Vanacore N Meco G Anticipation of onset age in familial Parkinson's disease Neurology 1994 44 1978 1979 7936260
Payami H Bernard S Larsen K Kaye J Nutt J Genetic anticipation in Parkinson's disease Neurology 1995 45 135 138 7824103
Markopoulou K Wszolek ZK Pfeiffer RF A Greek-American kindred with autosomal dominant, levodopa-responsive parkinsonism and anticipation Ann Neurol 1995 38 373 378 7668822 10.1002/ana.410380306
Gwinn-Hardy K Chen JY Liu H Liu TY Boss M Seltzer W Adam A Singleton A Koroshetz W Waters C Hardy J Farrer M Spinocerebellar ataxia type 2 with parkinsonism in ethnic Chinese Neurology 2000 55 800 805 10993999
Payami H Nutt J Gancher S Bird T McNeal MG Seltzer WK Hussey J Lockhart P Gwinn-Hardy K Singleton AA Singleton AB Hardy J Farrer M SCA2 may present as levodopa-responsive parkinsonism Mov Disord 2003 18 425 429 12671950 10.1002/mds.10375
Shan DE Soong BW Sun CM Lee SJ Liao KK Liu RS Spinocerebellar ataxia type 2 presenting as familial levodopa-responsive parkinsonism Ann Neurol 2001 50 812 815 11761482 10.1002/ana.10055
Hughes AJ Daniel SE Kilford L Lees AJ Accuracy of clinical diagnosis of idiopathic Parkinson's disease: a clinico-pathological study of 100 cases. J Neurol Neurosurg Psychiatry 1992 55 181 184 1564476
Tsuang DW Dalan AM Eugenio CJ Poorkaj P Limprasert P La Spada AR Steinbart EJ Bird TD Leverenz JB Familial dementia with lewy bodies: a clinical and neuropathological study of 2 families Arch Neurol 2002 59 1622 1630 12374501 10.1001/archneur.59.10.1622
Lucking CB Durr A Bonifati V Vaughan J De Michele G Gasser T Harhangi BS Meco G Denefle P Wood NW Agid Y Brice A Association between early-onset Parkinson's disease and mutations in the parkin gene. French Parkinson's Disease Genetics Study Group N Engl J Med 2000 342 1560 1567 10824074 10.1056/NEJM200005253422103
Lincoln SJ Maraganore DM Lesnick TG Bounds R de Andrade M Bower JH Hardy JA Farrer MJ Parkin variants in North American Parkinson's disease: cases and controls Mov Disord 2003 18 1306 1311 14639672 10.1002/mds.10601
Oliveri RL Zappia M Annesi G Annesi F Spadafora P Pasqua AA Tomaino C Nicoletti G Bosco D Messina D Logroscino G Manobianca G Epifanio A Morgante L Savettieri G Quattrone A The parkin gene is not a major susceptibility locus for typical late-onset Parkinson's disease Neurol Sci 2001 22 73 74 11487208 10.1007/s100720170053
| 15725358 | PMC551608 | CC BY | 2021-01-04 16:28:53 | no | BMC Neurol. 2005 Feb 22; 5:4 | utf-8 | BMC Neurol | 2,005 | 10.1186/1471-2377-5-4 | oa_comm |
==== Front
BMC Pregnancy ChildbirthBMC Pregnancy and Childbirth1471-2393BioMed Central London 1471-2393-5-31572072010.1186/1471-2393-5-3Research ArticleCustomized birth weight for gestational age standards: Perinatal mortality patterns are consistent with separate standards for males and females but not for blacks and whites Joseph K S [email protected] Russell [email protected] Linda [email protected] Victoria M [email protected] Arne [email protected] Sylvie [email protected] Robert [email protected] Perinatal Epidemiology Research Unit, Departments of Obstetrics and Gynaecology and of Pediatrics, Dalhousie University, Halifax, Nova Scotia, Canada2 Health Analysis and Measurement Group, Statistics Canada, Ottawa, Ontario, Canada3 Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynaecology, Dalhousie University, Halifax, Nova Scotia, Canada4 Departments of Paediatrics and of Obstetrics and Gynecology, University of Toronto, Toronto, Ontario, Canada5 Department of Social and Preventive Medicine, Université Laval, Sainte-Foy, Quebec, Canada6 Department of Obstetrics and Gynecology, University of British Columbia, Vancouver, British Columbia, Canada2005 20 2 2005 5 3 3 4 5 2004 20 2 2005 Copyright © 2005 Joseph et al; licensee BioMed Central Ltd.2005Joseph 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
Some currently available birth weight for gestational age standards are customized but others are not. We carried out a study to provide empirical justification for customizing such standards by sex and for whites and blacks in the United States.
Methods
We studied all male and female singleton live births and stillbirths (22 or more weeks of gestation; 500 g birth weight or over) in the United States in 1997 and 1998. White and black singleton live births and stillbirths were also examined. Qualitative congruence between gestational age-specific growth restriction and perinatal mortality rates was used as the criterion for identifying the preferred standard.
Results
The fetuses at risk approach showed that males had higher perinatal mortality rates at all gestational ages compared with females. Gestational age-specific growth restriction rates based on a sex-specific standard were qualitatively consistent with gestational age-specific perinatal mortality rates among males and females. However, growth restriction patterns among males and females based on a unisex standard could not be reconciled with perinatal mortality patterns. Use of a single standard for whites and blacks resulted in gestational age-specific growth restriction rates that were qualitatively congruent with patterns of perinatal mortality, while use of separate race-specific standards led to growth restriction patterns that were incompatible with patterns of perinatal mortality.
Conclusion
Qualitative congruence between growth restriction and perinatal mortality patterns provides an outcome-based justification for sex-specific birth weight for gestational age standards but not for the available race-specific standards for blacks and whites in the United States.
==== Body
Background
Birth weight-specific perinatal mortality curves among male and female births intersect to produce a paradox: overall perinatal mortality rates and perinatal mortality rates at lower birth weights are relatively higher among male births, while at higher birth weights perinatal mortality rates are relatively higher among female births [1]. This puzzling observation reflects a general phenomenon that is also seen when birth weight- and gestational age-specific perinatal mortality curves are contrasted across race, plurality, maternal smoking status, parity, altitude, country, and other determinants of birth weight and gestational age [2-14]. We have previously presented a solution for this paradox of intersecting mortality curves that involves a reformulation of perinatal and neonatal mortality risk [15-20]. This reformulation, based on the fetuses at risk approach, eliminates the crossover phenomenon and provides several new insights into perinatal health issues.
In this paper, we demonstrate the paradoxical crossover of birth weight-specific perinatal mortality curves among male and female births and show how this phenomenon is resolved using the fetuses at risk approach. We also explore issues related to fetal growth restriction among males and females using the same approach. This latter issue is particularly important from a conceptual and clinical standpoint because the current literature on birth weight for gestational age standards (sometimes referred to as fetal growth standards) is confusing. Some standards provide unisex reference values [21-24], several are sex-specific [1,25-34] and yet others provide both sex-specific and unisex reference values [35-38]. Of equal concern is the fact that several standards are customized for different races [1,25,27-29], parity [25,27,29,34,36], plurality [24,30] and other characteristics [27], while others are not [21-23,26,31-33,35,37].
We used the fetuses at risk approach to contrast growth restriction and perinatal mortality rates among males and females in order to provide empirical justification for sex-specific (vs unisex) birth weight for gestational age standards. We also constructed and compared gestational age-specific growth restriction and perinatal mortality curves among whites vs blacks in order to evaluate currently available birth weight for gestational age standards (single standard vs separate standards for whites and blacks in the United States).
Methods
We used data on all reported live births and stillbirths in the United States in 1997 and 1998 (National Center for Health Statistics perinatal mortality data file for all states and the District of Columbia for 1997 and 1998). Live births and infant death records for these years have been previously linked and gestational duration has been calculated based on the last menstrual period (LMP). Missing or inconsistent information on gestational age has been imputed or replaced in a small fraction (approximately 7 percent) of records by the National Center for Health Statistics (Hyattsville, Maryland). Gestational age was imputed from the month and year of the LMP when the exact LMP day was missing [39]. LMP-based gestational age information was replaced by the clinical estimate [40] when the former was inconsistent with birth weight or when there was no information on LMP (approximately 5 percent of births).
Analyses were restricted to singleton live births and stillbirths ≥22 weeks gestational age and ≥500 g birth weight in order to eliminate potential problems arising from regional differences in birth registration. Male and females births were first contrasted in terms of their gestational age and birth weight distributions. Birth weights were categorized into 500 g intervals for this purpose (500–999 g, 1,000–1,499 g, 1,500–1,999 g and so on). Birth weight-specific perinatal mortality rates, calculated within these birth weight categories, were computed as per convention by dividing the number of stillbirths and early neonatal (0 to 6 days) deaths in any birth weight category by the number of total births (stillbirths and live births) in that birth weight category. Similarly, gestational age-specific perinatal mortality rates among male and female births were contrasted, with rates computed by dividing perinatal deaths at any given gestation by the number of total births at that gestation.
The numbers of fetuses at risk for stillbirth and early neonatal death at each gestation were then used to calculate a second set of perinatal mortality rates. Under this fetuses at risk formulation, the stillbirth rate at 28 weeks gestation was computed by dividing the number of stillbirths at 28 weeks by the number of live births and stillbirths at 28 or more completed weeks of gestation. This implies that fetuses who delivered at 29, 30, 31 and 32 or more weeks gestation were also at risk of stillbirth at 28 weeks [15-19,41-44]. The fetuses at risk formulation applies equally to early neonatal death since a fetus (unborn) at 28 weeks gestation is at risk of birth and early neonatal death at that gestation [15,17,18]. Thus gestational age-specific perinatal/neonatal mortality rates under this formulation were calculated with perinatal/neonatal deaths at any gestational age in the numerator and the fetuses at risk of perinatal/neonatal death at that gestation in the denominator. This represents a survival analysis model with censoring of subjects (fetuses) at death or birth which ever occurs earlier (for a schematic depiction of the survival analysis model, see reference 18). In this model, neonatal death (and, in other contexts, serious pregnancy-related morbidity such as cerebral palsy [16]) is assigned to the point of birth since the responsible pathologic event/process is present at birth [18]. Gestational age-specific 'birth rates' (i.e., the number of births at any particular gestational week divided by the number of fetuses at risk of birth at that gestation) and rates of gestational age-specific labor induction/cesarean delivery were also estimated using the fetuses at risk approach [15-18].
We also examined gestational age-specific patterns of fetal growth restriction using the fetuses at risk approach [15,17-19]. The number of small-for-gestational age (SGA) live births at each gestation was divided by the number of fetuses at risk at that gestation in order to obtain the gestational age-specific SGA rate (or the gestational age-specific fetal growth restriction rate). SGA live births were identified using the 10th percentile cut-off from a birth weight for gestational age standard based on live births in the United States [38]. Gestational age-specific SGA rates were calculated using both the unisex and sex-specific 10th percentile values provided by this standard [38] to evaluate how well patterns of gestational age-specific growth restriction correspond with patterns of gestational age-specific perinatal mortality. This evaluation was premised on the belief that fetal growth restriction patterns should be qualitatively congruent with gestational age-specific perinatal mortality patterns. Such an expectation is consistent with clinical understanding and studies which show that growth restricted fetuses have a substantially higher perinatal mortality than appropriate-for-gestational age fetuses. For instance, Williams et al [1] showed that perinatal mortality at each gestational week was much higher among growth restricted births at the 10th percentile of birth weight for gestational age (eg., perinatal mortality rate 138 per 1,000 total births at 34–35 weeks) compared with appropriate-for-gestational age births at the 50th percentile of birth weight for gestational age (eg., perinatal mortality rate 27 per 1,000 total births at 34–35 weeks). We also examined gestational age-specific growth restriction differences among males and females using rate ratios (eg., growth restriction rate among males at 35 weeks gestation divided by growth restriction rate among females at 35 weeks gestation) and contrasted these with gestational age-specific differences in stillbirth and neonatal mortality rates (also using rate ratios eg., stillbirth rate among males at 35 weeks divided by the stillbirth rate among females at 35 weeks; early neonatal death rate among males at 35 weeks divided by the early neonatal death rate among females at 35 weeks). This was done to ascertain the relationship between patterns of growth restriction and patterns in the two components of perinatal mortality (stillbirth and early neonatal death).
Comparisons of male and female gestational age-specific growth restriction and gestational age-specific perinatal mortality patterns were contrasted with similar comparisons according to maternal race. Specifically, live births and stillbirths ≥22 weeks of gestational age and ≥500 g birth weight in the United States in 1997 and 1998 were used to compare gestational age-specific growth restriction and perinatal mortality rates among whites vs blacks.
Identification of SGA live births among blacks and whites was carried out using a single standard for both races [38] and also a race-specific standard [29]. As with contrasts between males and females, the contrasts between whites and blacks were restricted to singleton births.
Differences in rates were assessed using rate ratios and excess risks. Taylor series 95% confidence intervals were calculated on all rate ratios. All p values presented are two-sided. Sensitivity analyses were carried out to assess the potential effect of gestational age errors on patterns of growth restriction and perinatal mortality among males and females. Specifically, we reassessed growth restriction and mortality patterns among males and females after excluding all births for whom menstrual-based gestational age was either imputed or replaced by the clinical estimate of gestation.
Results
There were 3,905,694 singleton male births in the United States in 1997 and 1998 (≥22 weeks gestational age and ≥500 g birth weight). The low birth weight (<2,500 g) rate among male live births was 5.5%, and 10.5% of male live births were born preterm (<37 weeks). There were 3,723,153 female births in the United States during the same period and relative to males, female live births had a higher rate of low birth weight (6.4%, p < 0.0001) but a lower rate of preterm birth (9.4%, p < 0.0001). Males had a 14% (95% confidence interval 12 to 16, p < 0.0001) higher perinatal mortality than females; perinatal mortality rates among males and females were 6.78 and 5.95 per 1,000 total births, respectively.
The gestational age distribution of male live births (Figure 1) was 'shifted to the left' relative to female live births (p < 0.0001), while the birth weight distribution of females was markedly 'shifted to the left' relative to that of male live births (p < 0.0001). Birth weight-specific perinatal mortality rates (conventional calculation, perinatal deaths per 1,000 total births in a given birth weight category) showed the crossover paradox with males having relatively higher rates of perinatal death at birth weights <4,000 g, while females had relatively higher perinatal mortality rates at higher birth weights (Figure 2a). In contrast, gestational age-specific perinatal mortality rates (conventional calculation, perinatal deaths per 1,000 total births at any gestational week) showed similar mortality patterns among males and females (Tables 1 and 2), with males having a slightly higher perinatal mortality rate at some gestational ages (Figure 2b).
Figure 1 Gestational Age and Birth Weight Distributions of Male and Female Singleton Live Births. Gestational age (1a) and birth weight (1b) distributions of male and female singleton live births ≥22 weeks and ≥500 g in the United States, 1997 and 1998.
Figure 2 Conventional Calculation: Birth Weight- and Gestational Age-Specific Perinatal Mortality Rates among Male and Female Births. Conventional calculation: birth weight-specific (2a) and gestational age-specific (2b) perinatal mortality rates per 1,000 total births among male and female singleton births in the United States, 1997 and 1998.
Table 1 Gestational Age-Specific Numbers and Rates of Perinatal Death among Male Singleton Births, United States, 1997 and 1998.
Gestational age Stillbirths Live births Early neonatal deaths Perinatal mortality rate (1)† Fetuses at risk Perinatal mortality rate (2)†
28 648 6,808 263 122.2 3,841,944 0.24
29 579 8,100 221 92.2 3,834,488 0.21
30 701 11,297 230 77.6 3,825,809 0.24
31 668 14,339 208 58.4 3,813,811 0.23
32 809 20,242 209 48.4 3,798,804 0.27
33 786 30,140 212 32.3 3,777,753 0.26
34 881 51,673 298 22.4 3,746,827 0.31
35 915 86,166 338 14.4 3,694,273 0.34
36 1,033 154,986 354 8.9 3,607,192 0.38
37 1,144 308,629 394 5.0 3,451,173 0.45
38 1,173 626,450 470 2.6 3,141,400 0.52
39 1,122 925,764 541 1.8 2,513,777 0.66
40 897 848,527 431 1.6 1,586,891 0.84
41 469 444,468 237 1.6 737,467 0.96
≥42* 454 292,076 229 2.3 292,530 2.33
Total‡ 17,680 3,888,014 8,800 6.8 3,905,694 6.78
† Total births at each gestational week served as the denominator for perinatal mortality rates (1), while perinatal mortality rates (2) were calculated using fetuses at risk as the denominator (see text). All rates are expressed per 1,000.
* Large increase in perinatal mortality (2) at ≥42 weeks is partly because the period of risk exceeds 1 week (see also Figures 3-5).
‡ All gestational ages, including those ≥22 weeks and those with missing gestational age.
Table 2 Gestational Age-Specific Numbers and Rates of Perinatal Death among Female Singleton Births, United States, 1997 and 1998.
Gestational age Stillbirths Live births Early neonatal deaths Perinatal mortality rate (1)† Fetuses at risk Perinatal mortality rate (2)†
28 614 5,838 184 123.7 3,665,497 0.22
29 530 7,000 158 91.4 3,659,045 0.19
30 611 9,742 179 76.3 3,651,515 0.22
31 578 12,493 173 57.5 3,641,162 0.21
32 632 17,168 195 46.5 3,628,091 0.23
33 654 25,282 187 32.4 3,610,291 0.23
34 747 44,275 221 21.5 3,584,355 0.27
35 796 75,238 234 13.5 3,539,333 0.29
36 874 133,386 286 8.6 3,463,299 0.33
37 927 267,501 337 4.7 3,329,039 0.38
38 1,071 563,676 361 2.5 3,060,611 0.47
39 1,072 885,523 419 1.7 2,495,864 0.60
40 926 851,848 376 1.5 1,609,269 0.81
41 505 455,313 222 1.6 756,495 0.96
≥42* 443 300,234 198 2.1 300,677 2.13
Total‡ 15,537 3,707,616 6,614 6.0 3,723,153 5.95
† Total births at each gestational week served as the denominator for perinatal mortality rates (1), while perinatal mortality rates (2) were calculated using fetuses at risk as the denominator (see text). All rates are expressed per 1,000.
* Large increase in perinatal mortality (2) at ≥42 weeks is partly because the period of risk exceeds 1 week (see also Figures 3-5).
‡All gestational ages, including those ≥22 weeks and those with missing gestational age.
Gestational age-specific perinatal mortality rates calculated using the fetuses at risk approach showed that perinatal mortality rates increased with increasing gestational age (Figure 3). Males had a higher perinatal mortality than females at virtually all gestational ages (Tables 1 and 2). Gestational age-specific 'birth rates' (Figure 3a), gestational age-specific labor induction rates (Figure 3b) and gestational age-specific labour induction and/or cesarean delivery rates (data not shown) were marginally (but consistently) higher among pregnancies with males as compared with pregnancies with females (Figure 3). For example, the birth rate among males at 35 weeks gestation was 23.6 per 1,000 fetuses at risk, while that among females at 35 weeks was 21.5 per 1,000 fetuses at risk (rate ratio 1.10, 95% confidence interval 1.09 to 1.11, p < 0.0001). The labour induction rates at 35 weeks among males and females were 3.6/1,000 and 3.1/1,000 fetuses at risk, respectively; rate ratio 1.10, 95% confidence interval 1.07 to 1.13, p < 0.0001.
Figure 3 Fetuses at Risk Approach: Gestational Age-Specific Birth, Labor Induction and Perinatal Mortality Rates among Male and Female Births. Fetuses at risk approach: Gestational age-specific birth rates (3a, primary Y-axis), labor induction rates (3b, primary Y-axis) and perinatal mortality rates (3a and 3b, secondary Y-axis) among male and female singleton births in the United States, 1997 and 1998.
Figure 4 compares gestational age-specific rates of fetal growth restriction among males and females. When growth restriction was determined using a sex-specific standard, growth restriction rates among males were higher than growth restriction rates among females at all gestational ages and this pattern was qualitatively congruent with sex differences in perinatal mortality (Figure 4a). For instance, males at 35 weeks gestation had an 8 percent (95% confidence interval 5 to 11, p < 0.0001) higher growth restriction rate than females at the same gestational week (sex-specific standard) and this was qualitatively congruent with a 17 percent (95% confidence interval 7 to 27 percent, p = 0.0003) higher perinatal death rate among males compared with females at 35 weeks gestation. On the other hand, when a unisex standard was used to identify growth restricted live births, males had a lower rate of growth restriction at all gestational ages and this was not qualitatively congruent with the higher gestational age-specific pattern of perinatal mortality among males (Figure 4b). For instance, at 35 weeks gestation, growth restriction rates determined using a single standard for both males and females showed that males had a 20 percent (95% confidence interval 18 to 22 percent, p < 0.0001) lower rate of growth restriction compared with females (not consistent with the 17% higher perinatal mortality rate).
Figure 4 Fetuses at Risk Approach: Gestational Age-Specific Growth Restriction and Perinatal Mortality Rates among Male and Female Births. Fetuses at risk approach: Gestational age-specific fetal growth restriction (primary Y-axis) and perinatal mortality rates (secondary Y-axis) among male and female singleton births, with growth restriction rates based on sex-specific (4a) and unisex (4b) birth weight for gestational age standards, United States, 1997 and 1998.
Overall growth restriction rates based on a sex-specific standard showed that rates were 3% (95% CI 2 to 3) higher among males. Stillbirth and early neonatal mortality differences (rate ratios) among male vs female births both favored females (Table 3), although the mortality differences were much larger for early neonatal mortality (27%, 95% CI 23 to 31) than for stillbirth (8%, 95% CI 6 to 11). Gestational age-specific differences in growth restriction between males and females based on a sex-specific standard (eg., rate ratio at 35 weeks 1.08, 95% CI 1.05 to 1.11, Table 3) tended to be similar to gestational age-specific differences in stillbirth rates (eg., rate ratio at 35 weeks 1.10, 95% CI 1.00 to 1.21, Table 3), while differences in gestational age-specific early neonatal mortality tended to be larger (eg., rate ratio at 35 weeks 1.38, 95% CI 1.17 to 1.63, Table 3). Sensitivity analyses carried out to examine the potential effect of gestational age errors (by excluding births among whom gestational age was imputed or for whom the clinical estimate of gestation was used) showed essentially the same patterns of growth restriction and perinatal mortality among males and females.
Table 3 Gestational Age-Specific Rates of Fetal Growth Restriction Based on a Sex-Specific Standard [38] and Differences in Growth Restriction, Stillbirth and Early Neonatal Mortality Among Males and Females, Singleton Births, United States, 1997 and 1998.
Gestational age Fetal growth restriction Stillbirth rate ratio(males vs females) Early neonatal mortality rate ratio(males vs females)
Males Females Rate ratio (males vs females)
Number Rate † Number Rate †
28 631 0.2 502 0.1 1.20 1.01 1.36
29 764 0.2 660 0.2 1.10 1.04 1.33
30 1,165 0.3 923 0.3 1.20 1.10 1.23
31 1,539 0.4 1,369 0.4 1.07 1.10 1.15
32 2,141 0.6 1,866 0.5 1.09 1.22 1.02
33 3,294 0.9 2,946 0.9 1.07 1.15 1.08
34 5,691 1.6 5,098 1.5 1.07 1.13 1.29
35 8,934 2.5 7,922 2.3 1.08 1.10 1.38
36 15,813 4.6 13,910 4.2 1.09 1.13 1.19
37 30,029 9.1 25,180 7.9 1.15 1.19 1.13
38 55,401 18.5 49,599 17.1 1.09 1.07 1.27
39 84,257 35.7 79,440 33.9 1.05 1.04 1.28
40 75,983 52.8 74,504 51.1 1.03 0.98 1.16
41 36,956 62.6 34,214 56.5 1.11 0.95 1.10
≥42 14,679 100.5 14,551 96.7 1.04 1.05 1.19
Total‡ 337,277 91.6 312,684 89.3 1.03 1.08 1.27
† Gestational age-specific growth restriction rates (based on a sex-specific standard [38]) were calculated by dividing the number of small-for-gestational age live births (<10th percentile) at any gestational age by the number of fetuses at risk at that gestation. Stillbirth and early neonatal mortality rates were also calculated using fetuses at risk as the denominator.
‡ All gestational ages ≥22 weeks, except for growth restriction indices which were based on live births between 28 and 42 weeks.
Patterns of gestational age-specific growth restriction among whites and blacks could not be reconciled with patterns of gestational age-specific perinatal mortality, when growth restriction was defined by a race-specific standard (Figure 5a). Growth restriction rates defined using the race-specific birth weight for gestational age standard showed a crossover with blacks having significantly higher growth restriction rates than whites below 39 weeks and significantly lower growth restriction rates at 39 weeks and over. For instance, rates of growth restriction as defined by the race-specific standard were significantly lower among blacks compared with whites at 40 weeks gestation (rate ratio 0.89, 95% confidence interval 0.88 to 0.91, p < 0.0001), despite a significantly higher perinatal mortality rate among blacks at 40 weeks gestation (rate ratio 1.43, 95% confidence interval 1.29 to 1.58, p < 0.0001). On the other hand, rates of gestational age-specific growth restriction were qualitatively congruent with patterns of gestational age-specific perinatal mortality when growth restriction among blacks and whites was defined using a single birth weight for gestational age standard (Figure 5b). For example, at 40 weeks gestation, the significantly higher rate of perinatal death among blacks was consistent with the significantly higher rate of growth restriction seen among blacks when a single standard was used to define growth restriction (rate ratio for growth restriction at 40 weeks among blacks vs whites 2.06, 95% confidence interval 2.04 to 2.09, p < 0.0001). Growth restriction (based on a single standard for both races) and perinatal mortality rates were substantially higher among births to black mothers as compared with births to white mothers at all gestational ages (Figure 5b).
Figure 5 Fetuses at Risk Approach: Gestational Age-Specific Fetal Growth Restriction and Perinatal Mortality Rates among White and Black Births. Fetuses at risk approach: Gestational age-specific fetal growth restriction (primary Y-axis) and perinatal mortality rates (secondary Y-axis) among white and black singleton births, with growth restriction rates based on a race-specific standard (5a) and on a single birth weight for gestational age standard (5b), United States, 1997 and 1998.
Discussion
We have confirmed previous observations that birth weight-specific perinatal mortality rates among male and female births exhibit a puzzling crossover paradox [1]. Gestational age-specific perinatal mortality rates among males and females were similar when mortality rates were calculated per convention (using total births at a particular gestation for calculating the perinatal mortality rate). On the other hand, use of the fetuses at risk formulation [15-19,41-44] showed that males have a consistently higher perinatal mortality rate at all gestational ages. Further, our study shows that gestational age-specific growth restriction and perinatal mortality rates both increase with advancing gestational age. Gestational age-specific rates of growth restriction among males and females are qualitatively congruent with gestational age-specific perinatal mortality patterns when growth restriction rates are based on a sex-specific birth weight for gestational age standard. Use of a single standard for males and females results in a gestational age-specific pattern of growth restriction that cannot be reconciled with gestational age-specific differences in perinatal mortality among males and females.
In contradistinction, contrasts between whites vs blacks show that use of a single birth weight for gestational age standard for both races is justified, while the use of a currently available race-based standard is not defensible. Gestational age-specific growth restriction patterns among whites vs blacks based on a single standard correspond qualitatively to patterns of gestational age-specific perinatal mortality among whites and blacks (Figure 5).
Birth weight for gestational age standards are modeled after infant and child growth standards and assume that fetal growth restriction occurs at a constant rate throughout pregnancy. This assumption is implicit in the use of the same, fixed cut-off (eg., the 3rd percentile or the 10th percentile cut-off of birth weight for gestational age) for identifying fetal growth restriction at all gestational ages. Our findings challenge the former assumption and show that in fact fetal growth restriction rates are better viewed as increasing with advancing gestational age (Figures 4 and 5). This contention is supported by the finding that gestational age-specific growth restriction rates follow the pattern of gestational age-specific perinatal mortality rates. Recent studies which show that the incidence of hypertensive disorders and chorioamnionitis increases with increasing gestational age provide at least a partial explanation for the gestational age-dependent rise in fetal growth restriction and perinatal mortality rates [45,46].
Table 3 shows that differences in stillbirth rates between males and females are smaller than differences in early neonatal mortality rates. The phenomenon of higher neonatal mortality differentials (relative to stillbirth differentials) between males and females has been previously noted [1] and is probably a consequence of obstetric intervention. Obstetric intervention (i.e., early delivery through labor induction and/or cesarean delivery) is typically prompted by signs of fetal compromise and will be more likely among pregnancies with male fetuses given the male fetuses' greater biological vulnerability. Such intervention leads to a reduction in the stillbirth differential, while having a smaller (or the opposite) effect on neonatal mortality differences between males and females. This explanation is supported by the higher rates of labor induction (and labour induction and/or cesarean delivery) observed among pregnancies with male fetuses (Figure 3b). Differences in rates of congenital anomalies that are lethal after birth and more frequent in males (eg., X-linked recessive conditions) may partly contribute to this phenomenon as well.
The slightly higher rate of gestational age-specific labor induction/cesarean delivery among males relative to females is encouraging since it suggests that the small mortality risk difference between males and females is already being addressed by modern obstetric practice (despite male sex not being formally identified as a factor in decision making related to obstetric intervention). This may be a consequence of the use of sex-specific birth weight for gestational age standards or sex-specific ultrasound-based fetal growth standards and, as mentioned, probably also reflects higher rates of suspected fetal compromise among pregnancies with male fetuses. Despite the marginally higher rates of labor induction among pregnancies with male fetuses, however, mortality differences persist. Research should be directed at ascertaining whether excess neonatal mortality among males can be successfully reduced through explicit recognition of male sex as a factor for altering the threshold for obstetric intervention.
Although contemporary birth weight for gestational age standards have substantial face validity [1,47,48], their development would benefit from greater empirical support and validation. For instance, it should be feasible to refine standards based on empirically observed (cause-specific) patterns of birth weight-specific perinatal mortality and serious neonatal morbidity (at each gestational age). This would represent an improvement over current standards which rely heavily on theoretical assumptions (eg., normality of birth weight at any given gestational age) and insufficiently on relevant empirical information (namely, perinatal morbidity and mortality related to growth restriction). Such cross-sectional information cannot address fetal growth in continuing pregnancies, however; the latter requires longitudinal information which is ideally obtained through ultrasonographic measurements. On the other hand, estimation of fetal weight through ultrasonography [31,49] needs to be improved [50,51] and diagnostic methods for identifying fetal growth restriction have tended to rely on other indicators of growth restriction besides estimated fetal weight.
Our study has limitations that are typical of studies that use large data bases. Errors in gestational age information are inevitable, although the magnitude of these errors is likely to be similar among male and female births. The overall rate of missing gestational age was low, however (0.9 percent among white live births and 0.8 percent among black live births). Our estimates of gestational age-specific fetal growth restriction rates are approximate. Ideally, estimation of the incidence of fetal growth restriction requires identification of fetal growth restriction on a longitudinal basis among continuing pregnancies [18]. The alternative measure of gestational age-specific growth restriction employed in our study represents an index of 'revealed' fetal growth restriction [18]. This approximation is unlikely to be a factor that seriously distorts patterns of gestational age-specific growth restriction since faltering of fetal growth typically leads to a spontaneous delivery or delivery following obstetric intervention. Other potential limitations of our study include the use of gestational age information on stillbirths. The gestational age at delivery of a stillbirth typically overestimates the gestational age at the time of fetal death, although this difference is unlikely to be large in recent years. Further, both male and female stillbirths would have been affected by this measurement error to a similar extent.
Conclusion
The fetuses at risk approach resolves the paradox of intersecting perinatal mortality curves. Male births have higher rates of gestational age-specific perinatal mortality than female births. There is empirical justification for using sex-specific standards of birth weight for gestational age since gestational age-specific growth restriction patterns based on such standards correspond qualitatively with gestational age-specific perinatal mortality patterns. On the other hand, a single birth weight for gestational age standard for whites and blacks in the United States appears more appropriate than currently available race-specific standards since gestational age-specific growth restriction patterns among blacks and whites (based on a single standard) are qualitatively congruent with gestational age-specific patterns of perinatal mortality.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
KSJ proposed the study, carried out the analyses and drafted the manuscript. The results of the analyses were presented and discussed at a meeting of the Fetal and Infant Health Study Group of the Canadian Perinatal Surveillance System. All authors contributed to revising the manuscript for intellectual content. All authors read and approved the final version.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We are grateful to National Center for Health Statistics for providing us with access to the data. This study was carried out under the auspices of the Fetal and Infant Health Study Group of the Canadian Perinatal Surveillance System. Dr. Joseph, Dr. Dodds and Dr. Allen are supported by Clinical Research Scholar awards from the Dalhousie University Faculty of Medicine. Dr. Joseph is a recipient of a Peter Lougheed New Investigator award of the Canadian Institutes of Health Research and Dr. Dodds is a New Investigator of the Canadian Institutes of Health Research.
==== Refs
Williams R Creasy R Cunningham G Hawes W Norris F Tashiro M Fetal growth and perinatal viability in California Obstet Gynecol 1982 59 624 632 7070736
Yerushalmy J The relationship of parents' cigarette smoking to outcome of pregnancy – implications as to the problem of inferring causation from observed associations Am J Epidemiol 1971 93 443 56 5562717
Meyer MB Comstock GW Maternal cigarette smoking and perinatal mortality Am J Epidemiol 1972 96 1 10 5039725
Wilcox AJ Russell IT Why small black infants have lower mortality than small white infants: the case for population-specific standards for birth weight J Pediatr 1990 116 7 10 2295966
Wilcox AJ Russell IT Birthweight and perinatal mortality: III. Towards a new method of analysis Int J Epidemiol 1986 15 188 96 3721681
Wilcox AJ Skjœrven R Birth weight and perinatal mortality: the effect of gestational age Am J Public Health 1992 82 378 82 1536353
English PB Eskenazi B Reinterpreting the effects of maternal smoking on infant birthweight and perinatal mortality: a multivariate approach to birth weight standardization Int J Epidemiol 1992 21 1097 1105 1483814
Wilcox AJ Birth weight and perinatal mortality: the effect of maternal smoking Am J Epidemiol 1993 137 1098 1104 8317439
Buekens P Wilcox A Why do small twins have a lower mortality than small singletons? Am J Obstet Gynecol 1993 168 937 41 8456906
Wilcox AJ Skjœrven R Buekens P Kiely J Birth weight and perinatal mortality: A comparison of the United States and Norway JAMA 1995 272 709 11 7853628 10.1001/jama.273.9.709
Hertz-Picciotto I Din-Dzietham R Comparisons of infant mortality using a percentile-based method of standardization for birthweight or gestational age Epidemiol 1998 9 61 7 10.1097/00001648-199801000-00009
Lie RT Invited commentary: Intersecting perinatal mortality curves by gestational age – are appearances deceiving? Am J Epidemiol 2000 152 1117 9 11130616 10.1093/aje/152.12.1117
Cheung YB Yip P Karlberg J Mortality of twins and singletons by gestational age: a varying-coefficient approach Am J Epidemiol 2000 152 1107 16 11130615 10.1093/aje/152.12.1107
Wilcox AJ On the importance – and the unimportance – of birthweight Int J Epidemiol 2001 30 1233 41 11821313 10.1093/ije/30.6.1233
Joseph KS Liu S Demissie K Wen SW Platt RW Ananth CV Dzakpasu S Sauve R Allen AC Kramer MS for the Fetal and Infant Health Study Group of the Canadian Perinatal Surveillance System A parsimonious explanation for intersecting perinatal mortality curves: understanding the effect of plurality and of parity BMC Pregnancy Childbirth 2003 3 3 12780942 10.1186/1471-2393-3-3
Joseph KS Allen AC Lutfi S Murphy-Kaulbeck L Vincer MJ Wood E Does the risk of cerebral palsy increase or decrease with increasing gestational age? BMC Pregnancy Childbirth 2003 3 8 14693037 10.1186/1471-2393-3-8
Joseph KS Demissie K Platt RW Ananth CV McCarthy BJ Kramer MS A parsimonious explanation for intersecting perinatal mortality curves: understanding the effects of race and of maternal smoking BMC Pregnancy Childbirth 2004 4 7 15090071 10.1186/1471-2393-4-7
Joseph KS Incidence based measures of birth, growth restriction and death can free perinatal epidemiology from erroneous concepts of risk J Clin Epidemiol 2004 57 889 97 15504632 10.1016/j.jclinepi.2003.11.018
Joseph KS Theory of obstetrics: the fetuses at risk approach as a causal paradigm J Obstet Gynaecol Can 2004 26 953 6 15560858
Platt RW Joseph KS Ananth CV Grondines J Abrahamowicz M Kramer MS A proportional hazards model with time-dependent covariates and time-varying effects for analysis of fetal and infant death Am J Epidemiol 2004 160 199 206 15257989 10.1093/aje/kwh201
Gruenwald P Growth of the human fetus. I. Normal growth and its variation Am J Obstet Gynecol 1966 94 1112 9 5293993
Usher R McLean F Intrauterine growth of live-born Caucasian infants at sea-level: standards obtained from measurements in 7 dimensions of infants born between 25 and 44 weeks gestation J Pediatr 1969 74 901 10 5781799
David R Population-based intrauterine growth curves from computerized birth certificates South Med J 1983 76 1401 6 6635732
Ananth CV Vintzileos AM Shen-Schwarz S Smulian JC Lai Y-L Standards of birth weight in twin gestations stratified by placental chorionicity Obstet Gynecol 1998 91 917 24 9610996 10.1016/S0029-7844(98)00052-0
Brenner W Edelman D Hendricks C A standard of fetal growth for the United States of America Am J Obstet Gynecol 1976 126 555 64 984126
Lawrence C Fryer J Karlberg J Niklasson A Ericson A Modeling of reference values for size at birth Acta Paediatr Scand 1989 350 55 69
Gardosi J Chang A Kalyan B Sahota D Symonds E Customized antenatal growth charts Lancet 1992 339 283 7 1346292 10.1016/0140-6736(92)91342-6
Amini S Catalano P Hirsch V Mann L An analysis of birth weight by gestational age using a computerized perinatal data base, 1975–1992 Obstet Gynecol 1994 83 342 52 8127523
Zhang J Bowes W Jr Birth-weight-for-gestational-age patterns by race, sex, and parity in the United states population Obstet Gynecol 1995 86 200 208 7617350 10.1016/0029-7844(95)00142-E
Arbuckle T Wilkins R Sherman G Birth weight percentiles by gestational age in Canada Obstet Gynecol 1993 81 39 48 8416459
Maršál K Persson P-H Larsen T Lilja H Selbing A Sultan B Intrauterine growth curves based on ultrasonically estimated foetal weights Acta Paediatr 1996 85 843 8 8819552
Beeby PJ Bhutap T Taylor LK New South Wales population-based birthweight percentile charts J Paediatr Child Health 1996 32 512 8 9007782
Kramer MS Platt RW Wen SW Joseph KS Allen A Abrahamowicz M Blondel B Breart G for the Fetal/Infant Health Study Group of the Canadian Perinatal Surveillance System A new and improved population-based Canadian reference for birth weight for gestational age Pediatrics 2001 108 E35 11483845 10.1542/peds.108.2.e35
Källén B A birth weight for gestational age standard based on data in the Swedish Medical Birth Registry, 1985–1989 Eur J Epidemiol 1995 11 601 6 8549738
World Health Organization Physical status: the use and interpretation of anthropometry Report of a WHO expert committee Technical Report Series No 854 1995 Geneva: WHO
Lubchenco L Hansman C Dressler M Boyd E Intrauterine growth as estimated from liveborn birth weight data at 24 to 42 weeks of gestation Pediatrics 1963 32 793 800 14075621
Thomson A Billewicz W Hytten F The assessment of fetal growth J Obstet Gynaecol Br Common 1968 75 903 16
Alexander G Himes J Kaufman R Mor J Kogan M A United States national reference for fetal growth Obstet Gynecol 1996 87 163 8 8559516 10.1016/0029-7844(95)00386-X
Taffel S Johnson D Heuse R A method of imputing length of gestation on birth certificates Vital Health Stat 2 1982 93 1 11 7112964
MacDorman MF Atkinson JO Infant mortality statistics from the linked birth/infant death data set – 1995 period data Monthly Vital Statistics Report 1998 46 Hyattsville, MD: National Center for Health Statistics
Yudkin PL Wood L Redman CWG Risk of unexplained stillbirth at different gestational ages Lancet 1987 1 1192 4 2883499
Ferguson R Myers SA Population study of the risk of fetal death and its relationship to birth weight, gestational age, and race Am J Perinatol 1994 11 267 72 7945620
Hilder L Costeloe K Thilaganathan B Prolonged pregnancy: evaluating gestation-specific risks of fetal and infant mortality Br J Obstet Gynaecol 1998 105 169 73 9501781
Kramer MS Liu S Luo Z Yuan H Platt RW Joseph KS Analysis of perinatal mortality and its components: time for a change? Am J Epidemiol 2002 156 493 7 12225996 10.1093/aje/kwf077
Caughey AB Stotland NE Escobar GJ What is the best measure of maternal complications of term pregnancy: ongoing pregnancies or pregnancies delivered? Am J Obstet Gynecol 2003 189 1047 52 14586353 10.1067/S0002-9378(03)00897-4
Caughey AB Musci TJ Complications of term pregnancies beyond 37 weeks of gestation Obstet Gynecol 2004 103 57 62 14704245
Battaglia F Frazier T Hellegers A Birth weight, gestational age, and pregnancy outcome, with special reference to high birth weight-low gestational age infant Pediatrics 1966 37 417 22 5906367
Platt RW Abrahamowicz M Kramer MS Joseph KS Mery L Blondel B Breart G Wen SW Detecting and eliminating erroneous gestational ages: a normal mixture model Stat Med 2001 20 3491 503 11746333 10.1002/sim.1095
Hadlock FP Harrist RB Carpenter RJ Deter RL Park SK Sonographic estimation of fetal weight Radiology 1984 150 535 40 6691115
Nahum GG Stanislaw H Ultrasonographic prediction of term birth weight: how accurate is it? Am J Obstet Gynecol 2003 188 566 74 12592273 10.1067/mob.2003.155
Lerner JP Fetal growth and well-being Obstet Gynecol Clin North Am 2004 31 159 76 15062452
| 15720720 | PMC551609 | CC BY | 2021-01-04 16:32:05 | no | BMC Pregnancy Childbirth. 2005 Feb 20; 5:3 | utf-8 | BMC Pregnancy Childbirth | 2,005 | 10.1186/1471-2393-5-3 | oa_comm |
==== Front
BMC Public HealthBMC Public HealthBMC Public Health1471-2458BioMed Central 1471-2458-5-171572071610.1186/1471-2458-5-17Research ArticleLow adherence with antihypertensives in actual practice: the association with social participation – a multilevel analysis Johnell Kristina [email protected]åstam Lennart [email protected] Thor [email protected] Jan [email protected] Juan [email protected] Centre for Family Medicine, Karolinska Institutet, Huddinge, Sweden2 Department of Community Medicine, Malmö University Hospital, Lund University, Malmö, Sweden3 Regional Office, Skåne County Council, Lund, Sweden2005 18 2 2005 5 17 17 27 9 2004 18 2 2005 Copyright ©2005 Johnell et al; licensee BioMed Central Ltd.2005Johnell 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
Low adherence is a key factor in explaining impaired effectiveness and efficiency in the pharmacological treatment of hypertension. However, little is known about which factors determine low adherence in actual practice.
The purpose of this study is to examine whether low social participation is associated with low adherence with antihypertensive medication, and if this association is modified by the municipality of residence.
Methods
1288 users of antihypertensive medication were identified from The Health Survey in Scania 2000, Sweden. The outcome was low adherence with antihypertensives during the last two weeks. Multilevel logistic regression with participants at the first level and municipalities at the second level was used for analyses of the data.
Results
Low social participation was associated with low adherence with antihypertensives during the last two weeks (OR = 2.05, 95% CI: 1.05–3.99), independently of low educational level. However, after additional adjustment for poor self-rated health and poor psychological health, the association between low social participation and low adherence with antihypertensives during the last two weeks remained but was not conclusive (OR = 1.80, 95% CI: 0.90–3.61). Furthermore, the association between low social participation and low adherence with antihypertensives during the last two weeks varied among municipalities in Scania (i.e., cross-level interaction).
Conclusion
Low social participation seems to be associated with low adherence with antihypertensives during the last two weeks, and this association may be modified by the municipality of residence. Future studies aimed at investigating health-related behaviours in general and low adherence with medication in particular might benefit if they consider area of residence.
==== Body
Background
The effectiveness [1,2] and efficiency [3] of antihypertensives may be questioned, as adherence with antihypertensives may be as low as 50% [1,4]. The efficacy of antihypertensives has been evaluated in randomised clinical trials (RCTs). The RCTs are often of short duration, the study population is usually carefully selected and patients with co-morbidity or advanced ages are often excluded from these trials [5-8]. Furthermore, even though drop-outs and lost to follow up occur in RCTs, adherence with medication treatment is often actively supported. In actual practice, however, many patients, who would be excluded from RCTs, receive medication [2] for a long time and may not be as adherent with medication as those included in RCTs [5].
Low adherence is one important cause of uncontrolled hypertension [9,10]. Yet, low adherence is sometimes unrecognised [11] and is often interpreted as treatment resistance [10,12,13]. However, little is known about which factors determine low adherence in actual practice [14,15]. The purpose of this study is to examine whether low social participation is associated with low adherence with antihypertensives during the last two weeks, and if this association is modified by the municipality of residence.
Social participation is an important concept for understanding the influence of social factors on individual health and behaviour, and can be viewed as a feature of individual social networks [16]. Good social networks have been suggested to influence health behaviours, possibly through information exchange and establishment of health-related group norms [17]. Accordingly, a high level of social participation may facilitate adherence [16] and this association could be modified by the area of residence [18,19].
Multilevel analysis handles both the micro-scale of people and the macro-scale of context simultaneously within one model [20]. This analytic approach has been suggested as an interesting tool in pharmacoepidemiology [21].
The first aim of this study is to examine whether low social participation is associated with low adherence with antihypertensives during the last two weeks, independently of low educational level and health status (i.e., poor self-rated health and poor psychological health). The second aim is to analyse whether the hypothesised association between low social participation and low adherence with antihypertensives during the last two weeks varies between municipalities in Scania.
Methods
Participants
The Health Survey in Scania 2000 (HSS-2000) was a postal self-administered questionnaire sent out to a random sample of 23 437 individuals born from 1919 to 1981 living in Scania. The purpose of the HSS-2000 was to obtain information about health conditions and different types of health hazards among the inhabitants of Scania [22]. The province of Scania in southern Sweden has a population of about 1.2 million inhabitants and is divided into 33 municipalities. In total, 59% participated, of which 98% had complete information about medication use. The present study focused on those 9.6% who indicated use of antihypertensives during the last year and who had complete information about social participation (n = 1288).
The Ethical Committee at the Medical Faculty of Lund University approved the study proposal of The HSS-2000, and all of the participants received written information about the survey.
Outcome variable
Use of antihypertensives was based on an affirmative answer to the question "Have you during the last year used medicine, which was bought at the pharmacy...?" and indicating "Medication for high blood pressure"
Low adherence with antihypertensives during the last two weeks (dichotomised) was based on the question "Have you used (this) medicine during the last year, but not during the last 2 weeks?" Those participants who answered yes were considered to have low adherence.
Explanatory variables
Age was categorised in five groups: <35 (reference), 35–44, 45–54, 55–64 and ≥ 65 years.
Low social participation (dichotomised) was assessed after the respondent stated involvement in three or fewer activities (lowest quartile) of 13 formal or informal activities (study circle/course at work place, other study circle/course, union meeting, meeting of other organisations, theatre/cinema, arts exhibition, church, sports event, letter to editor of a newspaper/journal, demonstration, night club/entertainment, large gathering of relatives, private party), which the respondent might have participated in during the previous 12 months [23].
Low educational level (dichotomised) was defined as having nine years of education or less.
Poor self-rated health (dichotomised) was defined as a value of ≤ 3 on an ordinal self-rated health scale ranging from 1 ("Very bad") to 7 ("Very good") [24].
Poor psychological health (dichotomised) was determined by giving three or more affirmative answers to the 12 items composing the Standardised General Health Questionnaire (GHQ-12) [25].
Statistical analysis
Because of the hierarchy in the data, with individuals nested in municipalities, we used multilevel logistic regression [26] with individuals at the first level and municipalities at the second level. The area of residence might affect a person's social participation [27], and, consequently, it may be possible that the influence of low social participation on low adherence with antihypertensives during the last two weeks may vary between municipalities. Therefore, we let the slopes of the association between low social participation and low adherence vary at the municipality level. This random slopes analysis gives information about whether the association between low social participation and low adherence is different in different municipalities.
The first model i was created to study the influence of low social participation on low adherence with antihypertensives during the last two weeks, adjusting for age and sex. The second model ii was extended to also include low educational level, because low educational level could be a confounder in the association between low social participation and low adherence with antihypertensives during the last two weeks. The third model iii additionally contained poor self-rated health and poor psychological health. Impaired health may affect both low social participation and low adherence with antihypertensives.
Fixed effects
The results are shown as odds ratios (OR) with 95% confidence intervals (CI)
Random effects
We calculated the second level variance (variation between municipalities) regarding prevalence of low adherence with antihypertensives during the last two weeks (i.e., the intercepts in the multilevel regression), and the second level variance regarding the association between low social participation and low adherence with antihypertensives during the last two weeks (i.e., the slope variance in the multilevel regression). We also calculated the covariance between intercept and slope residuals. The covariance gives information about whether the association between low social participation and low adherence with antihypertensives during the last two weeks depends on the prevalence of low adherence in the different municipalities (i.e., cross-level interaction).
Parameters were estimated using the Restricted Iterative Generalized Least Squares (RIGLS) and penalised quasilikelihood (PQL). Extra-binomial variation was explored systematically in all models and we found no evidence for under- or over-dispersion. The MLwiN, Version 1.1 software package [28] was used for the analyses.
Results
Low adherence with antihypertensives during the last two weeks was found among 11% (145/1 288) of the participants and 49% (635/1 288) were classified as having low social participation. The participants mean age was 63 years. Those participants classified as having low social participation more often reported low adherence with antihypertensives during the last two weeks, low educational level, poor self-rated health and poor psychological health than those who were not classified as having low social participation (Table 1).
Table 1 Characteristics of the participants (n = 1288) according to individual low social participation.
Low social participation
Yes (n = 635)
No. (%) No (n = 653)
No. (%) Total (n= 1288)
No. (%)
Age (mean years) 65 60 63
Women 340 (54) 346 (53) 686 (53)
Low adherence with antihypertensives 96 (15) 49 (8) 145 (11)
Low educational level 417 (70) 284 (45) 701 (57)
Poor self-rated health 112 (19) 64 (10) 176 (14)
Poor psychological health 124 (21) 95 (15) 219 (18)
Fixed effects
Participants with low social participation had on average a more than twofold higher probability of reporting low adherence with antihypertensives during the last two weeks than those who did not have low social participation (OR = 2.28, 95% CI: 1.16–4.49) (Table 2). This association between low social participation and low adherence with antihypertensives during the last two weeks persisted after adjusting for low educational level (OR = 2.05, 95% CI: 1.05–3.99). However, after additional adjustment for poor self-rated health and poor psychological health, the association between low social participation and low adherence with antihypertensives during the last two weeks was not conclusive using a 95% confidence interval (OR = 1.80, 95% CI: 0.90–3.61).
Table 2 Municipality variance and age adjusted odds ratios (95% confidence intervals) of low adherence with antihypertensives during the last two weeks in relation to sex, low social participation, low educational level, poor self-rated health and poor psychological health.
Model i Model ii Model iii
OR 95% CI OR 95% CI OR 95% CI
Women vs. men 1.21 (0.82–1.78) 1.15 (0.78–1.69) 1.17 (0.78–1.77)
Low social participation (yes vs. no) 2.28 (1.16–4.49) 2.05 (1.05–3.99) 1.80 (0.90–3.61)
Low educational level (yes vs. no) 1.87 (1.18–2.96) 1.76 (1.09–2.84)
Poor self-rated health (yes vs. no) 1.45 (0.83–2.54)
Poor psychological health (yes vs. no) 1.54 (0.92–2.59)
Variance SE Variance SE Variance SE
Municipality variance in low adherence with antihypertensives (intercept variance) 0.801 (0.461) 0.776 (0.450) 0.793 (0.467)
Municipality variance of the association between low social participation and low adherence with antihypertensives (slope variance) 1.812 (0.889) 1.720 (0.857) 1.799 (0.916)
Municipality covariance between intercepts and slopes -1.163 (0.609) -1.116 (0.591) -1.175 (0.625)
Random effects
We found a variance between the municipalities in both low adherence with antihypertensives during the last two weeks (intercept variance) and in the association between low social participation and low adherence with antihypertensives during the last two weeks (slope variance) (Table 2 and Figure 1). The negative covariance between intercepts and slopes (Table 2) suggested that the associations between low social participation and low adherence with antihypertensives during the last two weeks (slopes) in the 33 municipalities depended on different prevalence of low adherence with antihypertensives during the last two weeks in the different municipalities. The association between low social participation and low adherence with antihypertensives during the last two weeks (slope) was weaker in municipalities with higher prevalence than in municipalities with lower prevalence of low adherence with antihypertensives during the last two weeks (i.e., cross-level interaction).
Figure 1 Slope variance in the association between low social participation and low adherence with antihypertensives during the last two weeks among 33 municipalities in Scania, Sweden.
Discussion
Main findings
Our results suggest that low social participation is associated with low adherence with antihypertensives during the last two weeks, independently of low educational level. In other words, the association between low social participation and low adherence withstood adjustment for socio-economic position (expressed by educational level) in our analyses. Social participation might therefore be considered as a real construct, and not only a proxy for socio-economic position. However, the association between low social participation and low adherence with antihypertensives during the last two weeks was weakened after additional adjustment for poor self-rated health and poor psychological health. Furthermore, the association between low social participation and low adherence with antihypertensives during the last two weeks may vary between municipalities in Scania.
The weakening of the association between low social participation and low adherence with antihypertensives during the last two weeks after we adjusted for poor self-rated health and poor psychological health may be an expression for confounding. Impaired health may negatively affect both social participation and adherence with antihypertensives. On the other hand, the observed reduction of the association may instead be telling us that physical and mental health are in the pathway between low social participation and low adherence with antihypertensives during the last two weeks [29]. Social participation may be considered as an early factor in the causal pathway that determines individual health-related behaviour, such as low adherence with medication. Social networks, which are connected to social participation, may promote shared norms around health behaviours, as treatment adherence, which could explain the pathway between social networks and impaired health [16].
Our present finding that the association between low social participation and low adherence with antihypertensives during the last two weeks may vary between municipalities in Scania gives empirical support to the existence of cross-level interactions (i.e., between municipality and individual) associated with health-related behaviours, such as low adherence with medication. These behaviours may be a result of the interaction between a person and his or her area of residence [18]. In a previous analysis, we observed that both individual and neighbourhood social participation are associated with individual impaired health and with use of hormone replacement therapy in women [18].
The present study shows that multilevel regression analysis can be used for investigation of geographical disparities in health and health-related behaviour (e.g, adherence with medication), without analysing any specific area characteristic [30]. In multilevel analysis, area effects can be investigated by measures of variance and by examining how area boundaries modify individual level associations [31].
Limitations of the study
The rather low participation rate (59%) may increase the risk of selection bias and reduce the ability to generalise the results and compare them to other populations. Nevertheless, the participation rate for participants aged 51–80 was about 65% and the participants using antihypertensives had a mean age of 63 years.
The low adherence question we used was stated in a non-threatening manner, which might facilitate for participants to give an honest response and not underreport low adherence [32]. Self-reported adherence has been reported to correlate with clinical measures of disease activity and control [4]. Moreover, self-report offers a convenient and non-invasive estimate of adherence behaviour. Nevertheless, the procedure of measuring adherence is controversial. Self-report can be subject to self-presentational and recall biases. People may overestimate their adherence and their memory may be inaccurate [32]. We might have reduced memory bias in this study by restricting the recall time to two weeks. However, the prevalence of current low adherence (11%) in this study is lower than low adherence reported in a longer period of time, which may be as high as 50% [1,4]. Therefore, our results may underestimate the association between social participation and low adherence with antihypertensives. It is possible that some participants with high adherence in the last two weeks had low adherence in other periods of the year. If this kind of misclassification would be more frequent among participants with low social participation, there would be differential misclassification, and the association between low social participation and adherence with antihypertensives could be underestimated. Non-differential misclassification would also underestimate the association between low social participation and low adherence with antihypertensives. Other ways of measuring adherence may be more appropriate, such as Morisky's four-item scale [33], which will be used in the Health Survey for Scania 2004.
People with low social participation and low adherence with antihypertensives during the last two weeks may have been less inclined to respond to the HSS-2000 questionnaire. This possible selection bias could lead to an underestimation of the association between low social participation and low adherence with antihypertensives during the last two weeks.
Conclusion
Our results suggest that low social participation is associated with low adherence with antihypertensives during the last two weeks, independently of low educational level. In addition, the association between low social participation and low adherence with antihypertensives during the last two weeks seems to vary between the municipalities in Scania, which gives empirical support to the existence of cross-level interactions (i.e., between municipality and individual) associated with health-related behaviours, such as low adherence with medication. We have recently showed that factors related to the area of residence influence the individual blood pressure level, especially in people using antihypertensive medication [34], which is in concordance with the results of this present study.
Future studies aimed at investigating health-related behaviours in general and low adherence with medication in particular might benefit if they consider that area of residence may modify associations between individual variables.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JM and KJ developed the original idea, participated in the design of the study, performed the statistical analyses and drafted the manuscript. LR and JS participated in the design of the study and revised the manuscript. TL participated in the design of the study, helped to collect 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:
http://www.biomedcentral.com/1471-2458/5/17/prepub
Acknowledgements
This study was funded by a ALF-Government Grant, Dnr M: E 39 390/98 (Juan Merlo), the Swedish council for working life and social research, the Swedish Medical Research Council and National Institutes of Health (R01 HL71084-01). We thank the Swedish Network for Pharmacoepidemiology (NEPI); the Pharmacoepidemiological Council at the County of Scania and Min Yang for useful comments about multilevel modelling.
==== Refs
Garfield FB Caro JJ Compliance and hypertension Curr Hypertens Rep 1999 1 502 506 10981113
Thurmer HL Lund-Larsen PG Tverdal A Is blood pressure treatment as effective in a population setting as in controlled trials? Results from a prospective study J Hypertens 1994 12 481 490 8064174
Ambrosioni E Pharmacoeconomic challenges in disease management of hypertension J Hypertens Suppl 2001 19 Suppl 3 S33 40 11713849
Kravitz RL Hays RD Sherbourne CD DiMatteo MR Rogers WH Ordway L Greenfield S Recall of recommendations and adherence to advice among patients with chronic medical conditions Arch Intern Med 1993 153 1869 1878 10.1001/archinte.153.16.1869 8250648
Evans A Kalra L Are the results of randomized controlled trials on anticoagulation in patients with atrial fibrillation generalizable to clinical practice? Arch Intern Med 2001 161 1443 1447 10.1001/archinte.161.11.1443 11386894
Remington RD Potential impact of exclusion criteria on results of hypertension trials Hypertension 1989 13 I66 8 2490830
Black N Why we need observational studies to evaluate the effectiveness of health care Bmj 1996 312 1215 1218 8634569
Altman DG Bland JM Generalisation and extrapolation Bmj 1998 317 409 410 9694763
Joshi PP Salkar RG Heller RF Determinants of poor blood pressure control in urban hypertensives of central India J Hum Hypertens 1996 10 299 303 8817403
Burnier M Long-term compliance with antihypertensive therapy: another facet of chronotherapeutics in hypertension Blood Press Monit 2000 5 Suppl 1 S31 4 10904240
Mushlin AI Appel FA Diagnosing potential noncompliance. Physicians' ability in a behavioral dimension of medical care Arch Intern Med 1977 137 318 321 10.1001/archinte.137.3.318 843149
Cleemput I Kesteloot K Economic implications of non-compliance in health care Lancet 2002 359 2129 2130 10.1016/S0140-6736(02)09114-6 12090976
Miller NH Hill M Kottke T Ockene IS The multilevel compliance challenge: recommendations for a call to action. A statement for healthcare professionals Circulation 1997 95 1085 1090 9054774
Payne KA Esmonde-White S Observational studies of antihypertensive medication use and compliance: is drug choice a factor in treatment adherence? Curr Hypertens Rep 2000 2 515 524 11062596
Caro JJ Salas M Speckman JL Raggio G Jackson JD Persistence with treatment for hypertension in actual practice Cmaj 1999 160 31 37 9934341
Berkman LF Glass T Brissette I Seeman TE From social integration to health: Durkheim in the new millennium Soc Sci Med 2000 51 843 857 10.1016/S0277-9536(00)00065-4 10972429
Kawachi I Berkman LF Berkman LF and Kawachi I Social Cohesion, Social Capital, and Health Social Epidemiology 2000 New York, Oxford University Press 174 190
Merlo J Lynch JW Yang M Lindstrom M Ostergren PO Rasmusen NK Rastam L Effect of neighborhood social participation on individual use of hormone replacement therapy and antihypertensive medication: a multilevel analysis Am J Epidemiol 2003 157 774 783 10.1093/aje/kwg053 12727671
Kidd KE Altman DG Adherence in social context Control Clin Trials 2000 21 184S 7S 10.1016/S0197-2456(00)00076-3 11018573
Duncan C Jones K Moon G Context, composition and heterogeneity: using multilevel models in health research Soc Sci Med 1998 46 97 117 10.1016/S0277-9536(97)00148-2 9464672
McMahon AD Approaches to combat with confounding by indication in observational studies of intended drug effects Pharmacoepidemiol Drug Saf 2003 12 551 558 10.1002/pds.883 14558178
Merlo J Lithman T Noreen D Melander A Users of medication in Scania 2000 (in Swedish) 2001 Lund, Skåne County Council
Statistics Sweden Living conditions. Isolation and togetherness - an outlook on social participation 1976 1980 Report no 18. Statistics Sweden
Eriksson I Unden AL Elofsson S Self-rated health. Comparisons between three different measures. Results from a population study Int J Epidemiol 2001 30 326 333 10.1093/ije/30.2.326 11369738
Tait RJ French DJ Hulse GK Validity and psychometric properties of the General Health Questionnaire-12 in young Australian adolescents Aust N Z J Psychiatry 2003 37 374 381 10.1046/j.1440-1614.2003.01133.x 12780478
Snijders TAB Bosker RJ Multilevel analysis - an introduction to basic and advanced multilevel modeling 1999 Thousand Oaks, California, SAGE Publications
Lindstrom M Merlo J Ostergren PO Individual and neighbourhood determinants of social participation and social capital: a multilevel analysis of the city of Malmo, Sweden Soc Sci Med 2002 54 1779 1791 10.1016/S0277-9536(01)00147-2 12113435
Rasbash J Browne W Goldstein H Yang M Plewis I Healy M Woodhouse G Draper D Langford I Lewis T A user's guide to MLwiN. Version 2.1. 2000 United Kingdom, Multilevel Models Project, Institute of Education, University of London
Mitchell R Gleave S Bartley M Wiggins D Joshi H Do attitude and area influence health? A multilevel approach to health inequalities Health Place 2000 6 67 79 10.1016/S1353-8292(00)00004-6 10785349
Merlo J Multilevel analytical approaches in social epidemiology: measures of health variation compared with traditional measures of association J Epidemiol Community Health 2003 57 550 552 10.1136/jech.57.8.550 12883048
Merlo J Chaix B Yang M Lynch J Rastam L A brief conceptual tutorial of multilevel analysis in social epidemiology: linking the statistical concept of clustering to the idea of contextual phenomenon J Epidemiol Community Health (in press)
Horne R Weinman J Patients' beliefs about prescribed medicines and their role in adherence to treatment in chronic physical illness J Psychosom Res 1999 47 555 567 10.1016/S0022-3999(99)00057-4 10661603
Morisky DE Green LW Levine DM Concurrent and predictive validity of a self-reported measure of medication adherence Med Care 1986 24 67 74 3945130
Merlo J Asplund K Lynch J Rastam L Dobson A Population Effects on Individual Systolic Blood Pressure: A Multilevel Analysis of the World Health Organization MONICA Project Am J Epidemiol 2004 159 1168 1179 10.1093/aje/kwh160 15191934
| 15720716 | PMC551610 | CC BY | 2021-01-04 22:25:48 | no | BMC Public Health. 2005 Feb 18; 5:17 | utf-8 | BMC Public Health | 2,005 | 10.1186/1471-2458-5-17 | oa_comm |
==== Front
BMC Struct BiolBMC Structural Biology1472-6807BioMed Central London 1472-6807-5-41571004010.1186/1472-6807-5-4Research ArticleCα-H···O=C hydrogen bonds contribute to the specificity of RGD cell-adhesion interactions Bella Jordi [email protected] Martin J [email protected] Wellcome Trust Centre for Cell Matrix Research, Faculty of Life Sciences, University of Manchester2005 14 2 2005 5 4 4 18 8 2004 14 2 2005 Copyright © 2005 Bella and Humphries; 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 Arg-Gly-Asp (RGD) cell adhesion sequence occurs in several extracellular matrix molecules known to interact with integrin cell-surface receptors. Recently published crystal structures of the extracellular regions of two integrins in complex with peptides containing or mimicking the RGD sequence have identified the Arg and Asp residues as key specificity determinants for integrin recognition, through hydrogen bonding and metal coordination interactions. The central Gly residue also appears to be in close contact with the integrin surface in these structures.
Results
When hydrogen atoms are modelled on the central Gly residue with standard stereochemistry, the interaction between this residue and a carbonyl group in the integrin surface shows all the hallmarks of Cα-H···O=C hydrogen bonding, as seen in the collagen triple helix and in many crystal structures of small organic molecules. Moreover, molecular dynamic simulations of the docking of RGD-containing fragments on integrin surfaces support the occurrence of these interactions. There appears to be an array of four weak and conventional hydrogen bonds lining up the RGD residues with main chain carbonyl groups in the integrin surface.
Conclusions
The occurrence of weak Cα-H···O=C hydrogen bonds in the RGD-integrin interaction highlights the importance of the conserved Gly residue in the RGD motif and its contribution to integrin-ligand binding specificity. Our analysis shows how weak hydrogen bonds may also play important biological roles by contributing to the specificity of macromolecular recognition.
==== Body
Background
The Arg-Gly-Asp (RGD) sequence is one of the most easily recognised motifs in molecular biology [1]. Discovered in fibronectin in 1984 [2], this tripeptide appears to be conserved in the cell attachment sites of many proteins from the extracellular matrix (ECM). The later discovery that RGD is recognised by members of the integrin family of cell surface receptors [3], confirmed the central role of RGD and suggested that its presence in a protein sequence might be indicative of cell-adhesion functionality [4]. Integrins are ubiquitously expressed heterodimer cell surface molecules that act as receptors for ECM molecules and other cell-surface adhesins. Through these cell-matrix and cell-cell interactions integrins control diverse cell functions such as adhesion, shape, growth, differentiation and mobility, and therefore contribute to important physiological processes such as development, immune responses and cancer [5]. Integrins are complex signalling engines: their extracellular domains interact with the ECM while their cytoplasmic tails interact with the cytoskeleton and other intracellular signalling molecules. Current hypotheses suggest that conformational changes resulting from these interactions enable integrins to transmit signals across the membrane in both directions. Recent advances in the structural biology of several integrin domains and their interactions with ligands have begun to define possible working scenarios for the signalling mechanisms [6-13].
As a consequence of their role in so many fundamental processes, integrin defects have been implicated in many common diseases, from cancer to pathogen invasion. An ability to block a particular integrin-ligand interaction may be a possible route to the control of certain pathological states, hence it is not surprising that some integrins have become attractive targets for drug design. Understanding the molecular bases of the interaction of integrins with their ligands is therefore essential for effective protein-based design of inhibitors or activators of their function. A milestone was reached in 2002 with the determination of the crystal structure of the extracellular segment of αVβ3 integrin in complex with a cyclic peptide containing the prototypical RGD sequence [8]. In that structure, the amino acids defining the RGD sequence are seen to establish specific interactions with corresponding residues in the integrin heterodimer surface, spanning the interface between the αV and β3 subunits (Figure 1a). Very recently, another landmark paper has reported several crystal structures of the extracellular region of the fibrinogen-binding integrin αIIbβ3 [12]. In addition to providing an improved picture of the allosteric basis of integrin signal transmission, this new set of structures shows the molecular details of the interaction between the αIIbβ3 RGD-binding site and various ligand mimetics (Figure 1b). These interactions are remarkably consistent with those previously observed in the complex between the αVβ3 integrin fragment and the cyclic RGD peptide (cRGD) [8].
At first glance, two interactions consistently seen in these crystal structures appear to be key in defining the specific molecular recognition between the RGD sequence in an integrin ligand and the surface of its integrin receptor: the Asp residue of the RGD triad completes the coordination of a divalent metal ion bound to the β subunit, while the Arg side chain extends in the opposite direction to form salt-bridge hydrogen bonds with one or two Asp residues in the α subunit. These two specific interactions or their equivalent are seen both in the cRGD-αVβ3 structure and in the structures of αIIbβ3 in complex with ligand-mimetics (Figure 1). There are no significant hydrophobic "pockets" or exosites contributing to the binding specificity. For example, a large fraction of the cRGD peptide does not make any contact with the αVβ3 integrin surface (Figure 1a). In the broader context of RGD-containing ligands and their integrin receptors, it would seem that these interactions are mainly electrostatic and that the two charged residues in the RGD sequence are necessary and sufficient for attachment [14].
What about the central Gly residue? In their analysis of the cRGD-αVβ3 crystal structure, Xiong et al. report that the Gly central residue makes several hydrophobic interactions with the integrin surface, including a contact with the carbonyl oxygen of residue Arg216 in the β3 integrin subunit [8]. Such contact between the Gly methylene group and a main-chain carbonyl oxygen is also observed in the crystal structure of αIIbβ3 in complex with the peptidomimetic eptifibatide (EFB) [12] (Figure 1b), which is a cyclic heptapeptide containing a homoArg-Gly-Asp sequence. The particular geometry of these contacts is strikingly reminiscent of a motif previously described in the collagen triple helix: a hydrogen bonding arrangement where the α-carbon of the Gly residue acts as hydrogen bonding donor in Cα-H···O=C interactions (Figure 2) [15].
So-called "weak" hydrogen bonds, such as those between carbon and oxygen atoms, have been traditionally neglected in descriptions of three-dimensional structures of macromolecules. Yet, C-H···O hydrogen bonds are ubiquitous in protein structures: virtually every conventional N-H···O=C hydrogen bond in every β-sheet in every determined protein structure carries a companion Cα-H···O=C interaction [16,17]. This applies to both parallel and antiparallel β-sheets, and exactly the same topology is also observed in the collagen triple helix [15]. For collagen and the β-sheet structures, the occurrence of Cα-H···O=C interactions is indicative of a very tight fit between the molecules involved, a close-packed structure in which all groups participate in some form of hydrogen bonding interaction.
How important are Cα-H···O=C and other weak hydrogen bonds in shaping the three-dimensional structure of proteins and macromolecular complexes? The subject has stimulated considerable debate (see [18] and [19] for reviews), although theoretical studies leave no doubt about the cohesive nature of these interactions [20-23]. With a strength approximately one-half of that from conventional hydrogen bonds, it seems reasonable to assume that the large numbers of weak hydrogen bonds detected in proteins may contribute to their stability. Furthermore, several biochemical functions have been linked to specific C-H···O hydrogen bonds, where position is more important that numbers. One example is the Gly-X-X-X-Gly motif, known to favour helix-helix interactions in membrane [24] and soluble proteins [25]via position-specific Cα-H···O=C hydrogen bonds. Another is the proposed role of C-H···O hydrogen bonds from cytosine and thymine bases to amino acid side chains during DNA-protein recognition [26]. Weak C-H···O hydrogen bonds have also been surveyed at protein-protein interfaces [27], and have been reported to play specific roles in catalysis [28], and in substrate and inhibitor recognition [29-32]. Recently, a server to identify weak hydrogen bonding interactions in protein structures has been made publicly available [33].
The functional occurrence of weak C-H···O hydrogen bonds in protein-ligand, protein-protein, and protein-DNA recognition suggests that their presence should be examined in detail in the structures of macromolecules with biomedical or biotechnological interest. Their potential should not be neglected in rational drug design approaches [31]. With this in mind, we present here an analysis of possible Gly-Cα-H···O=C interactions between RGD motifs and the RGD-binding sites from the αVβ3 and αIIbβ3 crystal structures. We conclude that the mutual geometry of the interaction is consistent with Cα-H···O=C hydrogen bonding. We discuss the implications of these hydrogen bonds for the cell adhesion interactions between integrins and their RGD-containing ligands.
Results and discussion
Building standard-geometry Hα atoms on the Gly central residues of the cRGD and EFB peptides produces the geometric arrangements shown in Figure 2, clearly reminiscent of the hydrogen bonding pattern previously described in the collagen triple helix (Figure 2c). The metrics of these Gly-carbonyl contacts are shown in Table 1. The Cα···O distances in the cRGD-αVβ3 and EFB-αIIbβ3 structures appear to be longer than the mean Cα···O distance in collagen, but are well within the observed range in crystal structures of small organic molecules (see below). Both Hα atoms from the collagen Gly residues are in hydrogen bonding position (Cα–Hα···O > 90°), and their Cα-Hα···O=C hydrogen bonds adopt a three-centred and bifurcated configuration (Figure 2c and [15]), that is not seen in the integrin structures. Nevertheless, the central Gly residues in the cRGD and EFB peptides appear to have one and two Hα atoms respectively in hydrogen bonding position to the carbonyl group of Arg216, a residue on the surface of the β3 subunit and directly at the interface with the αV and αIIb subunits.
An obvious caveat to this analysis comes from the moderate resolution of the cRGD-αVβ3 and EFB-αIIbβ3 crystal structures (3.2 Å and 2.9 Å respectively). Positional errors inevitable at that resolution may affect the precision of the fitting of the cRGD and EFB peptides and the accuracy of the hydrogen bonding geometries for both weak and strong hydrogen bonds. For example, a close look at the salt-bridge interactions between the Arg guanidinium group from the cRGD peptide and two Asp side chains on the αV integrin surface (Asp150 and Asp218), shows less than "ideal" hydrogen bonding orientation, especially for Asp150 (not shown). Yet, the accumulated knowledge of hydrogen bonding geometries in high-resolution crystal structures and their significant variability leaves no doubt about the existence of these strong hydrogen bonds and their contribution to the specificity of binding.
A similar level of confidence can be achieved for the Gly-Cα-Hα···O=C hydrogen bond by analysing the metrics of equivalent interactions in high-resolution crystal structures of small organic and organometallic molecules. Figure 3 shows the two fragment probes used in a statistical search for Gly-Cα-Hα···O=C nonbonded interactions in the Cambridge Structural Database (see Methods). Figure 4 shows that single hydrogen bonding (only one angle Cα-Hα···O ≥ 90°) predominates over the bifurcated case, and that a broad maximum in the Cα···O distribution occurs at about 3.4 Å, which can be taken as the "hydrogen bonding distance" for this type of interaction. This value is consistent with the theoretical value of 3.34 Å for the Gly-Cα-Hα···OH2 hydrogen bond from ab initio quantum calculations [22]. Figure 5 shows the distributions of Hα···O distances and Cα-Hα···O angles for the single hydrogen bond. The Hα···O distribution shows a broad maximum around 2.7 Å, whereas the angular distribution is very broad with maxima around 110° and 140°. Average parameters for single and double Gly-Cα-Hα···O=C hydrogen bonding (Table 1) are perfectly compatible with those calculated for the cRGD-αVβ3 and EFB-αIIbβ3 structures respectively, even though the accuracy of the values shown in Table 1 is clearly overestimated with respect to the resolution of these crystal structures. Thus, strictly from a geometrical point of view, the contacts between the Gly residues in the cRGD and EFB peptides and the main chain carbonyl group from Arg216 in the integrin surface bear all the characteristics of Cα-Hα···O=C hydrogen bonding. This observation is consistent with the exceptionally high frequency of intermolecular Gly-Cα-H···O=C hydrogen bonds recently reported in high resolution crystal structures of protein-ligand complexes [32].
A simple molecular docking analysis further supports the occurrence of Gly-Cα-H···O=C hydrogen bonds between RGD-containing ligands and integrin binding sites. In a first set of calculations, an RGD tripeptide was docked into the binding sites of both αVβ3 and αIIbβ3 integrins using constrained molecular dynamics (MD). Two constraints were imposed in the docking calculations: the carboxyl group from the Asp residue had to complete the metal coordination on the β3 subunit, and the Arg side chain had to form a salt bridge with appropriate Asp residues in the αV and αIIb subunits (see Methods), as observed in the crystal structures of αVβ3 and αIIbβ3 with different ligand-mimetics. Ten slightly different RGD models were obtained from the NMR structures of the adhesion domain of fibronectin [34], and were placed about 10 Å away from the integrin surface. Then these RGD models were subject to MD simulations until they docked into the integrin binding sites. In a second set of calculations, a longer peptide fragment with sequence VTGRGDSPAS from the adhesion domain of fibronectin was also docked into the binding sites of the two integrins (Figure 6). Again, ten different models for this peptide were obtained from fibronectin NMR structures [34]. Most of the simulations converged to models with Cα···O contact distances between the central Gly residue and the carbonyl of Arg216 in the 2.7–3.7 Å range (Figure 6b), with either one or two Gly-Hα atoms in hydrogen bonding orientation. These models were also the most favourable energetically (Table 2). From these calculations it seems to emerge that the RGD binding sites of the αVβ3 and αIIbβ3 integrins are primed to place the central Gly residue in the RGD triad directly above the carbonyl group of Arg216 of the β3 subunit (as observed in the cRGD-αVβ3 and EFB-αIIbβ3 crystal structures), forming one or two Cα-H···O hydrogen bonds that complement the main metal-coordination and salt-bridge interactions from the Asp and Arg side chains.
How important are these weak C-H···O hydrogen bonds in stabilising the cRGD-αVβ3 and EFB-αIIbβ3 complexes? A quantitative analysis of C-H···O hydrogen bonding at protein-protein interfaces has shown that they have an important contribution to the association and stability of protein complexes, accounting for about one third of the total hydrogen bonding interaction energy [27]. In fact, some of the hydrophobic or van der Waals interactions usually invoked to explain stabilising close contacts between molecules can be described better as weak C-H···O hydrogen bonds. These occupy a middle ground between the highly directional, conventional hydrogen bonds, and the directionless van der Waals interactions [32].
The recurrent appearance of some weak hydrogen bonding topologies in many structures of proteins and at protein-protein interfaces also reinforces the notion that they have a significant contribution to macromolecular stability. The most common occurrence of C-H···O hydrogen bonds in protein structures is a widespread Cα-H···O=C hydrogen bond N-terminal to the conventional N-H···O=C hydrogen bond in β-sheets [16,17] and in the collagen triple helix [15]. In this structural motif (Figure 7a), the Cα-H donor group is in the residue immediately N-terminal to the one carrying the N-H donor group, and both share the same C = O group as acceptor, an arrangement sometimes referred as "bifurcated" hydrogen bond [17,18,27]. This bifurcated hydrogen bonding motif is also the most common occurrence of C-H···O hydrogen bond at protein-protein interfaces [27]. The situation in Figure 7b occurs when the residue N-terminal to the one carrying the N-H donor group is Gly, with one or two Hα from Gly being in hydrogen bonding position. The bifurcated hydrogen bond scenario also occurs in the cRGD-αVβ3 and EFB-αIIbβ3 structures, where the N-H group from the Asp residue in the RGD peptide donates a hydrogen bond to the main chain carbonyl group from Arg216 of β3. This hydrogen bond has very bad geometry in the cRGD-αVβ3 structure (distance H···O 2.69 Å, angle N-H···O 133°), but looks better in the EFB-αIIbβ3 structure (distance H···O 2.39 Å, angle N-H···O 144°). These deviations from ideal hydrogen bonding geometry might be consequence of the resolution of the crystal structures, but all the MD docking simulations described above result in N-H···O=C hydrogen bonds that are slightly longer (typical H···O distances 2.4–2.5 Å) and slightly less linear (typical N-H···O angles 140°-150°) than the average hydrogen bonds between peptide groups in protein secondary structures. Automatic computational docking calculations of known integrin ligands on structural models of αVβ3 and αVβ5 consistently predict this N-H···O=C hydrogen bond to occur whenever an N-H group is present in the proximity of the carboxylate moiety [35,36]. Had these simulations included all nonpolar hydrogens, the companion Cα-H···O=C hydrogen bonds from the Gly residues would also have been observed.
A quick inventory of hydrophobic interactions between the cRGD and EFB peptides and the αVβ3 and αIIbβ3 surfaces suggests an additional candidate for classification as C-H···O=C hydrogen bond, between the Cβ-Hβ group from the Asp residue of the cRGD and EFB peptides and the main chain C = O group from Asn215 in the β3 subunit. This weak hydrogen bond is adjacent to the stronger, conventional hydrogen bond between the main chain N-H group from Asn215 and Oδ2 from the Asp residue in the RGD motif. Thus, a total of four hydrogen bonds, weak and conventional, aligns the bottom of the cRGD and EFB peptides against the integrin surfaces (Figure 8), and complements the main interactions from the Asp carboxyl and Arg guanidinium groups to provide a higher binding specificity.
It is clear from the cRGD-αVβ3 and EFB-αIIbβ3 structures that any side chain other than Gly in the RGD triad would not allow it to fit snugly within the integrin binding site, with the resulting weakening of hydrogen bonding and van der Waals interactions. Furthermore the main chain conformation for the central Gly residue in the cRGD-αVβ3 structure falls in a region of the Ramachandran map that is not allowed to any L-amino acid residue. Thus, Gly residues at the centre of the RGD motif are essential for being small, for being able to adopt specific main chain conformations, and for being able to interact closely with the integrin surface via Cα-H···O=C hydrogen bonds. All three characteristics contribute to the integrin-binding specificity of Gly residues at the centre of RGD motifs.
Inasmuch as the cRGD-αVβ3 and EFB-αIIbβ3 structures remain valid models for the structural basis of integrin-RGD ligand-binding specificity, it is reasonable to assume that the weak Cα/β-H···O=C hydrogen bonds depicted in Figure 8 will also occur in RGD-based cell-adhesion interactions. A special feature of the integrin surface at the RGD-binding site is the presence of two main chain carbonyl groups exposed to the solvent in the β3 subunit: Asn215 and Arg216. In absence of ligands these groups will probably interact with water molecules through conventional hydrogen bonding interactions (as seen for example in the crystal structure of the cacodylate-bound form of αIIbβ3, PDB accession code 1TXV [12]). Upon ligand binding, the RGD residues will displace these waters and place one amide and two methylene groups in hydrogen bonding position to carbonyl groups, increasing the specificity of the RGD-integrin interaction through multipoint recognition (Figure 8). This strategy will obviously be exploited by many competitive inhibitors for the integrin RGD-binding site. For example it is possible to substitute the weaker Cα-H donors from the Gly residue by a conventional N-H group (Figure 7c). This strategy has been exploited already in the design of aza-peptide and azacarba-peptide RGD mimetics [37-39], several of them with nanomolar activity. Molecular modelling of the interaction of these peptides with αVβ3 and αVβ5 RGD binding sites predicts the hydrogen bonding topology shown in Figure 7c[36]. It is interesting to notice that even in the absence of a conventional hydrogen bonding donor, the carbonyl group Arg216 in the β3 subunit still may be acceptor for weak hydrogen bonds. In the crystal structure of αIIbβ3 in complex with tirofiban [12], a non-peptidomimetic inhibitor derived from L-tyrosine, the Cδ1 atom from the substituted Tyr ring is some 3.01 Å away from the carbonyl oxygen of the very same Arg216. If a hydrogen atom is built with standard geometry on Cδ1, the calculated Hδ1···O distance is 2.01 Å and the Cδ1-Hδ1···O angle is 172°, again hydrogen bonding-like metrics. How should this Cδ1···O=C contact be called? We think that a description in terms of weak C-H···O hydrogen bonding is in this case more accurate than referring to this interaction as simply hydrophobic.
Conclusions
We have analysed in detail recently published structural data on the interaction between the extracellular regions of two integrins and peptides containing or mimicking the RGD sequence [8,12]. From this analysis we conclude that Cα-H···O=C hydrogen bonds from the central Gly residue also contribute to the specificity of binding. Weak hydrogen bonds are traditionally overlooked when describing protein structures, although they probably contribute to their stability. We think that our analysis provides one of the most interesting examples of C-H···O hydrogen bonds playing an important biological role, and may contribute to reverse the current trend of neglect of these interactions. In a recent paper, Sarkhel and Desiraju suggest that Nature may take advantage of the weaker C-H···O hydrogen bonds to optimise the efficiency of protein-ligand interactions, with a larger number of interactions coming into play even at the expense of the strength of the individual interactions [32]. By using more interactions, they suggest, specificity of recognition is increased, and because individual interactions are weaker, reversibility is possible. Our analysis of the interaction between the cRGD and EFB peptides and the αVβ3 and αIIbβ3 integrin surfaces would seem to corroborate this suggestion.
Methods
Integrin binding sites and hydrogen building
The following crystal structure coordinates were downloaded from the Protein Data Bank [40]: αVβ3 integrin in complex with a cyclic RGD peptide (cRGD), PDB accession code 1L5G [8]; αIIbβ3 integrin structure at 2.7 Å resolution, PDB accession code 1TXV [12]; αIIbβ3 in complex with eptifibatide (EFB), PDB accession code 1TY6 [12]; αIIbβ3 in complex with tirofiban, PDB accession code 1TY5. Models for integrin RGD-binding sites on αVβ3 and αIIbβ3 were obtained by selecting coordinates from integrin residues within 10 Å from the bound peptides. For the αIIbβ3 binding site, coordinates of the corresponding residues in the 1TXV structure were used, as this crystal structure has a better resolution. Coordinates for metal ions and structural waters present in the RGD-binding sites but not interfering with the binding of cRGD or EFB were also maintained. Hydrogen atoms were built with standard stereochemistry for the cRGD and EFB peptides and for the integrin RGD-binding sites as defined above, using the program REDUCE [41]. For the purpose of the analysis presented here all hydrogen atoms discussed in this paper could be positioned with satisfactory accuracy and predictable orientation.
Molecular docking calculations
For the molecular docking calculations, conformational models for RGD and VTGRGDSPAS peptides were obtained from the NMR structures of the adhesion domain in fibronectin [34]. Ten conformational models were used for each peptide. Each model was first manually docked approximately into the coordinates of the binding sites of αVβ3 and αIIbβ3 integrins, using the cRGD-αVβ3 and EFB-αIIbβ3 structures for guidance. Then each docked model was pulled away to about 10 Å from the integrin surfaces, and was docked back into the integrin binding site via molecular dynamics (MD) simulations using the program CNS [42]. Five simulations were run for each model, to a total of 50 MD simulations for each peptide-integrin pairing. A set of distance restraints was applied to the docking MD simulations, as observed on the cRGD-αVβ3 and EFB-αIIbβ3 structures. The side chain of the Asp residue group was restrained to coordinate the bound metal ion in the RGD-binding sites and to receive a hydrogen bond from the amide group of Asn215, in the β3 subunit. The side chain of the Arg residue was restrained to form hydrogen bonds with residues Asp150 and Asp218 on the αV subunit or residue Asp224 on the αIIb subunit. Additional restraints were imposed in the MD simulations with the VTGRGDSPAS peptide: the ring of Pro172 was restrained to hydrophobic contact with the side chain of Lys125, in the β3 subunit, and the Cα atoms of the N- and C-terminal residues in the peptide model were restrained not to separate more than 5 Å from each other. The coordinates of the integrin binding sites were kept fixed in all the simulations, and only the peptides were allowed to refine by restrained MD and energy minimisation. All molecular models were analysed with the program CHAIN [43] in a Silicon Graphics workstation.
Analysis of hydrogen bonding geometry in crystal structures of organic molecules
A survey in the Cambridge Structural Database [44] (July 2003 release), was carried out for Cα···O contacts between glycine-like fragments and carbonyl groups (Figure 3).
List of abbreviations
ECM, extracellular matrix; RGD, Arg-Gly-Asp sequence; cRGD, cyclic pentapeptide with sequence Arg-Gly-Asp-D-Phe-N(Me)-Val; EFB, eptifibatide; PDB, Protein Data Bank; CSD, Cambridge Structural Database; MD, molecular dynamics.
Authors' contributions
J.B. conceived the study and carried out the analysis of the structural data and molecular docking calculations. Both authors participated in the design, coordination and writing of the manuscript. Both authors read and approved the final manuscript.
Acknowledgements
We acknowledge the authors of the original paper on the cRGD-αVβ3 structure, Jian-Ping Xiong, Thilo Stehle and M. Amin Arnaout for useful criticisms to an early version of this manuscript.
Figures and Tables
Figure 1 Binding of peptide ligands to the integrin surfaces. (a) Detail of the crystal structure of the extracellular region of αVβ3 integrin in complex with the cyclic pentapeptide Arg-Gly-Asp-D-Phe-N(Me)-Val [8]. The peptide (orange), sits across the interface between the αV (red) and β3 (green) integrin subunits, but only the three amino acids from the RGD triad make significant contact with the integrin surface. The Asp residue completes the coordination of one of the three Mn2+ ions (purple spheres) at the top of the β3 subunit. (b) Detail of the crystal structure of the extracellular region of αIIbβ3 integrin in complex with the cyclic peptide eptifibatide [12], showing very similar interactions. Hrg and Mpt indicate L-homoarginine and β-mercaptopropionic acid residues, respectively. Due to higher resolution, water molecules (cyan spheres) are seen in this structure to complete the coordination of the metal ions. Other colours as in panel a. Both figures have been prepared using SETOR [45].
Figure 2 Geometry of Gly-Cα···O=C interactions after Gly α-hydrogen atoms are placed in their stereochemically predicted positions for: (a) the cRGD-αVβ3 crystal structure, and (b) the EFB-αIIbβ3 crystal structure. Atoms are colour coded as follows: oxygen, red; nitrogen, blue; hydrogen, white; metal, purple; carbon from integrin in grey; and carbon from the cRGD and EFB peptides in orange. Cα-H···O hydrogen bonds are shown as green dashed lines. (c) Hydrogen bonding in the collagen triple helix [15]. Conventional hydrogen bonds are shown in yellow, Cα-H···O hydrogen bonds in green. The two Hα atoms in collagen Gly residues participate in a bifurcated and three-centred hydrogen bonding configuration. Naming of Gly-Hα atoms follows the convention that Hα1 is equivalent to Hα in L-amino acids.
Figure 3 Fragments used in searches for non-bonded interactions in the Cambridge Structural Database [44]. Å 3.8 Å Cα···O distance cutoff was applied in all searches. A hydrogen atom was deemed to be in hydrogen bonding position if the angle Cα-H···O ≥ 90°. Separate searches were conducted for the bifurcated hydrogen bond (both angles ≥ 90°), single Cα-H···O=C hydrogen bond (one angle ≥ 90°, the other < 90°) and no hydrogen bonds (both angles < 90°).
Figure 4 Distribution of Cα···O distances in the Cambridge Structural Database structures (July 2003 release), containing the motif depicted in Figure 2. Three cases are considered: single Cα-H···O hydrogen bond (light grey), bifurcated hydrogen bond (dark grey), and no hydrogen bond (white). The single hydrogen bonded case clearly dominates with 1688 hits overall, for 218 of the bifurcated case and 166 hits for the no hydrogen bond case. The maximum in the single hydrogen bond distribution around 3.4 Å suggests that value as the Cα···O hydrogen bonding distance, although a significantly large number of interactions can be still classified as hydrogen bonds at the longer Cα···O distances.
Figure 5 Distribution of H···O distances (a) and Cα-H···O angles (b) for the 1688 instances of single Gly-Cα-H···O hydrogen bonding in crystal structures of the Cambridge Structural Database (July 2003 release).
Figure 6 Results of molecular dynamics simulations of docking fibronectin RGD-containing peptides on to models of integrin surfaces. (a) Representation of the 15 lowest-energy models for the docking of the VTGRGDSPAS peptide on αIIbβ3 model surface. Integrin residues are shown in black whereas the 15 peptide models are shown in different colours. For peptide models, the only side chains shown are those from the RGD triad (indicated with red labels). In all models the central Gly residue in the RGD triad is located directly on top of the carbonyl group from Arg 216 in the β3 subunit (shown with blue label). (b) Distribution of Cα···O distances in the final models of the molecular dynamics simulations Distances computed between the carbonyl oxygen in Arg216 from the β3 subunit and the central Gly residue from the RGD triad.
Figure 7 Bifurcated hydrogen bonding topologies. (a) The ubiquitous bifurcated hydrogen bonding topology seen in β-sheets and also in the collagen triple helix. The peptide chain is depicted with the N-terminus to the left. (b) Variation of the same bifurcated topology when the residue N-terminal to the donor N-H group is Gly, as observed in the cRGD-αVβ3 and EFB-αIIbβ3 crystal structures when Gly Hα atoms are built with standard geometry. Either one or two of the Gly Hα atoms can be in hydrogen bonding position. (c) Variation of the same bifurcated topology when the CH2 group N-terminal to the donor N-H group is replaced by another N-H group. This situation occurs for example when Gly is substituted by aza-glycine [37].
Figure 8 An array of four hydrogen bonds, two N-H···O=C (in yellow) and two C-H···O=C (in green), line up the bottom of the cRGD peptide against the integrin surface. The two weak Cα/β-H···O=C interactions thus contribute to the specificity of binding and presumably also have a cooperative effect on stability. Colour scheme for atom types as in Figure 2.
Table 1 Interatomic distances and angles for proposed and observed Gly-Cα-Hα···O hydrogen bonding interactions.
PDB structures * CSD Average values †
1L5Ga 1TY6b 1CGDc CH2···O H-CH···O
distances (Å)
Cα···O 3.39 3.48 3.15 3.45 (0.19) 3.44 (0.20)
Hα1···O 3.28 3.22 2.79 3.06 (0.25) 3.94 (0.38)
Hα2···O 2.68 2.99 2.63 3.05 (0.23) 2.81 (0.31)
angles (°)
Cα–Hα1···O 86 94 100 106 (11) 50 (23)
Cα–Hα2···O 121 108 109 107 (11) 127 (23)
Hα1···O=C 136 139 110 118 (26) 119 (26)
Hα2···O=C 150 168 91 115 (27) 119 (24)
* Distances and angles measured on each structural model after standard-geometry Hα building on the central Gly residues. The accuracy of distances and angles in the integrin structural models is probably overestimated, due to the resolution of these structural determinations.
† Average values for instances of double (CH2···O) and single (H-CH···O) Gly-Cα-H···O hydrogen bonding in the Cambridge Structural Database (July 2003 release). Standard deviations in parentheses. See Methods and Figures 3, 4 and 5 for details of the search. In the single hydrogen bond case, column Hα 1 refers to the atom in hydrogen bonding position (Cα-Hα1···O ≥ 90°) and Hα 2 to the atom in non-bonding position (Cα-Hα2···O < 90°).
a RGD-αVβ3 crystal structure, PDB accession code 1L5G [8].
b EFB-αIIbβ3 crystal structure, PDB accession code 1TY6 [12].
c Average values from the crystal structure of the collagen-like peptide (Pro-Hyp-Gly)4-Pro-Hyp-Ala-(Pro-Hyp-Gly)5, PDB accession code 1CGD [15].
Table 2 Gly-Cα-H···O=C contact distances (Å) for the lowest-energy model in each set of molecular docking simulations. Underlined distances correspond to Hα atoms in hydrogen bonding orientation (angle Cα-H···O=C > 90°).
Cα···O Hα 1···O Hα2···O
RGD-αVβ3 3.23 2.19 3.90
RGD-αIIbβ3 3.21 4.03 2.29
VTGRGDSPAS-αVβ3 3.31 2.96 2.73
VTGRGDSPAS-αIIbβ3 3.00 2.51 2.66
==== Refs
Ruoslahti E The RGD story: a personal account Matrix Biol 2003 22 459 465 14667838 10.1016/S0945-053X(03)00083-0
Pierschbacher MD Ruoslahti E Cell attachment activity of fibronectin can be duplicated by small synthetic fragments of the molecule Nature 1984 309 30 33 6325925 10.1038/309030a0
D'Souza SE Ginsberg MH Burke TA Lam SC Plow EF Localization of an Arg-Gly-Asp recognition site within an integrin adhesion receptor Science 1988 242 91 93 3262922
Ruoslahti E RGD and other recognition sites for integrins Annu Rev Cell Dev Biol 1996 12 697 715 8970741 10.1146/annurev.cellbio.12.1.697
Hynes RO Integrins: bidirectional, allosteric signaling machines Cell 2002 110 673 687 12297042 10.1016/S0092-8674(02)00971-6
Emsley J Knight CG Farndale RW Barnes MJ Liddington RC Structural basis of collagen recognition by integrin α2β1 Cell 2000 101 47 56 10778855 10.1016/S0092-8674(00)80622-4
Xiong JP Stehle T Diefenbach B Zhang R Dunker R Scott DL Joachimiak A Goodman SL Arnaout MA Crystal structure of the extracellular segment of integrin αVβ3 Science 2001 294 339 345 11546839 10.1126/science.1064535
Xiong JP Stehle T Zhang R Joachimiak A Frech M Goodman SL Arnaout MA Crystal structure of the extracellular segment of integrin αVβ3 in complex with an Arg-Gly-Asp ligand Science 2002 296 151 155 11884718 10.1126/science.1069040
Takagi J Petre BM Walz T Springer TA Global conformational rearrangements in integrin extracellular domains in outside-in and inside-out signalling Cell 2002 110 599 611 12230977 10.1016/S0092-8674(02)00935-2
Mould AP Symonds EJ Buckley PA Grossmann JG McEwan PA Barton SJ Askari JA Craig SE Bella J Humphries MJ Structure of an integrin-ligand complex deduced from solution X-ray scattering and site-directed mutagenesis J Biol Chem 2003 278 39993 39999 12871973 10.1074/jbc.M304627200
Humphries MJ McEwan PA Barton SJ Buckley PA Bella J Mould AP Integrin structure: heady advances in ligand binding, but activation still makes the knees wobble Trends Biochem Sci 2003 28 313 320 12826403 10.1016/S0968-0004(03)00112-9
Xiao T Takagi J Coller BS Wang JH Springer TA Structural basis for allostery in integrins and binding to fibrinogen-mimetic therapeutics Nature 2004 432 59 67 15378069 10.1038/nature02976
Mould AP Humphries MJ Cell biology: adhesion articulated Nature 2004 432 27 28 15525967 10.1038/432027a
Gottschalk KE Kessler H The structures of integrins and integrin-ligand complexes: implications for drug design and signal transduction Angew Chem Int Ed Engl 2002 41 3767 3774 12386845 10.1002/1521-3773(20021018)41:20<3767::AID-ANIE3767>3.0.CO;2-T
Bella J Berman HM Crystallographic evidence for Cα-H···O=C hydrogen bonds in a collagen triple helix J Mol Biol 1996 264 734 742 8980682 10.1006/jmbi.1996.0673
Derewenda ZS Lee L Derewenda U The occurrence of C-H···O hydrogen bonds in proteins J Mol Biol 1995 252 248 262 7674305 10.1006/jmbi.1995.0492
Ho BK Curmi PMG Twist and shear in β-sheets and β-ribbons J Mol Biol 2002 317 291 308 11902844 10.1006/jmbi.2001.5385
Desiraju GR Steiner T The weak hydrogen bond in structural chemistry and biology 1999 Oxford: Oxford University Press
Steiner T The hydrogen bond in the solid state Angew Chem Int Ed 2002 41 48 76 10.1002/1521-3773(20020104)41:1<48::AID-ANIE48>3.0.CO;2-U
Gu Y Kar T Scheiner S Fundamental properties of the CH···O interaction: is it a true hydrogen bond? J Am Chem Soc 1999 121 9411 9422 10.1021/ja991795g
Vargas R Garza J Dixon DA Hay BP How strong is the Cα-H···O=C hydrogen bond? J Am Chem Soc 2000 122 4750 4755 10.1021/ja993600a
Scheiner S Kar T Gu Y Strength of the CαH··O hydrogen bond of amino acid residues J Biol Chem 2001 276 9832 9837 11152477 10.1074/jbc.M010770200
Cannizzaro CE Houk KN Magnitudes and chemical consequences of R3N+-C-H···O=C hydrogen bonding J Am Chem Soc 2002 124 7163 7169 12059242 10.1021/ja012417q
Senes A Ubarretxena-Belandia I Engelman DM The Cα-H···O hydrogen bond: a determinant of stability and specificity in transmembrane helix interactions Proc Natl Acad Sci U S A 2001 98 9056 9061 11481472 10.1073/pnas.161280798
Kleiger G Grothe R Mallick P Eisenberg D GXXXG and AXXXA: common α-helical interaction motifs in proteins, particularly in extremophiles Biochemistry 2002 41 5990 5997 11993993 10.1021/bi0200763
Mandel-Gutfreund Y Margalit H Jernigan RL Zhurkin VB A role for CH···O interactions in protein-DNA recognition J Mol Biol 1998 277 1129 1140 9571027 10.1006/jmbi.1998.1660
Jiang L Lai L CH···O hydrogen bonds at protein-protein interfaces J Biol Chem 2002 277 37732 37740 12119293 10.1074/jbc.M204514200
Derewenda ZS Derewenda U Kobos PM (His)Cε-H...O=C< hydrogen bond in the active sites of serine hydrolases J Mol Biol 1994 241 83 93 8051710 10.1006/jmbi.1994.1475
Kleiger G Eisenberg D GXXXG and GXXXA motifs stabilize FAD and NAD(P)-binding Rossmann folds through Cα-H···O hydrogen bonds and van der Waals interactions J Mol Biol 2002 323 69 76 12368099 10.1016/S0022-2836(02)00885-9
Pierce AC Sandretto KL Bemis GW Kinase inhibitors and the case for CH···O hydrogen bonds in protein-ligand binding Proteins 2002 49 567 576 12402365 10.1002/prot.10259
Klaholz BP Moras D C-H···O hydrogen bonds in the nuclear receptor RARγ – a potential tool for drug selectivity Structure (Camb) 2002 10 1197 1204 12220491 10.1016/S0969-2126(02)00828-6
Sarkhel S Desiraju GR N-H...O, O-H...O, and C-H...O hydrogen bonds in protein-ligand complexes: strong and weak interactions in molecular recognition Proteins 2004 54 247 259 14696187 10.1002/prot.10567
Babu MM NCI: a server to identify non-canonical interactions in protein structures Nucleic Acids Res 2003 31 3345 3348 12824323 10.1093/nar/gkg528
Copie V Tomita Y Akiyama S Aota S Yamada K Venable R Pastor R Krueger S Torchia D Solution structure and dynamics of linked cell attachment modules of mouse fibronectin containing the RGD and synergy regions: comparison with the human fibronectin crystal structure J Mol Biol 1998 277 663 682 9533887 10.1006/jmbi.1998.1616
Marinelli L Lavecchia A Gottschalk KE Novellino E Kessler H Docking studies on αvβ3 integrin ligands: pharmacophore refinement and implications for drug design J Med Chem 2003 46 4393 4404 14521404 10.1021/jm020577m
Marinelli L Gottschalk KE Meyer A Novellino E Kessler H Human integrin αVβ5: homology modeling and ligand binding J Med Chem 2004 47 4166 4177 15293989 10.1021/jm030635j
Gibson C Goodman SL Hahn D Hölzemann G Kessler H Novel solid-phase synthesis of azapeptides and azapeptoides via Fmoc-strategy and its application in the synthesis of RGD-mimetics J Org Chem 1999 64 7388 7394 10.1021/jo9906173
Sulyok GAG Gibson C Goodman SL Hölzemann G Wiesner M Kessler H Solid-phase synthesis of a nonpeptide RGD mimetic library: New selective αvβ 3 integrin antagonists J Med Chem 2001 44 1938 1950 11384239 10.1021/jm0004953
Goodman SL Hölzemann G Sulyok GAG Kessler H Nanomolar small molecule inhibitors for αvβ6, αvβ5, and αvβ3 integrins J Med Chem 2002 45 1045 1051 11855984 10.1021/jm0102598
Berman HM Westbrook J Feng Z Gilliland G Bhat TN Weissig H Shindyalov IN Bourne PE The Protein Data Bank Nucl Acids Res 2000 28 235 242 10592235 10.1093/nar/28.1.235
Word JM Lovell SC Richardson JS Richardson DC Asparagine and glutamine: using hydrogen atom contacts in the choice of side-chain amide orientation J Mol Biol 1999 285 1735 1747 9917408 10.1006/jmbi.1998.2401
Brünger AT Adams PD Clore GM DeLano WL Gros P Grosse-Kuntsleve RW Jiang J-S Kuszewski J Nilges M Pannu NS Read RJ Rice LM Simonson T Warren GL Crystallography and NMR system: a new software suite for macromolecular structure determination Acta Crystallogr D 1998 54 905 921 9757107 10.1107/S0907444998003254
Sack JS CHAIN – A crystallographic modeling program J Mol Graph 1988 6 224 225 10.1016/S0263-7855(98)80040-4
Allen FH The Cambridge Structural Database: a quarter of a million crystal structures and rising Acta Crystallogr 2002 B58 380 388 12037359
Evans SV SETOR: hardware lighted three-dimensional solid model representation of macromolecules J Mol Graphics 1993 11 134 138 10.1016/0263-7855(93)87009-T
| 15710040 | PMC551611 | CC BY | 2021-01-04 16:03:53 | no | BMC Struct Biol. 2005 Feb 14; 5:4 | utf-8 | BMC Struct Biol | 2,005 | 10.1186/1472-6807-5-4 | oa_comm |
==== Front
BMC Med Inform Decis MakBMC Medical Informatics and Decision Making1472-6947BioMed Central London 1472-6947-5-31571323110.1186/1472-6947-5-3Research ArticleComparison of artificial neural network and logistic regression models for prediction of mortality in head trauma based on initial clinical data Eftekhar Behzad [email protected] Kazem [email protected] Hassan Eftekhar [email protected] Mohammad [email protected] Ebrahim [email protected] Department of Neurosurgery, Sina Hospital, Tehran University, Tehran, Iran2 Department of Biostatistics and Epidemiology, Faculty of Public Health, Tehran University, Tehran, Iran3 Department of Public Health, Faculty of Public Health, Tehran University, Tehran, Iran2005 15 2 2005 5 3 3 24 8 2004 15 2 2005 Copyright © 2005 Eftekhar et al; licensee BioMed Central Ltd.2005Eftekhar 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 recent years, outcome prediction models using artificial neural network and multivariable logistic regression analysis have been developed in many areas of health care research. Both these methods have advantages and disadvantages. In this study we have compared the performance of artificial neural network and multivariable logistic regression models, in prediction of outcomes in head trauma and studied the reproducibility of the findings.
Methods
1000 Logistic regression and ANN models based on initial clinical data related to the GCS, tracheal intubation status, age, systolic blood pressure, respiratory rate, pulse rate, injury severity score and the outcome of 1271 mainly head injured patients were compared in this study. For each of one thousand pairs of ANN and logistic models, the area under the receiver operating characteristic (ROC) curves, Hosmer-Lemeshow (HL) statistics and accuracy rate were calculated and compared using paired T-tests.
Results
ANN significantly outperformed logistic models in both fields of discrimination and calibration but under performed in accuracy. In 77.8% of cases the area under the ROC curves and in 56.4% of cases the HL statistics for the neural network model were superior to that for the logistic model. In 68% of cases the accuracy of the logistic model was superior to the neural network model.
Conclusions
ANN significantly outperformed the logistic models in both fields of discrimination and calibration but lagged behind in accuracy. This study clearly showed that any single comparison between these two models might not reliably represent the true end results. External validation of the designed models, using larger databases with different rates of outcomes is necessary to get an accurate measure of performance outside the development population.
==== Body
Background
In recent years, outcome prediction studies have become the avante garde in many areas of health care research, especially in critical care and trauma. However acceptable models for outcome prediction have been difficult to develop [1]. According to Wyatt and Altman, to be useful, a predictive model must be simple to calculate, have an apparent structure and be tested in independent data sets with evidence of generality [2]. While this is a high standard, availability and popularity of portable computers, deprioritize the need for simplicity of the model and having an apparent structure.
Artificial neural networks (ANNs) are mathematical constructs modeled on interconnection of nodes (neurons) giving a loose association with the animal nervous system. [3]
ANNs employ nonlinear mathematical models to mimic the human brain's own problem-solving process. Just as humans apply knowledge gained from past experience to new problems or situations, a neural network takes previously solved examples to build a system of "neurons" that makes new decisions, classifications, and forecasts. [4]
ANNs are complex and flexible nonlinear systems with properties not found in other modeling systems. These properties include robust performance in dealing with noisy or incomplete input patterns, high fault tolerance, and the ability to generalize from the input data [5]. Neural networks excel at applications where pattern recognition is important, and precise computational answers are not required, such as forecasting weather, stock predicting, or speech recognition [6]. Reports in medical literature suggest that ANN models are applicable in diagnosing diseases such as myocardial infarction [7,8] pulmonary emboli detection [9], gastrointestinal hemorrhage [10], waveform analysis of EKGs [11], EEGs [12,13], and radiographic images [14]. ANNs have also been successfully applied in clinical outcome prediction of trauma mortality [1,15], surgical decision making on traumatic brain injury patients [16], recovery from surgery [17,18], outcome in pediatric meningococcal disease [19] and transplantation outcome [20].
Lang EW et al have compared ANN with Logistic Regression in prediction of outcome after severe head injury and concluded that the differences in the results obtained with the two models were negligible [21].
Almost all of the published articles indicate that the performance of ANN models and logistic regression models have been compared only once in a dataset and the essential issue of internal validity (reproducibility) of the models has not been addressed.
The objective of this study was to compare the performance of ANN and multivariate logistic regression models for prediction of mortality in head trauma based on initial clinical data and whether these models are reproducible. We used different variables even if they were interdependent.
Methods
Study population
Among 8452 trauma patients' records admitted to the emergency departments of six major university hospitals in Tehran from 23 August 1999 to 22 September 2000, the records of 1271 patients whose main trauma was head injury, were selected for this study. The selection of head trauma as the main trauma was based on the definition of principal diagnosis in the Uniform Hospital Discharge Data Set (UHDDS). It defines the principal diagnosis as "that condition established after study to be chiefly responsible for occasioning the admission of the patient to the hospital for care". For making determination of the main trauma more practical in the case of ambiguity, hospitalization in the neurosurgical ward was used as an additional guideline. The database was based upon the trauma data registry program began in 1996 in Trauma Research Center, Sina Hospital, a hospital affiliated with the Tehran University of Medical Sciences [22,23]. The study population for this study was comprised of all trauma victims who had been admitted in one of the hospitals for more than 24 hours during the data-gathering period. For dead patients, this time limitation was disregarded, that is, records of all dead patients who were referred to these hospitals were included in the study. We have excluded those transferred to other hospitals or with related missing values. Structured, closed-question data checklists were used for the data gathering process. Three major categories of injury-related information were collected, that is, demographic data, pre-hospital data (if they were available) and in-hospital data. Hospital related data included: vital signs, Glasgow Coma Scale (GCS), Abbreviated Injury Scale (AIS-90 [24]), clinical findings in accordance with the International Classification of Diseases 10th revision (ICD-10) as well as the outcome of the patients. Data collection was conducted by a group of trained physicians who had completed special training courses to become familiar with the process of extracting Abbreviated Injury Score (AIS-90) codes and filling out the relevant questionnaires. For quality control (QC) purposes each hospital had a physician, who was responsible for overseeing the data gathering process. In the intubated patients GCS were calculated according to the other portions of the scale by this physician. Finally, a medical practitioner examined all the checklists in order to evaluate and amend them if deemed necessary based on pre-arranged and fixed protocols.
Since we were trying to build and compare models for prediction of outcome mainly based on the initial clinical data, only data related to the GCS, tracheal intubation status, age, systolic blood pressure (SBP), respiratory rate(RR), pulse rate(PR), injury severity score (ISS)(upon admission) and outcome were used in our study. In order to prepare the data for the Neural Network software and to enhance the reliability of the data, three variables of systolic blood pressure, respiratory rate and pulse rate were transformed to dichotomous(1,0) variables. Low systolic blood pressure was defined according to the following cutoff points: up to 5 years of age, less than 80 mmHg; and 5 years of age or older, less than 90 mmHg. Respiratory rate of 35 per min and pulse rate of 90 per min were selected as limits for definition of tachypnea and tachycardia. Other variables including GCS, age, systolic blood pressure (SBP) and injury severity score (ISS) variables were also converted from decimal (Base 10) to binary (Base 2). This conversion was carried out in order to render the input data suitable for processing by our ANN software with its default settings. The data and the data format were similar for both ANN and logistic regression models.
Development of logistic regression models
The dataset was divided randomly into two sets, one set of 839 cases (66% of the whole dataset) for training and 432 cases for testing the model. A model was built using a training set with logistic regression. GCS, tracheal intubation status(dichotomous), age, SBP(dichotomous), RR(dichotomous), PR(dichotomous) and ISS were the independent variables and the outcome (death/survival) was the dependent variable. The logistic regression analyses were performed using Intercooled STATA for windows, Version 6 (STATA Corp., College Station, TX) "logistic" default options.
The built logistic model was tested using the testing dataset (432 cases). These steps (randomized division of dataset and regression analysis considering the same variables) were repeated 1000 times. This resulted in 1000 pairs of training and testing datasets (2/3 and 1/3 of the original dataset, respectively) which were saved for further processing by the neural networking.
Development of ANN models
The ANN used in this study was a standard feed-forward, back-propagation neural network with three layers: an input layer, a hidden layer and an output layer. The input layer consisted of 23 input neurons, the hidden layer consisted of fifteen hidden neurons, and the output layer consisted of one output neuron (Fig. 1). The learning rate and momentum for network training were set respectively to 0.25 and 0.9 and the models were run until a minimum average squared error < 0.063 was obtained. The number of the network layers, hidden neurons and the stopping criteria were determined through trial-and-errors process because no commonly accepted theory exists for predetermining the optimal number of neurons in the hidden layer [25].
Figure 1 Diagrammatic representation of Artifical Neural Network ANN) structure with 23 input nodes in the input layer, 15 nodes in the hidden layer and one node in the output layer.
The training and testing datasets were the same as those were used with regression models, thus there was an ANN and a logistic model for each training and testing dataset.
PDP++ version 3.1 was used for the artificial neural network analysis as it has a powerful built-in scripting language and is freely available. This software can be downloaded from .
Comparison of model performance
Discrimination and calibration (goodness of fit) were both measured. Discrimination refers to the ability of a model to distinguish those who die from those who survive. A perfectly discriminating model would assign a higher probability of death to all cases that died than to any case that survived. The discriminatory power of the models was analyzed using the area under the receiver operating characteristic curves (ROC-A(z)). ROC curves were constructed by plotting true positives (patients who died and whom the model predicted as dying [i.e. sensitivity]) versus the false positive fraction (fraction of the patients who lived and were incorrectly classified as dying [i.e. (1 – specificity)]).
A ROC-A(z) value of one corresponds to a test that perfectly separates two populations, whereas a ROC-A(z) value of 0.5 corresponds to a perfectly useless test that performs no better than chance.
The relative calibration of the models, that is, how accurately the models predicted over the entire range of severity, was compared using Hosmer-Lemeshow (HL) statistics. The HL statistics is a single summary measure of the calibration and is based on comparing the observed and estimated mortality for patients grouped by estimated mortality. The resulting statistic follows a chi-squared distribution, with degrees of freedom equal to two less than the number of groups (10 in this study). The smaller the HL statistics, the better the fit, with a perfectly calibrated model having a value of zero.
A probability cut point of 0.5 was used to classify observations as events or nonevents. The overall accuracy ([true positive+true negative]/total) of the final model was determined by comparing the predicted values with the actual events.
For each of one thousand pairs of ANN and logistic models (trained and tested on the same datasets), these indices (area under the ROC curves, HL statistics and accuracy rate) were calculated and compared using paired T-tests (P < 0.05).
All the statistical analyses were performed using Intercooled STATA for windows, version 6 STATA Corp., College Station, TX) and its downloadable add-on ado files.
Some of the scripts used both for the STATA and PDP++ and the designed neural networks can be downloaded from the site of the Tehran University of Medical Sciences . Further details are available from the corresponding author.
Results
Table 1 shows clinical characteristics of the dataset. The mean age of the study population was (28.5 ± 19) years. 76% of our patients were male and the overall mortality rate was 7.5%. 7.5% of the patients had GCS < 8.
Table 1 General Characteristics of the Dataset
GCS 13.5 ± 3
Sex 76% Male
ISS 9.5 ± 14
SBP 117 ± 23
RR 19 ± 7
PR 86 ± 17
Intubated 2%
Age 28.5 ± 19
Mortality 7.5%
GCS = Glasgow Coma Scale; ISS = Injury Severity Score; SBP = Systolic Blood Pressure; RR = Respiratory Rate; PR = Pulse Rate
As is seen in Table 2 ANN significantly outperformed logistic models in both senses of discrimination and calibration, although from the standpoint of accuracy (cutoff point 0.5), logistic models were superior to ANN models. In 77.8% of cases the area under the ROC curve and in 56.4% of cases the HL statistics for the neural network model were superior to that for the logistic model. In 68% of cases the accuracy of the logistic model was superior to the neural network model.
Table 2 Results of Comparing 1000 pairs of Artificial Neural Network (ANN) and Logistic Regression (LR) models
LR(95% Confidence Interval) ANN(95% Confidence Interval) P <
Area under ROC curve .9538 (.9527 – .9549) .9646 (.9627 – .9665) 0.0000
H-L Statistics 53.13 (46.04 – 60.23) 41.51 (32.92 – 50.11) 0.015
Accuracy Rate 96.37 (96.32 – 96.42) 95.09 (94.93 – 95.24) 0.0000
The confidence intervals of the all the measures show higher variation in the ANN models results.
Discussion
Prognostic evaluation of the patients without CT scan findings may have limited applicability considering the availability of CT scans in the majority of the trauma centers. Nevertheless, the idea of prognostic models based on initial clinical data at admission is still worth trying and seems to be of practical value in some situations. Although omission of CT data weakens our armamentarium significantly, patients without paraclinical and imaging data make the real trauma scenes at which medical staff should have an evaluation. Actually as neurosurgeons we have initial clinical judgments based on our growing experiences and in many situations they prove to be true. We can expect computers do similar things and help us. In fact, vague situations like what we see in primary evaluation of head trauma patients are where ANN may prove to be superior to traditional linear modeling. This is one of the reasons that although many paraclinical and imaging factors are known to be of significant predictive value in the outcome of the head trauma patients, this study only used clinical measures which are simply available to a physician in the emergency department.
Study limitations
Calculating ISS and AIS are not simple tasks and needs training. This can reduce the practicability of the results. The pupillary size and reactivity are one of the clinical signs with prognostic value that were not considered in the study due to defect of the main database in this regard.
As is seen the mortality rate and the percentage of patients with GCS < 8 are the same (7.5%). This is a coincidence, but emphasizes the need for further studies in the populations with different rates of outcome.
The mean GCS was 13.5 ± 3. The standard deviation is over the top score of GCS (15) and means that the GCSs were skewed towards higher levels of consciousness.
Not considering the exact time interval between head trauma and admission which is simply available for the medical staff is one of the weak points of this study.
The lower intubation rate in this population study related to the distribution of the GCS may show the lower quality of the pre-hospital care prevailing in the hospitals selected for this study. This may render the reproduction of the results using the same network (Downloadable from ) somewhat difficult in other situations where pre-hospital care services are more advanced.
Using ISS as an independent variable underlines the role of general trauma in the models used for this study. This was unavoidable, because the original dataset had been a subset of general trauma patients.
Comparison of two models
Currently, the logistic regression and the artificial neural networks are the most widely used models in biomedicine, as measured by the number of publications indexed in Pubmed as attested by 45646 cases for the logistic regression and 8015 for the neural network.
Logistic regression is a commonly accepted statistical tool, which can generate excellent models. Its popularity may be attributed to the interpretability of model parameters and ease of use, although it has limitations. For example, logistic regression models use linear combinations of variables and, therefore, are not adept at modeling grossly nonlinear complex interactions as has been demonstrated in biologic and complex epidemiologic systems.
Not withstanding its limitations, neural networks are appealing for a number of reasons, namely; they seem to "learn" without supervision, they can be created by workers with very little mathematical model building experience, and software for building neural networks is now readily available. Neural networks have perhaps a special appeal to the medical community because of their superficial resemblance to the human brain (a structure with which most physicians are comfortable), and seem to promise "prediction" without the difficulties associated with use of mathematics.
ANNs are rich and flexible nonlinear systems that show robust performance in dealing with noisy or incomplete data and have the ability to generalize from the input data. They may be better suited than other modeling systems to predict outcomes when the relationships between the variables are complex, multidimensional, and nonlinear as found in complex biological systems.
The difficulty in developing models using artificial neural networks is that there are no set methods for constructing the architecture of the network. The most common type of artificial neural networks is the feed-forward back propagation multiperceptron (used in this study).
Another limitation of neural network models is that standardized coefficients and odds ratios corresponding to each variable cannot be easily calculated and presented as they are in regression models. Neural network analysis generates weights, which are difficult to interpret as they are affected by the program used to generate them [26]. This lack of interpretability at the level of individual variables (predictors) is one of the most criticized features in neural network models [27].
Several early applications of neural networks in medicine reported an excellent fit of the ANN model to a given set of data. The impressive results usually were derived from over fitted models, where too many free parameters were allowed. Linear and logistic regression models have less potential for overfitting primarily because the range of functions they can model is limited.
Neural network models require sophisticated software. The complexity and unfamiliarity of ANN has been a major drawback of this technique so far. However, as palmtop computing becomes increasingly powerful and popular, the complexity of ANNs may become less onerous in real-time clinical settings. [4]
Furthermore, there are some theoretical advantages comparing a predictive ANN model over conventional models such as logistic regression.
One such advantage is that ANN model allows the inclusion of a large number of variables [28]. Another advantage of the neural network approach is that there are not many assumptions (such as normality) that need to be verified before the models can be constructed.
Although, one of the strengths of ANNs is their ability to still find patterns despite missing data, in this study a dataset with no related missing values was used.
Recently the task of comparison between these two models has been addressed from different points of view. Considering the publication bias, several published works in the medical literature have demonstrated the success of the ANN approaches. In a review carried out by Sargent on 28 major studies, ANN outperformed regression in 10 cases (36%), was outperformed by regression in 4 cases (14%) and the 2 methods had similar performance in the remaining cases. Sargent concluded that both methods should continue to be used and explored in a complementary manner. [29]
Gaudart et al. using simulated data, have compared the performance of ANN and linear regression models for epidemiological data and concluded that both had comparable performance and robustness and despite the flexibility of connectionist models (like ANN), their predictions were stable. [30]
This study was primarily designed to compare the performance of an ANN and a multivariable logistic regression analysis with the goal of developing a model for predicting the outcome in head injury and for studying their internal validity (reproducibility). Also setting up a standalone practical model for prediction of mortality in the head trauma patients was a secondary goal of this study.
Using freely downloadable software (PDP++) and making the networks and scripts accessible by the researchers can be perceived as an advantage of this study.
This study showed that ANN models significantly outperformed logistic models in both senses of discrimination and calibration, although lagged behind in accuracy. It is pointed out that the calibration values should be treated with some caution in this study, since according to Hosmer and Lemeshow, [31] in describing the statistic, HL statistic should be used where at least one of the predictor variables is continuous.
This study clearly shows that in a single comparison of these models based on the same data there is 22.2% chance of getting discrimination results contrary to our findings in majority of comparisons. This ratio is 43.6% for calibration and 32% for the accuracy results. These figures are practically important and imply that any single comparison between these two models cannot reliably represent their final performance.
Although considerable efforts, through many trial-and-errors, were made to optimize the design of the network, the designed ANN models could, and should, be further improved. In line with any other predictive models, likewise the findings of this study need to be externally validated. The networks are downloadable and the results can be studied in other study populations with divergent data and different survival ratios.
So far, there is no single algorithm that performs better than all other algorithms for any set of given head injury data. To this end, there is room for much more work to be done before a definite conclusion can be reached.
Potential clinical use
Should the results be reproducible in other populations, using a simple preprogrammed calculator (or other programmable computing devices) and minimal training of the personnel, this model and similar ones may emerge to be of considerable practical value in triage of the patients. At that time a dedicated instead of general purpose ANN software should be designed for this purpose.
The authors concur with the conclusion arrived by Tu [25] that logistic regression remains the clear choice when the primary goal of model development is to examine possible causal relationships among variables. However, it appears that ANNs or some form of hybrid technique incorporating the best features of both logistic regression and neural network models might lead to development of optimum prediction models for head injured patients.
Conclusions
In conclusion, this study compared models for the prediction of outcome in head injury using trauma data from hospital registries in Tehran, the data was applied to artificial neural network and multivariable logistic regression analysis. The predictive ability of the artificial neural network model was found to be comparable to that of the logistic regression model. Specifically, the ANN models significantly outperformed logistic models in both senses of discrimination and calibration but lagged behind in accuracy. Although the performance of the models were studied when the models were applied to the different samples of the original population study, external validation is necessary to get an accurate measure of performance outside the development population. Studies using larger databases with different rates of outcomes may further clarify the differences between artificial neural network and logistic regression models in head injury outcome prediction and their clinical implications.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BE carried out the data extraction, performed the analysis and drafted the manuscript. KM supervised the analysis and critically reviewed the statistical viewpoints. HEA, MG and EK supervised the study and participated in its coordination. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgment
The authors would like to express their gratitude to Mr. Ahmad Reza Eftekhar for his comments and assistance with the editing of this paper. We would also like to thank Mr. Sam Holden Research Assistant at the University of Sydney for his editorial help.
==== Refs
DiRusso SM Sullivan T Holly C Cuff SN Savino J An Artificial Neural Network as a Model for Prediction of Survival in Trauma Patients: Validation for a Regional Trauma Area J Trauma 2000 49 212 223 10963531
Wyatt JC Altman DG Prognostic models: clinically useful or quickly forgotten? Brit M J 1995 311 1539 1541 (comment)
Greenwood D An overview of neural networks Behav Sci 1991 36 1 33 1998494
Terrin N Schmid CH Griffith JL D'Agostino RB Selker HP External validity of predictive models: A comparison of logistic regression, classification trees, and neural networks Journal of Clinical Epidemiology 2003 56 721 729 12954463 10.1016/S0895-4356(03)00120-3
Patterson DW Artificial Neural Networks: Theory and Applications Englewood Cliffs, Prentice Hall 1996
Leondes CT Neural network systems, techniques, and applications 1998 San Diego, Academic Press
Baxt WG Use of an artificial neural network for data analysis in clinical decision-making: the diagnosis of acute coronary occlusion Neural Comput 1990 2 480 9
Baxt WG Use of an artificial neural network for the diagnosis of myocardial infarction Ann Intern Med 1990 115 843 8 1952470
Kemeny V Droste DW Hermes S Nabavi DG Schulte-Altedo-rneburg G Siebler M Ringelstein EB Automatic embolus detection by a neural network Stroke 1999 30 807 10 10187883
Das A Ben-Menachem T Cooper GS Chak A Sivak MV JrGonet JA Wong RC Prediction of outcome in acute lower-gastrointestinal hemorrhage based on an artificial neural network: internal and external validation of a predictive model Lancet 2003 362 1261 6 14575969 10.1016/S0140-6736(03)14568-0
Vijaya G Kumar V Verma HK ANN-based QRS-complex analysis of ECG J Med Eng Technol 1998 22 160 7 9680599
Kloppel B Application of neural networks for EEG analysis. Consideration and first results Neuropsychobiology 1994 29 39 46 8127423
Jando G Siegel RM Horvath Z Buzsaki G Pattern recognition of the electroencephalogram by artificial neural networks Electroencephalogr Clin Neuropsychol 1993 86 100 9 10.1016/0013-4694(93)90082-7
Penedo MG Carreira MJ Mosquera A Cabello D Computer-aided diagnosis: a neural-network-based approach to lung nodule detection IEEE Trans Med Imaging 1998 17 872 80 10048844 10.1109/42.746620
Izenberg SD Williams MD Luterman A Prediction of trauma mortality using a neural network Am Surg 1997 63 275 81 9036899
Li YC Liu L Chiu WT Jian WS Neural network modeling for surgical decisions on traumatic brain injury patients Int J Med Inf 2000 57 1 9 10.1016/S1386-5056(99)00054-4
Grigsby J Kooken R Hershberger J Simulated neural networks to predict outcomes, costs and length of stay among orthopedic rehabilitation patients Arch Phys Med Rehabil 1994 75 1077 81 7944911 10.1016/0003-9993(94)90081-7
Tu JV Guerriere MR Use of a neural network as a predictive instrument for length of stay in the intensive care unit following cardiac surgery Comput Biomed Res 1993 26 220 9 8325002 10.1006/cbmr.1993.1015
Nguyen T Malley R Inkelis S Kuppermann N Comparison of prediction models for adverse outcome in pediatric meningococcal disease using artificial neural network and logistic regression analyses Journal of Clinical Epidemiology 2002 55 687 695 12160917 10.1016/S0895-4356(02)00394-3
Dorsey SG Waltz CF Brosch L Connerney I Schweitzer EJ Barlett ST A neural network model for predicting pancreas transplant graft outcome Diabetes Care 1997 20 1128 33 9203449
Lang EW Pitts LH Damron SL Rutledge R Outcome after severe head injury: an analysis of prediction based upon comparison of neural network versus logistic regression analysis Neurol Res 1997 19 274 80 9192380
Zargar M Modaghegh MHS Rezaishiraz H Urban Injuries in Tehran: Demography of trauma patients and evaluation of trauma care Injury 2001 32 613 7 11587698 10.1016/S0020-1383(01)00029-8
Moini M Rezaishiraz H Zafarghandi MR Characteristics and outcome of injured patients treated in urban trauma centers in Iran J Trauma 2000 48 503 7 10744293
Des Plaines IL Association for the Advancement of Automotive Medicine. Abbreviated Injury Scale: 1990 Revision Association for the Advancement of Automotive Medicine 1990
Tu JV Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes J Clin Epidemiol 1996 49 1225 31 8892489 10.1016/S0895-4356(96)00002-9
Baxt WG Application of artificial neural networks to clinical medicine Lancet 1995 346 1135 8 7475607 10.1016/S0140-6736(95)91804-3
Ohno-Machado L Rowland T Neural network applications in physical medicine and rehabilitation Am J Phys Med Rehab 1999 78 392 8 10.1097/00002060-199907000-00022
Bishop C Neural networks for pattern recognition 1995 Oxford University Press
Sargent DJ Comparison of artificial neural networks with other statistical approaches Cancer (Supplement) 2001 91 1636 42 10.1002/1097-0142(20010415)91:8+<1636::AID-CNCR1176>3.0.CO;2-D
Gaudart J Comparison of the performance of multi-layer perceptron and linear regression for epidemiological data Computational Statistics and Data Analysis 2004 44 547 570 10.1016/S0167-9473(02)00257-8
Hosmer D Lemeshow S Applied logistic regression 2000 New York, Wiley
| 15713231 | PMC551612 | CC BY | 2021-01-04 23:52:12 | no | BMC Med Inform Decis Mak. 2005 Feb 15; 5:3 | utf-8 | BMC Med Inform Decis Mak | 2,005 | 10.1186/1472-6947-5-3 | oa_comm |
==== Front
BMC BiolBMC Biology1741-7007BioMed Central London 1741-7007-3-31570748310.1186/1741-7007-3-3Research ArticleEngineered G protein coupled receptors reveal independent regulation of internalization, desensitization and acute signaling Scearce-Levie Kimberly [email protected] Michael D [email protected] Heather H [email protected] Bruce R [email protected] The Gladstone Institute of Neurological Disease and the Gladstone Institute of Cardiovascular Disease, San Francisco CA 94158 USA2 Departments of Medicine and Molecular and Cellular Pharmacology, University of California, San Francisco, CA, 94143 USA2005 11 2 2005 3 3 3 23 8 2004 11 2 2005 Copyright © 2005 Scearce-Levie 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 physiological regulation of G protein-coupled receptors, through desensitization and internalization, modulates the length of the receptor signal and may influence the development of tolerance and dependence in response to chronic drug treatment. To explore the importance of receptor regulation, we engineered a series of Gi-coupled receptors that differ in signal length, degree of agonist-induced internalization, and ability to induce adenylyl cyclase superactivation. All of these receptors, based on the kappa opioid receptor, were modified to be receptors activated solely by synthetic ligands (RASSLs). This modification allows us to compare receptors that have the same ligands and effectors, but differ only in desensitization and internalization.
Results
Removal of phosphorylation sites in the C-terminus of the RASSL resulted in a mutant that was resistant to internalization and less prone to desensitization. Replacement of the C-terminus of the RASSL with the corresponding portion of the mu opioid receptor eliminated the induction of AC superactivation, without disrupting agonist-induced desensitization or internalization. Surprisingly, removal of phosphorylation sites from this chimera resulted in a receptor that is constitutively internalized, even in the absence of agonist. However, the receptor still signals and desensitizes in response to agonist, indicating normal G-protein coupling and partial membrane expression.
Conclusions
These studies reveal that internalization, desensitization and adenylyl cyclase superactivation, all processes that decrease chronic Gi-receptor signals, are independently regulated. Furthermore, specific mutations can radically alter superactivation or internalization without affecting the efficacy of acute Gi signaling. These mutant RASSLs will be useful for further elucidating the temporal dynamics of the signaling of G protein-coupled receptors in vitro and in vivo.
==== Body
Background
The specificity, diversity, and physiological importance of G protein-coupled receptors (GPCR) have made these receptors excellent drug targets. It is becoming clear that the regulation of the GPCR itself – its location, stability, and signal duration – is a key component of the signaling process [1,2] The length of a GPCR signal can be modulated by receptor desensitization (decrease in receptor responsiveness) and receptor internalization (trafficking of receptors to endocytotic vesicles). The cell can also respond to prolonged activation by upregulating compensatory pathways. For example, prolonged signaling through a Gi-coupled receptor inhibits adenylyl cyclase (AC), while paradoxically increasing the ability of the Gs-coupled pathway to stimulate AC, a phenomenon known as AC superactivation [3]. Such regulatory mechanisms may contribute to the development of drug tolerance and dependence, including the response to chronic opiate use [4].
The complex effects of drugs at multiple receptor subtypes in multiple tissues have made it difficult to isolate the relative contributions of GPCR regulation, ligand binding, effector coupling, drug metabolism, and cellular downregulation machinery. Even if two receptors couple to the same signaling pathway, the physiological effects of their activation could vary tremendously depending on the pharmacokinetics of the ligands, the cell type expressing the receptors, and the interactions with desensitization mechanisms. An engineered family of receptors that share the same ligand binding and effector coupling, yet have discrete mutations that cause them to internalize or desensitize differentially, would help pinpoint the physiological consequences of GPCR desensitization. This is especially important in the light of recent evidence showing dramatically different endocytosis and signaling profiles of mu opioid receptors (MOR) in response to different ligands [5,6]. In addition, understanding the signals that allow transmembrane proteins to be more or less resistant to endocytosis will improve our understanding of endocytosis as a general regulatory mechanism, as it has been implicated in the regulation of signaling of growth factor receptors [7,8] and ion channels [9-11].
Our laboratory has engineered a Gi-coupled receptor that is insensitive to endogenous ligands but can still respond to the synthetic small-molecule agonist spiradoline [12,13]. This receptor activated solely by a synthetic ligand (RASSL) was based on the kappa opioid receptor (KOR). In the original RASSLs, exchanging the second extracellular loop of the KOR with the corresponding sequence from the delta opioid receptor, and making an additional point mutation (Q297E), resulted in a receptor with 1/2,000 of the response to dynorphin and other endogenous peptides relative to the wild-type KOR. However, the response of this RASSL to spiradoline was not altered. RASSLs can be expressed in a tissue-specific manner in transgenic mice, allowing direct control of Gi-mediated physiological responses such as heart rate [14]. It has also been recently used to help identify the mammalian sweet receptor by expressing it in mouse taste buds [15]. To investigate the endocytosis and desensitization of GPCRs, we have since developed four new RASSLs. To visualize these RASSLs in living cells, we fused the green fluorescent protein (GFP) to the N-terminus (outer portion) of the RASSL resulting in Rog (RASSL opioid green). Given the well-documented role of the C-terminal region in the desensitization and internalization of GPCRs [1,16-18], we made a series of C-terminal mutant RASSLs designed to desensitize and internalize at different rates.
This novel receptor system offers an opportunity to test specific hypotheses about the relationship between receptor sequence and receptor regulation, without requiring the use of multiple ligands that might have different effects on the signal and the regulation of receptors. Because RASSLs lack endogenous agonists, they allow greater control of the timing and specificity of activation than is possible with endogenous receptors. In these studies, we test how the removal of phosphorylation sites from the C-terminal regions of a Gi-coupled RASSL alters the receptor's internalization, desensitization, and induction of AC superactivation. Since it is well established that the endogenous mu and kappa opioid receptors differ in these properties, we also explore the regulation of kappa opioid RASSLs bearing specific portions of the mu opioid receptor C-terminal sequence. The cell culture experiments presented here provide a basis for in vivo studies in complex tissues such as the nervous system. Insight gained through these experiments may help explain the differences seen in vivo between different endogenous Gi-coupled receptors, improving our understanding of the contribution of receptor regulation to the physiological response to agonists and our overall understanding of GPCR signal regulation.
Results
Rog, a GFP-tagged RASSL, signals appropriately
Although an N-terminal GFP tag does not interfere with the function of the human KOR [19], we wanted to confirm that this tag does not modify the signaling properties of Rog, a KOR-based RASSL. Rog was transiently transfected into CHO cells along with a chimeric Gqi5 protein [20] that couples to Gi-coupled receptors but signals through the Gq pathway and therefore stimulates calcium release. Using this transient calcium response as a measure of Gi activation, we showed by FLIPR assay that Rog responded dose-dependently to spiradoline, but not to a range of doses of dynorphin, the endogenous ligand that activates the wild-type KOR (Figure 2A). Therefore, Rog, like its predecessors Ro1 and Ro2 [12], meets the criteria for a RASSL.
We evaluated all of the engineered receptors for their ability to affect cAMP formation. Gi signaling decreases intracellular cAMP levels by directly inhibiting adenylyl cyclase. Activation of these Gi-coupled receptors with spiradoline was therefore expected to inhibit forskolin-induced cAMP accumulation in a dose-dependent manner. Indeed, under basal conditions, 15-min treatment with spiradoline activated Rog, Rog-A, Rog-μ and Rog-μA, as indicated by inhibition of forskolin-induced accumulation of cAMP (Figure 5, Table 1). Despite differences in C-terminal amino acid sequence, the RASSLs showed no significant differences in their ability to inhibit cAMP accumulation after acute activation with spiradoline (Table 1). EC50 and cAMP inhibition values for all of the RASSLs were similar to those seen with the human KOR in the same assay. In a representative experiment, the EC50 for KOR was 0.91 nM spiradoline and cAMP inhibition was 68.6%.
Rog is internalized by agonist treatment
The GFP tag on Rog allows direct observation of agonist-induced receptor internalization by confocal microscopy (Figure 2B). In untreated cells, the receptor was visible primarily on the plasma membrane. One hour after spiradoline treatment (10–100 μM), the receptor was observed in bright, punctate intracellular vesicles. Dynorphin (100 μM), in contrast, did not lead to significant internalization of the receptor (Figure 2B). An ELISA that detects only cell-surface receptors was used to quantify the extent of receptor internalization after spiradoline treatment. With increasing doses of spiradoline, fewer receptors were detected on the cell surface, culminating in an approximately 45% loss of cell-surface receptors at the maximal dose of 100 μM (Figure 2C). The same dose of spiradoline resulted in a similar 47% loss of KOR from the cell surface (not shown). The time course of internalization in response to 1 μM spiradoline was relatively rapid, with significant receptor loss apparent within 5 min (Figure 2C). Maximal receptor loss was detected approximately 20 min after agonist treatment began.
Rog-A is resistant to agonist-induced internalization
To determine the role of C-terminal phosphorylation sites in receptor regulation, we examined spiradoline-induced internalization of Rog-A, a mutated version of Rog in which four C-terminal phosphorylation sites were mutated to alanine (Figure 1). HEK293 cells stably transfected with Rog-A were treated with 10 μM spiradoline, a dose sufficient to cause internalization of most Rog receptors (Figure 3A, left). One hour after spiradoline treatment, most Rog-A receptors appeared to remain in the membrane (Figure 3A, center). Quantification by cell-surface ELISA showed significantly less loss of cell-surface receptors for Rog-A than for Rog at spiradoline doses of 0.1–100 μM (Figure 3B). ANOVA indicated a main effect of drug dose (F10,20 = 66.53, p < 0.0001) and a main effect of receptor type (F1,20 = 55.29, p < 0.0001). As observed with Rog, maximal internalization of Rog-A in response to 1 μM spiradoline occurred after 20 min of drug treatment (Figure 3B). However, in contrast to Rog, fewer than 10% of the Rog-A receptors were internalized at that time point. ANOVA of the time course data indicated a main effect of length of treatment (F11,47 = 11.39, p < 0.0001), a main effect of receptor type (F1,47 = 203.16, p < 0.0001), and an interaction between receptor type and treatment length (F11,47 = 2.27, p < 0.02). These results suggest that C-terminal phosphorylation promotes receptor internalization. Activation of Rog-A inhibits cAMP as fully as Rog (Table 1), indicating that the reduced internalization does not alter acute signaling.
Rog-μ more readily internalizes in response to spiradoline
Since the mu opioid receptor (MOR) internalizes more readily than the KOR, we made Rog-μ, a chimeric receptor in which the entire intracellular portion of the C-terminus was replaced with the corresponding MOR sequence (Figure 1). Rog-μ was expected to internalize to a greater extent than Rog in response to spiradoline. Confocal microscopy showed nearly complete internalization of Rog-μ after one hour of treatment with 10 μM spiradoline (Figure 3A, right). A cell-surface ELISA revealed 25–30% internalization of Rog-μ at low doses of spiradoline, ranging from 0.01 to 0.1 μM (Figure 3C). Little internalization of Rog or Rog-A has been observed at these doses (Figures 3B and 3C). At higher doses of spiradoline, no difference in internalization between Rog and Rog-μ was observed. ANOVA indicated a main effect of drug dose (F9,52 = 55.79, p < 0.0001) and an interaction between receptor type and drug dose (F9,52 = 3.89, p < 0.0008). Post-hoc Scheffé analysis shows significant differences between Rog and Rog-μ at 0.01 μM (p = .024) and 0.1 μM (p = .016) doses of spiradoline. When cells were treated with 1 μM spiradoline for differing lengths of time, there was no detectable difference in the time course of internalization of Rog and Rog-μ (Figure 3C). ANOVA indicated a main effect of time (F11,48 = 53.20, p < 0.0001), with no effect of receptor type. There was also no difference in cAMP inhibition after spiradoline activation of Rog and Rog-μ (Table 1).
Constitutive internalization of Rog-μA
A variant of Rog-μ, known as Rog-μA, also has MOR sequence at the C-terminus, but five serine and glutamic acid residues at the C-terminus were mutated (Figure 1). These mutations were predicted to render Rog-μA more resistant to internalization than Rog-μ [18]. However, in several independent stably and transiently transfected cell lines (HEK293 and rat1a), Rog-μA always had significantly lower cell-surface expression than other receptors. As shown by cell-surface ELISA of one representative group of stably transfected HEK293 lines, Rog-μA was expressed at 28% of the level of Rog, and 41% of Rog-μ (Figure 4A). Despite the low cell-surface expression, Rog-μA signals as well as the other RASSLs after acute spiradoline treatment (Table 1).
Since cell-surface expression of a GPCR can be stabilized by the addition of antagonist [21,22], we examined the effect of the KOR antagonist norBNI on cell-surface expression of Rog-μA. Antagonist treatment nearly doubled the amount of Rog-μA detected in the membrane (Figure 4A; p < 0.005; F1,6 = 27.00). It also increased the cell-surface expression of Rog-μ, but to a lesser degree (p < 0.05, F1,6 = 11.54). In contrast, it had no effect on the cell-surface expression of Rog. The increase in membrane expression of Rog-μA after antagonist treatment was confirmed by confocal microscopy. Under basal conditions, little Rog-μA was seen in the plasma membrane, although the receptor was readily detected in other areas of the cell (Figure 4B). After overnight treatment with 10 μM norBNI, most of the Rog-μA was seen in the plasma membrane (Figure 4B). In contrast, untreated Rog receptors were primarily located in the plasma membrane (Figures 2B, 4B), and norBNI treatment had little effect on their localization (Figure 4B). These experiments suggest that Rog-μA may be constitutively downregulated and rapidly cycled in and out of the plasma membrane.
Rog-A is more resistant to desensitization
In addition to regulation by internalization, a GPCR signal can be modulated by desensitization: uncoupling from the signaling effectors after continuous agonist stimulation. To explore desensitization directly, we briefly pretreated each RASSL with 1 nM spiradoline for 15 min, and examined inhibition of cAMP accumulation in response to a variety of doses of spiradoline. The low pretreatment dose had caused no receptor internalization detectable by ELISA-based assays. Pretreatment reduced the responsiveness of Rog receptors to spiradoline (Figure 5A, Table 1). The same maximal inhibition of cAMP accumulation was observed, but the dose response curve was shifted approximately 10-fold, with the EC50 for Rog shifting from 0.41 nM to 4.32 nM spiradoline after spiradoline pretreatment. Pretreatment of Rog-A with the same dose of spiradoline, however, did not significantly affect the response of the cells to subsequent treatment (Figure 5A, Table 1). The EC50 for spiradoline after 1 nM spiradoline pretreatment of Rog-A was 0.42 nM, compared to 0.41 nM for vehicle-treated cells. Spiradoline pretreatment shifted the EC50 of Rog-μ to 2.05 nM (Figure 5A, Table 1). Notably, the maximal response of pretreated Rog-μ-expressing cells was less than half that of untreated cells, indicating a decreased efficacy of Rog-μ signaling through the Gi pathway. Spiradoline pretreatment strongly reduced the response of Rog-μA to further spiradoline treatment (Figure 5A, Table 1). In fact, it appears that the response to spiradoline in pretreated Rog-μA cells is so low that the dose range tested (up to 100 nM) does not yield a maximal inhibition of cAMP, and no sigmoidal dose-response curve can be fitted to these data. Therefore, we cannot calculate an accurate EC50 for desensitized Rog-μA receptors. However, assuming that the maximal response occurs at doses higher than 100 nM spiradoline, we can estimate that the EC50 would be at least 13.44 nM. This indicates that Rog-μ and Rog-μA receptors desensitize readily. For these receptors, the dose of spiradoline required to achieve the EC50 is significantly lower than the dose required to internalize the cell-surface receptors (Figures 4, 5). While only a fraction of the total surface receptor pool needs to be activated to activate Gi maximally, a much larger fraction of the receptor population must be internalized before it can be accurately measured.
AC superactivation is independent of receptor internalization
Chronic treatment of cells expressing Gi-coupled receptors with agonist results in a compensatory increase in the activity of AC and, therefore, an increased accumulation of cAMP in response to the same dose of forskolin [3]. We examined the development of this AC superactivation in cell lines transiently expressing the RASSL variants. Forskolin (10 μM) stimulates twice as much cAMP in Rog-expressing cells treated with 10 nM spiradoline for 18 hours, compared to cells acutely treated with forskolin alone (Figure 5B; p < 0.005, F1,10 = 17.72). A similar degree of superactivation was seen in cells transfected with the wild-type KOR, indicating the same cellular response to prolonged Gi signaling through both Rog and KOR. Overnight treatment of Rog-A-expressing cells with spiradoline, followed by stimulation with 10 μM forskolin, resulted in a slightly smaller increase in cAMP (Figure 5B; p < 0.05, F1,10 = 9.52). Notably, cells expressing Rog-μ and Rog-μA receptors showed no evidence of AC superactivation after 18 h of spiradoline pretreatment. These data show that receptors that desensitize and internalize more readily at the receptor level, such as Rog-μ and Rog-μA, do not induce compensations in an opposing signaling pathway.
Although little internalization of these RASSLs has been observed at these low doses of spiradoline, we wanted to ensure that the AC superactivation data could not be explained by differences in receptor internalization. Therefore, we performed an analysis of cell-surface receptor expression in parallel with the cAMP response experiment. Cells were plated and treated with 10 nM spiradoline for 18 hours exactly as described above. Then the cells were fixed and a cell-surface ELISA was performed. In general, 7–10% of the receptors were internalized by this treatment, but there were no significant differences between receptor types (Figure 5C). ANOVA showed a significant treatment effect (F1,18 = 8.22, p < 0.01), but no effect of receptor type (p > 0.99).
Discussion
We have engineered a series of RASSLs that inhibit cAMP after acute activation by spiradoline with equal efficacy but differ dramatically in cellular location and responses to chronic drug treatment.
Mutation of phosphorylation sites on the C-terminus to alanines resulted in a receptor that was relatively resistant to internalization in the presence of moderate doses of agonist (Rog-A, Figure 3) and showed no significant desensitization after pretreatment with a low dose of agonist (Figure 5A). This is consistent with recent findings that mutation of a single serine to alanine is sufficient to block internalization and desensitization of the KOR, since this mutation removes a residue that is required for G protein receptor kinase (GRK2) phosphorylation [23]. While these studies highlight the importance of GRK phosphorylation of GPCRs in mediating receptor internalization and desensitization, it is notable that partial internalization of Rog-A was still detected in response to higher doses of spiradoline, indicating that the receptor can be internalized through different mechanisms. Reports of GPCR endocytosis in the absence of GRK phosphorylation [24-26] suggest that the removal of C-terminal phosphorylation sites may reduce the affinity of the receptor for proteins that mediate endocytosis without preventing the protein-protein interactions that are essential for internalization.
Rog-μ, the MOR/KOR chimeric mutant, was more sensitive to agonist-induced desensitization (Figure 5A) and internalization (Figure 3) than Rog. This is consistent with observations that the MOR internalizes and desensitizes more readily than the KOR. Surprisingly, mutation of C-terminal phosphorylation sites on the MOR/KOR chimera to form Rog-μA did not inhibit desensitization (Figure 5A). Rog-μA has low basal surface expression, although the GFP-tagged receptor can be seen throughout the cell (Figure 4B). It is unlikely that the extracellular mutations in Rog-μA are responsible for the unusual internalization pattern of this receptor. The extracellular mutations in Rog-μA are identical to the ones in Rog, which was shown to reach the cell surface and respond to agonist the same as GFP-tagged wild-type KOR (Figure 2). The intracellular pool of Rog-μA is not due to receptor misfolding or abnormal sorting, because some receptor was detected on the membrane (Figure 4A) and the receptor showed normal agonist-induced signaling (Figure 5A, Table 1). The observation that the addition of an antagonist can "rescue" the low cell-surface expression of Rog-μA (Figure 4A) further suggests that misfolding is not responsible for the decrease in cell surface expression.
There are several potential mechanisms for the increase in cell-surface expression after antagonist treatment. One possibility is that the antagonist, norBNI, acts as a molecular chaperone, entering the cell, binding to the receptor in intracellular compartments, and bringing it to the membrane. Ligands can act as pharmacological chaperones for the delta opioid receptor, facilitating receptor maturation and export from the endoplasmic reticulum [21]. However, there are no reports that the norBNI antagonist is cell-permeable. Another possibility is that norBNI acts as an inverse agonist, stabilizing cell-surface receptors in an "off" conformation, making them inaccessible to GRKs and arrestins, which usually interact only with active receptors. This would suggest that in the absence of norBNI, Rog-μA may be constitutively active. However, the receptor still signals robustly in response to spiradoline (Figure 5A), so it cannot be fully active in the absence of ligand. It is also possible that, under basal conditions, Rog-μA has a higher than normal affinity for GRK or arrestin, but not the G proteins. This would result in constitutive turnover – the receptor constantly cycling in and out of the membrane – in the absence of constitutive signaling. The idea that this receptor is especially sensitive to the desensitization and internalization machinery is borne out by the observation of extensive desensitization in response to pretreatment with a low dose of agonist (Figure 5A). It will be interesting to investigate the physiological consequences of this apparent constitutive internalization and rapid desensitization in animal models.
Although most of the C-terminal sequence of Rog-μA is derived from the MOR, the MOR does not exhibit either constitutive turnover or abnormally low membrane expression. There is some evidence for partial basal internalization of the similar Rog-μ receptor (Figure 4A), although Rog-μ appears to be expressed predominantly at the cell surface under basal conditions. Since Rog-μ and Rog-μA differ at only five amino acids, some of those five residues must be responsible for the increased turnover of Rog-μA. Although phosphorylation of T394 has been reported to be required for desensitization of the MOR [18], subsequent reports have shown that mutating T394 to alanine facilitates the internalization and resensitization of the receptor [27]. This suggests that phosphorylation of T394 may be a membrane retention signal and that the T394A mutation in Rog-μA is responsible for the constitutive endocytosis observed in this study. Mutagenesis of individual amino acids in this region may allow the identification of the specific residue(s) responsible for the constitutive internalization of Rog-μA.
The differences in AC superactivation between our RASSLs indicate another layer of complexity in GPCR signaling. Previous studies suggest an inverse correlation between the ability of an opioid receptor to undergo ligand-activated endocytosis and its ability to induce AC superactivation by chronic signaling [28]. The induction of superactivation by Rog is consistent with this idea. Rog-μ, which internalizes and desensitizes readily, failed to induce any AC superactivation after chronic activation (Figure 5B). Similarly, Rog-μA, which may undergo constitutive endocytosis, did not induce AC superactivation after chronic administration of spiradoline. However, Rog-A, which was predicted to have enhanced superactivation due to its resistance to endocytosis, had levels of superactivation comparable to those seen with Rog. It is possible that Rog induces maximal superactivation, and the cell cannot respond to Rog-A signaling with any additional superactivation. Our results indicate that internalization does not directly induce AC superactivation. The dose of spiradoline (10 nM over 18 hours) used to induce strong AC superactivation in these experiments causes only minimal internalization of all of the receptors (Figure 5C). Moreover, there is no significant difference in degree of internalization among the different receptor types, although they show profound differences in superactivation. This suggests that the cellular mechanisms underlying AC superactivation and receptor internalization are independent.
AC superactivation has been attributed to upregulation of AC proteins induced by Gβγ protein subunits interacting directly with GRK2/3 proteins [29,30]. Therefore, the same mutations that prolong Rog-A signaling and inhibit endocytosis may also prevent the receptor from interacting with GRK2/3 proteins and subsequently activating Gβγ-mediated signaling events. Alterations in desensitization characteristics are unlikely to alter AC superactivation because of the drastic differences in time course underlying these distinct phenomena. Receptor desensitization happens on a scale of minutes, while AC superactivation is the result of much longer term chronic receptor activation. Therefore, a decrease in Gi signaling due to a more desensitized receptor is unlikely to have a significant effect on AC superactivation over the much longer time course used in these experiments. The additional possibility exists that altering C-terminal residues on the RASSL could increase the ability of receptors to couple to Go, resulting in perceived changes in AC superactivation [31]. However, if this were the case, one would expect to see a shift in the dose response curve for cAMP inhibition between Rog, Rog-A, Rog-μ and Rog-μA that is not observed in any of our experiments. Further studies with this engineered receptor system in vivo may clarify the complex relationship between ligand dependent endocytosis, interaction of a GPCR C-terminus with GRK2/3, desensitization, and superactivation of AC.
The ability of these RASSLs to induce different degrees of AC superactivation may have important physiological consequences in vivo. Interestingly, when a RASSL with a C-terminus corresponding to the wild-type human KOR was expressed at high levels in the hearts of transgenic mice, the mice developed a lethal cardiomyopathy [32]. One possible explanation is that basal signaling of the RASSL in mouse heart may increase Gs signaling through AC superactivation. Gs signaling has long been associated with heart failure, so AC superactivation may be responsible for the cardiomyopathy. Rog-μ and Rog-μA, RASSLs that do not induce superactivation, could be used to test this hypothesis and to study the consequences of AC superactivation in other tissues.
Internalized opioid receptors can be either degraded or recycled back to the membrane. The receptors that return to the membrane are stripped of arrestin, phosphates, and ligand, and are resensitized to ligand. Recent evidence demonstrates that the C-terminus is crucial for directing internalized receptors either into the degradative lysosomal pathway or back to the plasma membrane [27,33]. We expect that several of our engineered RASSLs should also differ in their post-endocytotic fate.
It is likely that the complex mechanisms governing GPCR endocytosis, recycling, desensitization and AC superactivation will be regulated differently in different cell types. The RASSLs described here exhibit similar properties in several different mammalian cell lines we tested (rat1a, CHO and HEK293), but their properties may change in specific cell types or under specific physiological conditions. One potentially fruitful avenue for future investigations would be to target different RASSLs to particular cell types in vivo. This would allow a thorough investigation of the interplay between receptor sequence and cell-type specific mechanisms of receptor regulation.
The development of a toolbox of engineered RASSLs that differ in internalization and desensitization raises several possibilities for future research and clinical investigations. Growing evidence points to a link between receptor dynamics and the potential for drugs to elicit tolerance or dependence, especially for opioid receptors [28,34]. Here, we present a system that allows the same drug to activate different receptors that have small, well-defined variations in sequence. The efficacy and potency of spiradoline is similar for Rog, Rog-A, Rog-μ and Rog-μA (Table 1); only the desensitization and internalization responses vary. Specifically, agonist-induced internalization is reduced in Rog-A and enhanced in Rog-μ, while Rog-μA shows agonist-independent internalization. Rog-A shows no desensitization after a brief spiradoline pretreatment, while the same treatment reduces the potency of spiradoline at Rog-μA and reduces efficacy at Rog-μ. Longer spiradoline pretreatment induces normal AC superactivation at Rog-A, but does not affect the AC response in Rog-μ or Rog-μA cells.
A family of engineered GPCRs that do not respond to endogenous ligands has enormous potential for selectively controlling G protein signaling in specific tissues without interfering with endogenous processes. Rog-A, a long-signaling RASSL, has several interesting implications for in vivo signal engineering. The basic RASSL, Ro2, shows rapid and extensive physiological desensitization [14], making it difficult to use in any therapeutic context where repeated activation of the receptor would be necessary. Rog-A, with its reduced desensitization, could allow continued physiological efficacy of repeated drug treatments. Comparison of the regulation and signaling of Rog, Rog-A, Rog-μ, Rog-μA, and future variants will contribute to the growing understanding of how GPCR signals are dynamically modulated. Study of these RASSLs in vivo will help solidify the elusive links between the receptor amino acid sequence, cell biology, and complex physiology.
Methods
Construction of mutant receptors
All receptors were based on the human kappa-opioid RASSL called Ro2 [12]. The GFP-tagged version of the RASSL has been named "Rog" for RASSL opioid with GFP tag. Rog was made by inserting the coding sequence for emerald GFP (Packard) at the N-terminus of the receptor, after a FLAG tag (DYKDDDDV) and the first eight amino acids of the RASSL. To create Rog-A, two serines and two threonines in the C-terminal region of the receptor were mutated to alanines (Figure 1). In the human kappa opioid receptor (KOR), the mutated residues correspond to S356A, T357A, S358A and T363A. The exact location of those residues in our RASSL construct, and the complete sequence of all RASSL variants can be found on our web site . For both Rog-μ and Rog-μA, the final 35 amino acids (345–380) of Rog were replaced by 47 C-terminal residues from the rat mu opioid receptor (MOR). Rog-μA contains the following additional modifications to the rat MOR C-terminus: T383A, E388Q, E391Q, E393Q and T394A. For each receptor, a schematic design and a C-terminal amino acid sequence alignment is shown in Figure 1. All constructs were sequenced to verify the mutations.
Expression of RASSLs in mammalian cells
HEK293 cells were grown in culture to 60–80% confluence and then transfected using Lipofectamine Plus (Invitrogen, Carlsbad, CA). The RASSL construct contained a cytomegalovirus promoter to drive mammalian expression, and a neomycin-resistance gene to allow selection of stable cell lines. Experiments on transiently transfected cells were performed approximately 48 h after transfection. To create stable cell lines, transfected cells were selected with G418 (500 μg/ml, Invitrogen) for 10–14 days. Individual colonies showing green fluorescence were selected and grown under maintenance doses of G418 (250 μg/ml). Receptor expression was confirmed visually by fluorescence microscopy, and by an enzyme-linked immunoadsorbent assay (ELISA, see below).
Cell-surface ELISA
Cell-surface expression of receptors was confirmed by an ELISA that detects only extracellular FLAG tag, which labels the N terminus of all RASSLs. This assay therefore quantifies only receptors that are in the membrane at the time of labeling, without providing detailed localization data about those receptors. Cells were plated at 100,000 cells/well on to 24-well plates coated with poly-d-lysine. Cultured cells were fixed in 4% paraformaldehyde for 10 min at 4°C, washed in phosphate-buffered saline (PBS), and then incubated in 1 μg/ml M1 anti-FLAG antibody (Sigma, St. Louis, MO) for 1 h at room temperature. They were washed again in PBS with 1 mM CaCl2 and incubated for 30 min at room temperature in secondary antibody (1:1000 goat anti-mouse conjugated with horseradish peroxidase, Biorad, Chicago, IL), then washed three times in PBS plus CaCl2. To develop the reaction, 0.25 ml of 2,2-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) liquid substrate (Sigma) was added to each well. After 15–60 min, 200 μl from each well was transferred to a 96-well plate and the optical density was read at 410 nm. To quantify agonist-induced internalization, cells were treated with various doses of spiradoline in medium for 10–120 min. The medium was removed and the cells were fixed and processed as described above. Vehicle treated cells on each plate were used to calculate a "maximum" cell-surface expression for that plate. All other treatment conditions on that plate were then normalized to this maximum to determine "percent internalization." Each experiment included 3–6 replicates per condition and was repeated at least 3 times. After values for cell-surface expression of each receptor were calculated and normalized, receptor expression was compared using two-way ANOVA (StatView v. 5.0, SAS Institute, Cary, NC). For dose response studies, receptor and dose were independent factors. For time course studies, receptor and length of treatment were independent factors.
cAMP accumulation assay
The degree of cAMP inhibition in spiradoline-treated HEK-293 cells transiently expressing RASSL variants was measured with the CatchPoint cAMP ELISA kit (Molecular Devices, Sunnyvale, CA). Cells were plated at 5 × 104/well into 96-well plates coated with poly-d-lysine. The next day, cells were rinsed in Krebs-Ringer bicarbonate buffer with glucose (KRBG, Sigma). Cells were then incubated in pre-stimulation buffer containing a phosphodiesterase inhibitor (0.75 mM 3-isobutyl-1-methylxanthine in KRBG buffer) for 10 min at room temperature to inhibit cAMP degradation. cAMP production was stimulated by the addition of 50 μM forskolin to all cells. At the same time, various doses of spiradoline in PBS were added to cells to establish a dose-response curve. After a 15-min drug treatment at 37°C, the cells were lysed and cAMP accumulation was assayed according to the CatchPoint protocol. Inhibition of cAMP by spiradoline was determined by comparison to cells treated with forskolin alone. For each experiment, 3–6 wells per condition were averaged, and EC50 and percent inhibition values for each receptor were determined by fitting curves for each independent experiment using SOFTmax PRO v. 4.0.1 (Molecular Devices). Data for 3–8 independent experiments were averaged to determine the EC50 and maximum cAMP inhibition for each receptor and condition. For desensitization assays, a pretreatment dose (1 nM) of spiradoline diluted in sterile PBS was added to the cells, which were then incubated for 10 min at 37°C. The cells were rinsed 4 times in KRBG and then stimulated and assayed for cAMP as described above. AC superactivation was determined by measuring forskolin-stimulated (10 μM) cAMP after treatment with either 10 nM spiradoline or vehicle for 18 h. AC superactivation data were expressed as a percent increase in forskolin-stimulated cAMP relative to vehicle-pretreated cells expressing the same receptor. Conditions and receptors were compared using a two-way ANOVA with receptor and treatment condition as independent factors.
Confocal microscopy
HEK293 cells stably expressing receptor constructs were plated at a density of 500,000 cells/ml onto glass Labtek II chamber slides (Fisher Scientific, Pittsburgh, PA) coated with poly-d-lysine. The following day, the cells were treated with agonist (spiradoline or dynorphin A 1–13) for typically one hour, or antagonist (NorBNI) overnight and briefly washed in PBS. The PBS was removed and replaced with 1 ml of cold 4% paraformaldehyde in PBS. The cells were fixed at room temperature for 10 min, washed with PBS, and then mounted in Vectashield (Vector Laboratories, Burlingame, CA) under cover slips. For confocal imaging on a Bio-Rad MRC 600 microscope, typical images were taken with a 40–60× oil immersion objective lens, subject to 5× Kalman filtering. The microscope operator was blind to both cell line and treatment condition.
Fluorometric imaging plate reader (FLIPR) assay
All receptors tested were transiently transfected using Lipofectamine into CHO cells in conjunction with the chimeric G protein Gqi5 [20] in a 5:1 molar ratio of DNA. Gqi5 is a chimeric G protein alpha subunit that has the Gαq wild-type sequence except for the C-terminal 5 residues, which were changed to the corresponding Gαi sequence. This allows Gi-coupled receptors to signal through the Gq pathway, resulting in a signal that can be detected using calcium-sensitive reagents. The following day, the cells were plated in a 96-well plate (50,000 cells per well) and allowed to grow for 24 h before being incubated with the calcium-sensitive dye fura-3 for 1 h. The assay was performed as described [12] using a range of dilutions of either spiradoline or dynorphin A 1–13 peptide (Sigma). All experiments were performed in triplicate.
Authors' contributions
KSL and MDL created the constructs and cell lines used here, designed and conducted the experiments; analyzed the data; and drafted the manuscript. HHE maintained cell lines and participated in making constructs, internalization assays and confocal microscopy. BRC conceived of the study and participated in its design and interpretation. All authors read and approved the final manuscript.
Acknowledgements
The authors are grateful for valuable technical assistance from David Sanan (confocal microscopy), Jin Liao (FLIPR assay), Sam Davis (assembly of receptor constructs), and Taeryn Kim (ELISA assays and cell culture). We also thank Drs. Lennart Mucke, Nathalie Cotte, Kam Dahlquist, Supriya Srinivasan, Whit Tingley and Alex Zambon for their insightful discussions and careful reading of this manuscript. Gary Howard, Stephen Ordway and Bethany Taylor provided editorial assistance in revision and preparation of this manuscript. This work was supported by NIH HL6064 (BRC) and a NIDA NRSA fellowship (KSL).
Figures and Tables
Figure 1 Design of RASSL variants. The design of each RASSL variant is shown on the left. Differences in amino acid sequences of the receptor variants are shown in the C-terminal alignments, right. Sequence derived from the MOR are dark, while KOR sequences are light. All mutated residues are underlined.
Figure 2 Agonist-induced signaling and internalization of Rog. (A) Maximum calcium response plotted as a function of spiradoline dose for cells transfected with Rog or the wild-type KOR and treated with dynorphin or spiradoline. (B) Confocal micrographs show representative internalization of GFP-tagged Rog receptors 1 h after treatment with 10 μM or 100 μM spiradoline. Dynorphin at 100 μM (far right) caused little internalization in this assay. (C) ELISA for FLAG-tagged cell-surface receptors shows dose-dependent loss of receptors from cell surface one h after spiradoline treatment. After a 1 μM dose of spiradoline, internalization is evident within 15 minutes. Data are expressed as a percentage of receptors detected on surface of untreated cells expressing Rog. Error bars represent SEM for three replicates.
Figure 3 Agonist-induced internalization is reduced in Rog-A, but not Rog-μ. (A) Confocal micrographs showing localization of GFP-tagged Rog (left), Rog-A (center), or Rog-μ (right) stably expressed in HEK293 cells and treated with 10 μM spiradoline for 1 h before fixation. (B) A dose-response ELISA shows less internalization of Rog-A in response to 1 h of spiradoline at doses of 0.3–100 μM spiradoline. After treatment with 1 μM spiradoline, Rog-A showed less internalization up to 1 h after treatment. (C) A dose-response ELISA shows more internalization of Rog-μ in response to 1 h of low doses of spiradoline ranging from 0.01 to 0.1 μM. After treatment with 1 μM spiradoline, there was no difference in the time course of internalization between Rog and Rog-μ. ELISA data are expressed as a percentage of receptors detected on surface of untreated cells expressing the same receptor.
Figure 4 Antagonist treatment increases cell-surface expression of Rog-μA. (A) ELISA comparing cell-surface expression of receptors stably expressed in HEK293 cell lines, either untreated (white bars) or treated for 18 h with the antagonist norBNI (10 μM, dark bars). Error bars represent SEM for three replicates. NorBNI significantly increased cell-surface expression of both Rog-μ and Rog-μA. OD, optical density. (B) Confocal micrographs showing that the localization of GFP-tagged Rog-μA is primarily intracellular in untreated cells (left), but the receptor moves to the membrane after 18 hours of antagonist treatment (right). There is relatively little change in Rog after norBNI treatment relative to untreated cells.
Figure 5 Desensitization of cAMP inhibition and superactivation of AC after pretreatment with spiradoline (A) Spiradoline (1 nM pretreatment) inhibited forskolin-induced cAMP formation in HEK293 cells transiently expressing Rog, Rog-A, Rog-μ, and Rog-μA. Data are expressed as inhibition of forskolin-induced cAMP. The baseline (0) represents maximum forskolin-induced cAMP accumulation in control cells. Pretreatment with 1 nM spiradoline for 10 min causes a shift in the dose-response curve for later spiradoline treatment in Rog and Rog-μA cells, but not Rog-A. Spiradoline pretreatment lowered the maximal response of Rog-μ to subsequent spiradoline treatment. (B) HEK293 cells transiently expressing receptors were treated 18 h with 10 nM spiradoline, and assayed for cAMP accumulation in response to a 15-min treatment with 10 μM forskolin. Spiradoline pretreatment significantly increased forskolin-induced cAMP in cells expressing KOR, Rog and Rog-A. Pretreatment of cells expressing Rog-μ and Rog-μA had not effect on response to forskolin. Data are expressed relative to the amount of cAMP accumulated after 10 μM forskolin treatment in cells pretreated with vehicle. Bars represent mean ± SEM for six replicates per condition. (C) HEK293 cells transiently expressing receptors were treated 18 h with 10 nM spiradoline, and assayed for loss of cell-surface expression by ELISA. Long-term spiradoline treatment significantly reduces cell-surface expression of all receptors, but the degree of internalization does not differ among different receptors. Bars represent mean ± SEM for three replicates per condition.
Table 1 EC50 and Emax values for inhibition of cAMP accumulation by spiradoline
Rog Rog-A Rog-μ Rog-μA
EC50 (nM spiradoline)
Acute activation 0.41 ± 0.08 0.41 ± 0.15 0.42 ± 0.23 0.50 ± 0.22
1 nM spiradoline pretreatment 4.32 ± 1.96 0.42 ± 0.21 2.05 ± 0.03 > 13.44 ± 5.41
% Inhibition
Acute activation 64.4 ± 4.8 56.1 ± 5.1 58.0 ± 0.9 64.1 ± 4.8
1 nM spiradoline pretreatment 55.7 ± 3.7 53.3 ± 0.9 43.6 ± 1.5 51.5 ± 6.0
Values are mean ± SEM for 3–8 experiments for each condition.
==== Refs
Ferguson SSG Evolving concepts in G protein-coupled receptor endocytosis: The role in receptor desensitization and signaling Pharmacol Rev 2001 53 1 24 11171937
Paing MM Stutts AB Kohout TA Lefkowitz RJ Trejo J β-arrestins regulate protease-activated receptor-1 desensitization but not internalization or down-regulation J Biol Chem 2002 277 1292 1300 11694535 10.1074/jbc.M109160200
Avidor-Reiss T Nevo I Levy R Pfeuffer T Vogel Z Chronic opioid treatment induces adenylyl cyclase V superactivation J Biol Chem 1996 271 21309 21315 8702909 10.1074/jbc.271.35.21309
Kieffer BL Evans CJ Opioid tolerance – In search of the holy grail Cell 2002 108 587 590 11893329 10.1016/S0092-8674(02)00666-9
Borgland SL Connor M Osborne PB Furness JB Christie MJ Opioid agonists have different efficacy profiles for G protein activation, rapid desensitization, and endocytosis of mu-opioid receptors J Biol Chem 2003 278 18776 18784 12642578 10.1074/jbc.M300525200
Alvarez VA Arttamangkul S Dang V Salem A Whistler JL Von Zastrow M Grandy DK Williams JT mu-Opioid receptors: Ligand-dependent activation of potassium conductance, desensitization, and internalization J Neurosci 2002 22 5769 5776 12097530
Zimmer M Palmer A Kohler J Klein R EphB-ephrinB bi-directional endocytosis terminates adhesion allowing contact mediated repulsion Nat Cell Biol 2003 5 869 878 12973358 10.1038/ncb1045
Haglund K Sigismund S Polo S Szymkiewicz I Di Fiore PP Dikic I Multiple monoubiquitination of RTKs is sufficient for their endocytosis and degradation Nat Cell Biol 2003 5 461 466 12717448 10.1038/ncb983
Nong Y Huang YQ Ju W Kalia LV Ahmadian G Wang YT Salter MW Glycine binding primes NMDA receptor internalization Nature 2003 422 302 307 12646920 10.1038/nature01497
Braithwaite SP Xia H Malenka RC Differential roles for NSF and GRIP/ABP in AMPA receptor cycling Proc Natl Acad Sci USA 2002 99 7096 7101 12011465 10.1073/pnas.102156099
St John PA Gordon H Agonists cause endocytosis of nicotinic acetylcholine receptors on cultured myotubes J Neurobiol 2001 49 212 223 11745659 10.1002/neu.1076
Coward P Wada HG Falk MS Chan SDH Meng F Akil H Conklin BR Controlling signaling with a specifically designed Gi-coupled receptor Proc Natl Acad Sci USA 1998 95 352 357 9419379 10.1073/pnas.95.1.352
Scearce-Levie K Coward P Redfern CH Conklin BR Engineering receptors activated solely by synthetic ligands (RASSLs) Trends Pharmacol Sci 2001 22 414 420 11479004 10.1016/S0165-6147(00)01743-0
Redfern CH Coward P Degtyarev MY Lee EK Kwa AT Hennighausen L Bujard H Fishman GI Conklin BR Conditional expression and signaling of a specifically designed Gi-coupled receptor in transgenic mice Nat Biotechnol 1999 17 165 169 10052353 10.1038/6165
Zhao GQ Zhang Y Hoon MA Chandrashekar J Erlenbach I Ryba NJ Zuker CS The receptors for mammalian sweet and umami taste Cell 2003 115 255 266 14636554 10.1016/S0092-8674(03)00844-4
Capeyrou R Riond J Corbani M Lepage J-F Bertin B Emorine LJ Agonist-induced signaling and trafficking of the μ-opioid receptor: Role of serine and threonine residues in the third cytoplasmic loop and C-terminal domain FEBS Lett 1997 415 200 205 9350996 10.1016/S0014-5793(97)01124-1
El Kouhen R Burd AL Erickson-Herbrandson LJ Chang C-Y Law P-Y Loh HH Phosphorylation of Ser363, Thr370, and Ser375 residues within the carboxyl tail differentially regulates μ-opioid receptor internalization J Biol Chem 2001 276 12774 12780 11278523 10.1074/jbc.M009571200
Pak Y O'Dowd BF George SR Agonist-induced desensitization of the μ opioid receptor is determined by threonine 394 preceded by acidic amino acids in the COOH-terminal tail J Biol Chem 1997 272 24961 24965 9312100 10.1074/jbc.272.40.24961
Schulz R Wehmeyer A Schulz K Visualizing preference of G protein-coupled receptor kinase 3 for the process of κ-opioid receptor sequestration Mol Pharmacol 2002 61 1444 1452 12021406 10.1124/mol.61.6.1444
Conklin BR Farfel Z Lustig KD Julius D Bourne HR Substitution of three amino acids switches receptor specificity of Gqα to that of Giα Nature 1993 363 274 276 8387644 10.1038/363274a0
Petäjä-Repo UE Hogue M Bhalla S Laperrière A Morello J-P Bouvier M Ligands act as pharmacological chaperones and increase the efficiency of δ opioid receptor maturation EMBO J 2002 21 1628 1637 11927547 10.1093/emboj/21.7.1628
Morello J-P Salahpour A Laperrière A Bernier V Arthus M-F Lonergan M Petäjä-Repo U Angers S Morin D Bichet DG Bouvier M Pharmacological chaperones rescue cell-surface expression and function of misfolded V2 vasopressin receptor mutants J Clin Invest 2000 105 887 895 10749568
McLaughlin JP Xu M Mackie K Chavkin C Phosphorylation of a carboxy-terminal serine within the kappa opioid receptor produces desensitization and internalization J Biol Chem 2003 278 34631 34640 12815037 10.1074/jbc.M304022200
Bouvier M Hausdorff WP De Blasi D O'Dowd BF Kobilka BK Caron MG Lefkowitz RJ Removal of phosphorylation sites from the β2-adrenergic receptor delays onset of agonist-promoted desensitization Nature 1988 333 370 373 2836733 10.1038/333370a0
Ferguson SSG Ménard L Barak LS Koch WJ Colapietro A-M Caron MG Role of phosphorylation in agonist-promoted β2-adrenergic receptor sequestration J Biol Chem 1995 270 24782 24789 7559596 10.1074/jbc.270.42.24782
Qiu Y Law PY Loh HH μ-opioid receptor desensitization: Role of receptor phosphorylation, internalization and representation J Biol Chem 2003 278 36733 36739 12860981 10.1074/jbc.M305857200
Wolf R Koch T Schulz S Klutzny M Schröder H Raulf E Bühling F Höllt V Replacement of threonine 394 by alanine facilitates internalization and resensitization of the rat μ opiod receptor Mol Pharmacol 1999 55 263 268 9927617
Finn AK Whistler JL Endocytosis of the mu opioid receptor reduces tolerance and a cellular hallmark of opiate withdrawal Neuron 2001 32 829 839 11738029 10.1016/S0896-6273(01)00517-7
Chakrabarti S Oppermann M Gintzler AR Chronic morphine induces the concomitant phosphorylation and altered association of multiple signaling proteins: A novel mechanism for modulating cell signaling Proc Natl Acad Sci USA 2001 98 4209 4214 11274443 10.1073/pnas.071031798
Pitcher JA Freedman NJ Lefkowitz RJ G protein-coupled receptor kinases Annu Rev Biochem 1998 67 653 692 9759500 10.1146/annurev.biochem.67.1.653
Watts VJ Molecular mechanisms for heterologous sensitization of adenylate cyclase J Pharmacol Exp Ther 2002 302 1 7 12065693 10.1124/jpet.302.1.1
Redfern CH Degtyarev MY Kwa AT Salomonis N Cotte N Nanevicz T Fidelman N Desai K Vranizan K Lee EK Coward P Shah N Warrington JA Fishman GI Bernstein D Baker AJ Conklin BR Conditional expression of a Gi-coupled receptor causes ventricular conduction delay and a lethal cardiomyopathy Proc Natl Acad Sci USA 2000 97 4826 4831 10781088 10.1073/pnas.97.9.4826
Whistler JL Enquist J Marley A Fong J Gladher F Tsuruda P Murray SR von Zastrow M Modulation of postendocytic sorting of G protein-coupled receptors Science 2002 297 615 620 12142540 10.1126/science.1073308
Bohn LM Gainetdinov RR Lin FT Lefkowitz RJ Caron MG Mu-opioid receptor desensitization by beta-arrestin-2 determines morphine tolerance but not dependence Nature 2000 408 720 723 11130073 10.1038/35047086
| 15707483 | PMC551613 | CC BY | 2021-01-04 16:02:56 | no | BMC Biol. 2005 Feb 11; 3:3 | utf-8 | BMC Biol | 2,005 | 10.1186/1741-7007-3-3 | oa_comm |
==== Front
BMC Blood DisordBMC Blood Disorders1471-2326BioMed Central London 1471-2326-5-31571322910.1186/1471-2326-5-3Research ArticleAn in vitro evaluation of standard rotational thromboelastography in monitoring of effects of recombinant factor VIIa on coagulopathy induced by hydroxy ethyl starch Engström Martin [email protected] Peter [email protected]ött Ulf [email protected] Department of Anaesthesia and Intensive Care, Lund University Hospital, Sweden2 Department of Anaesthesia and Intensive Care, Halmstad County Hospital, Sweden2005 15 2 2005 5 3 3 13 7 2004 15 2 2005 Copyright © 2005 Engström et al; licensee BioMed Central Ltd.2005Engström 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
Rotational thromboelastography (ROTEG) has been proposed as a monitoring tool that can be used to monitor treatment of hemophilia with recombinant factor VIIa (rFVIIa). In these studies special non-standard reagents were used as activators of the coagulation. The aim of this study was to evaluate if standard ROTEG analysis could be used for monitoring of effects of recombinant factor VIIa (rFVIIa) on Hydroxy Ethyl Starch-induced dilutional coagulopathy.
Methods
The study was performed in vitro on healthy volunteers. Prothrombin time (PT) and ROTEG analysis were performed after dilution with 33% hydroxy ethyl starch and also after addition of rFVIIa to the diluted blood.
Results
PT was impaired with INR changing from 0.9 before dilution to 1.2 after dilution while addition of rFVIIa to diluted blood lead to an overcorrection of the PT to an International Normalized Ratio (INR) value of 0.6 (p = 0.01). ROTEG activated with the contact activator ellagic acid was impaired by hemodilution (p = 0.01) while addition of rFVIIa had no further effects. ROTEG activated with tissue factor (TF) was also impaired by hemodilution (p = 0.01) while addition of rFVIIa lead to further impairment of the coagulation (p = 0.01).
Conclusions
The parameters affected in the ROTEG analysis were Clot Formation Time and Amplitude after 15 minutes while the Clotting Time was unaffected. We believe these effects to be due to methodological problems when using standard activators of the coagulation in the ROTEG analysis in combination with rFVIIa.
==== Body
Background
Patients undergoing massive hemorrhage experience dilutional coagulopathy with crystalloid and/or colloid resuscitation. If hemorrhage progresses, packed red cells (RBC) are transfused together with crystalloids and/or colloids. Regarding the coagulation this is not optimal, and the patients often develop a dilutional coagulopathy, sometimes worsened by hypothermia. In addition a coagulopathy caused by the administration of dextrane or hydroxy ethyl starch (HES) may be induced. The common approach to this is transfusion of fresh frozen plasma (FFP) and platelets, but bleeding might continue, with often fatal outcome. Prophylactic use of fresh frozen plasma (FFP) or platelet transfusion is not proven beneficial to prevent hemorrhage in massively transfused patients[1]. Hemorrhage, complicated by the development of coagulopathy, is therefore still the major cause of death in trauma patients arriving alive in the hospital.[2,3].
A novel approach to treat these patients is the use of recombinant factor VIIa (rFVIIa) to improve the coagulation.[4]. Recently there have been several case publications of successful treatment of coagulopathic trauma patients and surgical patients. [5-7].
Dilutional coagulopathy can be detected by the use of thromboelastographic measurements [8-11]. Rotational thromboelastography (ROTEG) is a recent development of thromboelastography. ROTEG gives a viscoelastic measurement of clot strength in whole blood. It is presented as a graph representing clot strength during the build-up of a clot (figures 1 and 2). From the graph several variables describing different parts of the coagulation process are derived and measured numerically. It thus gives a more comprehensive picture of the coagulation than standard tests, but is on the other hand less validated and standardised than the more common coagulation tests. An advantage of the method is that it can be used as a point-of-care analysis. One condition to be fulfilled if it should be used as a point-of-care method is that the commercially available kits can be used for analysis and that special preparations and/or dilutions should not be necessary, as that would require the skills and competencies of a full laboratory.
Figure 1 The figure shows the 3 representative tracings of the EXTEG analysis from one of the participants in the study. Above the normal tracing with a short CT and CFT is shown. It can be seen that the clot strength is rapidly increasing after initiation of the clotting. In the middle the tracing after hemodilution with HES is found. It can be seen that the CFT is prolonged and that the strength of the clot is increasing slower. Below the tracing after hemodilution and addition of rFVIIa is found. The clotting is then severely impaired, the clot strength is increasing slower and the maximum strength is also severely impaired.
Figure 2 The figure shows the result of a ROTEG analysis. Time is represented on the X axis and clot strength on the Y axis. The clot strength is arbitrarily measured in mm where maximum clot strength is 100 mm. The Clotting Time (CT) is the time from initiation of the analysis until the clot strength is 2 mm. The Clot Formation Time (CFT) is the time from clot strength 2 mm until clot strength 20 mm. A15 is the clot strength at 15 minutes.
Previous studies have shown that hemodilution with HES impairs the coagulation already at a dilution of 33% while crystalloid hemodilution does not appear to give a readily detectable impairment of the coagulation until the level of hemodilution reaches 50% [10,12]. In this study we have investigated if ROTEG can be used with the commercially available kits to monitor effects of rFVIIa and if rFVIIa is able to improve the coagulopathy caused by hemodilution with HES.
Methods
The local ethics committee of the University of Lund approved this study on healthy volunteers. Volunteers were not allowed to take any medication 14 days prior to the study. Informed consent was obtained from the participants and a total of eight were recruited. All participants had an indwelling intravenous catheter placed into the brachiocephalic vein, from which the blood samples were drawn with sterile disposable 5-ml syringes (Luer; Codan Medical Aps, Rödby, Denmark). A first 5-ml blood sample was discarded before every blood sample for the experiment described below. No tourniquet was used on the arm when samples were drawn.
Dilution of the blood samples was performed with HES 130/0.4 (Voluven®, Fresenius Co., Bad Homburg, Germany), the HES preparation found to cause the least pronounced coagulopathy after hemodilution [12]. Three different preparations of blood were examined. The first preparation contained 5 ml of undiluted blood (normal). The second preparation contained 3.3 ml of blood and 1.7 ml of HES thereby achieving a 33% dilution (dilution). The third preparation contained 3.3 ml of blood, 1.7 ml of HES and 50μl of rFVIIa at the concentration 0.12 μg/μl (dilution + rFVIIa). The latter concentration of rFVIIa was equivalent to the concentration achieved when the dose 90 μg/kg body weight is administered in vivo. 90 μg/kg is the recommended dose in hemophilia and well within the range suggested for treatment of acute hemorrhage in non hemophilia patients.[7,13]. The reason for the chosen dilution (33%) was that it is a clinically relevant dilution that is readily achieved during resuscitation of a patient. Before initiating this study we have also tested 50% dilution in a single person and the results were similar to the results with 33% dilution, but more pronounced. We then decided to study the 33% dilution systematically. Further on, this dilution did not induce unphysiologic changes in Ca or pH as tested in a pilot volunteer. The dilution was performed in a polypropylene test tube and the tube was gently turned to mix the blood with the added HES and rFVIIa. The HES was warmed in a heating block (Grant Instrumentation Ltd, Cambridge, UK) to 37°C prior to hemodilution in order to avoid hypothermia as a confounding factor after dilution.
The tests performed on the different preparations were hemoglobin concentration (Hb), Prothrombin time (PT) and ROTEG analysis. All tests were performed at normal body temperature (37°C). For the Hb measurements a Hemocue (HemoCue Co., Ängelholm, Sweden) was used. PT measurements were performed with a Rapidpoint Coag Analyzer (Bayer AB Diagnostics, Gothenburg, Sweden) with PT-ONE test cards. The ROTEG analyses were performed on a Rotational Thromboelastograph (ROTEG, Pentapharm, Munich, Germany) and the samples were analysed 120 seconds after the blood was drawn from the intravenous catheter. Both INTEG and EXTEG analyses were performed according to standard procedure recommended by the manufacturer. In INTEG analysis the coagulation is initiated with the addition of 20 μl of the contact activator ellagic acid (Pentapharm, Munich, Germany) to 320 μl of blood pipetted from the test tube to a reaction cup used in the ROTEG. In EXTEG analysis the coagulation is activated by the addition of 20 μl of a preparation containing tissue factor (TF) (Pentapharm, Munich, Germany) to 320 μl of blood pipetted from the test tube to the reaction cup. TF activates the coagulation through binding to Factor VIIa and this is believed to be the important interaction when in vivo coagulation occurs. The parameters obtained from the ROTEG analysis were Clotting Time (CT) reflecting the initiation of the coagulation, Clot Formation Time (CFT) reflecting the rate of clot formation once the formation is initiated and A15 describing the strength of the clot 15 minutes after initiation of the coagulation (figure 2).
Statistical analysis was performed with initial Kruskal-Wallis test and Wilcoxon's paired test was used when the Kruskal-Wallis test indicated a significant difference. All values are given as median (range). A p value of < 0.05 was considered statistically significant.
Results
The insertion of venous catheters and the blood sampling were performed uneventfully. The Hb values decreased as an expected sign of dilution (table 1). PT values increased in the dilution group compared to normal and decreased to below normal in the dilution + rFVIIa group (table 1).
Table 1 Effects of dilution and addition of rFVIIa on Hb and PT values. A lowering of Hb and an increase in the PT were seen as signs of dilution while the addition of rFVIIa lead to a decrease of the PT (n = 8).
Normal Dilution Dilution + rFVIIa
Hb (g/l) 136 (127–147) 88 (81–99)* 88 (81–99)*
PT (INR) 0.9 (0.7–1.2) 1.2 (0.9–1.3)§ 0.6 (0.5–0.7)*¶
* p = 0.01 compared to normal. ¶p = 0.01 compared to dilution. §p = 0.02 compared to normal.
In neither INTEG nor EXTEG analysis we found any change in the CT between normal and dilution groups, while CFT and A15 were impaired in the dilution group (tables 2 and 3). There were no differences between the dilution and the dilution + rFVIIa groups when analysed with the INTEG analysis (table 2). However, when rFVIIa was added to the dilution a prolongation of the CFT with > 200% and an impairment of the A15 with 40% were found (table 3 and figure 1).
Table 2 Coagulation variables as assessed by INTEG. Obvious signs of dilution are found in the CFT and the A15, while the addition of rFVIIa to the diluted blood does not affect the coagulation parameters (n = 8).
Normal Dilution Dilution + rFVIIa
CT (s) 93.5 (82–104) 108.5 (87–136) 97.5 (68–118)
CFT (s) 85.5 (59–111) 216 (158–310)* 190.5 (167–368)*
A15 (mm) 56 (52–61) 43 (36–47)* 43.5 (33–50)*
* p = 0.01 compared to normal.
Table 3 Coagulation variables as assessed by EXTEG. Obvious signs of dilution are found in the CFT and the A15. The addition of rFVIIa to the diluted blood leads to a prolongation of the CFT and the A15 (n = 8).
Normal Dilution Dilution + rFVIIa
CT (s) 51.5 (30–69) 63 (41–81) 58.5 (37–99)
CFT (s) 91 (67–105) 227 (171–332)* 558.5 (308–998)*¶
A15 (mm) 57 (53–63) 42 (33–49)* 25.5 (19–37)*¶
* p = 0.01 compared to normal. ¶p = 0.01 compared to dilution.
Discussion
Treatment of dilutional coagulopathy is challenging and the primary monitoring tools are measurement of PT, activated partial thromboplastin time (APTT), platelet count and fibrinogen [14-16]. These tests provide us with information regarding the activation of the coagulation process and about the absolute number of platelets. However, they do not give us information regarding the dynamic properties of blood clotting and the rate at which the clot is formed once the clotting is initiated. New monitoring methods are needed and ROTEG is a monitoring tool that could potentially be of value in these situations. To this end, it has been suggested that treatment of haemophilia patients and liver transplant patients with rFVIIa can be monitored with ROTEG where a shortening of the CT and CFT has been found in case series. [17-19].
In our study we found that hemodilution in vitro with HES lead to augmentation of the PT and the addition of rFVIIa results in a prompt decrease of the PT. This is in line with previous studies, which have shown that dilution with HES lead to readily detectable changes in the coagulation system[9,12,20]. Former studies have also shown a decrease or a normalisation of the PT after administration of rFVIIa[21,22]. In this study we found an overcorrection of the PT to values below normal range.
When analysing the ROTEG parameters we found that the 33% dilution with HES resulted in a prolongation of CFT values and impairment in A15 values in accordance with previous studies[9,11,12,20]. The CT was not significantly prolonged even though there was a trend towards a prolongation of the CT in both INTEG and EXTEG. After rFVIIa had been added to the diluted blood, coagulation variables remained unchanged when assessed with INTEG, but were markedly affected when assessed with EXTEG. CFT and A15 reflect the dynamic interplay between platelets and fibrin polymerisation, both being disturbed by HES hemodilution as can be seen in table 2 and 3. Addition of rFVIIa in vitro, worsening both these parameters according to EXTEG analysis, suggested that platelets or fibrinogen became dysfunctional in contrast to the clinical effect of administration of rFVIIa, where rFVIIa has been found successful in case stories of bleeding patients [4,7,13,23], even though these cases may not have been bleeding due to a HES induced coagulopathy.
The lack of effect in the INTEG analysis when adding rFVIIa to the diluted blood is most likely due to the fact that a contact activator is used to activate the coagulation in the INTEG and therefore insensitive to rFVIIa as rFVIIa initiates coagulation through interaction with TF.
The impairment of CFT and A15 in the EXTEG analysis after addition of rFVIIa to the diluted blood is harder to explain. We expected addition of rFVIIa to the diluted blood to result in an improvement of the TEG parameters measured in the EXTEG analysis. This was expected partly because TF is used as an activator of the coagulation in the EXTEG analysis and the first step in the initiation of the coagulation system is the interaction between TF and FVIIa. [17,18,24]. In our study it is likely that the coagulopathy induced was at least to some extent caused by a platelet dysfunction caused by the colloid hemodilution. As bleeding caused by Glanzmann's thrombastenia and other thrombocytopathias is frequently and successfully treated with rFVIIa, is a reason why rFVIIa potentially could be effective in the treatment of this HES-induced coagulopathy[25-28].
The previously reported improvements of coagulation in hemophilia and liver transplant patients as evaluated with ROTEG after administration of rFVIIa also lead us to believe that ROTEG parameters would be improved after addition of rFVIIa to the diluted blood[17,18,24]. It is however important that these studies were performed on hemophilia patients suffering from a severe deficiency of factor VIII or IX and on liver transplant patients suffering from a very complex coagulopathy. It also seems important to dilute TF extensively to detect the effects of rFVIIa on ROTEG. Dilutions of TF up to 1:17000 have been performed by the Ingerslev group in Denmark [18,24]. These dilutions are, however, not made with commercially available reagents that are ready to use immediately and therefore not suitable for use outside research laboratories.
Conclusion
In conclusion we found that 33% dilution of blood with HES 130/0.4 lead to impairment of the coagulation when evaluated with ROTEG or PT. Addition of rFVIIa lead to overcorrection of the prolonged PT. ROTEG analysis revealed that INTEG analysis was insensitive to effects of addition of rFVIIa and that EXTEG analysis was dramatically impaired by the addition of rFVIIa. It may well be that rFVIIa is not effective in improving the coagulopathy induced by HES hemodilution, but the further impairment of the coagulation seen in the EXTEG analysis is likely due to methodological problems. These problems make the commercially available EXTEG analyses inappropriate for monitoring of rFVIIa effects under circumstances of HES dilution.
Competing interests
The author(s) declare that they have no competing interests.
Authors contributions
ME contributed to the design of the study, performed the analyses and drafted the manuscript. PR contributed to the design of the study, to data interpretation and to preparing the manuscript. US contributed to the design of the study, performed the analyses and participated in manuscript preparation. All authors have read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Counts RB Haisch C Simon TL Maxwell NG Heimbach DM Carrico CJ Hemostasis in massively transfused trauma patients Ann Surg 1979 190 91 99 464685
Acosta JA Yang JC Winchell RJ Simons RK Fortlage DA Hollingsworth-Fridlund P Hoyt DB Lethal injuries and time to death in a level I trauma center J Am Coll Surg 1998 186 528 533 9583692 10.1016/S1072-7515(98)00082-9
Sauaia A Moore FA Moore EE Moser KS Brennan R Read RA Pons PT Epidemiology of trauma deaths: a reassessment J Trauma 1995 38 185 193 7869433
Kenet G Walden R Eldad A Martinowitz U Treatment of traumatic bleeding with recombinant factor VIIa Lancet 1999 354 1879 10584732 10.1016/S0140-6736(99)05155-7
Dutton RP Hess JR Scalea TM Recombinant factor VIIa for control of hemorrhage: early experience in critically ill trauma patients J Clin Anesth 2003 15 184 188 12770653 10.1016/S0952-8180(03)00034-5
Tobias JD Synthetic factor VIIa to treat dilutional coagulopathy during posterior spinal fusion in two children Anesthesiology 2002 96 1522 1525 12170072 10.1097/00000542-200206000-00039
Martinowitz U Kenet G Segal E Luboshitz J Lubetsky A Ingerslev J Lynn M Recombinant activated factor VII for adjunctive hemorrhage control in trauma J Trauma 2001 51 431 8; discussion 438-9 11535886
Petroianu GA Liu J Maleck WH Mattinger C Bergler WF The effect of In vitro hemodilution with gelatin, dextran, hydroxyethyl starch, or Ringer's solution on Thrombelastograph Anesth Analg 2000 90 795 800 10735778 10.1097/00000539-200004000-00005
Niemi TT Kuitunen AH Hydroxyethyl starch impairs in vitro coagulation Acta Anaesthesiol Scand 1998 42 1104 1109 9809097
Ekseth K Abildgaard L Vegfors M Berg-Johnsen J Engdahl O The in vitro effects of crystalloids and colloids on coagulation Anaesthesia 2002 57 1102 1108 12428635 10.1046/j.1365-2044.2002.02782_1.x
Jamnicki M Zollinger A Seifert B Popovic D Pasch T Spahn DR Compromised blood coagulation: an in vitro comparison of hydroxyethyl starch 130/0.4 and hydroxyethyl starch 200/0.5 using thrombelastography Anesth Analg 1998 87 989 993 9806670 10.1097/00000539-199811000-00002
Entholzner EK Mielke LL Calatzis AN Feyh J Hipp R Hargasser SR Coagulation effects of a recently developed hydroxyethyl starch (HES 130/0.4) compared to hydroxyethyl starches with higher molecular weight Acta Anaesthesiol Scand 2000 44 1116 1121 11028733 10.1034/j.1399-6576.2000.440914.x
Aldouri M The use of recombinant factor VIIa in controlling surgical bleeding in non-haemophiliac patients Pathophysiol Haemost Thromb 2002 32 Suppl 1 41 46 12214147 10.1159/000057301
Aucar JA Norman P Whitten E Granchi TS Liscum KR Wall MJ Mattox KL Intraoperative detection of traumatic coagulopathy using the activated coagulation time Shock 2003 19 404 407 12744481 10.1097/01.shk.0000048905.46342.6b
Garrison JR Richardson JD Hilakos AS Spain DA Wilson MA Miller FB Fulton RL Predicting the need to pack early for severe intra-abdominal hemorrhage J Trauma 1996 40 923 7; discussion 927-9 8656478
Keller MS Fendya DG Weber TR Glasgow Coma Scale predicts coagulopathy in pediatric trauma patients Semin Pediatr Surg 2001 10 12 16 11172565 10.1053/spsu.2001.19381
Hendriks HG Meijer K de Wolf JT Porte RJ Klompmaker IJ Lip H Slooff MJ van der Meer J Effects of recombinant activated factor VII on coagulation measured by thromboelastography in liver transplantation Blood Coagul Fibrinolysis 2002 13 309 313 12032396 10.1097/00001721-200206000-00006
Ingerslev J Christiansen K Calatzis A Holm M Sabroe Ebbesen L Management and monitoring of recombinant activated factor VII Blood Coagul Fibrinolysis 2000 11 Suppl 1 S25 30 10850560
Sorensen B Ingerslev J Thromboelastography and recombinant factor VIIa-hemophilia and beyond Semin Hematol 2004 41 140 144 14872435 10.1053/j.seminhematol.2003.11.024
Konrad CJ Markl TJ Schuepfer GK Schmeck J Gerber HR In vitro effects of different medium molecular hydroxyethyl starch solutions and lactated Ringer's solution on coagulation using SONOCLOT Anesth Analg 2000 90 274 279 10648306 10.1097/00000539-200002000-00007
Erhardtsen E Nony P Dechavanne M Ffrench P Boissel JP Hedner U The effect of recombinant factor VIIa (NovoSeven) in healthy volunteers receiving acenocoumarol to an International Normalized Ratio above 2.0 Blood Coagul Fibrinolysis 1998 9 741 748 9890717
Deveras RA Kessler CM Reversal of warfarin-induced excessive anticoagulation with recombinant human factor VIIa concentrate Ann Intern Med 2002 137 884 888 12458988
Lynn M Jeroukhimov I Klein Y Martinowitz U Updates in the management of severe coagulopathy in trauma patients Intensive Care Med 2002 28 Suppl 2 S241 7 12404093 10.1007/s00134-002-1471-7
Sorensen B Johansen P Christiansen K Woelke M Ingerslev J Whole blood coagulation thrombelastographic profiles employing minimal tissue factor activation J Thromb Haemost 2003 1 551 558 12871465 10.1046/j.1538-7836.2003.00075.x
Poon MC d'Oiron R Hann I Negrier C de Lumley L Thomas A Karafoulidou A Demers C Street A Huth-Kuhne A Petrini P Fressinaud E Morfini M Tengborn L Marques-Verdier A Musso R Devecioglu O Houston DS Lethagen S Van Geet C von Depka M Berger C Beurrier P Britton HA Gerrits W Guthner C Kuhle S Lorenzo JJ Makris PE Nohe N Paugy P Pautard B Torchet MF Trillot N Vicariot M Wilde J Winter M Chambost H Ingerslev J Peters M Strauss G Use of recombinant factor VIIa (NovoSeven) in patients with Glanzmann thrombasthenia Semin Hematol 2001 38 21 25 11735106 10.1053/shem.2001.29509
Caglar K Cetinkaya A Aytac S Gumruk F Gurgey A Use of recombinant factor VIIa for bleeding in children with Glanzmann thrombasthenia Pediatr Hematol Oncol 2003 20 435 438 14631616
Poon MC Use of recombinant factor VIIa in hereditary bleeding disorders Curr Opin Hematol 2001 8 312 318 11604567 10.1097/00062752-200109000-00008
d'Oiron R Menart C Trzeciak MC Nurden P Fressinaud E Dreyfus M Laurian Y Negrier C Use of recombinant factor VIIa in 3 patients with inherited type I Glanzmann's thrombasthenia undergoing invasive procedures Thromb Haemost 2000 83 644 647 10823254
| 15713229 | PMC551614 | CC BY | 2021-01-04 16:31:42 | no | BMC Blood Disord. 2005 Feb 15; 5:3 | utf-8 | BMC Blood Disord | 2,005 | 10.1186/1471-2326-5-3 | oa_comm |
==== Front
BMC Med GenetBMC Medical Genetics1471-2350BioMed Central London 1471-2350-6-61570116710.1186/1471-2350-6-6Research ArticleFamilial hypercholesterolemia in St.-Petersburg: the known and novel mutations found in the low density lipoprotein receptor gene in Russia Zakharova Faina M [email protected] Dorte [email protected] Michail Y [email protected] Valery I [email protected] Peter H [email protected] Gitte G [email protected] Anette [email protected] Boris M [email protected] Vladimir O [email protected] Alexander D [email protected] Vadim B [email protected] Ole [email protected] Department of Molecular Genetics, Institute of Experimental Medicine, St.-Petersburg, Russia2 Department of Medicine and Cardiology, Aarhus Sygehus, Aarhus University Hospital, Aarhus, Denmark3 Department of Clinical Biochemistry, Aarhus Sygehus, Aarhus University Hospital, Aarhus, Denmark4 Institute of Human Brain, St.-Petersburg, Russia5 Department of Biochemistry, Institute of Experimental Medicine, St.-Petersburg, Russia2005 8 2 2005 6 6 6 4 10 2004 8 2 2005 Copyright © 2005 Zakharova 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
Familial hypercholesterolemia is a human monogenic disease caused by population-specific mutations in the low density lipoprotein (LDL) receptor gene. Despite thirteen different mutations of the LDL receptor gene were reported from Russia prior to 2003, the whole spectrum of disease-causing gene alterations in this country is poorly known and requires further investigation provided by the current study.
Methods
Forty-five patients with clinical diagnosis of FH were tested for the apolipoprotein B (apoB) mutation R3500Q by restriction fragment length analysis. After exclusion of R3500Q mutation high-sensitive fluorescent single-strand conformation polymorphism (SSCP) analysis and automatic DNA sequencing were used to search for mutations in the LDL receptor gene.
Results
We found twenty one rare sequence variations of the LDL receptor gene. Nineteen were probably pathogenic mutations, and two (P518P, T705I) were considered as neutral ones. Among the mutations likely to be pathogenic, eight were novel (c.670-671insG, C249X, c.936-940del5, c.1291-1331del41, W422X, c.1855-1856insA, D601N, C646S), and eleven (Q12X, IVS3+1G>A, c.651-653del3, E207X, c.925-931del7, C308Y, L380H, c.1302delG, IVS9+1G>A, V776M, V806I) have already been described in other populations. None of the patients had the R3500Q mutation in the apoB gene.
Conclusions
Nineteen pathogenic mutations in the LDL receptor gene in 23 probands were identified. Two mutations c.925-931del7 and L380H are shared by St.-Petersburg population with neighbouring Finland and several other mutations with Norway, Sweden or Denmark, i.e. countries from the Baltic Sea region. Only four mutations (c.313+1G>A, c.651-653del3, C308Y and W422X) were recurrent as all those were found in two unrelated families. By this study the number of known mutations in the LDL receptor gene in St.-Petersburg area was increased nearly threefold. Analysis of all 34 low density lipoprotein receptor gene mutations found in St.-Petersburg argues against strong founder effect in Russian familial hypercholesterolemia.
==== Body
Background
Familial hypercholesterolemia (FH) (OMIM #143890) is one of the most common monogenic human diseases. It is inherited as an autosomal dominant trait with the prevalence of the heterozygous form conventionally considered to be about 1 of 500 in most populations [1]. Elevated blood serum cholesterol is due to impaired removal of low-density lipoproteins (LDL) from blood by LDL receptors, and it is associated with early onset coronary artery disease and myocardial infarction. Impairment of LDL receptor function usually results either from the absence or deficiency of the LDL receptor (OMIM *606945) itself or from a common mutation (R3500Q) in the gene of the receptor's ligand, apolipoprotein B (OMIM +107730) causing type B of the autosomal dominant hypercholesterolemia (OMIM #144010) [2]. Furthermore, a third type of monogenic autosomal dominant hypercholesterolemia (OMIM #603776) is due to the recently discovered defects in the proprotein convertase PCSK9 gene (OMIM *607786) [3]. A form of recessively inherited hypercholesterolemia (OMIM #603813) has a prevalence of less than 1:10 000000 and is due to defects in the LDL receptor adaptive protein ARH (OMIM *605747) [4,5]. Even though many forms of monogenic hypercholesterolemia are known, only apoB gene and LDL receptor gene variations seem to contribute significantly to the CHD morbidity in most populations [6].
The spectrum of LDL receptor mutations varies between different human populations and more than 900 mutations in the LDL receptor gene have been characterized worldwide [7-9]. A recent review [9] shows a clear difference in the LDL receptor gene mutations spectra for Western European countries, but this review gives nearly no data on the genetics of FH in the Eastern European countries and Russia. However, many mutations were described from Poland [10] and Bulgaria [11] and already thirteen different LDL receptor gene mutations have been published from the Russian population prior to 2003 [12]. In the present research we expand the study of the molecular genetic basis for FH in St.-Petersburg known to be the most well studied region of Russia in respect to FH-causing mutations and report 21 mutations, previously unknown in Russia.
Methods
Patients
Patients with FH were recruited from two lipid clinics of St.-Petersburg, namely Institute of Human Brain and Institute of Experimental Medicine. The clinical diagnosis of familial hypercholesterolemia was based on the following criteria: highly elevated plasma total cholesterol and LDL-cholesterol, presence of tendon xanthomata, corneal arcus or both, and positive family history of myocardial infarction and hypercholesterolemia with at least one first-degree relative affected. Forty-five probands fulfilling at least two of three criteria listed above were selected for the study. Full data of patients including their lipid data are given in the Discussion section (see Table 2). Informed consent was obtained in each case for DNA testing procedures.
Biochemical procedures
Genomic DNA was extracted from blood white cells using a standard method [13]. The patients were initially tested for the apoB mutation R3500Q by restriction fragment length analysis [14]. Exons of the LDL receptor gene were then amplified [15], and PCR products were subjected to gel-electrophoresis followed by ethidium bromide or silver staining of DNA to exclude non-specific amplification. Single-strand conformation polymorphism (SSCP) analysis was performed in an ABI 377 DNA Sequencer (PE-Applied Biosystems) sequencing device using 4,25% non-denaturing MDE polyacrylamide gels (Cambrex) at 20°C. The gels were run under two different conditions, i.e. standard MDE gel or with the addition of 5% glycerol to the MDE gel. Automated DNA sequencing was performed in part in the ABI 377 DNA Sequencer (PE-Applied Biosystems), in part in an ALFExpress-2 DNA sequencer (Amersham Life Sciences), using primers for routine PCR DNA amplification. In the case of samples bearing the deletions, PCR products were cloned into a commercially available vector (TACLONE, Medigen, Novosibirsk) and sequenced using universal and reverse primers.
Results
None of the 45 patients had the apoB R3500Q mutation, whereas 25 patients had mutations in the LDL receptor gene. Large genomic rearrangements in the LDL receptor gene were previously shown to be an uncommon cause of FH in St.-Petersburg [16] and had been excluded in most patients of the current group. Therefore we restricted our search to point mutations and to minor deletions and insertions. We identified 21 sequence variations (Table 1) of which two probably were not pathogenic (in the following considerations we give numbering of aminoacids in the LDL receptor according to Yamamoto's nomenclature, [17] ). The T705I mutation (known also as FH Paris-9) is not associated with elevated serum cholesterol [18], and the transition c.1617C>T (P518P) is synonymous. We consider the remaining 19 variations to be pathogenic. Eleven mutations have been described in other populations, but to our knowledge the remaining 8 mutations have not been described so far [7,8]. From two of these nucleotide substitutions (C249X and W422X) the stop codon arises, four are frame-shift mutations leading to premature stop codons (FsK202:S205X; FsK290:N309X; FsV409:S423X; FsV597:A622X), and two result in amino acid substitutions (D601N and C646S). Sequencing of the cloned mutant allele bearing the five-nucleotide deletion (c.936-940del5 or FsE291) is demonstrated on Fig. 1. Mutations c.313+1G>A, c.651-653del3, C308Y and W422X were found in two probands each.
Rapid tests were developed for most of the mutations and all of those were confirmed by using these methods (Table 1). Various detection methods, including heteroduplex analysis, restriction enzyme tests and SSCP for several mutations from the list are illustrated by Figures 2, 3. Cosegregation of the mutations and elevated blood serum cholesterol was demonstrated in seventeen families (see Fig. 2, 3 and Table 1). Totally the mutations were confirmed in 9 relatives and excluded in 27 members of the proband's families. Most important the diagnosis of FH was excluded in 9 and set in 2 children of probands before adulthood, i.e. prior to age 18. In these children mutation detection was of crucial importance to set the diagnosis and to suggest further life style.
In case of the Q12X mutation, the c.97C>T transition leads to occurrence of a new Mae I restriction site (CTAG) in exon 2 that allows a simple detection of this mutation. Restriction enzyme test was performed for the proband and her 4 descendants (Fig. 2). The presence of the mutation was confirmed in the son of the proband and excluded in the daughter and two grandchildren.
In some other mutations no restriction enzyme tests can be developed for their rapid detection. For example, it is true for the recurrent mutation IVS3+1G>A (c.313+1G>A) that was identified by means of SSCP and verified by sequencing in probands from two unrelated families. SSCP patterns on silver-stained gels differ strikingly in patients with and without the mutation. The presence of the mutation was confirmed in the son of the proband by SSCP.
In heterozygotes a deletion of 41 nucleotides at c.1291-1331 in exon 9 (mutation FsV409:S423X) results in occurrence of specific heteroduplexes. Besides, a PCR fragment of smaller size as compared to the normal allele is revealed in silver-stained polyacrylamide gels (Fig. 3). This mutation was identified in the son and daughter of the proband and excluded in the second daughter and in her child (Fig. 3).
Discussion
Previously 13 mutations in the LDL receptor gene have been reported in FH patients residing in St.-Petersburg [12] (see Table 3). Eight of those have not been reported in other populations. Our study revealed 21 mutations in the LDL receptor gene, out of which only T705I was previously reported from St.-Petersburg [12]. Nineteen of those 21 mutations are likely to be pathogenic. We exclude P518P (CCC>CCT; c.1617 C>T) from the list of pathogenic mutations since the transition c.1617 C>T results neither in an amino acid substitution nor in appearance of the new consensus splicing sequences. P518P mutation was found in the proband with mutation C646S, but it was not clarified if the mutations were in cis- or trans- position. Also we consider the T705I variant not to be a primary cause of FH, since the I705 allele itself is not associated with elevated cholesterol level [18] and the probable hypercholesterolemic effect of this mutation may be due to its linkage with other pathogenic LDL receptor gene mutations. Indeed, a variation in intron 7 (c.1061-8T) of unclear functional significance was shown to be very tightly linked to the I705 allele [19,20]. We have not searched for this intronic variant in the LDL receptor in the patient with the T705I substitution.
Among pathogenic mutations reported here, eleven have been described in other populations (Table 1) [7,8], and eight are novel. Six of the novel mutations lead to premature stop codons and result in truncated protein chains that probably lose their function. One out of these truncating mutations (c.936-940del5 or FsE291) also changes invariant nucleotides nearby exon-intron junction and thus may affect splicing. The D601N missense mutation, causing substitution of aspartic acid by asparagine has not been reported before. It seems likely that such a substitution might cause the loss of function since one other mutation in the same codon (D601Y) was described in familial hypercholesterolemia subjects [7]. The C646S also has not been reported before, but 5 other mutations affecting this codon have been described, one of which, C646Y (FH French Canadian-2), results in a transport defective protein (mutation class 2A) [21]. We, therefore, find it very likely that the C646S mutation is also pathogenic, but expression studies are needed to justify the effect of both missense mutations D601N and C646S. Previously described mutations were considered to be pathogenic due mostly to their listing in FH mutation databases even though functional studies were not systematically performed. Indeed, V806I mutation (known as variant FH New York -5) [22] occurs in the LDL receptor internalization signal NPVY, for which the consensus sequence NPxY is given (where X is not a conserved aminoacid) [23] and thus the substitution of isoleucin for valine may be not crucial for the LDL receptor function. The V776M mutation may have effect on LDL receptor mRNA splicing rather than to be realized on the protein level since the V776M mutation changes the invariant G at the 3' end of exon 16. However, this mutation is likely to be pathogenic since it was reported already in patients from La Habana, Cuba [24].
The apoB R3500Q mutation was not detected in any of our patients. This finding is in agreement with the previous observation that the R3500Q mutation had not been found in St.-Petersburg [25]. The mutation has been detected in another part of Russia only in 2 of 71 patients with symptoms of familial hypercholesterolemia [26]. The apoB R3500Q mutation is almost exclusively found in Caucasian individuals, and almost all subjects with the mutation carry the same haplotype. One of the highest frequencies of the mutation has been found in the Swiss population (approximately 1/200) [27], and Miserez and Muller [27] hypothesized that the mutation may have arisen in Switzerland 10,000 – 6,000 years ago. The prevalence of the mutation declines with increasing distance from the Central Europe [27,28], and the prevalence of the apoB R3500Q mutation is therefore expected to be low in the St.-Petersburg area (< 1/1000). A precise estimate would require a random sampling of the general population.
Mutations in LDL receptor gene typical for various ethnic groups were revealed in St.-Petersburg (Table 1). In particular, several mutations found in Denmark, Finland, Norway and Sweden are present in the St.-Petersburg population. The ethnic origin of these families is unclear and according to questioning and family names no evidence of a Scandinavian origin of probands was obtained. Only exception from our previous FH group was the proband with the E207X mutation who stated his German roots and indeed this mutation was previously reported from several families from Germany [29]. Interestingly, no specific Slavic or Eastern-European founder mutations were found in St.-Petersburg when comparing the LDL receptor mutation spectrum of Russia to those of Poland [10], Bulgaria [11] or Czech Republic [30,31]. G571E was found in Russia, Czech Republic and Poland but this mutation was also reported from many other countries worldwide. The C188Y mutation found in Russia [32] and in Czech Republic [30,31] cannot be considered a founder Slavic mutation, since it was reported from unique families in each country. Only one candidate for a Russian founder mutation is C139G identified in four unrelated families in different regions of the country, but this mutation is absent in related nations of the Eastern Europe as well as in other countries in the world.
In our study we were able to find LDL receptor gene pathogenic mutations in 23 of 45 patients. Many methods are considered to be superior to routine isotopic SSCP when screening for mutations in DNA (see e.g. [33]for comparison of SSCP and DHPLC). However, we believe that mutations could have been overlooked due to intrinsic limitations of SSCP method in our hands only in few cases. Fluorescent SSCP, used in the current study, can be superior even to DHPLC and is a sensitive method for mutation screening, especially when different gel electrophoresis conditions are applied [34]. Recent studies indicate, that SSCP run under two different conditions detect up to 96% of heterozygous variations (P.H. Nissen, unpublished results). In our study LDL receptor gene mutations were found in 56% (14 out of 26) patients selected by SSCP and in 53% (10 out of 19) patients which DNA was subjected to direct sequencing of all gene exons (Table 2). We do not find it likely that many patients are underdiagnosed due to presence of large genomic rearrangements. In a study conducted in St.-Petersburg, in the partly overlapping sample of FH patients, only one case of a large deletion was found in the sample of 50 probands, giving the rough estimate of genomic rearrangements 2% [16]. In fact in mixed populations (such as that of that of St. Petersburg) the contribution of large rearrangements to the spectrum of pathogenic mutations seems to vary from 6% in Great Britain [35] to 2.5% in English-speaking Canadians [36]. More important, some of intronic mutations leading to defects of splicing could be missed, since the design of primers [15] mostly following recommendations of Hobbs et al. [22] with the only exclusion for primers to amplify exon 3 allowed analyzing only 14 out of 34 intron-exon boundaries in the LDL receptor gene. Recent investigation [37] demonstrated that a high percent of previously missed LDL receptor mutations may be localized in introns. In the cited study [37] in the patient sample after modifying the gene analysis procedure nearly 27% of patients turned out to have intronic mutations, despite the functional significance of these mutations have still to be validated.
In the apoB gene we only looked for the R3500Q mutation, and the possibility that other mutations in that gene are responsible for hyperlipidemia cannot be ruled out. This possibility is unlikely, since the apoB gene has been studied extensively in other laboratories and no other pathogenic variants have been determined. To conclude we cannot definitely rule out the possibility that some undiscovered mutations in the LDL receptor gene have not been found. However, the possibility that the patients have mutations in a different gene involved in FH still remains since the mutation in the PCSK9 gene [3] were not tested in this study.
With the two exceptions, mutations previously reported from Russia [12,38] have been confined to a single family. The exceptions were the mutations G197del (FH-Lithuania), found in high percent of Ashkenazi Jewish families with FH from St.-Petersburg [39], and the C139G mutation found in two Slavic families from St.-Petersburg [40,12], one family in Novosibirsk [38] and one family in Moscow [41]. In this study 4 mutations, namely c.313+1G>A, c.651-653del3, C308Y and W422X were found in two unrelated families. Together with the previous findings we discovered 34 mutations in the LDL receptor gene in St. Petersburg of which only six were detected in more than one family. Fourteen LDL receptor gene mutations were discovered by the other Russian FH team from Moscow, Russia out of which only C139G is shared with St.-Petersburg population [12,41]. To date a total of 47 different FH mutations are known from Russia. The data presented here enlarge the spectrum of mutations found in the Russian population and make us regard it as genetically heterogeneous.
Conclusions
We identified nineteen pathogenic mutations in the LDL receptor gene in 23 probands and two probably neutral mutations. In our study only four mutations in the LDL receptor gene (c.313+1G>A, c.651-653del3, C308Y and W422X) were found to be recurrent, i.e. all of those were found in two apparently unrelated families. Together with the data obtained earlier [12,38] our results present an evidence against the strong founder effect among Russian FH patients, and it is likely that in the St.-Petersburg area there is as much genetic heterogeneity as in most other areas of the world.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
FMZ, MYM, VIG, PHN, GGN, AS performed cloning, SSCP, sequencing and familial analysis, BML, VOK, DD, ADD selected patients, VBV and OF participated in the study design and coordination. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgments
This research was supported by FEBS short-term fellowship, grant of the Russian Fund for Basic Research (05-04-48235) and by the grants of the Presidential Program for young scientists (MK-899.2003.04) and the Leading scientific schools of Russia (1730.2003.4). Authors thanks cordially Prof. Anne Soutar and Prof. Kimmo Kontula for critical reading of the manuscript and useful comments helping to improve the text.
Figures and Tables
Figure 1 Sequencing of the cloned mutant allele bearing the five nucleotide deletion (c.936-940del5 or FsE291). The sequence of the normal allele is shown at the top. The nucleotides absent in the mutant allele are underlined. The sequence of the mutant allele is shown at the bottom.
Figure 2 Detection of mutation Q12X in the LDL receptor gene by means of Mae I restriction enzyme test. Transition c.97 C>T in exon 2 of the LDL receptor gene leads to appearance of the new Mae I restriction site (CTAG) in the patients with the mutation. Mae I restriction enzyme test enables to confirm the presence of mutation Q12X identified by DNA sequencing in the proband (12-1) and in her son (12-2) and to exclude presence of the mutation in other relatives of the proband, including daughter (12-4) and grandchildren (12-3 and 12-5). Lengths of DNA restriction fragment are given at the left in bp and *total blood serum cholesterol figures of the patients – at the bottom of the gel in mg/dl.
Figure 3 Detection of mutation c.1291-1331del41 (mutation FsV409:S423X) of the LDL receptor gene by means of PCR product sizing and heteroduplex analysis. Deletion of 41 nucleotides results in significant change (indicated by C) of molecular weight of the PCR-amplified LDL receptor gene exon 9 fragment and in formation of specific heteroduplexes (indicated by A). Letter B indicates the PCR product of normal size. Current gel supports the presence of c.1291-1331del41 mutation in two children (3-2, 3-3) of the proband (3-1) and absence of this mutation in his daughter and grandson (3–4, 3–5). *total blood serum cholesterol figures of the patients – at the bottom of the gel in mg/dl.
Table 1 List of the LDL receptor gene mutations found in the current study
Mutation, systematic name Nucleotide change Exon/Intron Rapid test method Occurrence in other populations [7, 8] Number of families (patients) with the mutation
Missense-mutations
C308Y c.985 G>A exon 7 FblI China 2 (2)
L380H (FH Pori) c.1202 T>A exon 9 MnlI Finland 1 (1)
D601N c.1864 G>A exon 13 EcoRV None (New) 1 (2)
C646S c.1999 T>A exon 14 SSCP None (New) 1 (1)
V776M c.2389 G>A exon 16 SSCP China (Hong-Kong), Cuba, South Africa (Afrikaners) 1 (2)
V806I (FH New York-5) c.2479 G>A exon 17 The Netherlands, USA 1 (1)
Nonsense-mutations
Q12X (FH Turkey/Milan-4) c.97 C>T exon 2 MaeI Italy, France, Turkey 1 (2)
E207X (FH Morocco) c.682 G>T exon 4 MaeI China, Germany, Korea, Morocco, Norway, Sweden, UK, USA 1 (2)
C249X c.810 C>A exon 5 DdeI, Fnu4HI None (New) 1 (1)
W422X c.1328 G>A exon 9 MaeI None (New) 2 (3)
Splice site mutations
IVS3+1G>A (FH-Elverum/ Olbia) c.313+1G>A intron 3 SSCP Austria, Belgium, Denmark, Germany, Italy, Spain, Korea, Norway, The Netherlands, UK, Sweden, South Africa (black) 2 (3)
IVS9+1G>A c.1358+1G>A intron 9 AsuHPI The Netherlands 1 (1)
Frameshift mutations
FsK202: S205X c.670-671insG exon 4 SSCP None (New) 1 (2)
FsE287: V348X (FH North Karelia) c.925-931del7 exon 6 SSCP Finland, Sweden, USA 1 (1)
FsE291: N309X c.936-940del5 exon 6 SSCP None (New) 1 (1)
FsV409: S423X c.1291-1331del41 exon 9 Sizing None (New) 1 (3)
FsE414: M429X c.1302delG exon 9 SSCP Germany 1 (1)
FsV597: A622X c.1855-1856insA exon 13 None (New) 1 (1)
In-frame deletions
G197del c.651-653del3 exon 4 HA UK 2 (2)
Neutral mutations
T705I (FH Paris-9) c.2177 C>T exon 15 Denmark, France, The Netherlands, UK, USA etc. 1 (1)
Silent mutations
P518P c.1617 C>T exon 11 None (New) 1 (1)
Footnote: HA – heteroduplex analysis; SSCP – single-strand conformation polymorphism analysis. Numeration of nucleotides and aminoacids follows Yamamoto's nomenclature [17] and the letter c. before the number of nucleotide indicates that it was taken from cDNA sequence.
Table 2 Clinical features of St. Petersburg patients recruited for search of mutations in the LDL receptor gene
Patients ID Sex Age TC, mg/dl TG, mg/dl LDLC, mg/dl HDLC, mg/dl Xanthomas/Corneal arcus CHD CI Mutation name All exons sequencing
1 F 61 449 178 386 27 + + + V776M -
2 F 58 675 305 572 42 + + - T705I -
3 M 60 350 81 284 50 + + - FsV409: S423X -
4 F 49 326 203 248 37 - + - - -
5 F 36 414 201 321 53 + - - V806I -
6 M 68 404 192 333 33 + + - - -
7 F 76 522 161 444 46 - - - - -
8 F 54 446 165 353 60 - - - - -
9 F 52 500 ND ND ND - + + - -
10 F 62 370 120 303 43 + - - C308Y -
11 F 66 359 141 300 31 - + - - -
12 F 66 486 296 391 36 - + + Q12X -
13 F 52 469 228 389 34 + + - IVS9+1G>A -
14 F 53 325 101 286 19 - + + - -
15 F 58 446 192 334 74 - + - FsK202: S205X -
16 F 38 398 260 314 32 + - - FsE291: N309X -
17 M 46 280 117 220 37 - + + - -
18 F 58 519 179 446 37 - + - G197del -
19 F 54 364 135 300 37 + + + FsE287: V348X -
20 F 53 204 233 135 23 - + +++ - -
21 M 61 350 116 280 47 - - - - -
22 F 48 369 221 290 35 - + - - -
23 M 48 417 190 353 30 - + - C249X -
24 F 45 299 122 229 46 + - - IVS3+1G>A -
25 F 50 353 101 295 38 + - - FsE414: M429X -
26 F 45 558 210 478 38 + + + - -
27 M 62 469 136 398 44 + + ++ G197del +
28 F 8 915 179 850 29 + - - - +
29 F 40 367 73 283 69 - + - W422X +
30 F 60 480 104 393 66 - - - - +
31 M 41 314 247 226 39 - + + - +
32 M 37 344 292 262 24 + + + - +
33 F 52 360 744 179 32 + + + - +
34 M 40 312 70 258 40 + + ++ - +
35 M 29 253 79 195 42 - - - E207X +
36 M 38 520 168 447 39 + + + D601N +
37 F 19 447 156 362 54 - - - L380H +
38 F 63 347 84 266 64 + + + C646S +
39 F 56 432 106 374 37 + + - W422X +
40 M 47 540 226 446 49 + + + IVS3+1G>A +
41 F 70 433 123 370 38 - + + - +
42 M 34 534 ND ND ND + + + - +
43 F 52 654 115 573 58 + + + C308Y +
44 M 52 356 334 260 29 - + + - +
45 M 52 360 169 287 39 + + + FsV597: A622X +
Footnote: TC – total blood serum cholesterol, TG – triglycerides, LDLC – low density lipoprotein cholesterol, HDLC – high density lipoprotein cholesterol, CHD – coronary heart disease, CI – coronary infarction. Pluses in the CI column indicate number of CI survived. Minus in the right column indicates that the patient's LDL receptor gene was studied by high-sensitive fluorescent SSCP-analysis and not by frontal sequencing of all LDL receptor gene exons; the latter condition is indicated by plus.
Table 3 List of the LDL receptor gene mutations reported previously from St. Petersburg, Russia [12 and References therein]
Mutation, systematic name Nucleotide change Exon/Intron Rapid test method Occurrence in other populations [7, 8] Number of families (patients) with the mutation
Large rearrangements
delta 5kb Large deletion exons 3-5 Southern blot None 1 (2)
Missense-mutations
C127W c.444 T>G exon 4 Mva I None 1 (1)
G128G, A130P [c.447 T>C; c.451 G>C] exon 4 Cac 8 I, Bsu R I None 1 (1)
C139G c.478 T>G exon 4 Msp I None 1 (3)
C146R c.499 T>C exon 4 Apa I None 1 (3)
C188Y c.626 G>A exon 4 Rsa I Czech Republic 1 (3)
G571E c.1775 G>A exon 12 SSCP Italy, Germany, Poland, Czech Republic, Austria, Belgium, Greece 1 (1)
Nonsense-mutations
C74X c.285 C>A exon 3 DdeI Korea, Northern Japan 1 (1)
E397X c.1252 C>T exon 9 Alu I None 1 (7)
In-frame deletions
347delGCC c.347-349del exon 4 HA, Fnu4HI None 1 (1)
G197del c.652-654del exon 4 HA Israel, USA, UK, Poland, Czech Republic, Germany, South Africa, The Netherlands 7 (14)
Neutral mutations
T705I (FH Paris-9) c.2177 C>T exon 15 Denmark, France, The Netherlands, UK, USA etc. 1 (1)
Silent mutations
H229H c.750 C>T exon 5 Nco I None (New) 1 (1)
Footnote: HA – heteroduplex analysis; SSCP – single-strand conformation polymorphism analysis. Numeration of nucleotides and aminoacids follows Yamamoto's nomenclature [17] and the letter c. before the number of nucleotide indicates that it was taken from cDNA sequence.
==== Refs
Goldstein JL Hobbs HH Brown MS Scriver CR, Beaudet AL, Sly WS, Vale D Familial hypercholesterolaemia The metabolic and molecular basis of inherited disease 2001 III New York: McGraw Hill 2863 2914
Innerarity TL Mahley RW Weisgraber KH Bersot TP Krauss RM Vega GL Grundy SM Friedl W Davignon J McCarthy BJ Familial defective apolipoprotein B-100: a mutation of apolipoprotein B that causes hypercholesterolemia J Lipid Res 1990 31 1337 1349 2280177
Abifadel M Varret M Rabes JP Allard D Ouguerram K Devillers M Cruaud C Benjannet S Wickham L Erlich D Derre A Villeger L Farnier M Beucler I Bruckert E Chambaz J Chanu B Lecerf JM Luc G Moulin P Weissenbach J Prat A Krempf M Junien C Seidah NG Boileau C Mutations in PCSK9 cause autosomal dominant hypercholesterolemia Nat Genet 2003 34 154 156 12730697 10.1038/ng1161
Garcia CK Wilund K Arca M Zuliani G Fellin R Maioli M Calandra S Bertolini S Cossu F Grishin N Barnes R Cohen JC Hobbs HH Autosomal recessive hypercholesterolemia caused by mutations in a putative LDL receptor adaptor protein Science 2001 292 1394 1398 11326085 10.1126/science.1060458
Soutar AK Naoumova RP Traub LM Genetics, clinical phenotype, and molecular cell biology of autosomal recessive hypercholesterolemia Arterioscler Thromb Vasc Biol 2003 23 1963 1970 12958046 10.1161/01.ATV.0000094410.66558.9A
Goldstein JL Brown MS The cholesterol quartet Science 2001 292 1310 1312 11360986 10.1126/science.1061815
The low density lipoprotein receptor (LDLR) gene in familial hypercholesterolemia
Universal Mutation Database
Dedoussis GVZ Schmidt H Genschel J LDL-receptor mutations in Europe Hum Mutat 2004 24 443 459 15523646 10.1002/humu.20105
Gorski B Kubalska J Naruszewicz M Lubinski J LDL-R and Apo-B-100 gene mutations in Polish familial hypercholesterolemias Hum Genet 1998 102 562 565 9654205 10.1007/s004390050740
Mihaylov VA Horvath AD Savov AS Kurshelova EF Paskaleva ID Goudev AR Stoilov IR Ganev VS Screening for point mutations in the LDL receptor gene in Bulgarian patients with severe hypercholesterolemia J Hum Genet 2004 49 173 176 15015036 10.1007/s10038-004-0127-6
Low density lipoprotein receptor gene mutations in Russian patients with familial hypercholesterolemia
Kunkel LM Smith KD Boyer SH Borgaonkar DS Wachtel SS Miller OJ Breg WR Jones HW Rary JM Analysis of human Y-chromosome-specific reiterated DNA in chromosome variants Proc Natl Acad Sci U S A 1977 74 1245 1249 265567
Hansen PS Rudiger N Tybjaerg-Hansen A Faergeman O Gregersen N Detection of the apoB-3500 mutation (glutamine for arginine) by gene amplification and cleavage with MspI J Lipid Res 1991 32 1229 1233 1719111
Jensen HK Jensen LG Hansen PS Faergeman O Gregersen N High sensitivity of the single-strand conformation polymorphism method for detecting sequence variations in the low-density lipoprotein receptor gene validated by DNA sequencing Clin Chem 1996 42 1140 1146 8697568
Mandelshtam MJu Lipovetskyi BM Schwartzman AL Gaitskhoki VS A novel deletion in the low density lipoprotein receptor gene in a patient with familial hypercholesterolemia from Petersburg Hum Mutat 1993 2 256 260 8401534
Yamamoto T Davis CG Brown MS Schneider WJ Casey ML Goldstein JL Russell DW The human LDL receptor: A cysteine-rich protein with multiple Alu sequences in its mRNA Cell 1984 39 27 38 6091915 10.1016/0092-8674(84)90188-0
Lombardi P Sijbrands EJ Kamerling S Leuven JA Havekes LM The T705I mutation of the low density lipoprotein receptor gene (FH Paris-9) does not cause familial hypercholesterolemia Hum Genet 1997 99 106 107 9003505 10.1007/s004390050321
Heath KE Whittall RS Miller GJ Humphries SE I705 variant in the low density lipoprotein receptor gene has no effect on plasma cholesterol levels J Med Genet 2000 37 713 715 11182933 10.1136/jmg.37.9.713
Mozas P Cenarro A Civeira F Castillo S Ros E Pocovi M Mutation analysis in 36 unrelated Spanish subjects with familial hypercholesterolemia: identification of 3 novel mutations in the LDL receptor gene Hum Mutat 2000 15 483 484 10790219 10.1002/(SICI)1098-1004(200005)15:5<483::AID-HUMU19>3.0.CO;2-Q
Leitersdorf E Tobin EJ Davignon J Hobbs HH Common low-density lipoprotein receptor mutations in the French Canadian population J Clin Invest 1990 85 1014 1023 2318961
Hobbs HH Brown MS Goldstein JL Molecular genetics of the LDL receptor gene in familial hypercholesterolemia Hum Mutat 1992 1 445 466 1301956
Chen WJ Goldstein JL Brown MS NPXY, a sequence often found in cytoplasmic tails, is required for coated pit-mediated internalization of the low density lipoprotein receptor J Biol Chem 1990 265 3116 3123 1968060
Pereira E Ferreira R Hermelin B Thomas G Bernard C Bertrand V Nassiff H Mendez del Castillo D Bereziat G Benlian P Recurrent and novel LDL receptor gene mutations causing heterozygous familial hypercholesterolemia in La Habana Hum Genet 1995 96 319 322 7649549 10.1007/BF00210415
Shevtsov SP The APOB gene encoding putative low density lipoprotein receptor binding domain of the ApoB-100 protein shows no DNA polymorphisms Genetika 1996 32 295 7 (Translated from Genetika 32:295-297). 8713626 8713626
Krapivner SR Malyshev PP Rozhkova TA Poltaraus AB Kukharchuk VV Bochkov VN [Application of DNA analysis for differential diagnosis of familial hypercholesterolemia and familial defect of apolipoprotein B-100] Ter Arkh 2000 72 9 12 [Article in Russian] 10833789
Miserez AR Muller PY Familial defective apolipoprotein B-100: a mutation emerged in the mesolithic ancestors of Celtic peoples? Atherosclerosis 2000 148 433 436 10657582 10.1016/S0021-9150(99)00470-0
Horvath A Ganev V The mutation APOB-100 R3500Q in Eastern Europe Atherosclerosis 2001 156 241 242 11417523 10.1016/S0021-9150(01)00482-8
Nauck MS Koster W Dorfer K Eckes J Scharnagl H Gierens H Nissen H Nauck MA Wieland H Marz W Identification of recurrent and novel mutations in the LDL receptor gene in German patients with familial hypercholesterolemia Hum Mutat 2001 18 165 166 11462246 10.1002/humu.1171
Kuhrova V Francova H Zapletalova P Freiberger T Fajkusova L Hrabincova E Slovakova R Kozak L Spectrum of low density lipoprotein receptor mutations in Czech hypercholesterolemic patients Hum Mutat 2001 18 253 11524740 10.1002/humu.1185
Kuhrova V Francova H Zapletalova P Freiberger T Fajkusova L Hrabincova E Kozak L Slovakova R Spectrum of low density lipoprotein receptor mutations in Czech hypercholesterolemic patients Hum Mutat 2002 19 80 (Corrected and republished from Hum Mutat 2001, 18:253). 11754108 10.1002/humu.9000
Tatishcheva YuA Mandelshtam MYu Golubkov VI Lipovetsky BM Gaitskhoki VS Four new mutations and two polymorphic variants of the low-density lipoprotein receptor gene in familial hypercholesterolemia patients from St. Petersburg Rus JGenet 2001 37 1082 1086 (Translated from Genetika. 37:1290-1295). 11642133 10.1023/A:1011973817437
Bunn CF Lintott CJ Scott RS George PM Comparison of SSCP and DHPLC for the detection of LDLR mutations in a New Zealand cohort Hum Mutat 2002 19 311 11857755 10.1002/humu.9021
Ellis LA Taylor CF Taylor GR A comparison of fluorescent SSCP and denaturing HPLC for high throughput mutation scanning Hum Mutat 2000 15 556 564 10862085 10.1002/1098-1004(200006)15:6<556::AID-HUMU7>3.0.CO;2-C
Horsthemke B Dunning A Humphries S Identification of deletions in the human low density lipoprotein receptor gene J Med Genet 1987 24 144 147 3572996
Langlois S Kastelein JJP Hayden MR Characterization of six partial deletions in the low-density-lipoprotein (LDL) receptor gene causing familial hypercholesterolemia (FH) Am J Hum Genet 1988 43 60 68 2837085
Amsellem S Briffaut D Carrie A Rabes JP Girardet JP Fredenrich A Moulin P Krempf M Reznik Y Vialettes B de Gennes JL Brukert E Benlian P Intronic mutations outside of Alu-repeat-rich domains of the LDL receptor gene are a cause of familial hypercholesterolemia Hum Genet 2002 111 501 510 12436241 10.1007/s00439-002-0813-4
Mandelshtam MYu What were the outcomes of familial hypercholesterolemia studies for understanding of the hyperlipidemia genetics? Meditsinskaya Genetika 2003 2 509 519 [Article in Russian, with English Summary, p. 519]
Mandelshtam M Chakir Kh Shevtsov S Golubkov V Skobeleva N Lipovetsky B Konstantinov V Denisenko A Gaitskhoki V Schwartz E Prevalence of Lithuanian mutation among St.-Petersburg Jews with familial hypercholesterolemia Hum Mutat 1998 12 255 258 9744476
Chakir Kh Skobeleva NA Shevtsov SP Konstantinov VO Denisenko AD Schwartz EI Two novel Slavic point mutations in the low-density lipoprotein receptor gene in patients with familial hypercholesterolemia from St. Petersburg, Russia Mol Genet Metab 1998 63 31 34 9538514 10.1006/mgme.1997.2614
Meshkov AN Stambol'skii DV Krapivner SR Bochkov VN Kukharchuk VV Malyshev PP [Low density lipoprotein receptor gene mutations in patients with clinical diagnosis of familial hypercholesterolemia] Kardiologiia 2004 44 58 61 [Article in Russian] 15477777
| 15701167 | PMC551615 | CC BY | 2021-01-04 16:03:33 | no | BMC Med Genet. 2005 Feb 8; 6:6 | utf-8 | BMC Med Genet | 2,005 | 10.1186/1471-2350-6-6 | oa_comm |
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-91572071410.1186/1471-2156-6-9Research ArticleConstruction and analysis of tag single nucleotide polymorphism maps for six human-mouse orthologous candidate genes in type 1 diabetes Maier Lisa M [email protected] Deborah J [email protected] Adrian [email protected] Felicity [email protected] Jason D [email protected] Rebecca [email protected] Christopher [email protected] John [email protected] Luc J [email protected] Heather [email protected] Carolyn [email protected] Kara M [email protected] Giselle [email protected] Neil [email protected] Sarah [email protected] Dag E [email protected]ønningen Kjersti S [email protected] Cristian [email protected]îrgovişte Constantin [email protected] David A [email protected] David P [email protected] Laurence B [email protected] John A [email protected] Linda S [email protected] Rebecca C [email protected] Juvenile Diabetes Research Foundation/Wellcome Trust Diabetes and Inflammation Laboratory, Cambridge Institute for Medical Research, University of Cambridge, Wellcome Trust/MRC Building, Hills Road, Cambridge, UK2 Institute and Department of Medical Genetics, Ulleval University Hospital, University of Oslo, Oslo, Norway3 Laboratory of Molecular Epidemiology, Division of Epidemiology, Norwegian Institute of Public Health, Oslo, Norway4 Clinic of Diabetes, Institute of Diabetes, Nutrition and Metabolic Diseases 'N. Paulescu', Bucharest, Romania5 Department of Medical Genetics, Queen's University Belfast, Belfast City Hospital, Belfast, UK6 Department of Community Health Sciences, St George's Hospital Medical School, London, UK7 Department of Pharmacology, Merck Research Laboratories, Rahway, New Jersey, USA2005 18 2 2005 6 9 9 1 12 2004 18 2 2005 Copyright © 2005 Maier et al; licensee BioMed Central Ltd.2005Maier 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
One strategy to help identify susceptibility genes for complex, multifactorial diseases is to map disease loci in a representative animal model of the disorder. The nonobese diabetic (NOD) mouse is a model for human type 1 diabetes. Linkage and congenic strain analyses have identified several NOD mouse Idd (insulin dependent diabetes) loci, which have been mapped to small chromosome intervals, for which the orthologous regions in the human genome can be identified. Here, we have conducted re-sequencing and association analysis of six orthologous genes identified in NOD Idd loci: NRAMP1/SLC11A1 (orthologous to Nramp1/Slc11a1 in Idd5.2), FRAP1 (orthologous to Frap1 in Idd9.2), 4-1BB/CD137/TNFRSF9 (orthologous to 4-1bb/Cd137/Tnrfrsf9 in Idd9.3), CD101/IGSF2 (orthologous to Cd101/Igsf2 in Idd10), B2M (orthologous to B2m in Idd13) and VAV3 (orthologous to Vav3 in Idd18).
Results
Re-sequencing of a total of 110 kb of DNA from 32 or 96 type 1 diabetes cases yielded 220 single nucleotide polymorphisms (SNPs). Sixty-five SNPs, including 54 informative tag SNPs, and a microsatellite were selected and genotyped in up to 1,632 type 1 diabetes families and 1,709 cases and 1,829 controls.
Conclusion
None of the candidate regions showed evidence of association with type 1 diabetes (P values > 0.2), indicating that common variation in these key candidate genes does not play a major role in type 1 diabetes susceptibility in the European ancestry populations studied.
==== Body
Background
Type 1 diabetes is a common, multifactorial disease believed to be caused in a proportion of cases by an autoimmune destruction of pancreatic β-cells by an inflammatory infiltrate comprising T lymphocytes, dendritic cells and macrophages. This process results from a complex interaction between genetic and environmental risk factors. Genetically, it is under the control of the major histocompatibility complex (MHC) [1] and many other genes of smaller effect and mostly unknown identity.
A murine model of type 1 diabetes, the NOD mouse, spontaneously develops an autoimmune-mediated diabetes that has many similarities to the human disease. It is likely that components of the pathophysiology and genetic predisposition are conserved across species, and indeed two loci have already been shown to affect type 1 diabetes susceptibility in both species, namely the immunoregulatory MHC HLA class II and CTLA-4 genes. The other causative gene(s) in the known Idd regions controlling type 1 diabetes susceptibility in the NOD mouse could also determine susceptibility in humans, even though this depends on the frequency of susceptibility alleles in human populations, which affects statistical power, and that the correct candidate gene has been chosen from the Idd interval. These Idd intervals might contain many genes, including several involved in the immune response [2]. Nevertheless, in contrast to studies in humans based on linkage, the localisation of a type 1 diabetes locus to a specific chromosome region in the mouse genome using congenic strain breeding defines with certainty a set of genes, one or more of which is definitely a susceptibility gene [3,4].
The central importance of T cell development and function in type 1 diabetes is evident from the susceptibility genes identified so far. The MHC class II genes are important etiologically in two rat models of type 1 diabetes, the Biobreeding (BB) and KDP strains [5,6], the NOD mouse strain [3] and in humans [1], with their essential function not only in T cell activation and expansion but also in T cell repertoire formation in the thymus and clonal deletion of autoreactive cells. The BB rat type 1 diabetes susceptibility locus Ian4/Iddm1 [7] affects T lymphocyte development whereas the Cblb (KDP rat) [8] and CTLA4 [9] (in humans and NOD mice) susceptibility genes highlight the importance of the regulation of T cell activation, expansion and homeostasis in the periphery, and perhaps in the thymus as well.
In our selection of candidate genes within NOD congenic intervals, we have, therefore, biased our choice towards immune-related genes such as Il2 [2], Cd101 [10] and Nramp1/Slc11a1 [11]. From each of Idd5.2 [11], Idd9.2 [12], Idd9.3 [12], Idd10 [10], Idd13 [13] and Idd18 [14] we chose immune-associated functional candidate genes to study in human type 1 diabetes: Nramp1/Slc11a1 from Idd5.2 [11]; Frap1 from Idd9.2 (unpublished); 4-1bb/Cd137/Tnfrsf9 from Idd9.3 [12] (unpublished); Cd101/Igsf2 from Idd10 [10]; B2m from Idd13 [15] and Vav3 from Idd18 (note that very recent congenic strain mapping results indicate that the Idd18 interval contains only one gene with known immunological function, namely the VAV3 gene, and this will be published elsewhere). Table 1 summarises the main features of the six human candidate genes.
Table 1 NOD mouse Idd loci, location of their human orthologous regions, and selected functional candidate genes.
Idd Mouse chromosome Interval size (Mb) Number of genes Functional candidate genes in mouse Idd intervals Location of human orthologous region Human orthologue genes Known gene function and previously reported disease associations
Idd5.2 1 1.7 47 Idd5.2: Nramp1/Slc11a1 2q35 NRAMP1/SLC11A1 Endosomal/lysosomal acidification and associated with protection from infectious disease and susceptibility to autoimmune disease
Idd9.2 4 1.1 13 Idd9.2: Frap1 1p32 FRAP1 FKBP12-rapamycin associated protein of mTOR. Candidate tumour suppressor gene, whose function in apoptosis is influenced by allelic variation
Idd9.3 4 1.2 13 Idd9.3: 4-1bb 1p36 4-1BB Role in enhancing and regulating CD4+, CD8+ T cells and dendritic cells
Idd10 3 0.95 7 Idd10: Cd101 1p12 CD101 Co-stimulatory receptor of T cells
Idd13 2 6 cM > 50* Idd13: B2m 15q21 B2M Required for antigen presentation by MHC class I molecules and the development of diabetes in NOD mice
Idd18 3 0.7 2 Idd18: Vav3 1p13-p21 VAV3 Guanine nucleotide exchange factor involved in signalling of T and B cell receptors
*Estimated number of genes. A version of this table is provided in Additional file 1 with the supporting published references (see Additional file 1).
Results and discussion
A tag SNP approach to test for association was adopted for all genes, except for 4-1BB [16], in order to achieve cost-savings in genotyping. A multi-locus test was used to evaluate the association between type 1 diabetes and the tag SNPs due to linkage disequilibrium (LD) with one or more causal variants [17]. Coding and untranslated regions of NRAMP1 (MIM 600266), FRAP1 (MIM 601231), 4-1BB (MIM 602250), CD101 (MIM 604516), B2M (MIM 109700) and VAV3 (MIM 605541) were re-sequenced in 32 or 96 randomly chosen UK white patients with type 1 diabetes to identify SNPs and for the selection of tag SNPs. As LD between 4-1BB SNPs was weak, eight out of nine common SNPs were genotyped (minor allele frequency, MAF ≥ 0.03; one SNP could not be genotyped due to assay technical difficulties) and analysed using single-locus tests.
A total of 110 kb of re-sequenced regions yielded 220 SNPs, including six deletion/insertion polymorphisms (DIPs) (see Table 2 and Additional files 2, 3,4,5,6 and7). No coding changes or obvious candidates for variants that could change the function or expression of 4-1BB, FRAP1, or B2M were observed. A synonymous change was detected in exon 3 of NRAMP1 (MAF = 0.32) and a non-synonymous SNP (nsSNP) in exon 15 (MAF = 0.02), causing a conservative amino acid change: Asp543Asn (DIL5202/ss23142243). Interestingly, as in the case of its mouse orthologue [10], several nsSNPs were discovered in exons 3, 4, 5, and 8 of CD101 (see Additional file 5). Re-sequencing of the three alternative transcripts of VAV3, called VAV3 (27 exons), VAV3β (unique exon 1 and exons 4 to 27) and VAV3.1 (unique exon 18 and exons 19 to 27) yielded six exonic SNPs (see Additional file 7). Two SNPs, Pro611Ser (MAF = 0.13) and Gln613His (MAF = 0.13) are located in the SH3 domain of the VAV3 protein and, therefore, could result in VAV3 having altered protein interactions. In order to facilitate the computation of the selection of tag SNPs, VAV3 was divided into three sections as suggested by the pattern of LD across the gene.
Table 2 Summary of the re-sequencing study. Gene size, number of exons, amount of re-sequenced DNA for each gene (including 5' and 3' regions of gene), sequencing panel, and number of SNPs identified.
Locus Genomic size (kb) n exons Re-sequenced region (kb) n cases re-sequenced n SNPs
NRAMP1 13.58 15 (7,13)* 12.13 32 20
4-1BB 20.50 8 13.66 96 23
FRAP1 60.98 58 30.88 32 55
CD101 34.61 10 15.90 96 31
B2M 6.61 4 9.33 32 13
VAV3 393.70 27 (25,10)* 27.69 96 78
Total 529.98 122 109.59 - 220
* Number of exons in splice variants. n, number.
Two common nsSNPs (MAF ≥ 0.05; DIL1521/rs7528153 and DIL3809/ss23142432) from VAV3 and a microsatellite from NRAMP1 were genotyped a priori in the whole family collection (step 1 and 2) and a single nsSNP from CD101 in step 1 families only (DIL3794/rs3754112). The nsSNP DIL3810/ss23142433 in VAV3 was not tested because it was in quite strong LD with DIL3809/ss23142432 (R2 = 0.64), so that only DIL3809/ss23142432 was genotyped. Note that in our tag approach, the two VAV3 nsSNPs (DIL1521/rs7528153 and DIL3809/ss23142432) were chosen deliberately as tag SNPs.
In a pragmatic, phased genotyping strategy, in step 1, the multi-locus test P values for association between type 1 diabetes and candidate gene tag SNPs all exceeded 0.2, as did the single-locus test P values for 4-1BB SNPs. Consequently, we did not proceed to genotype in step 2 samples for any of the candidate genes (Table 3 and 4). Note that none of the nsSNPs of VAV3 and CD101 or the microsatellite of NRAMP1 showed evidence of association (Table 5). Allele A3 of the NRAMP1 microsatellite promoter (GT)n has previously shown linkage and association with autoimmune disease, and allele A2 with infectious disease susceptibility [18-20]. The relative risks of allele A3 and genotype A3/A3 in our type 1 diabetes samples was 0.96 (95% CI = 0.94 – 1.17) and 0.90 (95% CI = 0.70 – 1.16), respectively.
Table 3 Study design. Lengths of re-sequenced genomic regions, and number of tag SNPs or single SNPs genotyped in a pragmatic two-step genotyping design for NRAMP1, 4-1BB, FRAP1, CD101, B2M, and VAV3.
Locus Re-sequenced region (kb) n common SNPs* n tag SNPs Genotyping strategy (step 1 → step 2)
NRAMP1 12.13 12 4 Case-control → Family set 1+2
4-1BB 13.66 8 DIL4279/ss23142250 Family set 1
DIL4277/rs226476
DIL4569/rs226478
DIL4274/ss23142263
DIL4570/ss23142264
DIL4571/rs679563
DIL4273/ss23142265
DIL4272/ss23142270
FRAP1 30.88 21 6 Family set 1
CD101 15.9 18 8 Family set 1
B2M 9.33 10 8 Case-control → Family set 1
VAV3 27.69 19 (block 1)
18 (block 2)
15 (block 3) 7 (block 1)
11 (block 2)
10 (block 3) Family set 1
*For the selection of tag SNPs, minor allele frequencies of 0.03 were used for 4-1BB, CD101 and FRAP1, and 0.05 for NRAMP1, B2M and VAV3. Note that the numbers of attempted and actual genotypes are given in Additional file 8. n, number.
Table 4 Disease association results. Multi-locus test P values, lengths of re-sequenced genomic regions, and number of tag SNPs or single SNPs genotyped in a two-step genotyping design for NRAMP1, 4-1BB, FRAP1, CD101, B2M, and VAV3.
Locus Multilocus test P value/ Single-locus TDT P value Case-control Combined test P value
Family set 1 Family set 1 + 2
NRAMP1 - 0.56 0.20 0.68
4-1BB 0.71 - - -
0.88 - - -
0.52 - - -
0.35 - - -
0.53 - - -
0.29 - - -
0.95 - - -
0.24 - - -
FRAP1 0.44 - - -
CD101 0.68 - - -
B2M 0.90 - 0.11 0.75
VAV3 0.26 (block 1) - - -
0.80 (block 2) - - -
0.86 (block 3) - - -
Table 5 Association analysis of non-synonymous SNPs. SNPs with allele frequencies above 0.05 and the NRAMP1 (GT)n microsatellite in up to 1,476 families with at least one affected offspring. N, number; T, number of transmissions; NT, number of untransmitted alleles; %T, percentage transmission of minor allele from heterozygous parents to type 1 diabetes offspring (obtained by transmission/disequilibrium test (TDT)); GTRR, genotype relative risk; P, probability value (two-sided).
Locus Marker ID Amino acid change/ alleles Minor allele frequency N families T NT %T PTDT PGTRR
VAV3 DIL1521 Thr293Ser/T>A 0.27 1 476 834 840 49.64 0.77 0.90
DIL3809 Pro611Ser/G>A 0.13 1 476 417 429 49.29 0.68 0.84
CD101 DIL3794 Asn225Ser/A>G 0.32 652 517 515 49.9 0.95 0.96
NRAMP1 (GT)n - - 1 476 - - - - 0.36
With regards to our association study in humans, intronic and potential regulatory regions were not sequenced in the candidate genes since these cover large genomic regions, which will have to wait for much more extensive polymorphism maps [21]. For example, for VAV3, which spans almost 400 kb, less than 10% of the genomic region of VAV3 was re-sequenced to identify SNPs. The general importance of intronic and intergenic regulatory sequences as candidates for disease susceptibility is well recognised. Hence, potential unidentified causal variants in introns or flanking regions of the genes may have been missed, and remain a target for future analyses. Despite finding no evidence of association, it remains possible that there exists a common disease variant in one or more of the six candidate genes tested, which either has an effect smaller than would be detected with this study or is in much weaker LD with the tag SNPs than any other SNP known to us [22].
Finally, the possibility of one or more rare disease variants in a locus needs to be considered [23]. The best candidates for rare disease variants in the six genes studied here were thus genotyped in an expanded case-control collection of up to 3,704 type 1 diabetes cases and 3,930 controls: DIL5202/ss23142243 causes a non-conservative change in NRAMP (Asp543Asn, MAF = 0.02) and DIL3799/ss23142349 in CD101 (Val839Ile; MAF = 0.03). For both SNPs, P values above 0.05 were obtained (P = 0.19 for DIL5202/ss23142243 and P = 0.80 for DIL3799/ss23142349), therefore, making it less likely that these rare variants contribute to susceptibility to type 1 diabetes. Nevertheless, causal variants with MAFs less than 0.01 [24] may well remain undetected in our re-sequencing panels of 32 or 96 case DNAs. However, the re-sequencing of several hundred cases and controls is beyond the scope of the present study in which we have investigated variants with MAF ≥ 0.03.
Conclusion
Taken together, these data make an association between type 1 diabetes and common variation in coding and untranslated regions of the six functional candidate genes in the investigated human-mouse orthologue regions less likely. Several possibilities may account for this. A gene (or several genes) in an Idd interval may account for disease susceptibility in the NOD mouse, but the human orthologous region may lack this susceptibility variant. The scenario, in which candidate genes in the NOD Idd interval may not necessarily be harbouring a functional, causal variant in their human orthologue genes, was discussed previously [25]. It is also possible that the selected candidate gene in the Idd interval may not be the gene causing susceptibility to disease.
The tag SNP maps described here will be useful for association studies of other diseases. They will be integrated into future SNP maps encompassing the entire orthologous regions and all regulatory sequences and genes encoded within them.
Methods
Subjects
All family members were white and of European ancestral origin. The type 1 diabetes families comprised two parents and a least one affected child. The 748 type 1 diabetes families used in 'step 1' were as described previously [26]: 472 UK Warren 1 multiplex and 276 multiplex Human Biological Data Interchange families ascertained in the U.S.A. The case-control DNA set for the tag SNP approach consisted of 1,709 Caucasian type 1 diabetes cases, which were recruited from across Britain in the Juvenile Diabetes Research Foundation/Wellcome Trust funded UK Genetic Resource Investigating Diabetes (GRID) study [27], and 1,829 population-based controls from the 1958 British Birth Cohort (BBC) [28]. The mean age-at-onset of the cases, with almost all under 16 years of age at diagnosis, is 7.5 years (with a standard deviation of 4 years). The 1958 BBC controls are part of an ongoing longitudinal study and the subjects are British citizens born in a particular week in 1958. In order to test association for type 1 diabetes susceptibility and the rare variants in CD101 and NRAMP1, DIL3799/ss23142349 and DIL5202/ss23142243, a total of 3,704 type 1 diabetes cases and 3,930 controls were used.
For 'step 2' genotyping of NRAMP1, the 748 type 1 diabetes families described above were used in addition to 343 multiplex/simplex families from the UK, 159 Norwegian simplex families, 322 Romanian simplex families, and 60 multiplex families from the USA totalling the combined DNA sets to 1,632 type 1 diabetes families, as described previously [26].
Sequencing
Nested PCR products from DNA from 96 or 32 type 1 diabetes patients were sequenced using an Applied Biosystems (ABI) 3700 capillary sequencer (Foster City, CA), and SNPs identified using the Staden Package [29].
Genotyping
SNPs were genotyped using the Invader® assay (Third Wave Technologies, Inc. Madison WI) [30] and TaqMan MGB chemistry (ABI) [31]. The NRAMP1 microsatellite was genotyped on an ABI3700 sequencer using fluorescent primers as previously described [32]. Full details of primers and probes used for genotyping are available upon request. All genotyping data was double-scored independently.
Annotation
Annotation of NRAMP1 (European Molecular Biology Laboratory [EMBL] accession numbers D50402, D50403, BC041787, L32185, BC033754), FRAP1 (UO88966), 4-1BB (UO3387), CD101 (Z33642), B2M (BC032589) and VAV3 (AF118887, VAV3; AF118886, VAV3β; AF118887, VAV3.1) was performed by importing Ensembl information into a temporary ACeDB database as described in Burren et al. [33]. After confirmation of gene structures by BLAST analysis, these were re-extracted in GFF format and submitted to a local Gbrowse database (National Center for Biotechnology Information build 34) (DIL annotations viewable at T1DBase [34].
Statistical analysis
The program for the selection of tag SNPs [17] and association analysis used here are implemented in the Stata statistical system and may be downloaded from our website [35]. All genotyping data were in Hardy-Weinberg equilibrium (P > 0.05).
Authors' contributions
LMM and DJS contributed equally to this work by performing the genetic studies and writing the manuscript. AV, FP, RP, CL, JH, HF, CM, KMH, GC carried out the genetic studies and collated data. JDC performed the statistical analysis and participated in the design of the study. LJS participated in the sequence analysis. NW participated in design and collated data. KSR, CG, C I-T, DAS, DPS and LBP participated in the study design and coordination. JAT, LSW and RCT helped to draft the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 2
SNPs, including two deletion/insertion polymorphisms, identified in NRAMP1. Novel SNPs are denoted by "ss" numbers and previously published SNPs are denoted by "rs" numbers. Minor allele frequencies are based on the sequencing panel of 32 type 1 diabetes subjects. R2 values for non-typed SNPs. UTR, untranslated region.
Click here for file
Additional File 5
SNPs identified in CD101. Novel SNPs are denoted by "ss" numbers and previously published SNPs are denoted by "rs" numbers. Minor allele frequencies are based on the sequencing panel of 96 type 1 diabetes subjects. R2 values for non-typed SNPs. Note that DIL3969 has an allelic R2 < 0.80 due to technical difficulties with the assay. UTR, untranslated region.
Click here for file
Additional File 7
SNPs, including four in/dels identified in VAV3. Novel SNPs are denoted by "ss" numbers and previously published SNPs are denoted by "rs" numbers. Minor allele frequencies are based on the sequencing panel of 96 type 1 diabetes subjects. R2 values for non-typed SNPs. Note that DIL6496 and DIL6488 in block 1, and DIL1526 have allelic R2 values < 0.80, which was due to technical difficulties with those assays. UTR, untranslated region.
Click here for file
Additional File 1
Supporting published references for Table 1. NOD mouse Idd loci, the location of their human orthologous regions, and selected functional candidate gene within the Idd interval. *Estimated number of genes.
Click here for file
Additional File 3
SNPs identified in 4-1BB. Minor allele frequencies are based on the sequencing panel of 96 type 1 diabetes subjects. Novel SNPs are denoted by "ss" numbers and previously published SNPs are denoted by "rs" numbers. Note that DIL4247/rs6694557 could not be genotyped due to assay technical difficulties. UTR, untranslated region.
Click here for file
Additional File 4
SNPs identified in FRAP1. Novel SNPs are denoted by "ss" numbers and previously published SNPs are denoted by "rs" numbers. Minor allele frequencies are based on the sequencing panel of 32 type 1 diabetes subjects. R2 values for non-typed SNPs. UTR, untranslated region.
Click here for file
Additional File 6
SNPs identified in B2M. Novel SNPs are denoted by "ss" numbers and previously published SNPs are denoted by "rs" numbers. Minor allele frequencies are based on the sequencing panel of 32 type 1 diabetes subjects. R2 values for non-typed SNPs. UTR, untranslated region.
Click here for file
Additional File 8
Genotyping counts Numbers of attempted subjects for genotyping and of subjects with genotypes.
Click here for file
Acknowledgements
This work was funded by the Wellcome Trust and the Juvenile Diabetes Research Foundation International. L.M.M. was the recipient of a Wellcome Trust Prize Studentship. A.V. was a Mayo Foundation Scholar. We thank the Human Biological Data Interchange and Diabetes U.K. for USA and U.K. multiplex families, respectively and the Norwegian Study Group for Childhood Diabetes for the collection of Norwegian families. We acknowledge use of DNA from the 1958 British Birth Cohort collection, funded by the Medical Research Council grant G0000934 and Wellcome Trust grant 068545/Z/02. DNA samples were prepared by Jayne Hutchings, Gillian Coleman, Trupti Mistry, Kirsi Bourget, Sally Clayton, Matthew Hardy, Jennifer Keylock, Pamela Lauder, Meeta Maisuria, William Meadows, Meera Sebastian, Sarah Wood, The Avon Longitudinal Study of Parents and Children laboratory in Bristol, including Susan Ring, Wendy McArdle, Richard Jones, for preparing DNA samples.
==== Refs
Todd JA Bell JI McDevitt HO HLA-DQ beta gene contributes to susceptibility and resistance to insulin-dependent diabetes mellitus Nature 1987 329 599 604 3309680 10.1038/329599a0
Lyons PA Armitage N Argentina F Denny P Hill NJ Lord CJ Wilusz MB Peterson LB Wicker LS Todd JA Congenic mapping of the type 1 diabetes locus, Idd3, to a 780-kb region of mouse chromosome 3: identification of a candidate segment of ancestral DNA by haplotype mapping Genome Res 2000 10 446 453 10779485 10.1101/gr.10.4.446
Wicker LS Todd JA Peterson LB Genetic control of autoimmune diabetes in the NOD mouse Annu Rev Immunol 1995 13 179 200 7612220 10.1146/annurev.iy.13.040195.001143
Serreze DV Leiter EH Genes and cellular requirements for autoimmune diabetes susceptibility in nonobese diabetic mice Curr Dir Autoimmun 2001 4 31 67 11569409
Colle E Guttmann RD Seemayer T Spontaneous diabetes mellitus syndrome in the rat. I. Association with the major histocompatibility complex J Exp Med 1981 154 1237 1242 7026724 10.1084/jem.154.4.1237
Jacob HJ Pettersson A Wilson D Mao Y Lernmark A Lander ES Genetic dissection of autoimmune type I diabetes in the BB rat Nat Genet 1992 2 56 60 1303251 10.1038/ng0992-56
MacMurray AJ Moralejo DH Kwitek AE Rutledge EA Van Yserloo B Gohlke P Speros SJ Snyder B Schaefer J Bieg S Jiang J Ettinger RA Fuller J Daniels TL Pettersson A Orlebeke K Birren B Jacob HJ Lander ES Lernmark A Lymphopenia in the BB rat model of type 1 diabetes is due to a mutation in a novel immune-associated nucleotide (Ian)-related gene Genome Res 2002 12 1029 1039 12097339 10.1101/gr.412702
Yokoi N Komeda K Wang HY Yano H Kitada K Saitoh Y Seino Y Yasuda K Serikawa T Seino S Cblb is a major susceptibility gene for rat type 1 diabetes mellitus Nat Genet 2002 31 391 394 12118252
Ueda H Howson JM Esposito L Heward J Snook H Chamberlain G Rainbow DB Hunter KM Smith AN Di Genova G Herr MH Dahlman I Payne F Smyth D Lowe C Twells RC Howlett S Healy B Nutland S Rance HE Everett V Smink LJ Lam AC Cordell HJ Walker NM Bordin C Hulme J Motzo C Cucca F Hess JF Metzker ML Rogers J Gregory S Allahabadia A Nithiyananthan R Tuomilehto-Wolf E Tuomilehto J Bingley P Gillespie KM Undlien DE Ronningen KS Guja C Ionescu-Tirgoviste C Savage DA Maxwell AP Carson DJ Patterson CC Franklyn JA Clayton DG Peterson LB Wicker LS Todd JA Gough SC Association of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune disease Nature 2003 423 506 511 12724780 10.1038/nature01621
Penha-Goncalves C Moule C Smink LJ Howson J Gregory S Rogers J Lyons PA Suttie JJ Lord CJ Peterson LB Todd JA Wicker LS Identification of a structurally distinct CD101 molecule encoded in the 950-kb Idd10 region of NOD mice Diabetes 2003 52 1551 1556 12765969
Wicker LS Chamberlain G Hunter K Rainbow D Howlett S Tiffen P Clark J Gonzalez-Munoz A Cumiskey AM Rosa RL Howson JM Smink LJ Kingsnorth A Lyons PA Gregory S Rogers J Todd JA Peterson LB Fine mapping, gene content, comparative sequencing, and expression analyses support Ctla4 and Nramp1 as candidates for Idd5.1 and Idd5.2 in the nonobese diabetic mouse J Immunol 2004 173 164 173 15210771
Lyons PA Hancock WW Denny P Lord CJ Hill NJ Armitage N Siegmund T Todd JA Phillips MS Hess JF Chen SL Fischer PA Peterson LB Wicker LS The NOD Idd9 genetic interval influences the pathogenicity of insulitis and contains molecular variants of Cd30, Tnfr2, and Cd137 Immunity 2000 13 107 115 10933399 10.1016/S1074-7613(00)00012-1
Serreze DV Bridgett M Chapman HD Chen E Richard SD Leiter EH Subcongenic analysis of the Idd13 locus in NOD/Lt mice: evidence for several susceptibility genes including a possible diabetogenic role for beta 2-microglobulin J Immunol 1998 160 1472 1478 9570569
Lyons PA Armitage N Lord CJ Phillips MS Todd JA Peterson LB Wicker LS Mapping by genetic interaction: high-resolution congenic mapping of the type 1 diabetes loci Idd10 and Idd18 in the NOD mouse Diabetes 2001 50 2633 2637 11679445
Hamilton-Williams EE Serreze DV Charlton B Johnson EA Marron MP Mullbacher A Slattery RM Transgenic rescue implicates beta2-microglobulin as a diabetes susceptibility gene in nonobese diabetic (NOD) mice Proc Natl Acad Sci U S A 2001 98 11533 11538 11572996 10.1073/pnas.191383798
Johnson GC Esposito L Barratt BJ Smith AN Heward J Di Genova G Ueda H Cordell HJ Eaves IA Dudbridge F Twells RC Payne F Hughes W Nutland S Stevens H Carr P Tuomilehto-Wolf E Tuomilehto J Gough SC Clayton DG Todd JA Haplotype tagging for the identification of common disease genes Nat Genet 2001 29 233 237 11586306 10.1038/ng1001-233
Chapman JM Cooper JD Todd JA Clayton DG Detecting disease associations due to linkage disequilibrium using haplotype tags: a class of tests and the determinants of statistical power Hum Hered 2003 56 18 31 14614235 10.1159/000073729
Searle S Blackwell JM Evidence for a functional repeat polymorphism in the promoter of the human NRAMP1 gene that correlates with autoimmune versus infectious disease susceptibility J Med Genet 1999 36 295 299 10227396
Sanjeevi CB Miller EN Dabadghao P Rumba I Shtauvere A Denisova A Clayton D Blackwell JM Polymorphism at NRAMP1 and D2S1471 loci associated with juvenile rheumatoid arthritis Arthritis Rheum 2000 43 1397 1404 10857800 10.1002/1529-0131(200006)43:6<1397::AID-ANR25>3.0.CO;2-6
Esposito L Hill NJ Pritchard LE Cucca F Muxworthy C Merriman ME Wilson A Julier C Delepine M Tuomilehto J Tuomilehto-Wolf E Ionesco-Tirgoviste C Nistico L Buzzetti R Pozzilli P Ferrari M Bosi E Pociot F Nerup J Bain SC Todd JA Genetic analysis of chromosome 2 in type 1 diabetes: analysis of putative loci IDDM7, IDDM12, and IDDM13 and candidate genes NRAMP1 and IA-2 and the interleukin-1 gene cluster. IMDIAB Group Diabetes 1998 47 1797 1799 9792551
Consortium TIH The International HapMap Project Nature 2003 426 789 796 14685227 10.1038/nature02168
Lowe CE Cooper JD Chapman JM Barratt BJ Twells RC Green EA Savage DA Guja C Ionescu-Tirgoviste C Tuomilehto-Wolf E Tuomilehto J Todd JA Clayton DG Cost-effective analysis of candidate genes using htSNPs: a staged approach Genes Immun 2004 5 301 305 15029236 10.1038/sj.gene.6364064
Cox NJ Wapelhorst B Morrison VA Johnson L Pinchuk L Spielman RS Todd JA Concannon P Seven regions of the genome show evidence of linkage to type 1 diabetes in a consensus analysis of 767 multiplex families Am J Hum Genet 2001 69 820 830 11507694 10.1086/323501
Cohen JC Kiss RS Pertsemlidis A Marcel YL McPherson R Hobbs HH Multiple rare alleles contribute to low plasma levels of HDL cholesterol Science 2004 305 869 872 15297675 10.1126/science.1099870
Risch N Ghosh S Todd JA Statistical evaluation of multiple-locus linkage data in experimental species and its relevance to human studies: application to nonobese diabetic (NOD) mouse and human insulin-dependent diabetes mellitus (IDDM) Am J Hum Genet 1993 53 702 714 8352278
Vella A Howson JM Barratt BJ Twells RC Rance HE Nutland S Tuomilehto-Wolf E Tuomilehto J Undlien DE Ronningen KS Guja C Ionescu-Tirgoviste C Savage DA Todd JA Lack of association of the Ala(45)Thr polymorphism and other common variants of the NeuroD gene with type 1 diabetes Diabetes 2004 53 1158 1161 15047635
UK Genetic Resource Investigating Diabetes (GRID) study
1958 British Birth Cohort
Staden Package
Olivier M Chuang LM Chang MS Chen YT Pei D Ranade K de Witte A Allen J Tran N Curb D Pratt R Neefs H de Arruda Indig M Law S Neri B Wang L Cox DR High-throughput genotyping of single nucleotide polymorphisms using new biplex invader technology Nucleic Acids Res 2002 30 e53 12060691 10.1093/nar/gnf052
Ranade K Chang MS Ting CT Pei D Hsiao CF Olivier M Pesich R Hebert J Chen YD Dzau VJ Curb D Olshen R Risch N Cox DR Botstein D High-throughput genotyping with single nucleotide polymorphisms Genome Res 2001 11 1262 1268 11435409
Graham AM Dollinger MM Howie SE Harrison DJ Identification of novel alleles at a polymorphic microsatellite repeat region in the human NRAMP1 gene promoter: analysis of allele frequencies in primary biliary cirrhosis J Med Genet 2000 37 150 152 10712108 10.1136/jmg.37.2.150
Burren OS Healy BC Lam AC Schuilenburg H Dolman GE Everett VH Laneri D Nutland S Rance HE Payne F Smyth D Lowe C Barratt BJ Twells RC Rainbow DB Wicker LS Todd JA Walker NM Smink LJ Development of an integrated genome informatics, data management and workflow infrastructure: a toolbox for the study of complex disease genetics Hum Genomics 2004 1 98 109 15601538
Smink LJ Helton EM Healy BC Cavnor CC Lam AC Flamez D Burren OS Wang Y Dolman GE Burdick DB Everett VH Glusman G Laneri D Rowen L Schuilenburg H Walker NM Mychaleckyj J Wicker LS Eizirik DL Todd JA Goodman N T1DBase, a community web-based resource for type 1 diabetes research Nucleic Acids Res 2005 33 D544 549 15608258 10.1093/nar/gki095
Juvenile Diabetes Research Foundation/Wellcome Trust Diabetes and Inflammation Laboratory
| 15720714 | PMC551616 | CC BY | 2021-01-04 16:38:18 | no | BMC Genet. 2005 Feb 18; 6:9 | utf-8 | BMC Genet | 2,005 | 10.1186/1471-2156-6-9 | oa_comm |
==== Front
J Neuroengineering RehabilJournal of NeuroEngineering and Rehabilitation1743-0003BioMed Central London 1743-0003-2-61574062110.1186/1743-0003-2-6ResearchA wireless body area network of intelligent motion sensors for computer assisted physical rehabilitation Jovanov Emil [email protected] Aleksandar [email protected] Chris [email protected] Groen Piet C [email protected] Electrical and Computer Engineering Department, University of Alabama in Huntsville, Huntsville, Alabama 35899, USA2 Division of Biomedical Informatics, Mayo Clinic College of Medicine, Rochester, Minnesota 55905, USA2005 1 3 2005 2 6 6 28 1 2005 1 3 2005 Copyright © 2005 Jovanov et al; licensee BioMed Central Ltd.2005Jovanov 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 technological advances in integrated circuits, wireless communications, and physiological sensing allow miniature, lightweight, ultra-low power, intelligent monitoring devices. A number of these devices can be integrated into a Wireless Body Area Network (WBAN), a new enabling technology for health monitoring.
Methods
Using off-the-shelf wireless sensors we designed a prototype WBAN which features a standard ZigBee compliant radio and a common set of physiological, kinetic, and environmental sensors.
Results
We introduce a multi-tier telemedicine system and describe how we optimized our prototype WBAN implementation for computer-assisted physical rehabilitation applications and ambulatory monitoring. The system performs real-time analysis of sensors' data, provides guidance and feedback to the user, and can generate warnings based on the user's state, level of activity, and environmental conditions. In addition, all recorded information can be transferred to medical servers via the Internet and seamlessly integrated into the user's electronic medical record and research databases.
Conclusion
WBANs promise inexpensive, unobtrusive, and unsupervised ambulatory monitoring during normal daily activities for prolonged periods of time. To make this technology ubiquitous and affordable, a number of challenging issues should be resolved, such as system design, configuration and customization, seamless integration, standardization, further utilization of common off-the-shelf components, security and privacy, and social issues.
==== Body
Introduction
Wearable health monitoring systems integrated into a telemedicine system are novel information technology that will be able to support early detection of abnormal conditions and prevention of its serious consequences [1,2]. Many patients can benefit from continuous monitoring as a part of a diagnostic procedure, optimal maintenance of a chronic condition or during supervised recovery from an acute event or surgical procedure.
Important limitations for wider acceptance of the existing systems for continuous monitoring are: a) unwieldy wires between sensors and a processing unit, b) lack of system integration of individual sensors, c) interference on a wireless communication channel shared by multiple devices, and d) nonexistent support for massive data collection and knowledge discovery. Traditionally, personal medical monitoring systems, such as Holter monitors, have been used only to collect data for off-line processing. Systems with multiple sensors for physical rehabilitation feature unwieldy wires between electrodes and the monitoring system. These wires may limit the patient's activity and level of comfort and thus negatively influence the measured results. A wearable health-monitoring device using a Personal Area Network (PAN) or Body Area Network (BAN) can be integrated into a user's clothing [3]. This system organization, however, is unsuitable for lengthy, continuous monitoring, particularly during normal activity [4], intensive training or computer-assisted rehabilitation [5]. Recent technology advances in wireless networking [6], micro-fabrication [7], and integration of physical sensors, embedded microcontrollers and radio interfaces on a single chip [8], promise a new generation of wireless sensors suitable for many applications [9]. However, the existing telemetric devices either use wireless communication channels exclusively to transfer raw data from sensors to the monitoring station, or use standard high-level wireless protocols such as Bluetooth that are too complex, power demanding, and prone to interference by other devices operating in the same frequency range. These characteristics limit their use for prolonged wearable monitoring. Simple, accurate means of monitoring daily activities outside of the laboratory are not available [12]; at the present, only estimates can be obtained from questionnaires, measures of heart rate, video assessment, and use of pedometers [13] or accelerometers [14]. Finally, records from individual monitoring sessions are rarely integrated into research databases that would provide support for data mining and knowledge discovery relevant to specific conditions and patient categories.
Increased system processing power allows sophisticated real-time data processing within the confines of the wearable system. As a result, such wearable system can support biofeedback and generation of warnings. The use of biofeedback techniques has gained increased attention among researchers in the field of physical medicine and tele-rehabilitation [5]. Intensive practice schedules have been shown to be important for recovery of motor function [22]. Unfortunately, an aggressive approach to rehabilitation involving extensive therapist-supervised motor training is not a realistic expectation in today's health care system where individuals are typically seen as outpatients about twice a week for no longer than 30–45 min. Wearable technology and biofeedback systems appear to be a valid alternative, as they reduce the extensive time to set-up a patient before each session and require limited time involvement of physicians and therapists. Furthermore, wearable technology could potentially address a second factor that hinders enthusiasm for rehabilitation, namely the fact that setting up a patient for the procedure is rather time-consuming. This is because tethered sensors need to be positioned on the subject, attached to the equipment, and a software application needs to be started before each session. Wearable technology allows sensors that will be positioned on the subject for prolonged periods, therefore eliminating the need to position them for every training session. Instead, a personal server such as a PDA can almost instantly initiate a new training session whenever the subject is ready and willing to exercise. In addition to home rehabilitation, this setting also may be beneficial in the clinical setting, where precious time of physicians and therapists could be saved. Moreover, the system can issue timely warnings or alarms to the patient, or to a specialized medical response service in the event of significant deviations of the norm or medical emergencies. However, as for all systems, regular, routine maintenance (verifying configuration and thresholds) by a specialist is required.
Typical examples of possible applications include stroke rehabilitation, physical rehabilitation after hip or knee surgeries, myocardial infarction rehabilitation, and traumatic brain injury rehabilitation. The assessment of the effectiveness of rehabilitation procedures has been limited to the laboratory setting; relatively little is known about rehabilitation in real-life situations. Miniature, wireless, wearable technology offers a tremendous opportunity to address this issue.
We propose a wireless BAN composed of off-the-shelf sensor platforms with application-specific signal conditioning modules [10]. In this paper, we present a general system architecture and describe a recently developed activity sensor "ActiS". ActiS is based on a standard wireless sensor platform and a custom sensor board with a one-channel bio amplifier and two accelerometers [11]. As a heart sensor, ActiS can be used to monitor heart activity and position of the upper trunk. The same sensor can be used to monitor position and activity of upper and lower extremities. A wearable system with ActiS sensors would also allow one to assess metabolic rate and cumulative energy expenditure as a valuable parameter in the management of many medical conditions. An early version of the ActiS has been based on a custom developed wireless intelligent sensor and custom wireless protocols in the license-free 900 MHz Scientific and Medical Instruments (ISM) band [15]. Our initial experience indicated the importance of standard sensor platforms with ample processing power, minute power consumption, and standard software support. Such platforms were not available on the market during the design of our first prototype system. The recent introduction of an IEEE standard for low-power personal area networks (802.15.4) and ZigBee protocol stack [16], as well as new ZigBee compliant Telos sensor platform [17], motivated the development of the new system presented in this paper. TinyOS support for the selected sensor platform facilitates rapid application development [18]. Standard hardware and software architecture facilitate interoperable systems and devices that are expected to significantly influence next generation health systems [19]. This trend can also be observed in recently developed physiological monitors systems from Harvard [20] and Welch-Allen [21].
System Architecture
Continuous technological advances in integrated circuits, wireless communication, and sensors enable development of miniature, non-invasive physiological sensors that communicate wirelessly with a personal server and subsequently through the Internet with a remote emergency, weather forecast or medical database server; using baseline (medical database), sensor (WBAN) and environmental (emergency or weather forecast) information, algorithms may result in patient-specific recommendations. The personal server, running on a PDA or a 3 G cell phone, provides the human-computer interface and communicates with the remote server(s). Figure 1 shows a generalized overview of a multi-tier system architecture; the lowest level encompasses a set of intelligent physiological sensors; the second level is the personal server (Internet enabled PDA, cell-phone, or home computer); and the third level encompasses a network of remote health care servers and related services (Caregiver, Physician, Clinic, Emergency, Weather). Each level represents a fairly complex subsystem with a local hierarchy employed to ensure efficiency, portability, security, and reduced cost. Figure 2 illustrates an example of information flow in an integrated WBAN system.
Figure 1 Wireless Body Area Network of Intelligent Sensors for Patient Monitoring
Figure 2 Data flow in an integrated WWBAN
Sensor level
A WBAN can include a number of physiological sensors depending on the end-user application. Information of several sensors can be combined to generate new information such as total energy expenditure. An extensive set of physiological sensors may include the following:
• an ECG (electrocardiogram) sensor for monitoring heart activity
• an EMG (electromyography) sensor for monitoring muscle activity
• an EEG (electroencephalography) sensor for monitoring brain electrical activity
• a blood pressure sensor
• a tilt sensor for monitoring trunk position
• a breathing sensor for monitoring respiration
• movement sensors used to estimate user's activity
• a "smart sock" sensor or a sensor equipped shoe insole used to delineate phases of individual steps
These physiological sensors typically generate analog signals that are interfaced to standard wireless network platforms that provide computational, storage, and communication capabilities. Multiple physiological sensors can share a single wireless network node. In addition, physiological sensors can be interfaced with an intelligent sensor board that provides on-sensor processing capability and communicates with a standard wireless network platform through serial interfaces.
The wireless sensor nodes should satisfy the following requirements: minimal weight, miniature form-factor, low-power operation to permit prolonged continuous monitoring, seamless integration into a WBAN, standard-based interface protocols, and patient-specific calibration, tuning, and customization. These requirements represent a challenging task, but we believe a crucial one if we want to move beyond 'stovepipe' systems in healthcare where one vendor creates all components. Only hybrid systems implemented by combining off-the-shelf, commodity hardware and software components, manufactured by different vendors promise proliferation and dramatic cost reduction.
The wireless network nodes can be implemented as tiny patches or incorporated into clothes or shoes. The network nodes continuously collect and process raw information, store them locally, and send them to the personal server. Type and nature of a healthcare application will determine the frequency of relevant events (sampling, processing, storing, and communicating). Ideally, sensors periodically transmit their status and events, therefore significantly reducing power consumption and extending battery life. When local analysis of data is inconclusive or indicates an emergency situation, the upper level in the hierarchy can issue a request to transfer raw signals to the upper levels where advanced processing and storage is available.
Personal server level
The personal server performs the following tasks:
• Initialization, configuration, and synchronization of WBAN nodes
• Control and monitor operation of WBAN nodes
• Collection of sensor readings from physiological sensors
• Processing and integration of data from various physiological sensors providing better insight into the users state
• Providing an audio and graphical user-interface that can be used to relay early warnings or guidance (e.g., during rehabilitation)
• Secure communication with remote healthcare provider servers in the upper level using Internet services
The personal server can be implemented on an off-the-shelf Internet-enabled PDA (Personal Digital Assistant) or 3 G cell phone, or on a home personal computer. Multiple configurations are possible depending on the type of wireless network employed. For example, the personal server can communicate with individual WBAN nodes using the Zigbee wireless protocol that provides low-power network operation and supports virtually an unlimited number of network nodes. A network coordinator, attached to the personal server, can perform some of the pre-processing and synchronization tasks. Other communication scenarios are also possible. For example, the personal server running on a Bluetooth or WLAN enabled PDA can communicate with remote upper-level services through a home computer; the computer then serves as a gateway (Figure 1).
Relying on off-the-shelf mobile computing platforms is crucial, as these platforms will continue to grow in their capabilities and quality of services. The challenging tasks are to develop robust applications that provide simple and intuitive services (WBAN setup, data fusion, questionnaires describing detailed symptoms, activities, secure and reliable communication with remote medical servers, etc). Total information integration will allow patients to receive directions from their healthcare providers based on their current conditions.
Medical services
We envision various medical services in the top level of the tiered hierarchy. A healthcare provider runs a service that automatically collects data from individual patients, integrates the data into a patient's medical record, processes them, and issues recommendations, if necessary. These recommendations are also documented in the electronic medical record. If the received data are out of range or indicate an imminent medical condition, an emergency service can be notified (this can also be done locally at the personal server level). The exact location of the patient can be determined based on the Internet access entry point or directly if the personal server is equipped with a GPS sensor. Medical professionals can monitor the activity of the patient and issue altered guidance based on the new information, other prior known and relevant patient data, and the patient's environment (e.g., location and weather conditions).
The large amount of data collected through such services will allow quantitative analysis of various conditions and patterns. For example, suggested targets for stride and forces of hip replacement patients could be suggested according to the previous history, external temperature, time of the day, gender, and current physiological parameters (e.g., heart rate). Moreover, the results could be stored in research databases that will allow researchers to quantify the contribution of each parameter to a given condition if adequate numbers of patients are studied in this manner. Again, it is important to emphasize that the proposed approach requires seamless integration of large amounts of data into a research database in order to be able to perform meaningful statistical analyses.
ActiS – Activity Sensor
The ActiS sensor was developed specifically for WBAN-based, wearable computer-assisted, rehabilitation applications. With this concept in mind, we integrated a one-channel bio-amplifier and three accelerometer channels with a low power microcontroller into an intelligent signal processing board that can be used as an extension of a standard wireless sensor platform. ActiS consists of a standard sensor platform, Telos, from Moteiv and a custom Intelligent Signal Processing Module – ISPM (Figure 3). A block diagram of the sensor node is shown in Figure 4.
Figure 3 Telos wireless platform with intelligent signal processing daughtercard ISPM
Figure 4 Block diagram of the activity sensor (Telos platform and ISPM module)
The Telos platform is an ideal fit for this application due to small footprint and open source system software support. A second generation of the Telos platform features an 8 MHz MSP430F1611 microcontroller with integrated 10 KB of RAM and 48 KB of flash memory, a USB (Universal Serial Bus) interface for programming and communication, and an integrated wireless ZigBee compliant radio with on-board antenna [11]. In addition, the Telos platform includes humidity, temperature, and light sensors that could be used as ambient sensors. The Telos platform features a 10-pin expansion connector that allows one UART (Universal Asynchronous Receiver Transmitter) and I2 C interface, two general-purpose I/O lines, and three analog input lines.
The ISPM extends the capabilities of Telos by adding two perpendicular dual axis accelerometers (Analog Devices ADXL202) and a bio-amplifier with a signal conditioning circuit. The ISPM has its own MSP430F1232 processor for sampling and low-level data processing. This microcontroller was selected primarily for its compact size and ultra low power operation. Other features that were desirable for this design were the 10-bit ADC and the timer capture/compare registers that are used for acquisition of data from the accelerometers. The F1232 has hardware UART that is used for communications with Telos.
The ISPM's two ADXL202 accelerometers cover all three axes of motion. One ADXL202 is mounted directly on the ISPM board and collects data for the X and Y axes in the same plane. The second ADXL202 is mounted on a daughter card that extends vertically from the ISPM.
The user's physiological state is monitored using an on-board bio-amplifier implemented using an instrumentation amplifier with a signal conditioning circuit. The bio-amplifier could be used for electromyogram (EMG) or electrocardiogram (ECG) monitoring. The output of the signal conditioning circuit is connected to the local microcontroller as well as to the microcontroller on the Telos board via the expansion connector. The AD converter on the Telos board has a higher resolution (12 bit) than the F1232 on the ISPM (10 bit). This configuration gives flexibility of utilizing either microcontroller to process physiological signals.
An example application of the ActiS sensor as motion sensor on an ankle is given in Figure 5. This figure also visualizes the main components of acceleration during slow movements as projections of the gravity force (g) on the accelerometer's reference axes – Ax and Ay. Rotations of the sensor in the vertical plane (Θ) can be estimated as Θ = arctan(Ax / Ay). A compensation for non-ideal vertical placement can be achieved using the second accelerometer (not mounted in this photo) at 90-degree angle. Instead of calculating the angular position, many systems use off-the-shelf gyroscopes to measure angular velocity for the detection of gait phases [32]. A typical example of step detection is illustrated in Figure 6.
Figure 5 Activity sensor on an ankle with symbolic representation of acceleration components
Figure 6 Accelerometer based step detection using ankle sensors
Issues and Applications
WBAN systems can capitalize on recent technological advances that have enabled new methods for studying human activity and motion, making extended activity analysis more feasible. However, before WBAN becomes a widely accepted concept, a number of challenging system design and social issues should be resolved. If resolved successfully, WBAN systems will open a whole range of possible new applications that can significantly influence our lives.
System Design Issues
The development of pedometers and Micro-ElectroMechanical Systems (MEMS) accelerometers and gyroscopes show great promise in the design of wearable sensors. The main system design issues include:
• types of sensors
• power source
• size and weight of sensors
• wireless communication range and transmission characteristics of wearable sensors
• sensor location and mounting
• seamless system configuration
• automatic uploads to the patient's electronic medical record
• intuitive and simple user interface
Types of sensors
As for sensors, accelerometers and gyroscopes offer greater sensitivity and are more applicable for monitoring of motion since they generate continuous output. Bouten et al [27] found that frequency of human induced activity ranges from 1 to 18 Hz. Sampling rates in the existing projects vary from 10 – 100 Hz. Almost all projects in the last five years use MEMS accelerometers or a combination of accelerometers and gyroscopes [34,35]. As examples of full sets of sensors for research purposes, "MIThril" and Shoe Integrated Gait Sensor (SIGS) [26] systems feature 3 axes of gyroscopes, 3 axes of accelerometers, two piezoelectric sensors, two electric field sensors, two resistive band sensors, and four force sensitive resistors. These sensors can be mounted on the back of a shoe and in a shoe insole, respectively. Researchers at University of Washington School of Nursing have used off-the-shelf tri-axis accelerometer modules to study physical movement in COPD (Chronic Obstructive Pulmonary Disease) patients [2]. Both Lancaster University, UK, and ETH Zurich, Switzerland, have developed custom hardware realizing arrays of inertial sensor networks [24]. Lancaster used an array of 30 two-axis accelerometers. Similarly, ETH Zurich used a modular harness design [25].
The majority of foot-contact pedometers are designed to count steps only. Although they have been studied for use in complex energy estimation and have even shown a high degree of accuracy for walking / running activities [2] they are not well suited for rehabilitation.
Power source, size/weight, and transmission characteristics
To be unobtrusive, the sensors must be lightweight with small form factor. The size and weight of sensors is predominantly determined by the size and weight of batteries. Requirements for extended battery life directly oppose the requirement for small form factor and low weight. This implies that sensors have to be extremely power efficient, as frequent battery changes for multiple WBAN sensors would likely hamper users' acceptance and increase the cost. In addition, low power consumption is very important as we move toward future generations of implantable sensors that would ideally be self-powered, using energy extracted from the environment.
The radio communication poses the most significant energy consumption problem. Intelligent on-sensor signal processing has the potential to save power by transmitting the processed data rather than raw signals, and consequently to extend battery life. A careful trade-off between communication and computation is crucial for an optimal design. It appears that the most promising wireless standard for WBAN applications is ZigBee, as it represents an emerging wireless technology for the low-power, short-range, wireless sensors.
Location of Sensors
Although the purpose of the measurement does influence sensor location, researchers seem to disagree on the ideal body location for sensors. A motion sensor attached to an ankle is the most discriminative single position for state recognition, while a combination of hip and ankle sensors discriminates the states even more [25]. In a study of the relationship between metabolic energy expenditure and various activities, researchers at Eindhoven University of Technology, the Netherlands, placed tri-axial accelerometers on a subject's back waistline [27]. Krause et al use two accelerometers on the SenseWear armband [31]. Lee et al [2] placed accelerometer sensors in the subject's thigh pocket in order to measure angular position and velocity of the thigh. Doing so, they were able to accurately monitor a subject's activity and with the assistance of gyroscopes and compass headings were able to successfully estimate a subject's change in location. Some systems employ large arrays of wearable sensors. Laerhoven et al developed a loose fitting lab coat and trousers [24] consisting of 30 sensors; Kern et al [25]developed tighter fitting modular harnesses including a total of 48 sensors. Sensor attachment is also a critical factor, since the movement of loosely attached sensors creates spurious oscillations after an abrupt movement that can generate false events or mask real events.
Seamless system configuration
The intelligent WBAN sensors should allow users to easily assemble a robust ad-hoc WBAN, depending on the user's state of health. We can imagine standard off-the-shelf sensors, manufactured by different vendors, and sold "over-the-counter" [19]. Each sensor should be able to identify itself and declare its operational range and functionality. In addition, they should support easy customization for a given application.
Algorithms
Application-specific algorithms mostly use digital signal pre-processing combined with a variety of artificial intelligence techniques to model user's states and activity in each state. Digital signal processing include filters to resolve high and low frequency components of a signal, wavelet transform algorithms to correlate heel-strike and toe-off (steps) to angular velocity measured via gyroscopes [30], power spectrum analysis and a Gaussian model to classify activity types [26]. Artificial intelligence techniques may include fuzzy logic [28] and Kohonen self-organizing maps [31]. Some systems use physiological signals to improve context identification [31]. It has been shown that the activity-induced energy expenditure (AEE) is well correlated with the sum of integrals of the high frequency component of each individual axis [27].
Most of the algorithms in the open literature are not executed in real-time, or require powerful computing platforms such as laptops for real-time analysis.
Social Issues
Social issues of WBAN systems include privacy/security and legal issues. Due to communication of health-related information between sensors and servers, all communication over WBAN and Internet should be encrypted to protect user's privacy. Legal regulation will be necessary to regulate access to patient-identifiable information.
Possible applications
The WBAN technology can be used for computer-assisted physical rehabilitation in ambulatory settings and monitoring of trends during recovery. An integrated system can synergize the information from multiple sensors, warn the user in the case of emergencies, and provide feedback during supervised recovery or normal activity. Candidate applications include post-stroke rehabilitation, orthopaedic rehabilitation (e.g. hip/knee replacement rehabilitation), and supervised recovery of cardiac patients [36]. In the case of orthopaedic rehabilitation the system can measure forces and accelerations at different points and provide feedback to the user in real-time. Unobtrusive monitoring of cardiac patients can be used to estimate intensity of activities in user's daily routine and correlate it with the heart activity.
In addition, WBAN systems can be used for gait phase detection during programmable, functional electrical stimulation [33], analysis of balance and monitoring of Parkinson's disease patients in the ambulatory setting [32], computer supervision of health and activity status of elderly, weight loss therapy, obesity prevention, or in general promotion of a healthy, physically active, lifestyle.
Conclusion
A wearable Wireless Body Area Network (WBAN) of physiological sensors integrated into a telemedical system holds the promise to become a key infrastructure element in remotely supervised, home-based patient rehabilitation. It has the potential to provide a better and less expensive alternative for rehabilitation healthcare and may provide benefit to patients, physicians, and society through continuous monitoring in the ambulatory setting, early detection of abnormal conditions, supervised rehabilitation, and potential knowledge discovery through data mining of all gathered information.
Continuous monitoring with early detection likely has the potential to provide patients with an increased level of confidence, which in turn may improve quality of life. In addition, ambulatory monitoring will allow patients to engage in normal activities of daily life, rather than staying at home or close to specialized medical services. Last but not least, inclusion of continuous monitoring data into medical databases will allow integrated analysis of all data to optimize individualized care and provide knowledge discovery through integrated data mining. Indeed, with the current technological trend toward integration of processors and wireless interfaces, we will soon have coin-sized intelligent sensors. They will be applied as skin patches, seamlessly integrated into a personal monitoring system, and worn for extended periods of time.
==== Refs
Istepanian RSH Jovanov E Zhang YT Guest Editorial Introduction to the Special Section on M-Health: Beyond Seamless Mobility and Global Wireless Health-Care Connectivity IEEE Transactions on Information Technology in Biomedicine 2004 8 405 414 15615031
Wearable Technology Special Issue of the IEEE Engineering in Medicine and Biology Magazine 2003 22
Park S Jayaraman S Enhancing the Quality of Life Through Wearable Technology IEEE Engineering in Medicine and Biology Magazine 2003 22 41 48 12845818
Martin T Jovanov E Raskovic D Issues in Wearable Computing for Medical Monitoring Applications: A Case Study of a Wearable ECG Monitoring Device Proc of The International Symposium on Wearable Computers ISWC Atlanta, Georgia 2000 43 50
Winters JM Wang Y Winters JM Wearable Sensors and Telerehabilitation: Integrating Intelligent Telerehabilitation Assistants With a Model for Optimizing Home Therapy IEEE Engineering in Medicine and Biology Magazine 22 56 65
Otis BP Rabaey JM A 300-μW 1.9-GHz CMOS Oscillator Utilizing Micromachined Resonators IEEE Journal of Solid-State Circuits 2003 38 1271 1274
Ghovanloo M Najafi K A BiCMOS Wireless Stimulator Chip for Micromachined Stimulating Microprobes Proceedings of the Second Joint EMBS/BMES Conference 2002 2113 2114
Center for Wireless Integrated Microsystems (WIMS)
Raskovic D Martin T Jovanov E Medical Monitoring Applications for Wearable Computing The Computer Journal 2004 47 495 504
Jovanov E Price J Raskovic D Kavi K Martin T Adhami R Wireless Personal Area Networks in Telemedical Environment Proc 3rd International Conference on Information technology in Biomedicine ITAB-ITIS 2000 22 27
Otto C Gober JP McMurtrey RW Milenkoviæ A Jovanov E An Implementation of Hierarchical Signal Processing on Wireless Sensor in TinyOS Environment 43rd Annual ACM Southeast Conference ACMSE 2005
Steele BG Belza B Cain K Warms C Coppersmith J Howard J Bodies in motion: Monitoring daily activity and exercise with motion sensors in people with chronic pulmonary disease Journal of Rehabilitation Research & Development 2003 40 45 58 15074453
Digi-Walker step counter
Aminian K Robert P Buchser EE Rutschmann B Hayoz D Depairon M Physical activity monitoring based on accelerometry: validation and comparison with video observation Medical & Biological Engineering & Computing 1999 37 304 308 10505379
Milenkovic M Jovanov E Chapman J Raskovic D Price J An Accelerometer-Based Physical Rehabilitation System The 34th Southeastern Symposium on System Theory (SSST) 2002 57 60
ZigBee Alliance
Moteiv
TinyOS
Warren S Beyond Telemedicine: Infrastructures for Intelligent Home Care Technology Pre-ICADI Workshop on Technology for Aging, Disability, and Independence 2003 The Royal Academy of Engineering, Westminster, London
Malan D Fulford-Jones TRF Welsh M Moulton S CodeBlue: An Ad Hoc Sensor Network Infrastructure for Emergency Medical Care Proc of the MobiSys 2004 Workshop on Applications of Mobile Embedded Systems (WAMES 2004) 2004 12 14
Welch J Guilak F Baker SD A Wireless ECG Smart Sensor for Broad Application in Life Threatening Event Detection Proc of the 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2004 3447 3449
Taub E Uswatte G Pidikiti RD Constraint-induced (CI) movement therapy: a new family of techniques with broad application to physical rehabilitation-a clinical review J Rehabil Res Dev 1999 36 237 51 10659807
Tharion WJ Yokota M Buller MJ DeLany JP Hoyt RW Total Energy Expenditure Estimated Using a Foot-Contact Pedometer Medical Science Monitor 2004 10 504 509
Van Laerhoven K Kern N Gellersen HW Schiele B Towards A Wearable Inertial Sensor Network IEE EuroWearable 2003 (EuroWearable) 2003 Birmingham, UK
Kern N Schiele B Schmidt A Multi-Sensor Activity Context Detection for Wearable Computing European Symposium on Ambient Intelligence(EUSAI) 2003 Eindhoven, The Netherlands
Pentland S Healthwear: Medical Technology Becomes Wearable Computer 2004 37 34 41
Bouten CVC Koekkoek KTM Verduin M Kodde R Janssen JD A Triaxial Accelerometer and Portable Data Processing Unit for the Assessment of Daily Physical Activity IEEE Transactions On Biomedical Engineering 1997 44 136 147 9216127
Lee SW Mase K Activity and Location Recognition Using Wearable Sensors Pervasive Computing 2002 1 24 32
Morris SJ Paradiso JA Shoe-Integrated Sensor System For Wireless Gait Analysis And Real-Time Feedback Proc 2nd Joint EMBS/BMES Conference October 23–26, 2002
Aminian K Najafi B Büla C Leyvraz PF Robert P Ambulatory Gait Analysis Using Gyroscopes 25th Annual Meeting of the American Society of Biomechanics 2001 San Diego
Krause A Siewiorek DP Smailagic A Farringdon J Unsupervised, Dynamic Identification of Physiological and Activity Context in Wearable Computing Proc 7th International Symposium on Wearable Computers 2003 White Plains, NY 88 97
Melnick ME Radtka S Piper M Gait Analysis and Parkinson's Disease Rehab Management, The Interdisciplinary Journal of Rehabilitation 2002
Pappas IPI Keller T Mangold S Popovic MR Dietz V Morari M A Reliable Gyroscope-Based Gait-Phase Detection Sensor Embedded in a Shoe Insole IEEE Sensors Journal 2004 4 268 274
Analog Devices, MEMS and Sensors
Murata Piezoeletric Gyroscopes
CardioNet
| 15740621 | PMC552302 | CC BY | 2021-01-04 16:37:40 | no | J Neuroengineering Rehabil. 2005 Mar 1; 2:6 | utf-8 | J Neuroeng Rehabil | 2,005 | 10.1186/1743-0003-2-6 | oa_comm |
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-221572071710.1186/1471-2164-6-22Research ArticleUsing purine skews to predict genes in AT-rich poxviruses Da Silva Melissa [email protected] Chris [email protected] Department of Biochemistry and Microbiology, University of Victoria, Victoria, BC, V8W 3P6, Canada2005 18 2 2005 6 22 22 29 9 2004 18 2 2005 Copyright © 2005 Da Silva and Upton; licensee BioMed Central Ltd.2005Da Silva and Upton; 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
Clusters or runs of purines on the mRNA synonymous strand have been found in many different organisms including orthopoxviruses. The purine bias that is exhibited by these clusters can be observed using a purine skew and in the case of poxviruses, these skews can be used to help determine the coding strand of a particular segment of the genome. Combined with previous findings that minor ORFs have lower than average aspartate and glutamate composition and higher than average serine composition, purine content can be used to predict the likelihood of a poxvirus ORF being a "real gene".
Results
Using purine skews and a "quality" measure designed to incorporate previous findings about minor ORFs, we have found that in our training case (vaccinia virus strain Copenhagen), 59 of 65 minor (small and unlikely to be a real genes) ORFs were correctly classified as being minor. Of the 201 major (large and likely to be real genes) vaccinia ORFs, 192 were correctly classified as being major. Performing a similar analysis with the entomopoxvirus amsacta moorei (AMEV), it was found that 4 major ORFs were incorrectly classified as minor and 9 minor ORFs were incorrectly classified as major. The purine abundance observed for major ORFs in vaccinia virus was found to stem primarily from the first codon position with both the second and third codon positions containing roughly equal amounts of purines and pyrimidines.
Conclusion
Purine skews and a "quality" measure can be used to predict functional ORFs and purine skews in particular can be used to determine which of two overlapping ORFs is most likely to be the real gene if neither of the two ORFs has orthologs in other poxviruses.
==== Body
Background
In 1966, Szybalski first discovered that the mRNA synonymous strand of DNA contained a predominance of purine-rich clusters [1]; by convention, the top strand of a linear dsDNA molecule is viewed 5'→3', therefore when transcription of a gene is to the right, the top strand is considered the mRNA synonymous strand and if transcription is to the left, the top strand is the template strand. Chargaff's second parity rule states that for single-stranded DNA %A ≅ %T and %C ≅ %G [2,3] and implies that for regions with clusters of purines there must be local deviations from Chargaff's second parity rule favoring purines [4]. These local deviations from Chargaff's second parity rule also known as Chargaff differences have been seen in a variety of organisms including vaccinia virus; Bell et al. determined that Chargaff differences do correlate with direction of transcription and that the number of A nucleotides is greater than the number of T nucleotides in 83 of 92 vaccinia genes [4].
Many programs have been designed to predict genes, but few actually rate the "quality" or significance of the prediction and leave researchers to evaluate this themselves. In poxviruses, predicting which ORFs are likely to be expressed (genes) without the use of biochemical analysis usually involves simply choosing a minimum ORF length cut-off and excluding all ORFs that are smaller than the cut-off. Analysis may be extended to include manual inspection of each predicted ORF for the presence of promoter consensus sequences. Excluding ORFs that are smaller in size than the cut-off, however, risks missing genes that are unusually short; during annotation of vaccinia virus strain Copenhagen (VACV-COP) at least three recently verified genes (ranging from 162 bp – 231 bp) were not included in the initial annotation of the complete genome; these genes, VACV-COP A2.5L [5,6], A14.5L [7] and G5.5R [8] have now been included in our Poxvirus Orthologous Clusters (POCs) database [9].
Poxvirus genes are transcribed from both DNA strands and so far have never been shown to overlap more than a few nucleotides. Despite this knowledge, some poxvirus genomes have been liberally annotated so as to include all ORFs above a certain size, irrespective of whether they overlap larger well-characterized genes. Thus, the current GenBank file for VACV-COP contains 202 major (large and likely to be real genes) ORFs and 64 minor (small and unlikely to be real genes) ORFs [10,11]. The majority of these minor ORFs in VACV-COP overlap larger, major ORFs on the opposite DNA strand.
In this paper, it is shown that for the AT rich poxviruses, the purine skews can be used to help predict the synonymous (coding) strand, particularly in regions where smaller ORFs overlap each other on opposite strands of the genome and neither have orthologs in other poxvirus genomes. Furthermore, it is shown that the majority of minor ORFs found in VACV-COP are unlikely to be functional genes and that based on purine content, two of the three genes initially excluded from the annotation of the vaccinia virus genome due to their small size, fit our definition of a major ORF.
Results and discussion
Figure 1 shows the genomic purine skew (Figure 1a) and the direction of transcription (Figure 1b) for the major ORFs (genes) in VACV-COP. Since the major ORFs of VACV-COP are spread out evenly across the genome, and Figure 1b was created using only the major VACV-COP ORFs, the two figures (Figure 1a and 1b) follow very similar trends. A characteristic "W" shaped plot can be seen for both graphs; in Figure 1b, this is the result of a trend for large blocks of genes to be transcribed in the same direction (see arrows in Figure 1b). These data indicate a good correlation between the purine content of the genomic DNA and the direction of transcription; for example, for genes that are transcribed in the leftward direction, the bottom/synonymous strand is purine rich and the opposite is true for genes that are transcribed to the right. The correlation between purine content and the likelihood that an ORF is major is further supported by the fact that 180 of the 202 major ORFs of VACV-COP have a purine content greater than or equal to 50%. In this way, purine skews can be used to help annotate newly sequenced genomes by aiding in the determination of the mRNA synonymous strand.
Figure 1 Correlation between purine skew and direction of transcription of VACV-COP genome, excluding the non-coding terminal inverted repeats. (a) Purine skew drawn using DNAGrapher. Regions of the top strand that exhibit a purine bias will have a trend to the upward direction whereas regions that exhibit a pyrimidine bias will be drawn in the downward direction. Two example regions of changes in strand bias are shaded in green and marked (i) and (ii) (b) VACV-COP major ORFs drawn according to the strand of the genome on which each ORF is located. Beginning with a value of zero for the first major ORF of the genome, a numerical value of +1 or -1 is added to the value of the previous ORF depending on if the ORF is located on the top or bottom strand, respectively. (c) Gene orientation in two example regions demonstrating a change in strand bias. (i) Strand bias changes from a purine bias on the bottom strand, to a purine bias on the top strand that encompasses 1 gene on the top strand. (ii) Strand bias changes from a purine bias on the bottom strand, to a purine bias on the top strand that encompasses 4 genes located on the top strand.
When the purine skew (Figure 1a) slopes in the downward direction, this is due to a pyrimidine bias on the top strand and a commensurate purine bias on the bottom strand indicating that the major ORFs are located on the bottom strand. In regions where the purine skew changes direction from a downward slope to an upward slope or vice versa, these are regions on the genome where the transcription direction of the genes in the genome changes. For example, the purine skew appears to change direction from a downward slope to an upward slope at position 32,800 bp and then changes again from an upward slope to a downward slope at position 33,500. Figure 1c (i) shows that within this region (32,800–33,500 bp), there is one gene (VACV-COP K7R) that is located on the top strand (upwards slope on purine skew) and is flanked by genes that are located on the bottom strand (downward slope on purine skew). A second example can be seen in figure 1 c (ii) where an upward slope in the purine skew occurs between positions 52,400 and 57,000. In this case, the upward sloping region encompasses four genes (VACV-COP E5R, E6R, E7R and E8R) and the two downward sloping regions flanking each side of this region encompass genes that are located on the bottom strand.
It was previously shown that minor ORFs in VACV-COP tend to have higher than average serine content as well as lower than average aspartate and glutamate content [12]. Based on these observations and our current finding that the synonymous DNA strand is usually purine rich, we created a simple mathematical equation designed to provide a "quality" measure of each ORF. The results of the formula [Ser%-Asp%-Glu%+(50-AG%)], which essentially sums the trends in amino acid composition (3 amino acids) and purine content, are shown in Figure 2. If peptides are translated from ORFs on the non-synonymous strand, they tend to have a higher than average Ser%, but lower than average Asp% and Glu% (due to properties of the genetic code), and have a lower than average purine content. By subtracting the actual %purine from the genome average for VACV-COP (50%), if the ORF is major, the numerical result of the equation is negative and if the results of the equation are positive, the ORF is predicted to be minor.
Figure 2 Results of the "quality" measure for VACV-COP. Y-axis plots results of the "quality" calculation (Ser%-Asp%-Glu%+[50%-AG%]) and X-axis depicts rank of each ORF.
Plotting the results of this equation, we found that of the 266 ORFs originally predicted in VACV-COP, 6 ORFs (VACV-COP A ORF G, VACV-COP A ORF T, VACV-COP B ORF G, VACV-COP C ORF F, VACV-COP E ORF D, and VACV-COP F ORF A) were incorrectly classified as being major and 9 ORFs (VACV-COP A9L, VACV-COP A13L, VACV-COP A14L, VACV-COP A14.5L, VACV-COP A38L, VACV-COP A43R, VACV-COP C3L, VACV-COP I5L, VACV-COP I6L) were incorrectly classified as being minor.
It was found that the majority of incorrectly classified major ORFs were misclassified because they are small membrane proteins that had a lower aspartate and glutamate content than other major ORFs and that the majority of incorrectly classified minor ORFs were misclassified because they have a lower serine and higher purine percentage compared to other minor ORFs despite the fact that all but one minor ORF (VACV-COP A ORF T) overlap a major ORF on the opposite strand (Table 1). There were three genes that had initially been excluded from the annotation of VACV-COP due to their small size. Two of these genes (VACV-COP A2.5L and VACV-COP G5.5R) have a negative "quality" measure value indicating that they are major. One of these genes (VACV-COP A14.5L) was misclassified as minor likely due to the fact that it is a small membrane protein (Table 1).
Table 1 List of VACV-COP ORFs that were incorrectly classified.
Major ORFS incorrectly classified as minor
ORF name ORF size (bp) Serine content (%) Aspartate content (%) Glutamate content (%) Purine content (%) Explanation
VACV-COP A13L 210 11.43 1.43 2.86 48.82 Small, membrane protein
VACV-COP A14L 270 11.11 3.33 0 45.79 Small, membrane protein
VACV-COP A14.5L 159 7.55 1.89 1.89 44.45 Small, membrane protein
VACV-COP A38L 831 7.94 3.97 2.53 47.25 Membrane protein
VACV-COP A43R 582 10.31 5.67 1.55 51.11 Membrane protein
VACV-COP C3L 789 13.31 4.18 3.8 52.27 High Ser%, low Asp% and Glu%
VACV-COP I5L 237 5.06 2.53 1.27 49.58 Small, membrane protein
VACV-COP I6L 1146 10.99 4.45 4.45 49.7 High Ser%, low Asp% and Glu%
Minor ORFs incorrectly classified as major
ORF name ORF size (bp) Serine content (%) Aspartate content (%) Glutamate content (%) Purine content (%) Explanation
VACV-COP A ORF G 225 6.67 4 8 54.39 Low Ser%, high Asp% and AG%
VACV-COP A ORF T 243 1.23 3.7 2.47 51.63 Overlaps on same strand as major ORF
VACV-COP B ORF G 273 1.1 3.3 1.1 53.26 Low Ser%, high AG%
VACV-COP C ORF F 273 1.1 3.3 1.1 53.26 Low Ser%, high AG%
VACV-COP E ORF D 198 9.09 4.55 6.06 55.72 High Asp%, Glu%, AG%
VACV-COP F ORF A 201 4.48 4.48 0 50.49 Low Ser%
A similar analysis was repeated for the genome of amsacta moorei (AMEV), an extremely AT-rich (82%) entomopoxvirus [13]. The AMEV genome was chosen for two reasons: (1) because it is not closely related to any known poxviruses and therefore its genome contains a large number of genes with unknown function and (2) its genome was liberally annotated and therefore it is questionable which ORFs are likely to be functional genes. Thus, the "quality" measure was used to predict which AMEV ORFs are most likely to be minor. Figure 3 graphically depicts the results of the "quality" measure calculation for AMEV. Due to the extreme AT-richness of the AMEV genome, it was necessary to modify the "quality" measure to the following formula: [Ser%-Asp%-Glu%+(49%-AG%)]. 49% was chosen instead of 50% for the purine portion of this equation since the average purine content of the entire AMEV genome is 49%. As was the case with VACV-COP, if the ORF is minor, the results of the "quality" measure will be positive.
Figure 3 Results of the "quality" measure for amsacta moorei virus (AMEV). Y-axis plots results of the "quality" calculation (Ser%-Asp%-Glu%+[49%-AG%]) and X-axis depicts rank of each ORF.
It was found that there were 51 ORFs that had a positive "quality" value and are therefore considered minor. Of these 51 ORFs, 41 ORFs further fit our definition of a minor ORF as they overlapped another larger ORF on the opposite strand and 4 major ORFs (AMEV-161, AMEV-164, AMEV-171, and AMEV-183) were incorrectly classified as minor even though they each have orthologs in other poxviruses and are therefore major (Table 2). The remaining 6 ORFs (AMEV-001, AMEV-089, AMEV-148, AMEV-198, AMEV-ITR02, and AMEV-ITR08) that were classified as minor using our "quality" measure were found not to overlap any ORFs on the opposite or same DNA strand and were further analyzed using the AMEV purine skew in order to try and determine the correct coding strand in each of these 6 regions (Table 3). It was found that for 5 (AMEV-001, AMEV-089, AMEV-148, AMEV-198, AMEV-ITR02) of these 6 ORFs, the purine skew indicates a coding strand opposite the strand on which these ORFs are located, or in other words, that these ORFs are minor. For 1 (AMEV-ITR08) ORF, the purine skew indicated a coding strand identical to the strand on which this ORF is located and therefore this ORF may actually be major. AMEV-ITR08 does not have any orthologs in other poxviruses but it does show a 73.6% amino acid identity with the AMEV-ITR07 ORF which was classified as being major using the "quality" calculation further supporting that AMEV-ITR08 is likely major. AMEV-ITR08 was predicted to contain a transmembrane domain [13] which could explain why it was misclassified.
Table 2 List of AMEV ORFs that were incorrectly classified.
Major ORFS incorrectly classified as minor
ORF name ORF size (bp) Serine content (%) Aspartate content (%) Glutamate content (%) Purine content (%) Explanation
AMEV-161 243 11.11 2.47 1.23 47.56 Membrane protein
AMEV-164 708 7.63 2.12 2.97 47.97 High Ser%, low Asp% and Glu%
AMEV-171 276 3.26 1.09 1.09 48.39 Low Asp% and Glu%
AMEV-183 675 6.67 3.11 1.33 51.18 Low AG% and low Glu%
Minor ORFs incorrectly classified as major
ORF name ORF size (bp) Serine content (%) Aspartate content (%) Glutamate content (%) Purine content (%) Explanation
AMEV-152 225 0 12 1.33 60.97 Overlaps on same strand as major ORF
AMEV-189 180 1.67 8.33 1.67 43.17 Low Ser%, high Asp%
AMEV-191 228 0 2.63 10.53 61.9 Overlaps on same strand as major ORF
Table 3 List of 6 AMEV ORFs classified as minor that do not fit the definition of a minor ORF and conclusions as to whether or not they are minor.
ORF name DNA strand on which ORF is located Direction of purine skew Conclusion
AMEV-001 Top Down Minor
AMEV-089 Top Down Minor
AMEV-148 Bottom Up Minor
AMEV-198 Bottom Up Minor
AMEV-ITR02 Top Down Minor
AMEV-ITR08 Top Up May be major
There were three ORFs that had been classified as major (negative value for the "quality" measure) yet overlapped a larger gene on the opposite or same DNA strand (Table 2). Two of these ORFs (AMEV-152 and AMEV-191) overlap a larger ORF on the same strand and therefore neither the purine skew nor the "quality" measure are capable of determining which ORF is major; and one ORF (AMEV-189) overlaps the much larger spheroidin gene on the opposite strand and was likely misclassified due to its lower than average serine content and higher than average aspartate content.
For the analyses shown in figures 2 and 3, the cut-off value used in both cases was zero. The value of zero was chosen in the training case (VACV-COP) because it represented a reasonable cut-off between genes that were known to be major and ORFs that were known to be minor with minimal misclassification of genes. With our test case (AMEV), since it was not known which ORFs were major or minor, a cut-off of zero was initially used with the presumption that the cut-off may need to be adjusted due to the extreme AT-richness of the AMEV genome. Analyzing the "quality" measure data obtained for AMEV with a cut-off of zero yielded satisfactory results in that the number of overlapping and therefore likely to be minor ORFs that were misclassified was relatively low and because of this we decided to maintain the zero cut-off. It is likely that a cut-off of zero worked well with AMEV despite its extremely AT-rich genome because the "quality" measure that was used reflected the average AG% of the genome. It is also likely that other poxvirus genomes that are analyzed using our method would use a cut-off of zero, provided the "quality" measure that was used was changed to reflect the average AG content of the genome, although we have yet to test whether this cut-off is universal throughout all poxviruses.
Thus far we have shown that purine skews can be used to predict the coding strand of poxvirus genomes and that major ORFs in VACV-COP and in AMEV usually contain greater than 50% and 49% purines, respectively. In order to explain this purine richness in genes, the purine (R) to pyrimidine (Y) ratio (R:Y) was calculated for each codon position of each coding and non-coding ORF in VACV-COP. A Student's T-test was used to compare the mean R/Y ratio values for the coding (genes) and non-coding ORFs at each codon position; means were considered statistically different when the p-value was less than 0.05. At the first nucleotide position in the codon, both VACV-COP major and minor ORFs are rich in purines but the major ORFs (genes) have significantly (p < 0.05) higher levels of purines at this position (Table 4). At the second nucleotide position the major ORFs have a R:Y ratio of approximately 1 and the minor ORFs have a significantly lower R:Y ratio (p < 0.05) indicating that minor ORFs are pyrimidine rich at the second codon position whereas major ORFs contain roughly equal amounts of purines and pyrimidines at this position. At position 3, no statistical difference was found, with both major and minor ORFs being rich in pyrimidines. Thus, for the first and second nucleotide positions of the codons, the major ORFs (genes) have significantly higher purine content than the minor ORFs.
Table 4 Mean purine to pyrimidine ratios for each codon position of vaccinia virus Copenhagen major and minor ORFs. Positions marked with an asterisk (*) are statistically different.
Purine/Pyrimidine (R/Y) ratio at each codon position
Position 1* Position 2* Position 3
Major ORFs 1.77 0.99 0.93
Minor ORFs 1.21 0.75 0.96
It is important to remember that the use of purine/amino acid content of the coding strand and predicted protein, respectively, are just two measures that can be used to help predict whether an ORF is likely to be a functional gene and that usually they are only useful in discriminating between coding and non-coding strands. Occasionally ORFs that are fragments of bone fide genes are also flagged as non-functional, this is probably because of unusual amino acid content in small protein sub-domains. An example of this is the A25L ORF of VACV-COP that was flagged as non-functional by this method even though it is a fragment of the ATI protein. In a similar way, fragmentation of genes into smaller ORFs can also lead to unusual isoelectric points in the resulting predicted proteins; the 14 ORFs with a predicted pI of >9.6 are all minor ORFs or gene fragments. Thus, multiple approaches that may also include promoter analysis must be applied to attempt to correctly annotate small orphan ORFs in these genomes and there is no guarantee that the process will be 100% successful.
Conclusion
We have successfully shown that in the case of AT-rich poxviruses, purine skews can be used to help predict the coding regions of the genome. This is particularly useful if predicted ORFs overlap each other and it is not apparently obvious which ORF is major (when neither ORF has an ortholog in another poxvirus genome). A second method that can be used in conjunction with purine skews is to calculate the "quality" of each predicted ORF using information from amino acid composition and purine content. For a given ORF, if the results of this calculation are negative the ORF is predicted to be a functional gene, and if the results of the calculation are positive, the ORF is predicted to be minor.
By comparing purine to pyrimidine (R/Y) ratios at each codon position of major and minor vaccinia virus ORFs, it was found that the purine abundance seen for major ORFs stems primarily from the first codon position with both the second and third codon positions containing equal amounts of purines and pyrimidines.
The software used to create the purine skews (DNAGrapher) and the VOCs database are both available for public use via the web [14,15].
Methods
Purine skews
Purine skews were created using the DNA Grapher feature in VOCs [9]. The DNA Grapher program implements the algorithm originally developed by Lobry [16]. The algorithm assigns a direction to each base encountered in the sequence. In the case of purine skews, the graph begins at position (0,0) and move upwards one unit if the base encountered is a purine (A or G) and moves downwards one unit if the base encountered is a pyrimidine, (C or T). The plot continues in this fashion until the end of the sequence is reached. A variable window size can also be set. In this case, the plot trend will be either upwards or downwards, depending on the average number of purines or pyrimidines in the window. The window then slides over the number of bases defined by the window size. For example, if the window size was defined as 10 bp, the window will slide over to the eleventh base and then count the average. The DNAGrapher program is integrated into the VOCs software and is also accessible as a Java WebStart program [14].
Graphing ORFs by strand
The 202 major ORFs (genes) of VACV-COP were ordered in ascending order according to their start positions on the genome and then plotted according to which strand they are located using Microsoft Excel. The first gene was plotted at position 0 of the y-axis of the graph and a value of either -1 or +1 was added to the next gene on the genome depending on if it was on the bottom or top strand respectively.
ORF "quality" calculation
The analysis of VACV-COP ORFs was performed by plotting the results of the following equation: Ser%-Asp%-Glu%+(50%-AG%) where Ser% is serine percentage, Asp% is aspartate percentage, Glu% is glutamate percentage, AG% is purine percentage and the value of 50% is the average purine content of the VACV-COP genome. The "quality measure" for AMEV ORFs used the following formula: Ser%-Asp%-Glu%+(49%-AG%) where the only modification of this formula from VACV-COP was the value of 49% which reflects the average purine content of the AMEV genome. The amino acid composition and purine data was obtained from the VOCs database which is available on the internet as a Java Web Start program [9,15].
The results of the equation for each ORF were tabulated, sorted in ascending order and assigned a rank from 1 being the ORF with the most negative value to either 266 in the case of VACV-COP or 292 in the case of AMEV being the ORF with the most positive value. These results of the calculation were plotted on the y-axis and the rank of each ORF was plotted on the x-axis using Microsoft Excel.
Purine/pyrimidine ratio comparison
To analyse the ratio of purines to pyrimidines at each codon position, the total number of each nucleotide at each codon position was first calculated using the codontree program with the BC=A option (calculate the base composition at all 3 codon positions) selected [17,18]. Once the base composition at each codon position was calculated, the purine to pyrimidine ratio (R/Y) was calculated for each ORF of the dataset. The mean values of the R/Y ratio for each dataset were compared using Student's T-Test to determine if the mean R/Y ratio for each dataset was statistically different. The null hypothesis for the Student's T-test was that the means were equal and the null hypothesis was rejected if the p-value was < 0.05. The two datasets used for this portion of the paper consisted of (1) all ORFs classified as major in VACV-COP and (2) all ORFs classified as minor in VACV-COP.
Authors' contributions
MD performed all analyses and wrote the manuscript. CU conceived of, and supervised the study, and edited the manuscript.
Acknowledgements
We thank Jamie Thomas, Clint Johnson and Angleika Ehlers for their help programming DNA Grapher, and David Esteban for critical reading of the manuscript.
==== Refs
Szybalski W Kubinski H Sheldrick P Pyrimidine clusters on the transcribing strand of DNA and their possible role in the initiation of RNA synthesis Cold Spring Harb Symp Quant Biol 1966 31 123 127 4966069
Karkas JD Rudner R Chargaff E Seapration of B. subtilis DNA into complementary strands. II. Template functions and composition as determined by transcription with RNA polymerase Proc Natl Acad Sci U S A 1968 60 915 920 4970113
Rudner R Karkas JD Chargaff E Separation of B. subtilis DNA into complementary strands, I. Biological properties Proc Natl Acad Sci U S A 1968 60 630 635 4973487
Bell SJ Forsdyke DR Deviations from Chargaff's second parity rule correlate with direction of transcription J Theor Biol 1999 197 63 76 10036208 10.1006/jtbi.1998.0858
Senkevich TG White CL Koonin EV Moss B Complete pathway for protein disulfide bond formation encoded by poxviruses Proc Natl Acad Sci U S A 2002 99 6667 6672 11983854 10.1073/pnas.062163799
Senkevich TG White CL Weisberg A Granek JA Wolffe EJ Koonin EV Moss B Expression of the vaccinia virus A2.5L redox protein is required for virion morphogenesis Virology 2002 300 296 303 12350360 10.1006/viro.2002.1608
Betakova T Wolffe EJ Moss B The vaccinia virus A14.5L gene encodes a hydrophobic 53-amino-acid virion membrane protein that enhances virulence in mice and is conserved among vertebrate poxviruses J Virol 2000 74 4085 4092 10756020 10.1128/JVI.74.9.4085-4092.2000
Amegadzie BY Ahn BY Moss B Characterization of a 7-kilodalton subunit of vaccinia virus DNA-dependent RNA polymerase with structural similarities to the smallest subunit of eukaryotic RNA polymerase II J Virol 1992 66 3003 3010 1560534
Ehlers A Osborne J Slack S Roper RL Upton C Poxvirus Orthologous Clusters (POCs) Bioinformatics 2002 18 1544 1545 12424130 10.1093/bioinformatics/18.11.1544
Goebel SJ Johnson GP Perkus ME Davis SW Winslow JP Paoletti E The complete DNA sequence of vaccinia virus Virology 1990 179 247 266 2219722 10.1016/0042-6822(90)90294-2
Upton C Slack S Hunter AL Ehlers A Roper RL Poxvirus orthologous clusters: toward defining the minimum essential poxvirus genome J Virol 2003 77 7590 7600 12805459 10.1128/JVI.77.13.7590-7600.2003
Upton C Screening predicted coding regions in poxvirus genomes Virus Genes 2000 20 159 164 10872878 10.1023/A:1008126816295
Bawden AL Glassberg KJ Diggans J Shaw R Farmerie W Moyer RW Complete genomic sequence of the Amsacta moorei entomopoxvirus: analysis and comparison with other poxviruses Virology 2000 274 120 139 10936094 10.1006/viro.2000.0449
DNAGrapher DNAGrapher
Viral Orthologous Clusters (VOCS)
Lobry JR A simple vectorial representation of DNA sequences for the detection of replication origins in bacteria Biochimie 1996 78 323 326 8905151 10.1016/0300-9084(96)84764-X
Pesole G Attimonelli M Liuni S A backtranslation method based on codon usage strategy Nucleic Acids Res 1988 16 1715 1728 3281142
Codontree Codontree
| 15720717 | PMC552303 | CC BY | 2021-01-04 16:39:52 | no | BMC Genomics. 2005 Feb 18; 6:22 | utf-8 | BMC Genomics | 2,005 | 10.1186/1471-2164-6-22 | oa_comm |
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-111571004710.1186/1471-2202-6-11Research ArticleResponse of SII cortex to ipsilateral, contralateral and bilateral flutter stimulation in the cat Tommerdahl Mark [email protected] Stephen B [email protected] Joannellyn S [email protected] Vinay [email protected] Oleg [email protected] Barry [email protected] Department of Biomedical Engineering, University of North Carolina, Chapel Hill, NC 27599, USA2 Department of Cellular and Molecular Physiology, University of North Carolina, Chapel Hill, NC 27599, USA2005 14 2 2005 6 11 11 22 12 2004 14 2 2005 Copyright © 2005 Tommerdahl 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 distinctive property of SII is that it is the first cortical stage of the somatosensory projection pathway that integrates information arising from both sides of the body. However, there is very little known about how inputs across the body mid-line are processed within SII.
Results
Optical intrinsic signal imaging was used to evaluate the response of primary somatosensory cortex (SI and SII in the same hemisphere) to 25 Hz sinusoidal vertical skin displacement stimulation ("skin flutter") applied contralaterally, ipsilaterally, and bilaterally to the central pads of the forepaws. A localized increase in absorbance in both SI and SII was evoked by both contralateral and bilateral flutter stimulation. Ipsilateral flutter stimulation evoked a localized increase in absorbance in SII, but not in SI. The SII region that responded with an increase in absorbance to ipsilateral stimulation was posterior to the region in which absorbance increased maximally in response to stimulation of the contralateral central pad. Additionally, in the posterior SII region that responded maximally to ipsilateral stimulation of the central pad, bilateral central pad stimulation approximated a linear summation of the SII responses to independent stimulation of the contralateral and ipsilateral central pads. Conversely, in anterior SII (the region that responded maximally to contralateral stimulation), bilateral stimulation was consistently less than the response evoked from the contralateral central pad.
Conclusions
The results indicate that two regions located at neighboring, but distinctly different A-P levels of the anterior ectosylvian gyrus process input from opposite sides of the body midline in very different ways. The results suggest that the SII cortex, in the cat, can be subdivided into at least two functionally distinct regions and that these functionally distinct regions demonstrate a laterality preference within SII.
==== Body
Background
There is general agreement that in cats and monkeys (and presumably in humans) the spike discharge activity a mechanical stimulus sets up in rapidly adapting (RA), slowly adapting (SA), and Pacinian (PC) skin mechanoreceptors is projected centrally, at short latency and with relatively minor transformation, to primary somatosensory cortex (both SI and SII) in the contralateral hemisphere. There is no consensus, however, about the way in which the stimulus-evoked response in the ipsilateral hemisphere contributes to cerebral cortical somatosensory information processing and somatosensation. A distinctive property of SII is that it is the first cortical stage of the somatosensory projection pathway that integrates information arising from both sides of the body. Similar to SI, SII possesses a clear topographic organization [1], but unlike SI, a significant fraction of SII neurons possess bilateral receptive fields (RFs). The fraction of SII neurons with bilateral RFs varies from one topographic region of SII to the next. While ipsilateral input to SII is widely accepted, the role(s) of this input in somatosensory information processing remains uncertain. To investigate the effects on SII input deriving from mechanoreceptors in ipsilateral skin regions, the technique of optical intrinsic signal (OIS) imaging was used to assess the impact on SII of ipsilateral input on the response of SII to a contralateral input. The responses to contralateral, ipsilateral and bilateral vibrotactile stimulation (25 Hz – "flutter") of the forepaw of the cat were quantified and compared to make this assessment. Although evoked responses in both SI and SII were imaged in the studies, the primary focus of this report is the response of SII to the aforementioned stimuli.
Results
Figure 1 shows the OIS responses evoked in SII of two exemplary subjects by contralateral, ipsilateral, and bilateral central pad stimulation. Visual inspection of the images for the three stimulus conditions in each subject shows that: (1) the optical response to contralateral stimulation occurs in a region more anterior in SII than does the response to ipsilateral stimulation; (2) the SII optical response to bilateral stimulation occupies both the anterior and posterior regions that responded to independent stimulation of the contralateral and ipsilateral central pads; and (3) the optical response to ipsilateral stimulation does not evoke a large absorbance change in SI.
To more accurately characterize the spatial disparity between SII loci activated by contralateral vs. ipsilateral stimulation, the absorbance values were obtained and plotted along the posterior-anterior axis of SII. Figure 2 shows an absorbance vs. distance plot for each of the 2 subjects whose images are shown in Figure 1. Note that the distance between the peak absorbance values obtained under the ipsilateral and contralateral conditions is approximately 2 mm for Subject 1 (plots on left) and approximately 3 mm for Subject 2 (plots on right). The average across-subject (n = 6) distance between the peaks of SII activation evoked by contralateral and ipsilateral stimulation is 2.4 +/- 0.46 mm [2]. Interestingly, for both subjects, in the posterior region of SII, the magnitude of the response to bilateral stimulation exceeds that of the response to either contralateral or ipsilateral stimulation, but, in the anterior region of SII, the magnitude of response to contralateral stimulation is greater than that of the response to bilateral stimulation. The center of the distribution of the peaks of absorbance evoked by contralateral and ipsilateral stimulation, such as those shown in Figure 2, were used to define the anterior and posterior regions of SII in subsequent analyses (e.g., the peaks were used as the center point of the sampled regions of interest).
A more comprehensive view of the SII response to the contralateral and bilateral stimulus conditions can be better appreciated with a multi-dimensional surface plot of the data. Figure 3 compares the stimulus evoked response of SII in the two subjects to contralateral and bilateral stimulation, and from these plots, it is quite apparent that a large area (in the medial-lateral dimension) of the anterior region of SII is suppressed in the bilateral stimulus condition, relative to the response evoked by the contralateral stimulus. The surface plots of Figure 3 enable direct comparison of specific modules that exhibit a particular profile in one condition (e.g. the contralateral response of Subject 2 – see locations marked 1, 2 and 3) with that of another condition (compare with modules marked 1', 2' and 3' for the bilateral response). In this comparison, the absorbance values at loci 1 and 2 (in anterior SII) are clearly larger than those at locus 3 (in posterior SII) in the contralateral response, and the absorbance values at locations 1' and 2' (from the bilateral response) are much smaller than those at loci 1 and 2 of the contralateral response. Additionally, the response at locus 3 (posterior SII) in the contralateral response is very weak, but is larger in the response to the bilateral stimulus. Thus, the increase in activity observed at locus 3' (i.e., 3'>>3), in the posterior region of SII, evoked by bilateral stimulation parallels a decrease in activity at loci 1' and 2', in the anterior region of SII. In the case of Subject 1, the decrease in activity in anterior SII is not as pronounced as that seen in Subject 2 (also note difference in Figure 2), although the concurrent increase in posterior SII activity (compare module P with P') with decreased anterior SII activity (compare module A with A') is consistent with the shift in activation along the posterior-anterior axis observed in Subject 2. Responses evoked by ipsilateral stimulation are also displayed in Figure 3. Note that in both Subjects 1 and 2, absorbance values in the posterior region are much greater than those in the anterior region of SII.
To directly compare the time course of the response of the anterior and posterior regions of SII to the three stimulus conditions, we determined the time course of the absorbance changes in each region under each stimulus condition (Figure 4). The plots in Figure 4 show that in each of these 2 subjects and under each stimulus condition, the magnitude of the absorbance change evoked in either the posterior or anterior region of SII by ipsilateral stimulation was less than that evoked in the same region by contralateral or bilateral stimulation. Moreover, in both subjects, the magnitude of the response of the posterior region to bilateral stimulation was greater than that evoked by contralateral stimulation, whereas in the anterior region of SII, the magnitude of the response evoked by bilateral stimulation is either less than or approximately equal to the response evoked by contralateral stimulation.
Cluster plots were used to directly compare the response of SII to the different conditions of ipsilateral and contralateral stimulation. In each plot in Figure 5, the absorbance value obtained at each pixel to the 2 different stimulus conditions is plotted against each other – i.e., the x-axis is the absorbance value evoked by the contralateral stimulus and the y-axis is the absorbance value evoked by the ipsilateral stimulus. The clusters reveal a distinct differentiation in the population of SII neurons to the different stimulus conditions. Additionally, it appears that this could be a time dependent process, as there is little difference in the behavior of the pixels localized to the SII region in the early stages of the response (t = 1 sec), there is some grouping after 2 seconds, and there are two distinct clusters formed after several seconds (t = 5 sec). It should be emphasized that this type of graphic does not necessarily reflect spatial differences in the responses of two different stimulus conditions, but rather, it emphasizes whether or not different members of a set respond differently to different stimulus conditions. Thus, the information demonstrated by the cluster plots in Figure 5 can be summarized by stating that with an increase in stimulus duration (from 1 to 5 seconds), there is an increase in the segregation of the behavior of the different groups of pixels whose values are more predominantly affected by a contralateral vs. an ipsilateral stimulus.
The contralateral and bilateral stimulus conditions can also be compared using cluster analysis. However, because the difference between the contralateral and ipsilateral responses is more robust than the difference between the contralateral and bilateral responses, Figure 6 displays the results of cluster analysis of the anterior and posterior regions of SII independently. Independent analysis of the two regions allows for better resolution of shifts in the behavior of activity within particular regions. For each subject, the peak response was identified in both the anterior and posterior regions. Cluster plots were obtained by plotting the response (contralateral along the x-axis, bilateral along the y-axis) at each location within a 1 × 1 mm2 boxel surrounding the peak. In anterior SII, the majority of the responses to contralateral stimulation are stronger than the responses to bilateral stimulation – hence, the majority of the points plotted fall below the reference line (which has a slope of 1). In the posterior region, the majority of the points plotted are above the reference line – indicating that the response to the bilateral stimulus was greater than the response to the contralateral stimulus.
Thus far, the results suggest that the anterior and posterior regions of SII are differentially activated by contralateral, ipsilateral and bilateral stimulation. To determine the across-subject consistency of these findings, the average absorbance values evoked by the 3 different stimulus conditions were determined for all 6 of the subjects (Figure 7). Clearly, in the anterior region of SII, the contralateral stimulus condition evoked the largest magnitude of response, and therefore, the values of the absorbance increase obtained in this and the posterior region under each stimulus condition were normalized to this absorbance value (thus, standard error for the contralateral/anterior region condition = 0). Whereas in the anterior region of SII, the response evoked by bilateral stimulation is approximately 35% less than that evoked by contralateral stimulation, in the posterior region of SII, the bilateral stimulus evoked a response approximately 25% larger than that evoked by the contralateral stimulus. Analysis of variance showed that, at a 95% confidence interval, the average bilateral:contralateral response ratio was between 0.42 and 0.83 in the anterior region of SII. In the posterior region of SII, the same analysis showed the bilateral:contralateral response ratio to be between 0.97 and 1.43 at a 95% confidence interval. Although the response of the posterior region to the bilateral stimulus is larger than that evoked by the contralateral stimulus, it is less than that predicted by summation of the responses to the ipsilateral and contralateral stimuli (computed values, Figure 7). Furthermore, while the response of neither the anterior or posterior regions of SII to bilateral stimulation approximate a linear summation of the responses to independent stimulation of each central pad, the approximation of the bilateral response by summation of responses evoked by independent stimuli is much closer to being accurate in the posterior region of SII.
Discussion
The findings of this study demonstrated clearly that the anterior and posterior regions of SII process bilateral inputs very differently. At the locus of the maximal OIS response evoked in the posterior region by an ipsilateral stimulus, bilateral stimulation evoked a response that was, on average, about 25% larger than that evoked from the contralateral stimulus site. Conversely, at the locus of the maximal OIS response evoked by contralateral stimulation in the anterior region, bilateral stimulation evoked a response that was, on average, 35% lower than the activity evoked by a contralateral stimulus. This discrepancy between the optical responses of the anterior and posterior regions could be related to neurophysiological observations reported in earlier studies. For example, Carreras and Andersson [3] found that for a sizable fraction of the cat SII neurons in their study, ipsilateral mechanical skin stimulation inhibited the response to contralateral stimulation, whereas in contrast, Picard et al. [4] found in their study of neurons in the distal forelimb regions of cat SII that simultaneous delivery of contralateral and ipsilateral mechanical skin stimuli led to strong facilitation of SII neuron response. In the study of Picard et al. [4], the responses of cells to bilateral stimulation were found to exceed the stronger of the responses to unilateral stimulation by, on average, 230%. Their study was limited, however, to the very low numbers of SII neurons that had bilateral RFs on the distal limbs. Burton, et al. [5], similar to Carreras and Andersson [3], reported that SII cells with bilateral receptive fields (monkey) exhibited a reduction in mean firing rate of 30% when the contralateral stimulus was preceded by an ipsilateral stimulus. Finally, other workers have found that callosally-transmitted inputs tend to have excitatory effects on SII neurons that have bilateral RFs, and exert inhibitory effects on SII neurons that have exclusively contralateral RFs [6-8]. Simoes et al. [9] showed significant suppression of the MEG SII response in humans, with simultaneous inputs delivered to the same skin sites, and Hoechstetter et al. [10] described "interactions" in SII cortex (a response that was not the summation of the ipsilateral and contralateral response) to simultaneous bilateral stimuli. Definitive establishment of the relationship between stimulus-evoked SII neuroelectrical and OIS activation, however, must await the performance of combined imaging and neurophysiological investigations which utilize both methodologies in the same subjects and under the same stimulus conditions.
The main, although not the only, route for ipsilateral input to SII is through the corpus callosum, from cells located in SI and SII of the opposite cerebral hemisphere [4,7,8,11]. Even those regions in SII that represent most distal parts of the limbs receive significant numbers of connections from the homologous zones of the contralateral SI and SII [12-15]. Graziosi [16] showed that separate populations of cells in SI provide callosal projections to SI and SII in the opposite hemisphere and ipsilateral projections to SII. Some separation within SII of the responses to ipsilateral and contralateral stimulation was also shown by Friedman et al. [17] and Juliano et al. [18].
The neurons in the distal limb regions of SII do receive substantial callosal connections, but these neurons have been reported to lack ipsilateral RFs [1], indicating that callosal inputs are not strong enough to generate action potentials (at least under the conditions used in RF mapping studies). This suggests that SII neurons do not use their sensory inputs from the ipsilateral side of the body to construct functional properties dependent on bilateral inputs; in other words, to extract information about higher-order properties of bi-manually contacted objects from coordinated patterns of sensory stimulation of the two hands. Instead, it could be postulated that neurons in the distal limb regions of SII use their ipsilateral peripheral inputs to modulate the responses to contralateral peripheral stimulation. On the other hand, Bennett et al. [19] found that bilateral convergence on SII neurons varies markedly with the different classes of tactile neurons, and modulation of the SII response by ipsilateral inputs may vary from one cortical area to another with different stimulus modalities.
A number of interactions between stimuli applied to both hands have been demonstrated in human psychophysical studies. Gilson [20] found that the threshold for detection of vibrotactile stimuli applied to a fingertip is elevated by parallel stimulation of the other hand's fingers. In addition, Gescheider and Verrillo [21] reported that the magnitude of vibrotactile sensation, elicited by brief 25 or 300 Hz stimuli applied to thenar eminence, was decreased by stimuli applied simultaneously to the opposite hand, but was enhanced when the contralateral stimulus was applied 150 msec prior to the test stimuli. Essick and Whitsel [22] reported that the perception of the direction of motion of brushing stimuli on the skin is enhanced by the presence of a simultaneous contralateral brushing stimulus when the two stimuli move in the same direction, but is weakened when the contralateral stimulus moves in a direction opposite to that on the other arm. While the above described reports provide possible perceptual correlates for bilateral interactions that might occur in SII, such as those identified in the present study, it will remain uncertain until anterior or posterior SII cortical activity is studied under conditions that permit direct correlations of perceptual performance and cortical activity under precisely controlled conditions of contralateral vs. bilateral skin stimulation.
A recent report [23] demonstrated 3 separate functional cortical fields along the anterior-posterior axis in the macaque. These functional fields were defined based on differential neural responses from three distinct cortical fields, and their report was unique in that it described cortical areas within SII based on functional properties of cortical areas. In this report, we demonstrate at least two functional subdivisions within SII in the cat based on functional properties as well. However, the modes of stimulation used to distinguish the functional differences along the anterior-posterior axis of SII were very different in this study (contralateral/ipsilateral/bilateral vs. proprioceptive/cutaneous inputs in the Fitzgerald study), and subsequent investigations using other stimulus modalities could reveal that SII of the cat is organized in a very similar fashion to SII of primates. The multiple fields found in SII, based on functional differences, could be, as suggested by Fitzgerald, et al. [23], indicative of the existence of a number of distributed processing streams. The significance of the presented work is that the response of these different cortical areas, which could represent information from so-called separate information streams, changes in a manner dependent upon the activity of neighboring cortical areas. Distinction of cortical areas within SII, identified by functional characteristics, demonstrates the nonlinearity of the integration of information from different sources (or information streams).
One question that the results suggest is whether or not SII can be segregated by laterality preference, in a manner similar to that observed in other sensory systems. Laterality has been demonstrated in the primary sensory cortex of both the visual system and the auditory system of both primates and cats, and the data in this report strongly suggest that there are cortical areas within SII that exhibit preference to ipsilateral or contralateral inputs. In terms of processing information from simultaneous contralateral and ipsilateral stimuli, there could be further similarities between the somatosensory, auditory and visual systems that have yet to be described. Future investigations will aim to further clarify the role of SII in integration of information from inputs across the body midline.
Conclusions
The responses evoked by contralateral and ipsilateral flutter stimulation of the central pad of the cat forepaw define functional subdivisions in SII: the two modes of stimulation maximally activate cortical regions that are anterior and posterior to one another, respectively. Bilateral stimulation, or providing simultaneous contralateral and ipsilateral stimulation, reveals, additionally, that the two adjacent cortical areas process bilateral inputs differently. In the posterior region, where ipsilateral stimulation evokes a maximal response, bilateral stimuli evoke a response that is greater than the response evoked by either the individual ipsilateral or contralateral response. In the anterior region of SII, where the contralateral stimulus evokes a maximal response, bilateral stimuli evoke responses that are smaller in magnitude than the responses evoked by the contralateral stimulus.
Methods
Subjects & preparation
Adult cats (males and females; n = 6) were subjects. All surgical procedures were carried out under deep general anesthesia (1 – 4% halothane in a 50/50 mixture of oxygen and nitrous oxide). After induction of general anesthesia the trachea was intubated with a soft tube and a polyethylene cannula was inserted in the femoral vein to allow administration of drugs and fluids (5% dextrose and 0.9% NaCl). For each subject, a 1.5 cm diameter opening was made in the skull overlying somatosensory cortex, a chamber was mounted to the skull over the opening with dental acrylic, and the dura overlying anterior parietal cortex was incised and removed. Following the completion of the surgical procedures all wound margins were infiltrated with long-lasting local anesthetic, the skin and muscle incisions were closed with sutures, and each surgical site outside the recording chamber was covered with a bandage held in place by adhesive tape.
Subjects were immobilized with Norcuron and ventilated with a gas mixture (a 50/50 mix of oxygen and nitrous oxide; supplemented with 0.1 – 1.0% halothane when necessary) delivered via a positive pressure respirator 1–3 hours prior to the data acquisition phase of the OIS imaging experiments. Respirator rate and volume were adjusted to maintain end-tidal CO2 between 3.0 – 4.0%; EEG and autonomic signs (slow wave content; heart rate, etc.) were monitored and titrated (by adjustments in the anesthetic gas mixture) to maintain levels consistent with light general anesthesia. Rectal temperature was maintained (using a heating pad) at 37.5°C.
Euthanasia was achieved by intravenous injection of pentobarbital (45 mg/kg) and by intracardial perfusion with saline followed by fixative (10% formalin). Following perfusion fiducial marks were placed to guide removal, blocking, and subsequent histological sectioning of the cortical region studied. All procedures were reviewed and approved in advance by an institutional committee and are in full compliance with current NIH policy on animal welfare.
Stimuli and stimulus protocols
Results were obtained during stimulation of the contralateral central pad of the forepaw and/or the ipsilateral central pad of the forepaw. The stimuli always consisted of sinusoidal vertical skin displacements (25 Hz, 400 microns, stimulus duration 5 – 20 sec, inter-stimulus interval 60 sec) and were applied using a servocontrolled transducer (Cantek Enterprises, Canonsburg, PA) that is capable of delivering sinusoidal stimuli in the range of 1–250 Hz at amplitudes in the range of 0–1000 microns. The stimuli were delivered independently to the ipsilateral and contralateral skin sites, and also were applied simultaneously to both sites (bilateral stimulation). The stimulus probes were positioned 500 microns beyond the point at which skin contact was detected (via force transducer on the Cantek). The bilateral stimulus protocols reported in this paper were synchronized to start and stop at the same time. The contralateral, ipsilateral and bilateral stimuli were interleaved on a trial-by-trial basis. This approach was used to control for temporal changes in cortical "state" unrelated to stimulus conditions which, if unrecognized, might obscure or modify any differences between the optical responses evoked by the contralateral, ipsilateral and bilateral stimulus conditions.
OIS imaging
Near-infrared (IR; 833 nm) OIS imaging was carried out using an oil-filled chamber capped with an optical window [24]. Images of the exposed cortical surface were acquired 200 msec before stimulus onset ("reference" or "prestimulus" images) and continuously thereafter ("poststimulus" images; at a resolution of one image every 0.5 to 1.5 sec) for 15–20 sec following stimulus onset. Exposure time was 200 msec. Absorbance images were generated by subtracting each prestimulus (reference) image from its corresponding poststimulus image and subsequently dividing by the reference image. Averaged absorbance images typically show regions of both increased absorption of IR light and decreased absorption of light (to a depth of approximately 1400 microns) which have been shown to be accompanied by increases and decreases in neuronal activation, respectively [24-29].
Histological procedures/identification of cytoarchitectural boundaries
At the conclusion of the experiment, the imaged cortical region was removed immediately following intracardial perfusion with saline and fixative. The region then was blocked, postfixed, cryoprotected, frozen, sectioned serially at 30 μm, and the sections stained with cresyl fast violet. The boundaries between adjacent cytoarchitectonic areas were identified by scanning individual sagittal sections separated by no more than 300 μm and were plotted at high resolution using a microscope with a drawing tube attachment. The resulting plots then were used to reconstruct a two-dimensional surface map of the cytoarchitectonic boundaries within the region studied with optical and neurophysiological recording methods. The locations of microelectrode tracks and electrolytic lesions evident in the histological sections were projected radially to the pial surface and transferred to the map of cytoarchitectonic boundaries reconstructed from the same sections. As the final step, the cytoarchitectonic boundaries (along with the locations of microelectrode tracks and lesions whenever present) identified in each brain were mapped onto the images of the stimulus-evoked intrinsic signal obtained from the same subject, using fiducial points (made by postmortem applications of india ink or needle stabs) as well as morphological landmarks (e.g., blood vessels and sulci evident both in the optical images and in histological sections). Locations of cytoarchitectonic boundaries were identified using established criteria [30-32].
Abbreviations
A-P = anterior-posterior
RA = rapidly adapting
SA = slowly adapting
PC = Pacinian
RF = receptive field
OIS = optical intrinsic signal
IR = near infrared
EEG = electro-encephalogram
MEG = magneto-encephalogram
Authors' contributions
BW and OF participated in the design of the experiments, the data collection, and drafting of the manuscript. SS, JC, and VT made significant contributions to the data collection and the analysis of the data. MT played a major role in all aspects of the development of the manuscript.
Acknowledgements
This work was supported, in part, by US Army Research Office grant P43077-LS (M. Tommerdahl, P.I.), NIH NS050587 (M. Tommerdahl, P.I.) and NIH NS35222 (B. Whitsel, P.I.).
Figures and Tables
Figure 1 Cat SI and SII optical responses to 25 Hz vibrotactile stimulation of the forepaws (2 subjects). A. View of the cortical surface, showing the vascular pattern and coronal (COR), ansate (ANS), and suprasylvian (SS) sulci. Exposed portions of SI and SII are indicated. Left hand column of Subject 1 and Subject 2 : Averaged difference images for responses evoked by (B) contralateral, (C) ipsilateral and (D) bilateral stimuli. Adjacent thresholded images of responses evoked by the three modes of stimulation – horizontal grid lines facilitate comparison of the position of loci of the evoked responses and are spaced 2 mm apart. Stimulus sites are indicated by figurines. Note that in both subjects, ipsilateral stimulation evokes a response posterior to the response evoked by contralateral stimulation. Scale bar is 2 mm. Orientation of images indicated by P (posterior), A (anterior), M (medial) and L (lateral) axes.
Figure 2 Spatial distribution of response along anterior-posterior axis of SII. Graphs obtained from OIS data in the anterior and posterior SII cortical regions evoked by flutter stimulus on the central pad of the two subjects shown in Figure 1. Orientation of segmentation indicated by figurines. Left Panel: Note that the response evoked by the bilateral stimulus is slightly smaller than the response evoked by the contralateral stimulus in the anterior region but larger than the response evoked by the contralateral stimulus in the posterior region. Right Panel: Note that in this case, the absorbance values evoked by the bilateral stimulus is significantly smaller than the response evoked by the contralateral stimulus in anterior SII, but approximates a summation of the ipsilateral and contralateral responses in posterior SII. The range of the responses to bilateral stimulation in anterior SII is typified by these 2 subjects – though the differences between the bilateral and contralateral response ranged from small (as in subject 1) to large (as in Subject 2) the bilateral response under the conditions studied were weaker than the contralateral response for all subjects (n = 6).
Figure 3 Comparison of contralateral vs. bilateral response in SII. Data displayed is a subset of the data displayed in Figure 1. Region of interest is indicated in figurines. X and Z axes are distance, indicated in mm (along either the anterior-posterior axis or medial-lateral axis). Y axis is absorbance. In Subject 1, note the increase in absorbance in the posterior region in the bilateral response, as compared to the contralateral response (P'>P). Also note the slight decrease in absorbance in the anterior region in the bilateral response relative to the contralateral response (A'<A). In Subject 2, note that in the modules identified in anterior SII (1 & 2), the absorbance value at loci 1 & 2 are greater than their counterparts in the bilateral response (1' and 2'). Module 3, on the other hand, is located in posterior SII (the region of maximal ipsilateral activation), and the absorbance values at 3' are much greater than the values at 3.
Figure 4 Graphs obtained from OIS data in the anterior and posterior SII cortical regions evoked by flutter stimulus on the central pad of 2 subjects. Figurines indicate regions of interest. (A) Top Panels: Time course of absorbance values from the posterior SII region obtained with ipsilateral, contralateral, and bilateral stimulus. Note that in this case, the absorbance evoked by the bilateral stimulus is larger than the response evoked by either the ipsilateral or the contralateral stimulus. (B) Bottom Panels: Time course of absorbance values from the anterior SII region obtained from data sampled during ipsilateral, contralateral, and bilateral stimulation. Note that the response evoked by the bilateral stimulus is smaller than the response evoked by the contralateral stimulus. Maximal differentiation of the time course of the response to the different stimulus conditions appears to occur between 1 and 3 seconds.
Figure 5 Cluster plots of ipsilateral vs. contralateral response of 2 subjects. For each cluster plot, values of individual pixels are plotted as a function of the response measured at that pixel to the ipsilateral stimulus (horizontal axis) vs. the response measured at that locus evoked by the contralateral stimulus (vertical axis). Colors depict the pixels that maximally responded to ipsilateral (green) and contralateral (red) stimulation. Cluster separation follows the same trend as the time course shown in Figure 4. After 5 seconds, the activity of the responding population has diverged into two distinct clusters. Reference images at top are at same orientation as the reference images in Figure 1.
Figure 6 Cluster plots of contralateral vs. bilateral response of 2 subjects. For each cluster plot, values of individual pixels are plotted as a function of the absorbance measured at that pixel to the contralateral stimulus (horizontal axis) vs. the response measured at the same locus (or pixel) evoked by the bilateral stimulus (vertical axis). Reference line, plotted at a slope of 1, indicates where pixels with equal values for both conditions lie. Note that in the anterior region, the majority of the pixels are below the reference line (response to bilateral stimulus was weaker than the response to the contralateral stimulus) and that in the posterior region, the majority of the pixels are plotted above the reference line (response to bilateral stimulus is greater than the response to the contralateral stimulus). Plots are normalized (minimum absorbance value scaled to 0; maximum scaled to 1).
Figure 7 Averaged normalized absorbance values (n = 6) obtained at 3 seconds after stimulus onset with respect to the contralateral stimulus in the anterior and posterior SII region. Computed response is the summation of the ipsilateral and contralateral responses. Left Panel: The contralateral stimulus in the anterior SII region evoked the greatest cortical response, while the ipsilateral stimulus was significantly lower. The bilateral response, however, evoked a cortical response that fell between the ipsilateral and the contralateral evoked responses. Note that the bilateral response is lower than the computed response. Right Panel: The ipsilateral response in the posterior SII region is greater than the ipsilateral response in the anterior SII region. The difference between the ipsilateral and contralateral responses is not as great in the posterior SII region as it was observed in the anterior SII region. Note that the bilateral response is larger than both the ipsilateral and contralateral responses, but still remains less than the computed value.
==== Refs
Burton H Jones E, Peters A Second somatosensory cortex and related areas Cerebral Cortex 1986 Plenum, New York
Tommerdahl M Favorov O Chiu J Whitsel B Optical intrinsic signal imaging of ipsilateral, contralateral, and bilateral forelimb inputs to cat SII Society for Neuroscience 34th Annual Meeting 2004 642 612
Carreras M Andersson S Functional properties of neurons of the anterior ectosylvian gyrus of the cat Journal of Neurophysiology 1963 26 100 126 14018881
Picard N Lepore F Ptito M Guillemot J Bilateral interaction in the second somatosensory area (SII) of the cat and contribution of the corpus callosum Brain Research 1990 536 97 104 2085764 10.1016/0006-8993(90)90013-2
Burton H Sinclair R Whang K Vibrotactile stimulus order effects in somatosensory cortical areas of rhesus monkeys Somatosensory Motor Research 1998 15 316 324 9875549 10.1080/08990229870727
Innocenti G Manzoni T Response patterns of somatosensory cortical neurones to peripheral stimuli. An intracellular study Archives Italiennes de Biologie 1972 110 322 347 4659134
Innocenti G Manzoni T Spidalieri G Cutaneous receptive fields of single fibers of the corpus callosum Brain Res 40 507 512 1972, May 26 5027175 10.1016/0006-8993(72)90153-9
Robinson D Electrophysiological analysis of interhemispheric relations in the second somatosensory cortex of the cat Exp Brain Res 18 131 144 1973, Sep 29 4766168
Simoes C Alary F Forss N Hari R Left-Hemisphere-Dominant SII Activation after Bilateral Median Nerve Stimulation NeuroImage 2002 15 686 690 11848711 10.1006/nimg.2001.1007
Hoechstetter K Meinck H Henningsen P Scherg M Rupp A Psychogenic sensory loss: magnetic source imaging reveals normal tactile evoked activity of the human primary and secondary somatosensory cortex Neurosci Lett 323 137 140 2002, Apr 26 11950512
Petit D Lepore F Picard N Guillemot J Bilateral receptive fields in cortical area SII: contribution of the corpus callosum and other interhemispheric commissures Somatosensory Motor Research 1990 7 97 112 2378194
Caminiti R Innocenti G Manzoni T The anatomical substrate of callosal messages from SI and SII in the cat Experimental Brain Research 35 295 314 1979, Apr 2
Manzoni T Barbaresi P Conti F Callosal mechanism for the interhemispheric transfer of hand somatosensory information in the monkey Behavioural Brain Research 1984 11 155 170 6704235 10.1016/0166-4328(84)90138-4
Manzoni T Conti F Fabri M Callosal projections from area SII to SI in monkeys: anatomical organization and comparison with association projections J Comp Neurol 252 245 263 1986, Oct 8 3782508 10.1002/cne.902520208
Barbaresi P Bernardi S Manzoni T Callosal connections of the somatic sensory areas II and IV in the cat J Comp Neurol 283 355 373 1989, May 15 2745745 10.1002/cne.902830305
Graziosi M Tucci E Barbaresi P Ugolini G Manzoni T Cortico-cortical neurones of somesthetic area SI as studied in the cat with fluorescent retrograde double-labelling Neurosci Lett 31 105 110 1982, Aug 16 7133546 10.1016/0304-3940(82)90100-8
Friedman D Jones E Burton H Representation pattern in the second somatic sensory area of the monkey cerebral cortex J Comp Neurol 192 21 41 1980, Jul 1 7410612 10.1002/cne.901920103
Juliano S Hand P Whitsel B Patterns of metabolic activity in cytoarchitectural area SII and surrounding cortical fields of the monkey Journal of Neurophysiology 1983 50 961 980 6631472
Bennett R Ferrington D Rowe M Tactile neuron classes within second somatosensory area (SII) of cat cerebral cortex Journal of Neurophysiology 1980 43 292 309 7381522
Gilson E Baddeley A Tactile short-term memory Quarterly Journal of Experimental Psychology 1969 21 180 184 5787977
Gescheider G Verrillo R Contralateral enhancement and suppression of vibrotactile sensation Perception & Psychophysics 1982 32 69 74 7133949
Essick G Whitsel B The capacity of human subjects to process directional information provided at two skin sites Somatosensory Motor Research 1988 6 1 20 3242341
Fitzgerald P Lane J Thakur P Hsiao S Receptive Field Properties of the Macaque Second Somatosensory Cortex: evidence for Multiple Functional Representations Journal of Neuroscience 24 11193 11204 2004, Dec 8 10.1523/JNEUROSCI.3481-04.2004
Tommerdahl M Whitsel B Franzen O, Johansson R, Terenius L Optical imaging of intrinsic signals in somatosensory cortex Somesthesis and the Neurobiology of Somatosensory Cortex 1996 Basel: Birkhauser Verlag AB 369 384
Grinvald A Real-time optical mapping of neuronal activity: from single growth cones to the intact mammalian brain Annual Review of Neuroscience 1985 8 263 305 3885828 10.1146/annurev.ne.08.030185.001403
Grinvald A Bonhoeffer T Malonek D Shoham D Bartfeld E Arierli A Hildesheim R Ratzlaff E Squire L, Weinberger N, Lynch G, McGaugh J Optical imaging of architecture and function in the living brain Memory Organization and Locus of Change 1991 NY: Oxford Univ Press 49 85
Grinvald A Lieke E Frostig R Hildesheim R Cortical point-spread function and long-range lateral interactions revealed by real-time optical imaging of macaque monkey primary visual cortex Journal of Neuroscience 1994 14 2545 2568 8182427
Tommerdahl M Delemos K Whitsel B Favorov O Metz C Response of anterior parietal cortex to cutaneous flutter versus vibration Journal of Neurophysiology 1999 82 16 33 10400931
Ebner T Chen G Use of voltage-sensitive dyes and optical recordings in the central nervous system Progress in Neurobiology 1995 46 463 506 8532849 10.1016/0301-0082(95)00010-S
Hassler R Muhs-Clement K Architektonischer Aufbau des sensomotorischen und parietalen Cortex der Katze Journal fur Hirnforschung 1964 6 377 420
McKenna T Whitsel B Dreyer D Metz C Organization of cat anterior parietal cortex: Relations among cytoarchitecture, single neuron functional properties and interhemispheric connectivity Journal of Neurophysiology 1981 45 667 697 7229676
Burton H Mitchell G Brent D Second somatic sensory area in the cerebral cortex of cats: somatotopic organization and cytoarchitecture Journal of Comparative Neurology 1982 210 109 135 7130474 10.1002/cne.902100203
| 15710047 | PMC552304 | CC BY | 2021-01-04 16:03:47 | no | BMC Neurosci. 2005 Feb 14; 6:11 | utf-8 | BMC Neurosci | 2,005 | 10.1186/1471-2202-6-11 | oa_comm |
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-71569137110.1186/1471-2334-5-7Research ArticleInternational outbreak of Salmonella Oranienburg due to German chocolate Werber Dirk [email protected] Johannes [email protected] Fabian [email protected] Treeck Ulrich [email protected] Gerhard [email protected] Steen [email protected] Anja M [email protected] Peter [email protected] Rita [email protected] Ian ST [email protected] Susanne C [email protected] Edda [email protected] Ekkehard [email protected] Andrea [email protected] Anja [email protected] Yvonne [email protected]äpe Helmut [email protected] Michael H [email protected] Andrea [email protected] Department of Infectious Disease Epidemiology, Robert Koch-Institut, Berlin, Germany2 Niedersächsisches Landesgesundheitsamt, Hannover, Germany3 Landesinstitut für den öffentlichen Gesundheitsdienst, Nordrhein-Westfalen, Germany4 Institute for Hygiene and Environment, Center for Infectious Disease Epidemiology, Germany5 Department of Bacteriology, Mycology and Parasitology, Statens Serum Institut, Copenhagen, Denmark6 Government Health Service Institute, Dillenburg, Germany7 National Reference Centre for Salmonella and other Bacterial Enteric Pathogens, Institute for Hygiene and Environment Hamburg, Germany8 National Reference Centre for Salmonella and other Bacterial Enteric Pathogens, Robert Koch-Institut, Wernigerode, Germany9 Enter-net surveillance hub, HPA Communicable Disease Surveillance Centre, London, United Kingdom10 Federal Institute for Risk Assessment, Berlin, Germany11 Foodborne, Waterborne and Zoonotic Diseases Division, PPHB, Health Canada12 Laboratory of Enteric Pathogens, Department of Microbiology, Helsinki, Finland13 Swedish Institute for Infectious Disease Control, Stockholm, Sweden2005 3 2 2005 5 7 7 30 8 2004 3 2 2005 Copyright © 2005 Werber et al; licensee BioMed Central Ltd.2005Werber 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
This report describes a large international chocolate-associated Salmonella outbreak originating from Germany.
Methods
We conducted epidemiologic investigations including a case-control study, and food safety investigations. Salmonella (S.) Oranienburg isolates were subtyped by the use of pulsed-field gel electrophoresis (PFGE).
Results
From 1 October 2001 through 24 March 2002, an estimated excess of 439 S. Oranienburg notifications was registered in Germany. Simultaneously, an increase in S. Oranienburg infections was noted in other European countries in the Enter-net surveillance network. In a multistate matched case-control study in Germany, daily consumption of chocolate (matched odds ratio [MOR]: 4.8; 95% confidence interval [CI]: 1.3–26.5), having shopped at a large chain of discount grocery stores (MOR: 4.2; CI: 1.2–23.0), and consumption of chocolate purchased there (MOR: 5.0; CI: 1.1–47.0) were associated with illness. Subsequently, two brands from the same company, one exclusively produced for that chain, tested positive for S. Oranienburg. In two other European countries and in Canada chocolate from company A was ascertained that also contained S. Oranienburg. Isolates from humans and from chocolates had indistinguishable PFGE profiles. No source or point of contamination was identified. Epidemiological identification of chocolate as a vehicle of infections required two months, and was facilitated by proxy measures.
Conclusions
Despite the use of improved production technologies, the chocolate industry continues to carry a small risk of manufacturing Salmonella-containing products. Particularly in diffuse outbreak-settings, clear associations with surrogates of exposure should suffice to trigger public health action. Networks such as Enter-net have become invaluable for facilitating rapid and appropriate management of international outbreaks.
==== Body
Background
Non-typhoidal Salmonella spp. is a substantive cause of human gastroenteritis in many parts of the world [1]. In Germany, non-typhoidal salmonellosis remains the most frequently reported infectious disease. For example, in 2001, the 77,185 Salmonella reports (incidence:94/100,000) received at the federal level by the Robert Koch-Institut (RKI) accounted for 31% of all notifications for the 54 notifiable conditions [2]. Salmonella enterica subspecies enterica serotype Enteritidis (S. Enteritidis) is the predominating serotype followed by S. Typhimurium. They represented 65% and 23% of the reported cases of non-typhoidal salmonelloses with known serotype in 2001. Thus, the remaining ~250 serotypes reported to the RKI in that year, including S. Oranienburg, accounted for only 12%.
In mid-October 2001, the National Reference Center for Salmonella and Other Enteric Pathogens (NRC) in Hamburg noted an unusual increase in the number of S. Oranienburg isolates received in October. At that time, no increase was noticeable in the national database for statutorily reportable infectious diseases; 50 S. Oranienburg notifications (median: 1 per week) had been registered for 2001. On November 19, the NRC in Wernigerode informed the RKI that it had received a S. Oranienburg isolate in September. The isolate was submitted by a private laboratory for serotyping and had come with the additional source information "confectionery sample". Upon inquiry, a large German chocolate manufacturer (company A), which produced a broad variety of chocolates and products made thereof, called the RKI on November 27, and confirmed that it had sent in the confectionery sample. According to company A, the positive sample originated from an in-house control of a chocolate product and the pertaining batch, due to be exported to the United States, was completely destroyed and not distributed. Notwithstanding, the number of statutory S. Oranienburg notifications had sharply increased and continued to rise. This report describes the epidemiologic, food safety, and microbiological investigations of this outbreak.
Methods
Epidemiologic investigation
Descriptive epidemiology
A standard exploratory questionnaire was distributed on 20 November 2001 via state health departments to all local health departments to aid the collection of data on food and environmental exposure from cases. In addition, local health departments were asked to immediately interview patients with newly reported S. Oranienburg infections about chocolate consumption in the seven days before disease onset, and also to send any remaining chocolate to a food safety laboratory.
International case-finding
A request was distributed to participants of the Enter-net surveillance network [6] on December 10, to see if other countries were affected or had relevant information.
Case-control study
On December 3, while exploration of patients were ongoing and results inconclusive, S. Oranienburg isolates from patients and from the in-house chocolate control were found to be indistinguishable by pulsed-field gel electrophoresis (PFGE). On the same day, a multistate case-control study was initiated and coordinated by RKI to test the hypothesis that at least one product from company A was associated with S. Oranienburg-infections. As we were denied a product list from company A, we resorted to the company's web-site and included in our food history evaluation all the products listed there. Some products from company A, e.g., bars of chocolate (brand A), were exclusively sold at a large chain of discount grocery stores (chain X). We found that the majority of chocolates sold at chain X were produced by company A. Therefore, for the analysis we constructed a variable for chocolate(s) purchased at chain X ("chain-X-chocolate") as a proxy for chocolate-products from company A because most patients could remember the flavor of the purchased chocolate, but seldom the brand name.
The hypothesis-testing questionnaire collected data on the consumption of chocolates, and some other foods, particularly those previously associated with outbreaks of S. Oranienburg in other countries [3-5]. Food-consumption history was evaluated for two different time periods, i.e., for the seven days prior to onset of symptoms in the case-patients and for the seven days before the interview. In addition, case-patients were asked about clinical symptoms, duration of illness and hospitalization. We defined a case-patient as a person with gastroenteritis starting after 1 October 2001 who had been reported with a S. Oranienburg infection to a public health department before December 6. Case-patients were excluded from the analysis, if they could have been secondary, i.e., if they reported to having had contact with a person with diarrhea in the seven days prior to symptom onset. Cases were selected from the national reportable database by simple random sampling; in Lower Saxony an attempt to interview all case-patients was launched. Case selection was done irrespective of whether patients also had been interviewed with an exploratory questionnaire. At least one age and telephone exchange-matched control subject was selected for each case-patient by sequentially adding 2 to each case-patients' telephone number. Control subjects were eligible if they were in the same age-group as their matched case-patient (0–5 years, 6–17 years, 18–59 years, 60 years or older), had no gastrointestinal symptoms after 1 October 2001, and had not traveled abroad in the seven days prior to the onset of symptoms of the matched case. Telephone interviews were conducted by state health departments, local health departments, and the RKI. Data were analyzed with Epi Info V6.04c (Centers for Disease Control and Prevention, Atlanta, GA).
Investigation by the food safety authorities
The local food safety authority inspected company A's production facility and took samples from already packaged ("in-house") chocolates, and from ingredients from its suppliers. Beginning December 11, a nationwide chocolate sampling of German chocolates in grocery stores was initiated by the Federal Ministry of Consumer Protection, Food and Agriculture, and was assisted by the Federal Institute for Risk Assessment (Bundesinstitut für Risikobewertung, "BfR"). On December 18, when a chocolate leftover from brand A tested positive, the investigations were tailored to German chocolates from company A. The BfR examined quantitatively four Salmonella positive chocolate leftovers and five chocolates from grocery stores using the most probable number technique [7].
Molecular subtyping
For comparison by the use of PFGE, S. Oranienburg isolated from stool specimens were sent to the NRC from laboratories in Germany and, on Enter-net request, from other countries. Furthermore, isolates from chocolates were submitted from state or private food laboratories in Germany as well as from Canada and the Czech Republic. PFGE-analysis was carried out according to Prager et al [8].
Results
Epidemiologic investigation
Descriptive epidemiology
In 2001, the RKI received 50 reports of S. Oranienburg up to reporting week 42, but 462 reports in the following 23 weeks (15 October 2001–24 March 2002, "outbreak period", figure 1). Thus, an excess of 439 S. Oranienburg reports were registered assuming a background rate of one report per week. The median age was 15 years (range: 0–92 yrs), 240 (52%) patients were female. There was no difference in the gender distribution within the single 10-year age-bands (P = .51). All 16 states of Germany reported S. Oranienburg cases during the outbreak period, with the highest incidence in the state of Schleswig-Holstein (1.78/100,000) bordering on Denmark. In total, 206 of the 440 German counties were affected with a median of one report and a maximum of 16 from the city of Hamburg during the outbreak period.
Figure 1 Disease onset (n = 362) of reported (n = 462) S. Oranienburg cases from reporting week 42/2001 to reporting week 12/2002 (outbreak period). The asterisk indicates the week when the (first) public warning was issued, and the incriminated chocolate products were recalled
Sixty exploratory questionnaires were received from eight states by the end of 2001. Forty-three (88%) of 49 patients with information on chocolate consumption had a symptom onset after 1 October 2001. Of the 34 who gave information as to where they bought the chocolate, 21 (62%) explicitly reported chain X. Some reported exclusively having eaten chocolate bars from brand A, among them a two-year-old child. On 18 December 2001, two months after the initial outbreak alert, a leftover consumed reportedly by this child in the seven days before symptom onset tested positive for S. Oranienburg.
International case-finding
On December 11, one day after the Enter-net request was distributed, Denmark was the first country to respond. Twelve cases of S. Oranienburg had been reported in Denmark from October 18 through December 10, compared with only two cases in 2001 before October 18. None of the clustered cases were travel-related [9]. Exploratory patient interviews had already been conducted at the time of the Enter-net request. At this point in time the investigators in Denmark, without knowledge of the German S. Oranienburg problem, independently suspected German chocolate bought in chain X as the source of the Danish outbreak. Chocolate was the only food item that all patients reported eating. The majority stated purchasing chocolate in chain X, which, although German, operates internationally [9]. In the next few days, an increase in the number of S. Oranienburg infections was reported from other countries such as Austria, Belgium, Finland, Sweden, The Netherlands (figure 2), and Canada [10]. As it became apparent that German chocolate was contaminated with S. Oranienburg, patient interviews were conducted that showed that several patients remembered having consumed German chocolate [10], except in Canada where all of the patients denied this consumption.
Figure 2 Number of S. Oranienburg infections reported to the Enter-net database from participating countries, except Germany
Case-control study
Sixty cases and 62 controls from five states were enrolled in the matched case-control study. Interviews were conducted with a median delay of 37 days (range: 12–64 days) after disease onset in case-patients. Twelve case-control pairs were excluded from chocolate-specific analysis, nine because the case-patients could have been secondary, one due to illness in September, and two where the control subjects could not remember whether they had eaten chocolate. Of the 48 cases and 50 controls that were analyzed, 24 (50%) case-patients and 32 (67%) control subjects were female, 22 (46%) case-patients were younger than 10 years. Ten (21%) of the case-patients reported to have suffered from bloody diarrhea and 14 (29%) were hospitalized (table 1). Results of the preliminary analysis were available on December 14. All 48 case-patients ate chocolate in the seven days before symptom onset, but this also applied to 43 (86%) of the control subjects. Three variables relating to the seven-day period prior to symptom onset of the case-patient were significantly associated with disease (table 2). The first variable was having shopped at chain X (matched odds ratio [MOR]: 4.2; 95% confidence interval [CI]: 1.2–23.0). The second variable was having consumed chain-X-chocolate (MOR: 5.0; 95% CI: 1.1–47.0). Eleven (25%) of 44 case-patients gave such an exposure history, six of whom reported having consumed either brand A chocolate exclusively (which could be inferred from the flavor of chocolate eaten), or were uncertain whether they had also eaten another kind of chocolate (n = 3). The third variable was having eaten (any kind of) chocolate on a daily basis (MOR: 4.8; 95% CI: 1.3–26.5). None of the other variables including all those relating to the seven days before the interview were significantly associated with illness.
Table 1 Clinical characteristics of S. Oranienburg cases (n = 48) analyzed in a case-control-study, December 2001
Symptoms Frequency, n (%)
Diarrhea 41 (85)
Fever > 38,5°C 27 (56)
Vomiting 17 (35)
Hospitalization 14 (29)
Antimicrobial medication 12 (25)
Visible blood in stool 10 (21)
Table 2 Significant risk factors for S. Oranienburg-associated illness in Germany, October-December 2001
Exposure Cases exposed (n/N, %) * Controls exposed (n/N, %)* MOR Exact 95% CI P-value
Ate chocolate bought at chain X 11/44 (25) 2/45 (4) 5.0 1.1, 47.0 0.04
Daily consumption of chocolate 22/48 (46) 12/50 (24) 4.8 1.3, 26.5 0.01
Shopped at chain X 31/44 (71) 19/45 (42) 4.2 1.2, 23.0 0.03
MOR = Matched odds ratio, CI = Confidence interval
* Proportion and percentages of cases and controls exposed ignoring matching
Public health action
On December 18, the finding of S. Oranienburg in a chocolate leftover of a patient led to an immediate public warning and recall of all chocolates of this brand with specific production numbers by company A. The recall was extended to other products from company A a few days later. Chocolates included in the German recall were promptly withdrawn from the market in other European countries as well as in Canada. In Canada, Finland, and Sweden, samples from withdrawn chocolates tested positive for S. Oranienburg [10].
Investigation by the food safety authorities
The local food safety authority in Germany did not identify hygienic deficiencies at the production facility. Samples obtained in the beginning of December 2001 from in-house chocolates (n = 12), as well as from cocoa (n = 3) and cocoa powder (n = 7) from a supplier of company A tested negative. This applied also to German chocolates sampled in grocery stores until 18 December (on that day a leftover tested positive). Overall, S. Oranienburg was found in 18 (5%) of 381 chocolates that were tested and reported to BfR during the outbreak period. S. Oranienburg was isolated from two different brands of company A; all positive chocolates were produced during the same week in August 2001. Estimates of the number of Salmonella in the tested chocolates ranged between 1.1 and 2.8 per gram.
Molecular subtyping
From October 2001 through January 2002, the NRC received 98 S. Oranienburg isolates from human cases of gastroenteritis originating in Germany (n = 52), Austria (n = 19), Belgium (n = 8), Canada (n = 6), Denmark (n = 4), The Netherlands (n = 4), Sweden (n = 4), and the Czech Republic (n = 1). Furthermore, 15 chocolate isolates were sent to the NRC for PFGE-analysis from Germany (n = 12), Canada (n = 2), and the Czech Republic (n = 1). They came from an in-house sample, from leftovers of chocolates consumed by patients in their incubation period, and from chocolates sampled in grocery stores in Germany. The PFGE profiles of S. Oranienburg isolates from patients with symptom onset after 1 October 2001 (outbreak period) in Germany and in the other countries mentioned above, except Canada, were indistinguishable (figure 3), but differed from S. Oranienburg isolates from German patients with symptom onset before October. All 15 chocolate isolates showed PFGE profiles indistinguishable from human isolates of the outbreak period.
Figure 3 Comparison of human S. Oranienburg isolates from the outbreak- period with strains of this serovar received sporadically before the outbreak by the use of PFGE (digested with XbaI, BlnI, and SpeI) lanes: 1–5: isolates from the outbreak period 6–13: isolates before the outbreak period S: molecular reference
Discussion
We describe an international outbreak of S. Oranienburg and present several lines of evidence that German chocolate from company A was the vehicle of infections. S. Oranienburg, a rare serotype in food as well as in humans in Germany, was isolated from retail-sampled chocolates of two brands produced by company A, from chocolate leftovers that had been consumed by patients before symptom onset, and from an in-house sample of company A obtained prior to the outbreak. In a case-control study, S. Oranienburg infection was significantly associated with the consumption of chain-X-chocolate (proxy for chocolate from company A) in the week prior to symptom onset, but not in the seven days before the interview. Case-patients were more likely than control subjects to report eating chocolate daily, likely indicating an increased probability of having been exposed to contaminated chocolate. Furthermore, patient isolates from the outbreak period shared a PFGE profile with isolates from chocolates but differed from isolates of patients who became sporadically diseased with S. Oranienburg before the outbreak. In addition, the food histories and microbiological results from S. Oranienburg patients in several other countries pointed to the same source [9,10].
Salmonella infections after consumption of contaminated chocolate, although rare, have been known since the 1960's [11]. Common to all reported chocolate-outbreaks, including ours, was that the epidemics were propagated in time, widely disseminated geographically, and affected large number of persons, predominantly children [12-17] (table 3). In addition, only very small numbers of Salmonella have been recovered from chocolates in these outbreaks, suggesting a very low infectious dose. Estimates of the number of S. Oranienburg cells per gram in this outbreak ranged from 1.1–2.8. However, we cannot exclude that the bacteria were unevenly distributed in the chocolate(products) and that those parts carrying many viable cells were not tested quantitatively. In chocolate, the low moisture (water activity aw: 0.4–0.5) and high sugar content does not favor bacterial growth, but significantly increases thermal resistance [11,18,19]. In addition, it has been speculated that the food matrix protects Salmonella against the acidic conditions of the stomach [11], which could imply that only few salmonellae are necessary to cause illness.
Table 3 Overview of published chocolate outbreaks due to Salmonella contamination
Year Country Serovar Vehicle* Source of contamination cfu/g No. of affected persons Peak of outbreak Age of
cases
1970 Sweden S. Durham Chocolate
products
(n>1), Cocoa powder / 110 Dec-May 53% ≤15 years
1973 – 1974 USA, Canada S. Eastbourne Chocolate balls
from Canada Cocoa beans 2.5 200 Dec-Feb. 3 years
(median)
1982 England,
Wales S. Napoli Chocolate bars
from Italy Unknown 2–23 272 May-Aug 58% ≤ 15 years
1985 – 1986 Canada S. Nima Chocolate coins
from Belgium Unknown / / Dec-Jan ?
1987 Norway,
Finland S. Typhimurium Chocolate
products,
(n = 3) from
Norway Avian
contamination speculated ≤1 349 Mar-May 6 years
(median)
2001– 2002 Germany,
other European
countries S. Oranienburg Two chocolate
brands from
Germany Unknown 1.1–2.8 439 Oct-Dec 15 years
(median)
* In each outbreak, the identified vehicles were traced to a single manufacturer
Company A produced several dozen tons of chocolate per day. All positive samples were produced in the same week. However, S. Oranienburg reports above an expected baseline of 1–2 reports per week in Germany were received for five months. The protracted nature of chocolate-associated outbreaks probably reflects both the long shelf-life of chocolate [20] and the long survival of Salmonella in these products [15,21]. S. Oranienburg was isolated from chocolates five months after manufacture. In an S. Napoli outbreak in England and Wales, this interval was 12 months. The number of affected persons reported in chocolate-associated Salmonella outbreaks has grown steadily over the years (table 3). Among other factors, this may parallel advances in food-processing technologies and improvements of national surveillance systems. Taken together, the chocolate industry faces a difficult situation because:
raw ingredients (e.g., cacao beans, milk powder) can carry Salmonella spp.,
the low water activity and high fat content in chocolate increases thermal resistance so that temperatures reached during chocolate production (even after considerable overheating [19]) do not necessarily destroy Salmonella,
a small number of Salmonella may be sufficient to cause disease,
even with low-level contamination, chocolate can affect large number of persons (often children) scattered over a wide area, and thus, has the potential to cause serious public health consequences.
It is noteworthy that the case-control study did not identify further products as risk factors. This applied also to the second contaminated brand of company A, which was included in the recall. The small proportion of study cases (25%) mentioning having eaten chain-X-chocolate lends support to the hypothesis of more contaminated brands (even from other manufacturers), which could be one explanation for the continuing case-occurrence. However, inaccuracies due to lack of brand awareness may have played a particular role in this outbreak, and the time-delay between disease onset and interview (median: 37 days) may have contributed to an inaccurate recall of cases and their guardians. Furthermore, cases with a disease onset in 2002 may have occurred as a result of a diminished impact of the public warning due to the Christmas season. For example, chocolate gifts received or given for Christmas may not have been thoroughly enough checked for best-before dates stated in the public warning.
Identification of vehicles in foodborne outbreaks can become difficult if the exposure is common. Consumption of a wide variety of German chocolates was reported by all case-patients (and 88% of explored patients), but also from 86% of the control subjects in the week prior to onset of illness. When groups are (nearly) universally exposed or a more specific hypothesis cannot be tested, often the best one can do is to establish a "dose-response-relationship" [22], i.e., unravel differences in the frequency of consumption of the incriminated food between cases and controls. Consequently, the observation that a higher proportion of cases reported eating chocolate on a daily basis added to the evidence that chocolate was the vehicle in this outbreak. Furthermore, the Danish data provided powerful supplementary evidence because consumption of German chocolate was particularly common in Germany but unusual in Denmark. Therefore, in multinational outbreaks, international collaboration provides an important means for disclosing the common source of infections, particularly when the contaminated food is very popular in one (likely the source) country (e.g., [23,24]). Multinational collaboration facilitated by Enter-net helped in preventing contaminated chocolate from entering the market in Canada, Finland and Sweden, thereby averting human illness. Furthermore, by rapid electronic exchange and comparison of PFGE profiles, the Canadian cluster of human cases could be classified as unrelated to this outbreak.
No source or point of contamination was identified. Hygienic deficiencies had not been observed at the production facility of company A, which used a modern production method. This included an extra heating of the milled cocoa beans by a special heat-steam treatment with 125–130°C as an additional safeguard. Samples from in-house chocolates and from ingredients tested negative. However, no environmental samples and very few samples of raw ingredients (n = 10) were obtained. In a S. Eastbourne outbreak in Canada/USA in 1973/74, 286 environmental samples and 98 chocolate samples from the production-line were examined. No in-line chocolate sample tested positive and overall only 6 (1.6%) samples were positive (bean processing rooms [n = 4], and samples from a molding plant [n = 2]] [12]. Therefore, source investigations in chocolate-outbreaks should include extensive sampling in the production environment to increase the likelihood of determining possible points of contamination. In this outbreak, it remains unclear whether the salmonellae survived the heating or (re)contaminated the chocolate afterwards. Consequently, long-term preventive measures to render chocolate-production safer could not be implemented.
An Enter-net urgent inquiry was sent after the first results of molecular subtyping suggested a link between human cases and chocolate from company A. Until then, investigators in Germany and Denmark had worked independently unaware that the outbreak extended outside of their respective countries. An earlier inquiry, ideally as early as an outbreak was suspected by the investigating countries (or as an increase was noted in the Enter-net database), may have speeded up hypothesis generating, and thus, may have helped in earlier identification of the vehicle, thereby preventing illnesses.
Finally, a public warning or recall of company A products did not occur before a brand A leftover tested positive although the confluence of information – the results of the case-control study, the Danish investigations, and the subtyping comparison between human isolates and the in-house sample – had already pointed to company A products as the source of the outbreak. Yet, no specific product or lot had been identified at the time. For this reason, a recall or a public warning were considered excessive responses by the German food safety authority. However, relying on microbiological confirmation in leftovers, if available for testing, is disputable (directionality of contamination unclear) and is dangerous in unopened food packages because critical time can elapse before a positive culture in food is obtained [25]. Therefore, it has been argued that public health action should be based on well-performed epidemiological investigations encompassing clear statistical associations with a specific exposure [25-27]. Such data are easiest to obtain when only one (ideally distinct) vehicle is involved that is infrequently consumed. Nonetheless, when food-production leads to more than one contaminated foodstuff, or when popular foods are vehicles of infection, hypothesis generating or testing can become intricate. Unfortunately, these instances appear conducive to affect large areas/populations. Therefore, we believe that clear associations even with surrogates of exposure suffice to justify public health actions (e.g., extensive source investigations) provided they plausibly fit other lines of evidence.
Conclusions
To our knowledge, this is the largest reported chocolate-associated outbreak, the seriousness being emphasized by the hospitalizations (29%) and self-reported bloody diarrhea (21%) of the study cases. Despite the use of improved production technologies, the chocolate industry continues to carry a small risk of manufacturing Salmonella-containing products. For the future, awareness among German food safety authorities must be heightened for the need to base public health action not exclusively on laboratory confirmation in food, and to conduct timely and comprehensive source investigations to enhance food safety in the long-run. The international scale of this outbreak shows how easy it is to distribute a contaminated product across many countries. This underlines the necessity of mechanisms for international surveillance and information dissemination such as Enter-net to ensure that international outbreaks can be dealt with rapidly and in an appropriate manner. Similar networks should be set up or, if existing, should be connected (possibly overseen by WHO), to allow rapid communications to other parts of the world when it is clear that a contaminated product is distributed internationally.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DW was the principal investigator of the German part of this outbreak; he carried out the statistical analysis of the case-control study, and drafted the manuscript. JD and FF were responsible for the outbreak investigation including the case-control study for the state ("Bundesland") of Lower Saxony. UvT was responsible for the outbreak investigation including the case-control study for the state ("Bundesland") of Northrhine-Westfalia. GF was responsible for the outbreak investigation including the case-control study for the state ("Bundesland") of Hamburg. SE conducted the Danish part of this outbreak investigation. AMH was responsible for the outbreak investigation including the case-control study for the state ("Bundesland") of Hesse. PR detected the outbreak and conducted microbiological investigations. RP and HT conducted the PFGE-analysis. ISTF coordinated the Enter-net inquiries and investigations. SB conducted interviews in the case-control study, designed the database and helped in the analysis. EB conducted quantitative analysis of Salmonella in chocolate. EW coordinated food safety investigations in this outbreak. AE conducted the Canadian part of this outbreak investigation. AS conducted the Finnish part of this outbreak investigation. YA conducted the Swedish part of this outbreak investigation. MHK was instrumental in the design of the case-control study. AA coordinated the German part of this outbreak investigation, broadened the scope of this outbreak by prompting an urgent Enter-net inquiry, and helped in designing the case-control study and drafting the manuscript.
All authors participated in revising the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors thank Christian Berghold (Austria) and Jean-Marc Collard (Belgium) for providing information on public health action in their countries in response to this outbreak. The authors also thank the Division of Foodborne, Waterborne and Zoonotic Diseases and the National Microbiology Laboratory of Health Canada and the Canadian Food Inspection Agency for their provision of data and specimens for this investigation; Ewa Kaske for editorial assistance.
The outbreak-investigation was conducted as part of the Project "Emerging Foodborne Pathogens in Germany" which is funded by the German Ministry of Education and Research under 01K10202.
==== Refs
Tauxe RV An update on Salmonella Health Environ Dig 1996 10 1 4
RKI RKI Salmonellen-Erkrankung Infektionsepidemiologisches Jahrbuch meldepflichtiger Krankheiten für 2001 2003 Mercedes-Druck Berlin 104 106
Gustavsen S Breen O Investigation of an outbreak of Salmonella Oranienburg infections in Norway, caused by contaminated black pepper Am J Epidemiol 1984 119 806 812 6720677
Hedberg CW Korlath JA D'Aoust JY White KE Schell WL Miller MR Cameron DN MacDonald KL Osterholm MT A multistate outbreak of Salmonella javiana and Salmonella oranienburg infections due to consumption of contaminated cheese JAMA 1992 268 3203 3207 1433759 10.1001/jama.268.22.3203
Tsuji H Hamada K Outbreak of salmonellosis caused by ingestion of cuttlefish chips contaminated by both Salmonella Chester and Salmonella Oranienburg Jpn J Infect Dis 1999 52 138 139 10508002
IST Fisher on behalf of the Enter-net participants The Enter-net international surveillance network-how it works Euro Surveill 1999 4 52 55 12631902
Garthright WE Most probable number of serial dilutions FDA Bacteriological Analytical Manual 1998 8 Gaithersburg (MD): AOAC International 1 12 App. 2
Prager R Liesegang A Rabsch W Gericke B Thiel W Voigt W Fruth A Karch H Bockemühl J Tschäpe H Clonal relationship of Salmonella enterica serovar Typhimurium phage type DT104 in Germany and Austria Zbl Bakteriol 1999 289 399 414
Ethelberg S International outbreak of Salmonella Oranienburg, October-December Part 2: Denmark Eurosurveillance Weekly 2002 6
Fisher IS de Jong B van Pelt W Aramini J Berghold C Matthys F Aramini J Berghold C Matthys F Powling J Siitonen A International outbreak of Salmonella Oranienburg, October-December Part 3: other countries Eurosurveillance Weekly 2002 6
D'Aoust J Salmonella and the chocolate industry J Food Protect 1977 40 718 727
Gastrin B Kampe A Nystrom KG Oden-Johanson B Wessel G Zetterberg B Salmonella durham epidemic caused by contaminated cocoa Lakartidningen 1972 69 5335 5338 4650740
Craven PC Mackel DC Baine WB Barker WH Gangarosa EJ International outbreak of Salmonella Eastbourne infection traced to contaminated chocolate Lancet 1975 1 788 792 48010 10.1016/S0140-6736(75)92446-0
D' Aoust J Aris BJ Thisdele P Durante A Brisson N Dragon D Salmonella eastbourne outbreak associated with chocolate J Inst Can Sci Technol Aliment 1975 8 181 184
Gill ON Sockett PN Bartlett CL Vaile MS Rowe B Gilbert RJ Dulake C Murrell HC Salmaso S Outbreak of Salmonella napoli infection caused by contaminated chocolate bars Lancet 1983 1 574 577 6131266 10.1016/S0140-6736(83)92822-2
Anon Salmonella nima in British Columbia CMAJ 1986 135 1286 3779559
Kapperud G Gustavsen S Hellesnes I Hansen AH Lassen J Hirn J Jahkola M Montenegro MA Helmuth R Outbreak of Salmonella typhimurium infection traced to contaminated chocolate and caused by a strain lacking the 60-megadalton virulence plasmid J Clin Microbiol 1990 28 2597 2601 2279988
Barrile JC Cone JF Effect of added moisture on the heat resistance of Salmonella anatum in milk chocolate Appl Microbiol 1970 19 177 178 5467125
Roberts TA Pitt JI Farkas J Grau FH (Eds) Microbial Ecology of Food Commodities Microorganisms in Foods 6 Blackie Academic & Professional 379 389 Chapter 10
Tamminga SK Beumer RR Kampelmacher EH van Leusden FM Survival of Salmonella eastbourne and Salmonella typhimurium in chocolate J Hyg (Lond) 1976 76 41 47 1107412
Barrile J Cone J Keeney P A study of salmonellae survival in milk chocolate Manufacturing confectioner 1970 50 34 39
Tauxe RV Hughes JM International investigation of outbreaks of foodborne disease BMJ 1996 313 1093 1094 8916684
Killalea D Ward LR Roberts D de Louvois J Sufi F Stuart JM Wall PG Susman M Schwieger M Sanderson PJ Fisher IS Mead PS Gill ON Bartlett CL Rowe B International epidemiological and microbiological study of outbreak of Salmonella Agona infection from a ready to eat savoury snack – I: England and Wales and the United States BMJ 1996 313 1105 1107 8916693
Shohat T Green MS Merom D Gill ON Reisfeld A Matas A Blau D Gal N Slater PE International epidemiological and microbiological study of outbreak of Salmonella Agona infection from a ready to eat savoury snack – II: Israel BMJ 1996 313 1107 1109 8916694
Tauxe RV Emerging foodborne diseases: an evolving public health challenge Emerg Infect Dis 1997 3 425 434 9366593
Majkowski J Strategies for rapid response to emerging foodborne microbial hazards Emerg Infect Dis 1997 3 551 554 9368788
Sobel J Griffin PM Slutsker L Swerdlow DL Tauxe RV Investigation of Multistate Foodborne Disease Outbreaks Public Health Rep 2002 117 8 19 12297677
| 15691371 | PMC552305 | CC BY | 2021-01-04 16:28:14 | no | BMC Infect Dis. 2005 Feb 3; 5:7 | utf-8 | BMC Infect Dis | 2,005 | 10.1186/1471-2334-5-7 | oa_comm |
==== Front
BMC NeurolBMC Neurology1471-2377BioMed Central London 1471-2377-5-31571793410.1186/1471-2377-5-3Research ArticleTai Chi and vestibular rehabilitation improve vestibulopathic gait via different neuromuscular mechanisms: Preliminary report McGibbon Chris A [email protected] David E [email protected] Stephen W [email protected] Donna M [email protected] Peter M [email protected] Steven L [email protected] Institute of Biomedical Engineering, University of New Brunswick, Fredericton, NB E3B5A3, Canada2 Biomotion Laboratory, Massachusetts General Hospital, Boston, MA 02114, USA3 MGH Institute of Health Professions, Boston, MA 02129, USA4 Harvard Medical School, Boston, MA 02115, USA5 Dept of Neurology, Massachusetts General Hospital, Boston, MA 02114, USA6 New England School of Acupuncture, Watertown, MA, 02472, USA7 Dept of Rehabilitation Medicine, Emory University School of Medicine, Atlanta, GA 30322, USA2005 18 2 2005 5 3 3 21 1 2004 18 2 2005 Copyright © 2005 McGibbon et al; licensee BioMed Central Ltd.2005McGibbon 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
Vestibular rehabilitation (VR) is a well-accepted exercise program intended to remedy balance impairment caused by damage to the peripheral vestibular system. Alternative therapies, such as Tai Chi (TC), have recently gained popularity as a treatment for balance impairment. Although VR and TC can benefit people with vestibulopathy, the degree to which gait improvements may be related to neuromuscular adaptations of the lower extremities for the two different therapies are unknown.
Methods
We examined the relationship between lower extremity neuromuscular function and trunk control in 36 older adults with vestibulopathy, randomized to 10 weeks of either VR or TC exercise. Time-distance measures (gait speed, step length, stance duration and step width), lower extremity sagittal plane mechanical energy expenditures (MEE), and trunk sagittal and frontal plane kinematics (peak and range of linear and angular velocity), were measured.
Results
Although gait time-distance measures were improved in both groups following treatment, no significant between-groups differences were observed for the MEE and trunk kinematic measures. Significant within groups changes, however, were observed. The TC group significantly increased ankle MEE contribution and decreased hip MEE contribution to total leg MEE, while no significant changes were found within the VR group. The TC group exhibited a positive relationship between change in leg MEE and change in trunk velocity peak and range, while the VR group exhibited a negative relationship.
Conclusion
Gait function improved in both groups consistent with expectations of the interventions. Differences in each group's response to therapy appear to suggest that improved gait function may be due to different neuromuscular adaptations resulting from the different interventions. The TC group's improvements were associated with reorganized lower extremity neuromuscular patterns, which appear to promote a faster gait and reduced excessive hip compensation. The VR group's improvements, however, were not the result of lower extremity neuromuscular pattern changes. Lower-extremity MEE increases corresponded to attenuated forward trunk linear and angular movement in the VR group, suggesting better control of upper body motion to minimize loss of balance. These data support a growing body of evidence that Tai Chi may be a valuable complementary treatment for vestibular disorders.
==== Body
Background
Vestibulopathy decreases whole body dynamic postural control and causes functional limitations [1-4]. Limitations in ambulation, dynamic balance and trunk control, for example, can lead to disability and contribute to decreased quality of life [5]. Vestibular rehabilitation (VR) is a well-accepted exercise program intended to remedy balance impairment caused by damage to the peripheral vestibular system [6]. Vestibulopathy impairs both the vestibulo-ocular reflex (VOR) and the vestibulo-spinal reflexes (VSR) [7]; hence, VR is designed to adapt the CNS to diminished vestibular input and to compensate for VOR and VSR loss, via gaze and balance retraining, which in turn should improve whole body dynamic stability [8-10].
Alternative therapies, such as Tai Chi (TC), have recently gained popularity as a treatment paradigm for a variety of human ailments, including balance impairment [11-13]. TC employs detailed regimens of physical movement, breathing techniques, and cognitive tools to strengthen the body, relax the mind, and balance the flow of life force [14]. The purported improvements in overall body control and mind-body focus with TC may offer an improved approach to treating balance dysfunction [11,13,15-17]. Where the explicit objective of many VR exercises is to improve gaze stability, TC emphasizes a 'soft' unfocussed gaze during the prescribed balance exercises. Although there is strong evidence that VR [4,8-10,18], and more recently TC [11,13,15-17], can benefit people with vestibulopathy, the degree to which gait improvements may be related to neuromuscular adaptations of the lower extremities are unknown. Thus, although the end result of both TC and VR should be improved dynamic stability during locomotor activities of daily living, including gait, we hypothesized that the mechanisms underlying such improvements should differ substantially. A better understanding of how balance dysfunction interventions affect lower extremity neuromuscular function during ADL may be useful for developing gait training exercises, and for providing a fuller understanding of the link between motor function and balance.
In this report, we present preliminary data from a blinded randomized clinical trial comparing the effects of VR and TC on gait function, joint kinetics and trunk kinematics in older adults. While the overall aim of this study was to determine the effects of balance rehabilitation on gait characteristics, we directed our efforts in this paper to better understand the relationship between mechanical energy transfers along the lower extremity kinematic chain (ankle-knee-hip), and forward and side-to-side velocity of the trunk.
Our general hypothesis was that adults with balance impairment from vestibulopathy who receive the VR or TC intervention will improve gait function as indicated by time-distance measures. Our specific hypotheses are based on the following rationale: Recent studies in healthy older adults [19-21], with general impairments such as strength loss [22], or pathologies such as knee arthritis [23,24], show that the hip musculature often aids, or compensates for, ankle plantar-flexor muscles in providing both forward propulsion and trunk stability [25]. These prior studies have shown a consistent decline in plantar-flexor muscle power during gait, with an increase in hip muscle power, in older adults with, and without, known mobility impairments. As shown recently by Neptune and colleagues [26], the ankle plantar flexors contribute significantly to both forward propulsion and vertical trunk stability. Thus, one would expect that improvements in lower extremity motor control, aimed at increasing forward propulsion and trunk stabilization, would be represented by decreases in hip mechanical energy expenditures and increases in ankle (and perhaps knee) mechanical energy expenditures, and be directly related to improved kinematics of the trunk.
Based on the above rationale, and the specific treatment programs described in the following sections, we hypothesized: 1) TC treatment will improve lower extremity motor control by increasing ankle mechanical energy expenditure (MEE) contribution, and decreasing hip MEE contributions, to total energy of the leg, more than VR; and 2) that improved trunk control following TC will be positively correlated with improvements in lower extremity motor control, while improvements in trunk control following VR will not be correlated with improvements in lower extremity motor control. The latter may indirectly implicate other mechanisms, most likely improvements in VOR/gaze stability [27].
Methods
Subjects
Fifty-three patients with balance impairment due to vestibular hypofunction were recruited and randomized into two treatment groups: VR, a group vestibular rehabilitation intervention, and TC, a group Tai Chi exercise. Of the 53 patients admitted, 15 dropped out or were excluded prior to completing the intervention. The majority of drop outs were due to a new medical condition unrelated to the study preventing participation (e.g. fractured foot, acute back pain) or due to the sudden need to care for an ill family member. Two subjects were eliminated because of lack of force plate data to use in the data analysis (see Gait Analysis section for more detail). Of the 36 subjects remaining for analysis, 17 subjects were randomized into the VR treatment group (12 unilateral and 5 bilateral) and 19 subjects in the TC group (11 unilateral, 8 bilateral).
Unilateral or bilateral vestibular hypofunction (UVH or BVH) diagnoses were obtained as previously described [10]. Briefly, all patients had gait unsteadiness without evidence of central nervous dysfunction. All patients were referred to the study because they had locomotor instability for which they sought treatment from project physicians. Patients with bilateral vestibular hypofunction had bilaterally decreased caloric responses (total slow phase velocity of ≤10 degrees·sec-1 for the sum of right and left ear caloric stimulation at 27 and 44°C warm water stimulation of both ears and ≤8°·sec-1 slow phase velocity for the sum of 35 cc of ice water stimulation in each ear) and decreased VOR gains during passive rotational testing at up to 50°·sec-1 (at least 2.5 SD below normal mean values at frequencies of rotational testing from .01 to 0.5 Hz). Patients with unilateral vestibular hypofunction had damage only on one side, including at least 30% unilaterally reduced caloric response, positional nystagmus while lying with the damaged ear down, and/or confirmatory abnormalities on rotational testing (mildly decreased low frequency gains, increased phase leads and asymmetrical rotation induced nystagmus, i.e., decreased vestibular time constant). Patients with bilateral deficits are typically more disabled than are those with unilateral deficits. The average time post-onset of vestibulopathy for the 36 subjects included in the analysis was 3.05 years (range 0.58 – 12 years). All subjects were community dwelling and reported varying degrees of limitations in locomotor activity. Twenty of the 36 subjects were female, and the 36 subjects were 59.5 ± 11.5 years old (range, 41–81), 1.70 ± .11 m tall and 83.6 ± 16.5 kg in weight (breakdown by treatment group is shown in the Results section). All subjects had at least one course of VR since the time of onset of their vestibular symptoms. Inclusion criteria required that each subject did not have VR for >6 months from study enrollment. The testing protocol was approved by MGH institutional review board, and all subjects provided written informed consent according to institutional guidelines on human research.
Interventions
The VR and TC treatment interventions were provided in a total of six (3 VR and 3 TC groups) small groups with an average of 8 subjects per group. Each intervention program met once weekly, on separate weekdays, for 10 weeks in the same exercise room. The weekly sessions for both intervention groups lasted approximately 70 minutes. Each treatment program was lead by the same instructor for the three treatment cohorts. The instructors were blinded to the exercises provided to the other treatment program. One or two assistants were available for each session for all treatment groups to insure participants' safety. All treatment sessions included time to: 1) review material introduced in prior sessions; 2) introduce new material; 3) ask questions and share personal experiences or concerns regarding the practices; and 4) cool down and rest.
Tai Chi intervention
The TC intervention incorporated three objectives outlined in a balance-related TC program developed by Wolf and colleagues [12,17]. First, it emphasized movements that are easily comprehensible. Second, the sequence of exercises introduced reflected a progression that increasingly challenges postural stability, with a shift in weight bearing from bilateral to unilateral support. Third, the program emphasized increasing the magnitude of trunk and arm rotation while diminishing the base of support. The five specific TC movements employed in this study – 'raising the power', 'withdraw and push', 'wave hand like clouds', 'brush knee twist step', and 'separate right and left legs' – are described and illustrated in a training manual for the Cheng Man-Ch'ing's Yang-style short form [28]. In addition to these five formal TC movements, the intervention also included a short set of traditional TC warm-up exercises focused on loosening up the physical body and incorporating mindfulness and imagery into movement. Warm-up exercises included: gentle stretches sequentially targeting the shoulders, necks, arms and legs; a torso stretching exercise that coordinated weight shifts with rotations of the trunk and passive arm swinging; and a 5 minute seated meditation emphasizing relaxed diaphragmatic breathing. Approximately 20 minutes of each class was devoted to warm-up exercises, of which 10–12 minutes was spent in standing. Following an additional 40 minutes of formal TC practice, 10 minutes was allowed for group discussion.
Vestibular rehabilitation intervention
The VR intervention used in this study was a comprehensive exercise program designed to improve the problems specifically associated with damage to the peripheral vestibular system [4,6,10,29]. Each treatment session focused on the three main objectives of the VR intervention. Firstly, a series of eye-head coordination exercises were performed to promote gaze stability during both quiet standing and dynamic functional activities (such as combining movement of an image across the retina with head movement). Subjects progressed to performing these eye-head exercises with the target on a more complex background (to simulate real world activities), at increasingly faster speeds of head movements (eg, 2–3 Hz), and during more dynamic standing and locomotor activities. A second treatment objective included VOR training in a group format with subjects standing. Target foveation objects (words of various sizes) were fixed to a large checkerboard background covering one wall of the exercise room, enabling us to provide the appropriate visual stimuli. Patients were progressed by increasing the speed (frequency and amplitude) of head movement to train the VOR more appropriately at speeds consistent with everyday locomotor activities. The third main component of the VR program was upright balance retraining exercises that enhance the use of various sensory cues for gaining posture control [8,29,30]. Examples of these exercises include subjects maintaining their balance while decreasing their base of support (such as standing on one foot, marching or walking heel to toe) and while walking on various floor surfaces (such as the pliable surfaces of a foam mat). Subjects were further challenged by incorporating head and trunk movements or with eyes closed during standing and walking exercises. All exercises were performed in an upright position (either standing or during locomotion) based on individual tolerance. If required, a seat was provided and the exercise performed in a seated position until the subject was able to tolerate the activity in standing. Each group treatment session lasted 60 minutes allowing 20 minutes for each of the 3 main exercise components. There was an additional 10 minutes for questions and answer time and for assistance with individual progression of home exercise programs.
Gait analysis
Subjects performed two-to-four gait trials along a 10 m level walkway at baseline testing, and at testing following the intervention program, at their freely selected pace upon the command "Please walk as you normally would, as if taking a brisk walk in the park". Body segment kinematics were acquired at 150 Hz with a four-camera Selspot optoelectric tracking system (Selective Electronics, Partille, Sweden), used to acquire position and orientation data of 11 segments (both feet, shanks, thighs and arms, and the pelvis, trunk and head). Collection of kinematic data and processing is described in more detail elsewhere [31]. Kinetic data consisted of ground reaction forces acquired from two adjacent piezoelectric force platforms (Kistler Instruments, Winterthur, Switzerland), synchronously sampled with body segment kinematic data at 150 Hz. Subjects were required to have at least one good gait trial, both at baseline and post-intervention testing sessions, to be included in the data analysis. A "good gait trial" was one that satisfied the following criteria: 1) one foot was required to be in whole contact with only one or both force platforms without interference from the other foot, 2) all body segments were visible, and tracked without artifact, during the stance portion of gait. Two subjects were excluded on the basis of failing one or both of the above criteria.
Data analysis
Parameters selected for data analysis consisted of dynamic gait function (time-distance measures), lower extremity neuromuscular control (sagittal plane mechanical energy expenditures, MEE), and trunk stability (sagittal and frontal plane kinematics).
Dynamic gait function
Gait function was assessed with standard time-distance measures [4,32], including: gait speed, step length, step width and stance duration. Gait speed was measured as the average anterior-posterior velocity component of the whole-body center of gravity over stance phase of gait. Step length was measured as the anterior-posterior distance between right and left ankle centers when each foot was flat on the floor during its respective mid stance portion. Stance duration was measured as the time elapsed between heel strike and toe off (duration of stance phase), and step width was measured as the lateral distance between ankle centers at the foot positions used for step length calculation.
Lower extremity neuromuscular function
Neuromuscular function of the lower extremities was assessed using mechanical energy expenditure (MEE) of the ankle, knee and hip, relative to the total MEE of the leg, and were computed as described previously [23,33]. Briefly, the mechanical power profile of the joint, the scalar product of net joint moment and angular velocity, is integrated over specific time intervals to arrive at mechanical energy expended, MEE, or work done. The intervals are defined by periods of concentric transfer (MEE(+), the amount of concentric mechanical energy expended with segment-to-segment energy transfer), eccentric transfer (MEE(-), the amount of eccentric mechanical energy expended with segment-to-segment energy transfer), and no-transfer (MEE(o), the amount of concentric and eccentric energy expended without segment-to-segment energy transfer) conditions. The total joint MEE is simply the sum of these components (MEE(t) = MEE(+) + MEE(-) + MEE(o)). Leg MEE is the sum of joint MEE (ankle, knee and hip) for different conditions, or totals. Percentage contribution of joint MEE (for each condition and total) to leg MEE (for each condition and total) was then calculated.
Trunk stability
Trunk stability was assessed using kinematics of the trunk center of mass [34], and consisted of anterior-posterior trunk velocity (peak and range) as well as lateral trunk velocity (peak and range); sagittal plane angular (pitch) velocity of the trunk (peak and range) and frontal plane angular (roll) velocity of the trunk (peak and range). The rationale for using kinematic measures of trunk stability, instead of overall stability (such as whole-body CG sway or kinematics), however, was to enable us to examine the relationship between lower extremity neuromuscular function (using the mechanical energy analysis as such a measure) and the kinematics of the upper body, as a mass to be controlled apart from the legs. Peaks and ranges were taken from stance phase of the same leg used for the mechanical energy analysis described above.
Statistical analysis
One-way ANCOVA was used to compare change scores between the two groups, using the baseline values as covariates. Paired samples t-tests compared the change in each variable for each group from baseline to post-intervention testing. Pearson correlations were used to examine associations between change scores in lower extremity MEE and change scores in trunk velocities, for each treatment group. Due to the large number of comparisons in this exploratory study, a Ryan-Holm step down Bonferonni approach was used to control for type I errors [35], using a family-wise α = .05. Using this scheme, families of three members (MEE(+), MEE(-) and MEE(o) contributions) required significance at α = .017 for at least one comparison, α = .025 for the second comparison, and α = .050 for the third comparison. Families of four members (anterior-posterior peak and range, and lateral peak and range of trunk velocity) required significance at α = .013 for at least one comparison, α = .017 for the second comparison, and so on. All p-values given will be unadjusted, but the adjusted α is given for each comparison where appropriate. SPSS for Windows (v10, SPSS Inc. Chicago, IL) was used for all statistical analyses.
Results
The two groups were not different in age (VR: 56.9 ± 11.6 yrs; TC: 61.7 ± 11.3 yrs; p = .223), height (VR: 1.69 ± .11 m; TC: 1.71 ± .11 m; p = .712) or weight (VR: 81.1 ± 19.3 kg; TC: 85.8 ± 13.6 kg; p = .399). There was no significant difference in proportion of UVH and BVH in the treatment groups (Chi-square, p = .429), or proportion of men and women in the treatment groups (Chi-square, p = .709).
There were no significant between-groups differences (using ANCOVA for controlling for baseline differences) for any of the variables examined. There were, however, significant changes pre- and post-treatment within each group. These latter results appear to suggest that clinically important differences in each group's response to the therapies exist. Thus, the remainder of the results presented will focus on the within-groups comparisons.
Time distance measures
Both groups improved (unadjusted α = .05) following intervention in time-distance measures (see Table 1), with the TC group showing greater overall improvements; the VR group improved significantly in stance duration (p = .044) and step length (p = .045), but not in gait speed (p = .060) or step width (p = .390); the TC group improved in gait speed (p = .009) and step length (p = .010), but not in stance duration (p = .055) or step width (p = .313).
Table 1 Time distance measures before and after intervention.
Variable Baseline Value Post-Treatment Value
Mean Standard Dev Mean Standard Dev p-value*
VR Gait speed (m/s) 1.180 .312 1.235 .229 .060
Step length (m) .616 .119 .639 .116 .045
Stance duration (s) .667 .065 .653 .047 .044
Step width (m) .093 .045 .096 .040 .390
TC Gait speed (m/s) 1.090 .275 1.170 .261 .009
Step length (m) .582 .110 .612 .118 .010
Stance duration (s) .715 .089 .684 .055 .055
Step width (m) .109 .046 .114 .042 .313
* Within-groups paired t-test significance, based on unadjusted α = .05.
Mechanical energy expenditures
Figure 1 shows the changes in joint and leg MEE(t) for each joint, and the sum of all the joints (leg). Although the total leg MEE(t) change was similar, the distribution of joint MEE(t) were quite different for the two treatment groups. Comparison of the change in percent contribution of MEE for each transfer condition for each joint to leg MEE showed that only the TC group had significantly reduced (p < .001, adjusted α = .017) relative hip concentric MEE(+) and increased (p = .019, adjusted α = .025) relative ankle concentric MEE(+), following training. These data are shown in Figure 2.
Figure 1 Change scores in ankle, knee and hip total MEE(t) and leg total MEE(t) for VR and TC groups (in J %BW). Error bars represent 95% confidence intervals on the mean.
Figure 2 Change scores in percent contribution of ankle, knee and hip concentric MEE(+) to leg concentric MEE(+) for VR and TC groups (in J %BW). Error bars represent 95% confidence intervals on the mean.
Trunk kinematics
TC group had significantly increased (p = .009, adjusted α = .013) peak trunk forward velocity during stance phase of gait following treatment, while the VR group's increase was similar though not statistically significant (p = .018, adjusted α = .013). There were no significant changes in forward velocity range, nor were there significant changes in peak or range of lateral trunk velocity for either group. The VR group, however, did show a significant increase in peak trunk angular velocity (p = .007, adjusted α = .017) and range of trunk angular velocity (p < .001, adjusted α = .013) in the frontal plane. There were no significant changes in trunk angular velocity in the frontal plane for the TC group, and neither group showed significant changes in peak and range of trunk angular velocity in the sagittal plane. These data are summarized in Figure 3.
Figure 3 Change scores in trunk velocity. (a) Linear velocity: anterior-posterior (A/P) velocity peak and range, and medial-lateral (M/L) velocity peak and range; (b) Angular velocity: pitch (sagittal plane) velocity peak and range, roll (frontal plane) velocity peak and range. Error bars represent 95% confidence intervals on the mean.
Relationships between MEE and trunk kinematics
Correlation analysis between changes scores in leg MEE and trunk kinematics revealed significant relationships for both treatment groups. Most striking was the consistent directional relationship between trunk velocity and leg MEE within each of the treatment groups. For the VR group, changes in range and peak of forward velocity of the trunk was negatively correlated with changes leg MEE (range: r = -.536, p = .013, adjusted α = .013; peak: r = -.431, p = .042, adjusted α = .017). For the TC group, however, changes in range and peak of forward velocity of the trunk was positively correlated with changes in leg MEE (range: r = .620, p = .003, adjusted α = .013; peak: r = .451, p = .026, adjusted α = .017). Figure 4 shows scatter plots depicting the positive and negative relationships between change scores in leg MEE and trunk velocity range for TC and VR groups, respectively. There were no significant relationships detected between change in joint or leg MEE and change in lateral linear velocity of the trunk, nor in sagittal or frontal plane angular velocity of the trunk.
Figure 4 Change scores in trunk forward velocity range (in m/s) versus change scores in total leg MEE(t) (in J %BW) for VR (top plot) and TC (bottom plot) groups. Dashed lines represent the 95% confidence intervals on the mean.
Discussion and conclusions
Little is known about the mechanisms of improved balance and postural control following rehabilitation in people with vestibulopathy. Although VR has shown promise for improving patients balance and gaze stability [4,9,10,29,36-40], just over 65% of people treated respond to the therapy [10]. Improvements in function are not ubiquitous with VR treatment. Alternative therapies, such as TC, offer a complementary approach to improving balance and postural control by teaching body control and awareness [12,15]. The purpose of the present study was to examine lower extremity neuromuscular function during gait in patients receiving either VR or TC treatment, and to examine how changes in neuromuscular patterns relate to changes in trunk control. Our first hypothesis, that patients receiving TC treatment will improve lower extremity motor control by increasing ankle MEE contribution and decreasing hip MEE contribution more so than the VR patients was not supported by analysis of between-group differences. However, in examining the within-groups changes (pre versus post-intervention), some potentially important biomechanical observations were made. Our second hypothesis, that trunk control in patients receiving TC will be positively correlated with improvements in lower extremity motor control, but trunk control in those receiving VR will not be positively correlated with improvements in lower extremity motor control, was supported.
Although an overall improvement in gait function (as indicated by time-distance measures) for both treatment groups was observed, confirming our general hypothesis, our data suggest that the mechanisms underlying those improvements differ, and appear to be linked to differences in neuromuscular responses of the lower extremities to the treatment programs. Specifically, our data suggest that changes in the relative contribution of individual joints to total leg mechanical energy expenditure (MEE), and the relationship between changes in lower extremity mechanical energy expenditure and changes in upper body kinematics, are different between TC and VR interventions (Figure 4). Further, these data highlight the importance of assessing gait not only with time-distance functional gait measures (Table 1), but also with measures that assess neuromuscular function of the lower extremities and control of the body's most massive segment, the trunk.
We found that TC patients significantly increased the contribution of ankle MEE to total leg MEE and decreased contribution of hip MEE to total leg MEE following the treatment program, while the VR group showed no significant change in ankle or hip MEE contributions following intervention (Figure 2). Although the changes were not statistically different between groups, the within groups comparisons suggest that clinically important trends may nonetheless be present in terms of biomechanical responses to the different therapies. Figure 1 shows the total joint MEE(t) (sum of all transfer components) for ankle, knee and hip, and total leg MEE(t), for each group. The increases in leg MEE(t), which were similar for both groups, were apparently achieved by different neuromuscular adaptations of the individual joints. Where the VR group appear to increase total MEE(t) of each joint to gain a total MEE increase of the leg, the TC group show a distinctive pattern of substituting ankle plantar-flexor contribution for hip extensor/flexor contribution. Prior studies on the relative roles of ankle and hip kinetics in gait [19,20,24] suggest that the result observed for TC patients indicates a trend toward a reduction in hip compensation and increased use of ankle muscles to provide both propulsion and stability.
Figure 2 shows the percent contribution of the changes to the concentric leg MEE(+) for the changes in concentric MEE(+) of individual joints. While both groups decreased the relative contribution of concentric hip MEE(+), only the TC group increased the contribution of ankle concentric MEE(+), while the VR group increased the contribution of knee concentric MEE(+). Concentric energy transfer represents the energy expended by muscles in concentric contraction when energy is being transferred between segments. Because concentric contraction represents work being done by the muscles (as opposed to eccentric contraction, which is work being done on the muscles), we can interpret the above finding as meaning that, for the TC group, a greater proportion in the change in concentric work done by the leg muscles is attributed to the change in concentric work of ankle plantar-dorsiflexors, while for the VR group, this contribution decreases.
One possible reason concentric ankle MEE contribution may have increased significantly in the TC group, but not the VR group, is because the TC and the warm-up exercises improved ankle flexibility. Tight ankles (limited range of motion, ROM) may preclude the optimal structural alignment to coordinate mechanical energy sufficiently to increase propulsion [41], perhaps at the expense of trunk stabilization. Ankle function is important for balance corrections in both healthy elderly and vestibulopathic subjects [42-44]. A study by Van Deusen et al. [45] found that Tai Chi-like exercises for elders with arthritis resulted in a significant increase in ankle plantar flexion; this finding supports the above contention that the TC group in our study may have increased ankle MEE contribution as a result of increased ankle ROM, ankle moment, or both. The tight coupling between ankle and hip power in gait [19] would also explain the neuromuscular adaptive decrease in hip MEE contribution. Given the importance of ankle-plantar flexors in both propulsion and trunk stability, we conclude that TC teaches optimization of MEE in an effort to control the trunk while improving lower extremity function. The relationship between lower extremity MEE and trunk kinematics for the two treatment groups lends further credibility to this conclusion.
As shown in Figure 4, the relationship between change in leg MEE and change in the range of forward trunk velocity was positive for the TC group, and negative for the VR group. Similar relationships were also observed between change in leg MEE and change in peak forward trunk velocity. The observed direct relationship for the TC group suggests that the redistribution of power among ankle, knee and hip joints, which resulted in a net increase in the total MEE of the leg, enabled these patients to attain a faster gait. This observation is expected based on the principles of TC, which emphasize a vertical alignment integrating the head, torso, hips and legs. This concept of integrated alignment is reflected in phrases from the TC classics such as ". suspend the spine like a necklace of pearls" and "movements are initiated in the feet, steered by the waist and administered through the hands." [46]. In contrast, for the VR group, however, the increase in leg MEE was associated with a decrease in both peak and range of trunk velocity. This finding suggests that VR subjects, when increasing power generation/absorption with their lower extremities, reduce trunk oscillations during gait, possibly as a way to stabilize the trunk and head. This corrective procedure may not be necessary for TC subjects as they learned to move the trunk more proportionately to total lower extremity MEE, without need to explicitly attend to additional factors or mechanisms to stabilize the head. Although speculative, these scenarios correspond with the observed high positive correlation between change in leg MEE and change in trunk velocity peak and range seen after TC training but not after VR rehabilitation.
Because the VR exercise program may increase subjects' awareness of eye and head movement strategies that cause dizziness and instability [8,47], a more rigid head and trunk strategy during dynamic activities such as gait would be expected. The VR program's balance retraining exercises do not emphasize dynamic whole body movement patterns that improve overall postural control [10]. Subjects practice maintaining balance in challenging postures (narrow base of support such as feet together and one-legged standing still) using a variety of self selected movement patterns. It is probable that subjects would make the trunk more rigid to lessen head movement during these tasks. The VR group's decrease in trunk velocity range with increase in mechanical energy of the lower extremities during gait appears to support this explanation. Within the TC group, the subjects practice series of movement patterns that include elements of controlled trunk rotation without instruction on eye fixation. The training of smoothly transitioning body segment motions may provide these subjects a different mode of compensation for their instability. Practice of the TC movements may promote more natural trunk movements similar to healthy persons as shown in our biomechanical findings during gait.
Although the preliminary results presented here suggest the lower extremities may play an important role in the ability of vestibulopathic patients to improve gait function, several limitations of the present study may prevent broad generalization of the results. Our small sample size was perhaps the most important limitation. It is probable that the lack of between-group significant changes, particularly in light of many significant within group differences (pre-post intervention), was due to high variances obscuring group mean differences. Indeed, a larger sample size for controlling type II errors (increasing power), and better control of type I errors for multiple statistical tests, is warranted for future full-scale studies. The large age range within groups may have also contributed to high variability, but note that more heterogeneous samples in fact enhance external validity, including generalizability, of the results. We also did not include a no-treatment control group in the experimental design of this preliminary study. Because vestibulopathic patients may learn to compensate spontaneously, a no-treatment or sham-treatment group would be necessary to determine if changes in gait function are truly a result of the interventions. This is unlikely, however, given the inclusion criteria that all subjects must have had stable symptoms for 6 months and were on average 3 year post-onset of vestibulopathy. Given that both groups improved gait function in our randomized design comparing substantially different interventions suggests that the effects observed were not spurious. It must be recognized, however, that assumed improvements in function, via increased gait speed for example, may be limited [48,49]. As well, we only analyzed the mechanics of the lower extremities in the sagittal plane. It is highly likely that compensations for lower limb power impairments occurred in frontal and transverse planes as well. Also, the different number of patients in each group having a diagnosis of UVH and BVH is a potential limitation. Although there were no significant differences in proportion of UVH and BVH between the two treatment groups, that the BVH patients were much more disabled than the UVH patients in this study may be important, even when the difference in proportions of diagnostic categories (BVH or UVH) within treatment groups is small. Lastly, although there were no significant differences in age and gender distribution between treatment groups, a larger study sample would allow such subgroup effects to be studied.
We conclude that VR and TC can successfully improve gait function, as determined by common time-distance measures, in patients with vestibulopathy. We further conclude, however, that TC improves lower extremity motor control more than VR, by selective redistribution of joint energetics, which appears to engender a more vigorous gait and better trunk control. TC, as a complementary treatment to VR, may allow for better control of the trunk through reorganization of lower-extremity motor patterns, elicited from the flowing, controlled TC exercises.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All authors participated in the overall study design, contributed to the interpretation of data and writing/editing of the manuscript, and have read and approved the final manuscript. CAM conceived the hypotheses for this manuscript and carried out the data analysis; DEK was the principal investigator of the project; SWP was the neurologist associated with the project; DMS conducted the patient testing and assisted in development of the vestibular rehabilitation program; PMW conducted the Tai Chi intervention; and SLW was the project consultant.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Supported in part by the NIH R21 AT00553-01. The authors thank Laura Busick, PT, Kathleen M. Gill-Body, DPT, MS, NCS, and Dr. Ted J. Kaptchuk, OMD for their assistance with the research design and interventions, and Lara Asmundson, MS, PT and Dov Goldvasser, MscE, for assistance with data collection.
==== Refs
Ishikawa K Edo M Yokomizo M Terada N Okamoto Y Togawa K Analysis of gait in patients with peripheral vestibular disorders ORL J Otorhinolaryngol Relat Spec 1994 56 325 330 7838484
Ishikawa K Edo M Yokomizo M Togawa K Characteristics of human gait related variables in association with vestibular system disorders Acta Otolaryngol Suppl 1995 520 Pt 1 199 201 8749118
Baloh RW Fife TD Zwerling L Socotch T Jacobson K Bell T Beykirch K Comparison of static and dynamic posturography in young and older normal people J Am Geriatr Soc 1994 42 405 412 8144826
Krebs DE Gill-Body KM Riley PO Parker SW Double-blind, placebo-controlled trial of rehabilitation for bilateral vestibular hypofunction: preliminary report Otolaryngol Head Neck Surg 1993 109 735 741 8233513
Brandt T Bilateral vestibulopathy revisited Eur J Med Res 1996 1 361 368 9360934
Herdman SJ Vestibular Rehabilitation 1994 Philadephia, F.A. Davis Company
Grossman GE Leigh RJ Instability of gaze during locomotion in patients with deficient vestibular function Ann Neurol 1990 27 528 532 2360793 10.1002/ana.410270512
Herdman SJ Role of vestibular adaptation in vestibular rehabilitation Otolaryngol Head Neck Surg 1998 119 49 54 9674514
Horak FB Jones-Rycewicz C Black FO Shumway-Cook A Effects of vestibular rehabilitation on dizziness and imbalance Otolaryngol Head Neck Surg 1992 106 175 180 1738550
Krebs DE Gill-Body KM Parker SW Ramirez JV Wernick MR Vestibular rehabilitation: Useful but not universally so Otolaryngol Head Neck Surg 2003 128 240 250 12601321 10.1067/mhn.2003.72
Hain TC Fuller L Weil L Kotsias J Effects of t'ai chi on balance Arch Otolaryngol Head Neck Surg 1999 125 1191 1195 10555688
Wolf SL Coogler C Xu T Exploring the basis for Tai Chi Chuan as a therapeutic exercise approach Arch Phys Med Rehabil 1997 78 886 892 9344312 10.1016/S0003-9993(97)90206-9
Wu G Evaluation of the effectiveness of Tai Chi for improving balance and preventing falls in the older population--a review J Am Geriatr Soc 2002 50 746 754 11982679 10.1046/j.1532-5415.2002.50173.x
Yang JM Yang Stule Tai Chi Chuan: I. Advanced Tai Chi Tehory and Tai Chi Jing 1985 1st Boston, MA, Yangs Martial Arts Academy
Wayne PM Krebs DE Parker SW McGibbon CA Kaptchuk TJ Gill-Body KM Wolf SL Can Tai Chi improve vestibulopathic postural control? Arch Phys Med Rehabil 2004 85 142 152 14970982 10.1016/S0003-9993(03)00652-X
Wolf SL Barnhart HX Ellison GL Coogler CE The effect of Tai Chi Quan and computerized balance training on postural stability in older subjects. Atlanta FICSIT Group. Frailty and Injuries: Cooperative Studies on Intervention Techniques Phys Ther 1997 77 371 381 9105340
Wolf SL Barnhart HX Kutner NG McNeely E Coogler C Xu T Reducing frailty and falls in older persons: an investigation of Tai Chi and computerized balance training. Atlanta FICSIT Group. Frailty and Injuries: Cooperative Studies of Intervention Techniques J Am Geriatr Soc 1996 44 489 497 8617895
Strupp M Arbusow V Maag KP Gall C Brandt T Vestibular exercises improve central vestibulospinal compensation after vestibular neuritis Neurology 1998 51 838 844 9748036
DeVita P Hortobagyi T Age causes a redistribution of joint torques and powers during gait J Appl Physiol 2000 88 1804 1811 10797145
Judge JO Davis RB Ounpuu S Step length reductions in advanced age: the role of ankle and hip kinetics J Gerontol A Biol Sci Med Sci 1996 51 M303 12 8914503
Kerrigan DC Todd MK Croce UD Lipstiz LA Collins JJ Biomechanical gait alterations independent of speed in the healthy elderly: Evidence for specific limiting impairments Arch Phys Med Rehabil 1998 79 317 322 9523785 10.1016/S0003-9993(98)90013-2
McGibbon CA Puniello MA Krebs DE Mechanical energy transfer during gait in relation to strength impairment and pathology in elderly women Clin Biomech 2001 16 324 333 11358620 10.1016/S0268-0033(01)00004-3
McGibbon CA Krebs DE Compensatory gait mechanics in patients with unilateral knee arthritis J Rheumatol 2002 29 2410 2419 12415602
McGibbon CA Krebs DE Puniello MS Mechanical energy analysis identifies compensatory strategies in disabled elder's gait J Biomech 2001 34 481 490 11266671 10.1016/S0021-9290(00)00220-7
McGibbon CA Toward a better understanding of gait changes with age and disability: Neuromuscular adaptation Exercise Sport Sci Rev 2003 31 102 108 10.1097/00003677-200304000-00009
Neptune RR Kautz SA Zajac FE Contributions of the individual ankle plantar flexors to support, forward progression and swing initiation during walking J Biomech 2001 34 1387 1398 11672713 10.1016/S0021-9290(01)00105-1
McGibbon CA Krebs DE Wolf SL Wayne PW Scarborough DM Parker SW Tai Chi and vestibular rehabilitation effects on gaze and whole-body stability J Vestib Res 2005 14 467 478 15735329
Cheng MC T'ai Chi Ch'uan: A Simplified Method of Calistenics for Health and Self Defense 1981 Berklely, CA, North Atlantic Books
Gill-Body KM Krebs DE Parker SW Riley PO Physical therapy management of peripheral vestibular dysfunction: two clinical case reports Phys Ther 1994 74 129 142 8290618
Cooksey FS Rehabilitation in vestibular injuries Proc R Soc Med 1946 39 273 278
Riley PO Mann RW Hodge WA Modelling of the biomechanics of posture and balance J Biomech 1990 23 503 506 2373723 10.1016/0021-9290(90)90306-N
Ishikawa K Cao ZW Wang Y Wong WH Tanaka T Miyazaki S Toyoshima I Dynamic locomotor function in normals and patients with vertigo Acta Otolaryngol 2001 121 241 244 11349787 10.1080/000164801300043668
McGibbon CA Krebs DE Puniello MS Mechanical energy analysis identifies compensatory strategies in disabled elders' gait J Biomech 2001 34 481 490 11266671 10.1016/S0021-9290(00)00220-7
Krebs DE Wong D Jevsevar D Riley PO Hodge WA Trunk kinematics during locomotor activities Phys Ther 1992 72 505 514 1409883
Ludbrook J Multiple comparison procedures updated Clin Exp Pharmacol Physiol 1998 25 1032 1037 9888002
Asai M Watanabe Y Shimizu K Effects of vestibular rehabilitation on postural control Acta Otolaryngol Suppl 1997 528 116 120 9288254
Goldvasser D McGibbon CA Krebs DE Vestibular rehabilitation outcomes: velocity trajectory analysis of repeated bench stepping. Clin Neurophysiol 2000 111 1838 1842 11018500 10.1016/S1388-2457(00)00387-4
Herdman SJ Clendaniel RA Mattox DE Holliday MJ Niparko JK Vestibular adaptation exercises and recovery: acute stage after acoustic neuroma resection Otolaryngol Head Neck Surg 1995 113 77 87 7603726
Shumway-Cook A Horak FB Vestibular rehabilition: An exercise approach to managing symptoms of vestibular dysfunction Seminars in Hearing 1989 10 194 207
Szturm T Ireland DJ Lessing-Turner M Comparison of different exercise programs in the rehabilitation of patients with chronic peripheral vestibular dysfunction J Vestib Res 1994 4 461 479 7850042
Zajac FE Understanding muscle coordination of the human leg with dynamical simulations J Biomech 2002 35 1011 1018 12126660 10.1016/S0021-9290(02)00046-5
Keshner EA Allum JH Honegger F Predictors of less stable postural responses to support surface rotations in healthy human elderly J Vestib Res 1993 3 419 429 8275275
Winter DA Patla AE Prince F Ishac M Gielo-Perczak K Stiffness control of balance in quiet standing J Neurophysiol 1998 80 1211 1221 9744933
Woollacott MH Shumway-Cook A Changes in posture control across the life span--a systems approach Phys Ther 1990 70 799 807 2236223
Van Deusen J Harlowe D The efficacy of the ROM Dance Program for adults with rheumatoid arthritis Am J Occup Ther 1987 41 90 95 3551621
Wile D Lost T'ai-Chi Classics From the Late Ch'ing Dynasty 1996 1st New York, NY, State University of New York
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
van den Bogert AJ Pavol MJ Grabiner MD Response time is more important than walking speed for the ability of older adults to avoid a fall after a trip J Biomech 2002 35 199 205 11784538 10.1016/S0021-9290(01)00198-1
Pavol MJ Owings TM Foley KT Grabiner MD Mechanisms leading to a fall from an induced trip in healthy older adults J Gerontol A Biol Sci Med Sci 2001 56 M428 37 11445602
| 15717934 | PMC552306 | CC BY | 2021-01-04 16:28:53 | no | BMC Neurol. 2005 Feb 18; 5:3 | utf-8 | BMC Neurol | 2,005 | 10.1186/1471-2377-5-3 | oa_comm |
==== Front
BMC OphthalmolBMC Ophthalmology1471-2415BioMed Central London 1471-2415-5-11570520710.1186/1471-2415-5-1Research ArticleLittle evidence for an epidemic of myopia in Australian primary school children over the last 30 years Junghans Barbara M [email protected] Sheila G [email protected] School of Optometry and Vision Science, University of New South Wales, Sydney, UNSW Sydney 2052. Australia2 School of Psychological Science, La Trobe University, Bundoora 3083, Australia2005 11 2 2005 5 1 1 4 6 2004 11 2 2005 Copyright © 2005 Junghans and Crewther; 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
Recently reported prevalences of myopia in primary school children vary greatly in different regions of the world. This study aimed to estimate the prevalence of refractive errors in an unselected urban population of young primary school children in eastern Sydney, Australia, between 1998 and 2004, for comparison with our previously published data gathered using the same protocols and other Australian studies over the last 30 years.
Methods
Right eye refractive data from non-cycloplegic retinoscopy was analysed for 1,936 children aged 4 to 12 years who underwent a full eye examination whilst on a vision science excursion to the Vision Education Centre Clinic at the University of New South Wales. Myopia was defined as spherical equivalents equal to or less than -0.50 D, and hyperopia as spherical equivalents greater than +0.50 D.
Results
The mean spherical equivalent decreased significantly (p < 0.0001) with age from +0.73 ± 0.1D (SE) at age 4 to +0.21 ± 0.11D at age 12 years. The proportion of children across all ages with myopia of -0.50D or more was 8.4%, ranging from 2.3% of 4 year olds to 14.7% of 12 year olds. Hyperopia greater than +0.50D was present in 38.4%. A 3-way ANOVA for cohort, age and gender of both the current and our previous data showed a significant main effect for age (p < 0.0001) but not for cohort (p = 0.134) or gender (p = 0.61).
Conclusions
Comparison of our new data with our early 1990s data and that from studies of over 8,000 Australian non-clinical rural and urban children in the 1970's and 1980's provided no evidence for the rapidly increasing prevalence of myopia described elsewhere in the world. In fact, the prevalence of myopia in Australian children continues to be significantly lower than that reported in Asia and North America despite changing demographics. This raises the issue of whether these results are a reflection of Australia's stable educational system and lifestyle over the last 30 years.
==== Body
Background
The prevalence of myopia is currently receiving worldwide attention as many recent studies report dramatic increases over the last 20 years [1,2]. Myopia and its aetiology is an interesting example of the intertwining of 'nature and nurture' with both genetics and life-style environment as important issues [3]. There is strong evidence indicating that genetic inheritance is a major contributor, both from the examination of prevalences across different racial backgrounds [4], from family pedigrees [5] and from twin studies [6]. However, there is increasing evidence suggesting that high heritability does not preclude rapid environmentally-induced increases in prevalence [7], rather, inherited factors are likely to both drive the susceptibility and resistance to environmentally-induced myopia [6,8].
Despite much research interest over the last half century, there have been surprisingly few well-designed epidemiological studies of refractive error with large numbers of randomly selected younger school children to form the basis of valid world wide comparisons of the earliest stages of development of myopia [3,9,10]. However, a group sponsored by the World Health Organisation in 2001 has devised a protocol to be used during studies of refractive error across different cultural and ethnic settings: the 'Refractive Error Study in Children' (RESC) [11].
In general, estimates of the prevalence of myopia have shown less increase in the Western world than in Asia, and less increase in rural than in urban populations [1,10,12-16]. Five very large studies across two decades and involving over 10,000 children in Taiwan are very important for understanding the changing prevalence of myopia in young Asian children (1.8% in 1986 rising to 12% in 1995 for 6 year olds, 40% rising to 56% for 12 year olds) [2]. A similar change is also reflected in Singaporean studies of myopia in military conscripts aged 17 years (26% to 83% from the late 1970s to the late 1990s as reviewed by [1]), of whom notably 82% were Chinese [17].
It has often been suggested that myopia is more prevalent in ethnic Chinese (reviewed [18]), but only relatively recent studies compare the prevalence of myopia in young ethnic Chinese children living either in China and in other countries [1,12,15,18-22]. For younger Chinese children aged around 5–7 years, the prevalence of myopia was found to range from under 5% in rural China [14,23] to 24% in Chinese Malays [20] and 30% in urban Hong Kong [19,22]. For older Chinese children aged 11–12 years, the prevalence ranged from 23% of rural Chinese[14,23] to 40% in urban China[12], 47% of Chinese Malays [20] and 57% in urban Hong Kong [22]. Japan has a similarly high prevalence of myopia in young school children estimated in recent times to be 43.5% of 12 year olds [24].
By comparison, the epidemiology of refractive error for young Australian school children is relatively well documented and presents a very different profile. A number of studies were carried out in the early 1970s and the 1980s on relatively large groups of unselected primary school children from the socio-economic extremes (generally aged 5 to 12 years), and indicated a prevalence of myopia ranging from approximately 3% to 13% (see Table 1) [25-29]. Two of those early studies investigated children largely from underprivileged, rural, families [26,27], and the other was of children from several, middle to upper socio-economic class private schools [25]. One smaller study was carried out in the mid 1980s on children from a representative selection of government schools in Brisbane [29], and would therefore have investigated children from a broader range of backgrounds. Interestingly, this latter study was the only Australian study to have determined refractive error under cycloplegia, yet yielded the highest prevalence of myopia. Thus, it has been difficult to determine whether the prevalence of myopia has increased in young school children in Australia as reported elsewhere. The majority of Australian residents are of Caucasian extraction living a very western lifestyle, leading one to expect the prevalence of myopia to be similar to that found in US or Europe. Yet, studies suggest that the prevalence of myopia in Australian primary school children is low by world standards [10].
In 2003 we reported the relative proportions of refractive errors in a large unselected primary school population of 2,535 children drawn from a very broad range of socio-economic backgrounds in Sydney, the largest city in Australia, in the early 1990s [30]. The children attended fourteen primary schools and two preschools. As in the earlier studies, the proportion of children with myopia greater than -0.50 DS spherical equivalence, as determined by non-cycloplegic retinoscopy, was found to be low by world standards (1.0% of 4 year olds rising to 8.3% of 12 year olds). We have now analysed the prevalence of refractive error in a new similar group of 1,936 children unselected primary school children drawn generally from the same area as our first study.
Methods
The study design is a retrospective examination of records of the Vision Education Centre (VEC) [31] school vision screenings (so named because parents were not present to ratify history) conducted in the Clinic of the School of Optometry and Vision Science, UNSW. Approvals for the study and permission to approach schools were obtained from the Committee for Use of Humans in Research at the University of New South Wales (UNSW), Sydney, Australia. The protocols adhered to the tenets of the Declaration of Helsinki. Parents or guardians were provided with an information sheet and requested an outline of known symptoms. Signed consent was required prior to a child's participation.
Sampling and recruitment
Permission was obtained from the NSW Department of Education and the NSW Catholic Education Office to approach all schools in the eastern region of Sydney (some thirty coeducational primary schools) to send entire classes to the VEC. A flyer was sent describing the VEC science excursion and age-appropriate eye examination, inviting Years 1, 3 and 5 particularly to participate.
The group of 1,936 children examined came from the eastern suburbs along the southern beaches of Sydney, and may be thought of as randomly selected with little likelihood of bias to the data as individual classes were free to respond. Children were drawn from twelve government and non-government primary schools and one pre-school and attended the clinic only once. During the 1996 Australian Bureau of Statistics census 14,785 children aged 4 to 12 years were recorded in this region (Randwick and Waverley precincts of Eastern Sydney) who came from a very broad range of ethnic and socio-economic backgrounds present, where 37 different languages might be spoken in the home [32]. This was reflected in the children attending VEC. Census data indicate approximately 9% of the children in the current study were likely to be of Asian origin [32], a figure supported by our interpretation of family name for each child [30]. Participation in the eye examinations was typically well over 90% for each class, with teachers reporting non-participation to be predominantly due to illness on the day. Less than 3% of parents intentionally prevented participation, even if eye care had previously been sought. This particularly high participation rate was largely due to the attraction of a an age-appropriate student-centred hands-on science lesson about eyes and vision [31] delivered alongside the eye examination.
Clinical examination
The comprehensive optometric examination by experienced paediatric practitioners included all age-appropriate tests meeting Australian Optometric Competency Standards, except that parents/guardians were not present to ratify history. Refractive error was determined by non-cycloplegic retinoscopy with optical fogging while the child maintained fixation on a distant non-accommodative (6 metre) target. In most cases refractive status was confirmed by subjective refraction. Other tests included letter visual acuity at 6 m and 33 cm, cover test for strabismus, motilities, saccades, pupil reactions, near point of convergence, heterophoria, stereopsis, accommodative facility, colour vision and ophthalmoscopy.
Justification of choice of testing procedures
Cycloplegic retinoscopy was not undertaken for many reasons including the fact that VEC studies started prior to the 2000 convention suggesting use of cycloplegic retinoscopy for studies of refractive error prevalence [11]. Secondly, the VEC visit was meant as an excursion and the children had to return to normal classes with near work demands after the morning outing. Thirdly, it was important for comparison purposes to use refractive data procured under the same conditions as that used for the earlier groups of children. Fourthly, an initial evaluation without cycloplegia is necessary in order to understand daily function. Fifthly, non-cycloplegic retinoscopy was only one component of the exam. Outcomes regarding a decision to refer would not alter for most children had a cycloplegic refraction been carried out, as several other near function tests that would also indicate the possible existence of latent hyperopia or pseudo-myopia were included. Lastly, the degree of refractive error may in fact be influenced by cycloplegia (see Discussion for elaboration [33-38]).
Autorefractors were not employed as hand-held versions were unavailable when the first cohort was seen. Equally as important, there is no convincing evidence that the proportion of myopes identified in the sample would have changed [39].
Comparison with earlier data
To compare the estimated prevalence of myopia in this urban population of 'Australian children' over the last decade, this more recent 2000s data set was analysed against data from an earlier cohort of 2,322 children with similar demographics seen in the early 1990's, using the same testing protocols and seen at the same venue [30]. The optometric results of that earlier cohort have previously been reported [40], and it was noted that 7.1% of those children were already wearing spectacles [30], indicating that our recruitment procedure did not preclude children already under the care elsewhere. The data for any child examined in both cohorts was deleted from the earlier data set to avoid bias in the analysis. The mean date of assessment for this last 2000s cohort was September 2000, and for the early 1990s cohort was June 1992. Thus, the average gap between assessments of children from the two cohorts was 8 years and 3 months.
Statistical analyses
Data was analysed by Analysis of Variance ANOVA (StatView software). Only refractive data from right eyes was used for the current refractive class analysis, as the correlation between right and left eye refractions was extremely high (p < 0.0005). The preferred criterion to define myopia in this study is that used clinically in Australia: a spherical equivalent equal to or more minus than -0.50 D. However, as myopia more minus than -0.50 D has occasionally been used to define myopia in epidemiological studies [13,19,41], analyses using the criterion 'myopia more minus than -0.50 D' were also performed for comparison. Hyperopia was defined as spherical equivalents greater than +0.50 D. Thus, emmetropia for this study was defined as refractions in the range -0.25 to +0.50 dioptres spherical equivalence inclusive. Means are quoted with the associated standard error.
Results
The records of 1,936 children aged 4 to 12 years from a non-clinical unselected population examined during the six years from March 1998 to May 2004 were analysed retrospectively to estimate the prevalence of different types of refractive error. Primary schools of their own choice sent more children from years 1, 3, and 5, which resulted in unequal numbers of children in each of the age groups. There were 925 boys and 951 girls, and the relative numbers for both males and females in each age group are shown in Table 2. For 59 children, the gender was not indicated on the record card and could not be inferred with certainty from the given name. The data not associated with gender has only been included in analyses entitled 'All' as shown in Tables 2 and 3. Mean age was 8.36 years. The relative proportions of the different classifications of refractive error for all children combined (including those of unknown gender) for each age group are shown in Table 2.
The mean spherical equivalent refraction of all 1,936 children was +0.45 ± 0.02 DS, however it should be noted that there is a preponderance of children aged 5–6, 9 and 11 years old corresponding with Years 1, 3, and 5 of primary school. Overall, there was no significant difference in spherical equivalent refractive error between girls and boys (p = 0.697). In general, mean refraction demonstrates a highly significant shift towards less hyperopia with increasing age (p < 0.0001) from 0.73 ± 0.1DS for 4 year olds to 0.21 ± 0.11 for 12 year olds, however this is more noticeable after the age of 9 years as seen in Fig. 1. With increasing age, more children are found in the emmetropic category and fewer in the low hypermetropic category.
A summary of the relative proportions of myopia and hyperopia for this cohort of children of all ages seen during the six years ('2000s' data) is given in Table 3. The majority of children screened are emmetropic by our criteria: 53.0% averaged across all ages. The proportion of children manifesting moderate to high degrees of hypermetropia (≥+1.50 DS) is 6.2% across all ages. Only 6.9% of children of all ages had refractive errors more minus than -0.50 DS, ranging from 2.3% of 4 year olds to 13.3% of 12 year olds (Fig. 2). If the more liberal definition of myopia is applied (myopia equal to or more minus than -0.50), then 8.4% of all children were myopic (ranging from 2.3% of 2 year olds to 14.7% of 12 year olds). Only 0.8% of the 1,936 children were more than -4.00 DS myopic.
Comparison with previous data
The number of children per age group for the two cohorts is shown in Fig. 3, and notably, the age profile differs slightly between cohorts, though the mean age of 8.37 years for the earlier cohort was similar to that of this later cohort. The mean and standard error for right eye spherical equivalent refractive error by age is shown for each cohort in Fig. 4.
A three-way analysis of variance was carried out for cohort, age and gender. There was a main effect for age (p < 0.0001), but no main effect for either cohort (p = 0.134) or gender (p = 0.61). However, there were age/gender/cohort interactions that indicate a trend towards an increasing shift away from the hyperopic refraction in the later cohort.
Discussion
An analysis of the prevalence of refractive errors in young school children in eastern Sydney during the last thirteen years has been presented. The latest data gathered from 1,936 unselected primary school-aged children in the last 6 years, indicates that the prevalence of myopia remains quite low compared to that reported for the western world and Asia, especially as refractive error was established by non-cycloplegic retinoscopy (as will be discussed later). These findings are not significantly different (p = 0.13) to our previous report [30] indicating that 6.5% of 2,535 unselected children aged 4 to 12 years seen in the early 1990s were myopic by at least 0.50 D. Notably, those children were of similar socio-economic and ethnic status drawn from the same region of Sydney and seen at the same Centre using the same testing protocol.
Therefore, if we take the total 4,258 children seen since 1990, the relative frequency of refractive error across all is: 54.2% emmetropic by our criteria, 32.3% low to moderate hyperopes, 5.3% myopic greater than -0.50D spherical equivalence and 7.4% myopic by at least -0.50 DS. The number with myopia of at least -4.00 DS was an extremely small 0.6%.
The prevalence of myopia in Sydney primary school children compared to the rest of the world
As alluded to in the introduction, the proportion of Sydney children with myopia is dramatically less than in Asia. Indeed, the proportion appears significantly lower than in the USA [41] and Canada [42] (4% and 6% of 6 year olds respectively, or 20% of 12 year olds in USA), but higher than urban India with only 4.4% of all school children under 16 years myopic [13] and higher particularly than in other less developed countries [10].
In the past, a lack of internationally accepted definitions for 'myopia' has hampered valid comparisons across the various studies [10]. Commonly the criteria 'greater than -0.50 DS' or 'at least -0.50 DS' are employed. However, our separate analyses using both of these criteria only resulted in a difference of 1.5% of all children included as myopic, in keeping with other dual analyses [13,41], and is low either way when compared with Asia or North America.
Comparison across studies is also difficult when only an 'overall' mean refraction is presented covering all children in a study, due to the well known increasing prevalence of myopia with age. Indeed, the comparison of data from our own two data sets is confounded to some extent by the slightly different age profiles for each cohort. However, in neither cohort was the age range nor mean significantly different, so the similar proportion of myopes is not unexpected.
Comparison of refractive error with and without a cycloplegic agent
The question of optimal ocular conditions for comparison of the prevalence of refractive errors remains controversial. A cycloplegic agent is typically proposed as the gold standard [3,43,44] in the belief that it will eliminate ciliary muscle action or spasm, and thus unmask latent hyperopia or pseudomyopia. Thus, the use of a cycloplegic would be firstly predicted to lead to a decrease in the prevalence of myopia, and an increase in the prevalence of hyperopia. However, as a cycloplegic also leads to associated mydriasis and the introduction of unpredictable spherical aberrations, it is arguable that cycloplegia will induce unpredictable errors. In fact, Gao et al [38] in 2002 reported significant changes in the refractive components of children's eyes under conditions of deep cycloplegia and mydriasis that were greatest in hyperopic eyes and smallest in myopic eyes, adding no definitive evidence as to the relative efficacy of cycloplegia.
Thus there appears to be no scientific concurrence regarding the efficacy of cycloplegia for studies on the prevalence of myopia [35-37], with several major studies electing to use cycloplegia (see review in [10,9,11]) and others not [18-21,23,42,45]. Presumably this design variability exists because there is no decisive evidence indicating a difference between refractions determined with and without a cycloplegic agent in eyes that have a myopic refraction. In general, a more positive retinoscopic finding is reported under cycloplegia, though considerable individual variation is seen including a myopic shift in some [33,35-37,46]. Not surprisingly, the differences noted decreased both with age and with less positive refraction.
As our refractive data was derived from non-cycloplegic retinoscopy we readily concede that mean refractive error may be less hyperopic than if a cycloplegic had been used. However, we suggest that as the influence of a cycloplegic is uncertain and is of least concern for myopes, the estimated prevalence of myopia will not be significantly altered by our decision to not use a cycloplegic. In support of this notion are new conference data from Rose et al [47,48] reporting refractive status ascertained by cycloplegic autorefraction in over 1,000 children aged 6–7 years from across the same city of Sydney. They reported values of 'around 3%' for the prevalence of myopia of at least 0.50D [47], and then the value of 1.5% for myopia of 'approximately 0.50D' [48] with a participation rate between 73 and 80%. From Table 2 it can be seen that 2.4% of our 6 year olds in the current study were at least 0.50D myopic – a value that is strikingly similar.
Demographics versus lifestyle
Worldwide patterns of the prevalence of myopia suggest significant differences are likely to be due to the different demographics and lifestyles [1,10,49]. Zadnik [41] concedes that the increase in numbers of myopic children in the US Orinda study may be due to changing ethnic demographics. The apparent slight increase in myopia in Australia reported in the current study may also be in part accounted for by our changing ethnic demographics in urban areas. However demographics and ethnic compositions are unlikely to be responsible for the large changes reported in Asian and some other western countries [1,50].
Whatever way it is argued, our results indicate little evidence for an epidemic of myopia although there is a developmental trend towards an earlier decrease in hyperopia to the point of myopia. Thus, the question of whether it is a matter of lifestyle, or perhaps familial environmental stress, or more, remains. Certainly, the education system and housing has changed little in Australia the last 30 years. By comparison, most Asian children participating in myopia epidemiological studies reportedly are more likely to live in high-rise residential blocks [17] and have strong demands at school to memorize along with parental and peer pressure to do well, and for some, a competitive entrance examination to enter school [19,51].
Conclusions
It is concluded that despite some differences in methodology across earlier studies, the prevalence of myopia in young Australian school children does not appear to have increased significantly over the last 30 years if one allows for the change in ethnic demographics. It is also proposed that an explanation for the large increase in prevalence of myopia reported in other countries must include questions relating to lifestyle in addition to genetic propensity.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SC conceived and designed the study, jointly analysed the results and worked the drafts through to the final version. BJ coordinated the study, collated the clinic records, jointly analysed the results, researched the background for the paper, prepared the draft manuscript and was responsible for presentation. Both authors read and approved the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Supported by a University of New South Wales GoldStar Grant to BM Junghans.
Figures and Tables
Figure 1 Histogram of mean right eye spherical equivalent refractive error by age and gender ('2000s' data). Error bars represent standard error.
Figure 2 Line graphs of the percentage Sydney children with myopia by age using the alternate criteria of more minus than -0.50 D spherical equivalence (blue line) or more minus than -0.25DS spherical equivalence (orange line), and, the percentage of Taiwanese children seen in the year 2000 more minus than -0.25D (aqua dashed line) taken from Lin et al, Ann Acad Med 2004 33:27–33.
Figure 3 Histogram of the number of children in each age group for data from the current study (2000s) and the previously reported early 1990s data [30].
Figure 4 The mean right eye spherical equivalent refractive error by age for data from the current study and the previously reported early 1990s data [30]. Error bars represent standard error.
Table 1 Prevalence of myopia in Australia school children.
Authors Year N Place Ages Method Criterion Prevalence %
Robbins & Bailey [25] 1975 1,243 Private schools 4, 8, 12 Non-cycloplegic retinoscopy Myopia≥-0.50D
"suspected of having myopia" 5.6% 4 yrs
6.4% 12 yrs
Amigo, McCarthy & Pye [26] 1976 1,166 Rural underprivileged 7 to 12 Non-cycloplegic retinoscopy "any tendency for myopia" 3.4%
Walters [27] 1981
1982
1983 2,055
1,187
1,722 Rural underprivileged 4 to 14 Non-cycloplegic retinoscopy Myopia≥-0.50D 4.40%
3.03%
2.67%
Macfarlane [29] 1987 877 Representative government schools 6 to 11 Cycloplegic retinoscopy Not stated 13%
Junghans et al [30] Early 1990s 2,535 14 government & church schools 4 to 12 Non-cycloplegic retinoscopy Myopia≥-0.50D 2.0% 4 yrs
10.9% 12 yrs
Mean 6.5%
Table 2 Estimated prevalence of refractive error (%) by age in children aged 4 to 12 years.
Age Myopia Emmetropia Hyperopia N
More than -3.75 -0.50 to -3.75 of -0.50 -0.25 to +0.50 +0.75 to +1.25 More than +1.25 Female Male All
4 0.0 2.3 0.0 43.2 50.0 4.5 24 16 44
5 0.4 1.5 1.2 49.8 41.3 5.8 121 129 259
6 0.0 1.7 0.7 47.6 43.4 6.6 145 133 286
7 0.0 2.5 2.5 54.8 33.1 7.0 72 80 157
8 1.0 5.6 0.5 49.0 29.6 14.3 100 85 196
9 0.8 6.5 1.5 56.3 29.7 5.3 133 128 263
10 2.4 9.0 2.4 54.7 27.8 3.8 149 88 212
11 0.5 10.9 2.1 58.0 24.5 4.0 146 189 376
12 2.1 11.2 1.4 57.3 22.4 5.6 61 77 143
Table 3 Mean estimated prevalence of myopia and hyperopia in children aged 4 to 12 years.
Female Male All
Refractive Category: N % N % N %
Myopia ≤-4 8 0.8 8 0.9 15 0.8
Myopia -3.75≤x≤-0.75 58 6.1 58 6.3 118 6.1
Myopia of -0.50 11 1.2 17 1.8 29 1.5
Emmetropia -0.25≤x≤0.50 504 53.0 501 54.2 1030 53.2
Hyperopia +0.75≤x≤+1.25 320 33.6 282 30.5 624 32.2
Hyperopia ≥+1.50 50 5.3 59 6.4 120 6.2
Total 951 100 925 100 1936 100
==== Refs
Seet B Wong TY Tan DT Saw SM Balakrishnan V Lee LK Lim AS Myopia in Singapore: taking a public health approach Br J Ophthalmol 2001 85 521 526 11316705 10.1136/bjo.85.5.521
Lin LLK Shih YF Hsiao CK Chen CJ Prevalence of myopia in Taiwanese schoolchildren: 1983 to 2000 Ann Acad Med 2004 33 27 33
Saw SM Katz J Schien OD Chew SJ Chan TK Epidemiology of myopia Epidem Rev 1996 18 175 187
Curtin BJ The Myopias. Basic Science and Clinical Management. 1985 Philadelphia , Harper and Row
Pacella R McLellan J Grice K Del Bono EA Wiggs JL Gwiazda JE Role of genetic factors in the etiology of juvenile-onset myopia based on a longitudinal study of refractive error Optom Vis Sci 1999 76 381 386 10416932
Lyhne N Sjolie AK Kyvik KO Green A The importance of genes and environment for ocular refraction and its determiners: a population based study among 20-45 year old twins Br J Ophthalmol 2001 85 1470 1476 11734523 10.1136/bjo.85.12.1470
Rose KA Morgan IG Smith W Mitchell P High heritability of myopia does not preclude rapid changes in prevalence Clin Exp Ophthalmol 2002 30 168 172 10.1046/j.1442-9071.2002.00521.x
Saw SM Chua WH Hong CY Wu HM Chan WY Chia KS Stone RA Tan D Nearwork in early-onset myopia Invest Ophthalmol Vis Sci 2002 43 332 339 11818374
Zadnik K Mutti DO Friedman NE Adams AJ Initial cross-sectional results from the Orinda Longitudinal Study of Myopia Optom Vis Sci 1993 70 750 758 8233371
Working group on myopia prevalence and progression Myopia: prevalence and progression 1989 Washington D.C. , National Research Council
Negrel AD Maul E Pokharel GP Zhao J Ellwein LB Refractive Error Study in Children: sampling and measurement methods for a multi-country survey.[comment] Am J Ophthalmol 2000 129 421 426 10764848 10.1016/S0002-9394(99)00455-9
Yap M Wu M Wang SH Lee FL Liu ZM Environmental factors and refractive error in Chinese children Clin Exp Optom 1994 77 8 14
Dandona R Dandona L Naduvilath TJ Srinivas M McCarty CA Rao GN Refractive errors in an urban population in Southern India: the Andhra Pradesh Eye Disease Study Invest Ophthalmol Vis Sci 1999 40 2810 2818 10549640
Zhang MZ Saw SM Hong RZ Fu ZF Yang H Shui YB Yap MK Chew SJ Refractive errors in Singapore and Xiamen, China--a comparative study in school children aged 6 to 7 years Optom Vis Sci 2000 77 302 308 10879787 10.1097/00006324-200006000-00010
Zhao J Pan X Sui R Munoz SR Sperduto RD Ellwein LB Refractive Error Study in Children: results from Shunyi District, China Am J Ophthalmol 2000 129 427 435 10764849 10.1016/S0002-9394(99)00452-3
Dandona R Dandona L Srinivas M Sahare P Narsaiah S Munoz SR Pokharel GP Ellwein LB Refractive error in children in a rural population in India Invest Ophthalmol Vis Sci 2002 43 615 622 11867575
Wu HM Seet B Saw SM Lim TH Chia KS Does education explain ethnic differences in myopia prevalence? A population based study of young adult males in Singapore Optom Vis Sci 2001 78 234 239 11349931 10.1097/00006324-200104000-00012
Lam CS Edwards M Millodot M Goh WS A 2-year longitudinal study of myopia progression and optical component changes among Hong Kong schoolchildren Optom Vis Sci 1999 76 370 380 10416931
Lam CSY Goh WSH The incidence of refractive errors among school children in Hong Kong and its relationship with the optical components Clin Exp Optom 1991 74 97 103
Chung KM Mohidin N Yeow PT Tan LL O'Leary D Prevalence of visual disorders in Chinese schoolchildren Optom Vis Sci 1996 73 695 700 8950751
Edwards MH The development of myopia in Hong Kong children between the ages of 7 and 12 years: a five-year longitudinal study Ophthalmic Physiol Opt 1999 19 286 294 10645384 10.1046/j.1475-1313.1999.00445.x
Fan DS Lam DS Lam RF Lau JT Chong KS Cheung EY Lai RY Chew SJ Prevalence, incidence, and progression of myopia of school children in Hong Kong Invest Ophthalmol Vis Sci 2004 45 1071 1075 15037570 10.1167/iovs.03-1151
Yap MK Cho J Woo G A survey of low vision patients in Hong Kong Clin Exp Optom 1990 73 19 22
Matsumura H Hirai H Prevalence of myopia and refractive changes in students from 3 to 17 years of age Surv Ophthalmol 1999 44 S109 115 10548123 10.1016/S0039-6257(99)00094-6
Robbins HG Bailey IL The organisation and findings of a school vision screening service Aust J Optom 1975 58 392 401
Amigo G McCarthy A Pye D Visual characteristics of an underprivileged group of Australian children Aust J Optom 1976 59 188 197
Walters J Portsea Modified Clinical Technique: results from an expanded optometric screening protocol for children Aust J Optom 1984 67 176 186
Nathan J Kiely PM Crewther SG Crewther DP Disease-associated visual image degradation and spherical refractive errors in children American Journal of Optometry & Physiological Optics 1985 62 680 688 4073201
Macfarlane DJ Fitzgerald WJ Stark DJ The prevalence of ocular disorders in 1000 Queensland primary school children ANZ J Ophthalmol 1987 15 161 174
Junghans BM Crewther SG The prevalence of myopia among primary school children in eastern Sydney Clin Exp Optom 2003 86 339 345 14558856
Junghans BM Crewther SG The Vision Education Centre: a multi-level educational tool J Optom Ed 1992 17 82 86
Ethnic Affairs Commission of NSW The people of New South Wales: Statistics from the 1996 census 1998 Sydney , NSW Government 149 151
Young FA Leary GA Baldwin WR West DC Box RA Harris E Johnson C Comparison of cycloplegic and non-cycloplegic refractions of Eskimos Am J Optom Arch Am Acad Optom 1971 48 814 824 5286432
Ludlum WM Weinberg SS Twarowski CJ Ludlum DP Comparison of cycloplegic and non-cycloplegic ocular component measurement in children Am J Optom Arch Am Acad Optom 1972 49 805 818 4508715
Hiatt RL Braswell R Smith L Patty JW Refraction using mydriatic, cycloplegic and manifest techniques Am J Ophthalmol 1973 76 739 744 4748194
Schultz L Variations in refractive change induced by cyclogyl upon children with differing degrees of ametropia Am J Optom Phsyiol Opt 1975 52 482 484
Chan OYC Edwards M Comparison of cycloplegic and noncycloplegic retinoscopy in Chinese pre-school children Optom Vi Sci 1994 71 312 318
Gao L Zhou X Kwok AKH Yu N Ma L Wang J The change in ocular refractive components after cycloplegia in children Jpn J Ophthalmol 2002 46 293 298 12063039 10.1016/S0021-5155(02)00479-3
Chat SW Edwards MH Clinical evaluation of the Shin-Nippon SRW-5000 autorefractor in children Ophthalmic Physiol Opt 2001 21 87 100 11261351
Junghans B Kiely P Crewther D Crewther S Referral rates for a functional vision screening among a large cosmopolitan sample of Australian children Ophthal Physiol Opt 2002 22 10 25 10.1046/j.1475-1313.2002.00010.x
Zadnik K The Glenn A. Fry Award Lecture (1995). Myopia development in childhood Optom Vis Sci 1997 74 603 608 9323731
Robinson BE Factors associated with the prevalence of myopia in 6-year-olds Optom Vis Sci 1999 76 266 271 10375239
Curtin BJ Adult myopia Acta Ophthalmol 1988 185S 78 79
Mutti DO Zadnik K Egashira S Kish L Twelker JD Adams AJ The effect of cycloplegia on measurement of ocular components Invest Ophthalmol Vis Sci 1994 35 515 527 8113002
Garner LF Meng CK Grosvenor TP Mohidin N Ocular dimensions and refractive power in Malay and Melanesian children Ophthalmic Physiol Opt 1990 10 234 238 2216470 10.1016/0275-5408(90)90004-I
Suryakumar R Bobier WR The manifestation of noncycloplegic refractive state in pre-school children is dependent on autorefractor design Optom Vis Sci 2003 80 578 586 12917577 10.1097/00006324-200308000-00012
Rose K Mai TQ Smith W Morgan IG Mitchell P Prevalance of myopia in school-aged children: comparative data from Sydney and Vietnam. In: XVI International Congress of Eye Reserch: Sydney, Australia.
2004 Academic Press
Rose K Smith W Morgan IC Mitchell P Preliminary results from the Sydney Myopia Study. In XVICER Satellite meeting on the Eye and Brain and Myopia: 4 - 7 September 2004; Fraser Island, Australia.
2004
Baldwin WR A review of statistical studies of relations between myopia and ethnic, behavioral, and physiological characteristics Am J Optom Phsyiol Opt 1981 58 516 527
Villarreal MG Ohlsson J Abrahamsson M Sjostrom A Sjostrand J Myopisation: the refractive tendency in teenagers. Prevalence of myopia among young teenagers in Sweden Acta Ophthalmologica Scand 2000 78 177 181 10.1034/j.1600-0420.2000.078002177.x
Goldschmidt E Lam CS Opper S The development of myopia in Hong Kong children Acta Ophthalmol Scand 2001 79 228 232 11401628 10.1034/j.1600-0420.2001.790303.x
| 15705207 | PMC552307 | CC BY | 2021-01-04 16:03:49 | no | BMC Ophthalmol. 2005 Feb 11; 5:1 | utf-8 | BMC Ophthalmol | 2,005 | 10.1186/1471-2415-5-1 | oa_comm |
==== Front
BMC PsychiatryBMC Psychiatry1471-244XBioMed Central London 1471-244X-5-101572072610.1186/1471-244X-5-10Research ArticleQuality of life in patients with personality disorders seen at an ordinary psychiatric outpatient clinic Narud Kjersti [email protected] Arnstein [email protected] Alv A [email protected] Department of Psychiatry, Aker University Hospital, Sognsvannsveien 21, 0310 Oslo, Norway2 Center for Health Promotion, Faculty of Psychology, University of Bergen Christiesgate 13, 5015 Bergen, Norway4 Department of Clinical Cancer Research, Rikshospitalet-Radiumhospitalet Trust, Montebello, 0310 Oslo, Norway2005 20 2 2005 5 10 10 12 9 2004 20 2 2005 Copyright © 2005 Narud et al; licensee BioMed Central Ltd.2005Narud 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
Epidemiological studies have found reduced health-related quality of life (QoL) in patients with personality disorders (PDs), but few clinical studies have examined QoL in PDs, and none of them are from an ordinary psychiatric outpatient clinic (POC). We wanted to examine QoL in patients with PDs seen at a POC, to explore the associations of QoL with established psychiatric measures, and to evaluate QoL as an outcome measure in PD patients.
Methods
72 patients with PDs at a POC filled in the MOS Short Form 36 (SF-36), and two established psychiatric self-rating measures. A national norm sample was compared on the SF-36. An independent psychiatrist diagnosed PDs and Axis-I disorders by structured interviews and rated the Global Assessment of Functioning (GAF). All measurements were repeated in the 39 PD patients that attended the 2 years follow-up examination.
Results
PD patients showed high co-morbidity with other PDs and Axis I mental disorders, and they scored significantly lower on all the SF-36 dimensions than age- and gender-adjusted norms. Adjustment for co-morbid Axis I disorders had some influence, however. The SF-36 mental health, vitality, and social functioning were significantly associated with the GAF and the self-rated psychiatric measures. Significant changes at follow-up were found in the psychiatric measures, but only on the mental health and role-physical of the SF-36.
Conclusion
Patients with PDs seen for treatment at a POC have globally poor QoL. Both physical and mental dimensions of the SF-36 are correlated with established psychiatric measures in such patients, but significant changes in these measures are only partly associated with changes in the SF-36 dimensions.
==== Body
Background
According to the DSM-IV [1] personality disorders (PDs) are characterized by enduringly deviating patterns of perceiving, relating to, and thinking about the environment and oneself that are exhibited in a wide range of social and personal contexts. Such patterns lead to "clinically significant distress or impairment in social, occupational, or other important areas of functioning". The DSM-IV does not indicate how "clinically significant distress or impairment" (page 633) should be evaluated, however, and a recent study showed that various formulations of this criterion hardly increased diagnostic validity [2].
Since the DSM-IV included the Global Assessment of Functioning Scale (GAF) as Axis V, it is reasonable to consider if the GAF should be used for evaluations of "significant distress or impairment". Two problems are implicit in such an approach: at what GAF cut-off score should "significant distress or impairment" be set, and the evaluation is only done by a professional. As to the first problem, Kessler et al [3] suggested a GAF score of 60 as a cut off for "serious mental illness". As to the second, in patients with somatic diseases "clinically significant distress or impairment" has for a long time been quantified by self-rating of the health-related quality of life (QoL) including mental health [4]. Several general instruments for the rating of QoL have been developed among which the Medical Outcome Study Short Form-36 (the SF-36) and its brief version the SF-12 have become the most popular [5,6].
Major treatment outcome studies of PDs like the Collaborative Longitudinal Personality Disorders Study [7], the Norwegian Network of Psychotherapeutic Day Hospitals [8], and the Cassel Hospital study [9] used the GAF, however, and did not included any QoL measurements, and the QoL was not included in a recommended core battery of instruments for measurement of changes in PD patients [10].
In contrast, two epidemiological studies of PDs have used the SF-12 as a measure of disability. In a national study from Australia, Jackson and Burgess [11] reported that the SF-12 Physical and Mental Component Summary Scales (PCS and MCS) were both significantly reduced in persons with one or more PDs diagnosed by a screening instrument, compared to persons without. When further examining the relationship between the SF-12 and PDs, they found that co-morbid Axis I and chronic physical conditions explained a considerable part of the MCS and PCS scores in PDs [12]. Mean MCS and PCS scores also became significantly lower with an increasing number of comorbid PDs present. A national study from the United States [13] confirmed the reduced MCS after controlling for co-morbid Axis I disorder in the avoidant, dependent, paranoid, schizoid, and antisocial PDs, but not in the histrionic PD, diagnosed by a more extensive diagnostic interview schedule than the Australian study.
A review of the literature showed that the QoL had only been used as a disability measure in a few clinical research studies of PD patients. In depressed elderly patients, Abrams et al [14] found that the presence of criteria for cluster B PDs predicted lower QoL. Since the cluster B criteria overlapped considerably with symptoms of depression, it was unclear if they made any independent contribution to reduced QoL. Swinton et al [15] reported that male PD patients in a high security forensic setting were less satisfied with their overall QoL than patients with schizophrenia. The authors emphasized that the high security setting was quite unusual. Hueston et al. [16] showed that primary care patients with high risk for PDs, scored significantly lower on overall QoL and on several subscales of the SF-36, compared to patients with a low risk for PDs. Since prevalence of depression and alcohol dependence was higher in the high-risk group, the influence of PDs alone on QoL was difficult to tease out in that study. Nakao et al [17] examined the relationship between PDs and the GAF in 136 Axis I patients mainly with mood and anxiety disorders and found that patients with any comorbid PDs were more disabled than those without. They did not adjust for the presence of Axis I disorders, however.
None of these clinical studies takes QoL as observed in PD patients seen at an ordinary psychiatric outpatient clinic (POC) as their point of departure. However PD patients are frequent at POC, and the QoL is an important self-rated measure of "clinically significant distress or impairment". Since QoL data on PD patients seen at at POC seems to be lacking from the literature, we found it relevant to study a consecutive sample of PD patients from a POC and collect QoL data with SF-36. We posed the following research questions: 1) How are the SF-36 dimensions mean scores in PD patients compared to age- and gender-adjusted norm data? 2) To what extent are the SF-36 scores in PD patients associated with co-morbid Axis I disorders? 3) How is the association between the SF-36 dimension and established patient- and professional-rated psychiatric measures in PD patients? and 4) What changes in the SF-36 dimensions of treated PD patients are observed from baseline to follow-up, and how are they related to changes in the psychiatric measures?
Methods
Setting
Furuset POC serves a communality of Oslo City, Norway with a population of 28.000 people. At the time of the study, the staff consisted of three psychiatrists, three clinical psychologists, two psychiatric nurses, and two social workers. The intake rate was approximately 400 new patients a year. The first author (KN) invited the staff to take part in the study by referring to her new patients with probable PDs. Six professionals were willing to participate, while four declined due to heavy clinical burden, or lack of interest.
At the start of the study in 1996, Furuset was a new suburb of Oslo, and the inhabitants were characterized by lower socio-economic conditions, high mobility, and a considerable prevalence of immigrants from Asian countries. The suburb had a high proportion of municipal housings, and the criterion for allotment to them was severe mental disorder and/or severe socio-economic problems. Many patients seen at Furuset POC were out of work due to mental disorders, and/or due to socio-economic circumstances.
Patients
Patients aged from 18 to 75 years were consecutively recruited from January 1, 1996 to June 30, 1998. The patients were referred from the local GPs, and physical examination and adequate treatment and follow-up of physical diseases were the responsibility of the GPs. The six therapists screened for probable PDs among new patients scheduled for treatment. Exclusion criteria were mental retardation, lifetime psychosis and bipolar disorder, organic mental disorders, current strong suicidal ideation, and insufficient knowledge of the Norwegian language. Eligible patients received oral and written information about the study from their therapists. Then the patients were invited to take part in the study, and they all gave written informed consent. The Ethical Review Board of Department of Psychiatry, Aker University Hospital approved the project.
The six therapists did not miss out any patients at screening, but 5 (4%) eligible patients declined to take part in the study. Among 110 eligible patients referred to the study, only 91 filled in the SF-36 at baseline due to administrative misunderstandings. However, when they were compared to the 19 who did not fill in, the non-attenders only had significantly fewer co-morbid Axis I-disorders (data not shown).
In order to answer the research questions, the sample was divided into three groups: cluster A+B PDs (n = 39), cluster C (n = 33), and Axis I-disorders (n = 19). The cluster A+B group could also contain co-morbid cluster C PDs and Axis I-disorders, and the cluster C co-morbid Axis I-disorders.
Follow-up procedure
Two years after baseline, the patients received a mailed written appointment for a follow-up interview. Those who did not show up were sent a written reminder. If they still did not meet, they were called by phone, and if there was no answer, their addresses and phone numbers were checked at the Census register. Appointments were mailed to new addresses, and phone-calls were made in case of non-response. Only a few patients responded to these extended search procedures.
Norm sample
Norm data on the SF-36 was obtained from the Survey of Level of Living in Norway 1998 [18] comprising 6.638 participants aged 23 to 75 years. The norm data were adjusted by gender and distribution into 5-year age groups in relation to the PD sample.
Assessments
At baseline, diagnoses of PDs were made with the use of the Personality Disorder Examination, and Axis I-disorders were diagnosed by the MINI-International Neuropsychiatric Interview. Anamnestic data were collected, and global assessment of function was rated. The professional-based interviews and examinations of all patients at baseline and follow-up were carried out by a single experienced psychiatrist (KN), who did not take part in any treatment given.
All patients also filled in the following self-rating instruments at baseline: the SF-36, the Social Adjustment Scale, and the Symptom Checklist 90-Revised Personality Severity Index.
At follow-up all these assessments were repeated, and additional information about treatment as well as job/education, social- and family changes was collected.
Measures
Professional-rated
The Personality Disorder Examination (PDE) [19] is a structured clinical interview for PDs according to the DSM-III-R with good inter-rater reliability, and wide international application. Findings are reported as PD diagnoses, and as dimensional PD scores based on the sum of the scoring on each PD criterion (0: not present, 1: probably present, and 2: definitely present). Dimensional scores for the PD clusters are used as a main psychopathology variable, and the numbers of PDs are also reported.
The MINI International Neuropsychiatric Interview [20] was used to diagnose Axis-I disorders according to DSM-IV. The MINI covers 18 Axis-I disorders, has been translated into many languages and has demonstrated good inter-rater reliability. Findings are reported as numbers and percentages of patients with positive Axis-I diagnoses, and as mean number of such diagnoses.
The Global Assessment of Functioning (GAF) is a rating scale for the current evaluation of the overall functioning of a subject on a continuum from severe mental disorder to complete mental health that was defined as Axis V of the DSM-IV. Scale values range from 1 (sickest individual) to 100 (the healthiest person). The scale is divided in ten equal intervals from 1 – 10 to 91 – 100. Most outpatients will be rated between 40 and 70, although some individuals rated above 70 may seek therapy. The GAF is a reliable instrument [21], and the cut-off score for 'minimal impairment' has been set at 70 points or higher [22] and for 'serious mental disorder' at lower than 60 [3].
Patient-rated
The SF-36 [5] was chosen for measurement of health-related QoL, since it is in widespread use, and has shown good psychometric properties in Norway [23]. The SF-36 has demonstrated sensitivity to change, and score changes can be interpreted as changes in the health-related quality of life of the patient. The SF-36 assesses eight dimensions of physical and mental health, and the range is from 100 (optimal) to zero (poorest): physical functioning (PF), physical role functioning (RP), bodily pain (BP), general health (GH), vitality (VT), social functioning (SF), emotional role functioning (RE), and mental health (MH).
The Social Adjustment Scale (SAS-SR) [24] contains 42 questions which investigate expressive or instrumental roles in six major areas of functioning: work, social and leisure activities, relationship with extended family, role as a spouse/ partner, role as a parent, and role as member of the family unit. Each area is measured as a continuous variable, and the scales for individual items range from 1 (best) to 5 (worst). The scores within each role area are summed, and a mean for each area is obtained. By adding up the scores of all items and dividing by the number of items actually scored, an overall adjustment score is obtained.
The Symptom Checklist 90-Revised (SCL-90-R), Personality Severity Index (PSI). The SCL-90-R is a 90-item self-report inventory assessing current levels of mental symptoms patterns. Each item is a description of a mental symptom rated on a five-point scale, and rates the degree of 'distress/discomfort' during the last week prior to its administration. Several indices based on the SCL-90-R scores have been defined, and the PSI is the mean value of 22 items covering the interpersonal sensitivity, anger/ hostility, and paranoid ideation subscales. The PSI reflects the presence and severity of relatively enduring characteristics of the patient, and is, therefore, relevant for the evaluation of severe PDs [25]. For the PSI, pathology is defined by a cut-off score of ≥ 1.0.
Statistics
Data were analyzed by SPSS version 12.0. Descriptive statistics were conducted with independent and paired-sample t-test as well as one-way ANOVA (with Bonferroni's correction for multiple comparisons) for metric variables, and with χ2 and Fisher's Exact Test for categorical variables. The Mann-Whitney test and the Wilcoxon signed-rank test were applied for metric variables when the data distribution violated parametric assumptions. Spearman's rank correlation was used for associations. One-sample t-tests were applied when the SF-36 dimension scores of the age- and gender-adjusted norm groups were compared to the means of the three diagnostic groups. The comparison of mean scores on the SF-36 dimensions between the three groups was conducted with oneway ANOVA with Bonferroni's correction. The influence on the QoL scores of patients with PDs of Axis I disorders and of cluster C PDs in the cluster A+B PDs group, was examined with linear regression analyses. All tests were two-sided, and the level of significance was set at p < .05.
Results
Sample characteristics
All the 91 patients included were Caucasian, and 48 (53%) were females. The mean age of the sample at baseline was 36.3 years (SD 10.5) and ranging from 19 to 74 years, with a median of 35 years. None of the patients had any significant somatic diseases as reported by their GPs.
Patients belonging to the cluster A+B, cluster C, and Axis I- disorders groups did not differ at baseline on demographic variables (Table 1).
Table 1 Demographic and psychopathological features at baseline in patients with cluster A+B, and cluster C personality disorders, and non-psychotic Axis I disorders.
Variables Cluster A+B (n = 39) Cluster C (n = 33) Axis I (n = 19) p
Age (mean, SD) 35.1 (11.2) 36.6 (10.9) 41.1 (10.4) .15
Gender (n, %) .74
Male 17 (44) 17 (52) 8 (42)
Female 22 (55) 16 (48) 11 (58)
Relationship (n, %) .53
Paired 23 (59) 18 (55) 7 (37)
Non-paired 16 (41) 15 (45) 12 (63)
Basic education level (n, %) .28
≤ 9 years 15 (33) 13 (32) 8 (33)
10 – 12 years 14 (31) 16 (39) 13 (54)
≥ 13 years 16 (36) 12 (29) 3 (13)
Early childhood loss (n, %) 12 (31) 7 (21) 1 (5) .22
Childhood sexual abuse (n, %) 7 (18) 7 (21) 1 (5) .22
Age of onset mental problems (mean, SD) 16.7 (8.9) 20.0 (8.4) 27.0 (11.8) <. 001 A+B, C vs Axis I
Work income (n, %) 23 (59) 21 (64) 13 (68) .78
Economic support 16 (41) 12 (36) 6 (32)
Income last year (1.000 NOK) (mean, SD) 154 (74) 178 (70) 179 (62) .27
On sickleave last year (n, %) .37
No 18 (46) 10 (30) 5 (26)
≤ 12 weeks 11 (28) 12 (37) 5 (26)
≥ 13 weeks 10 (26) 11 (33) 9 (52)
No of PD cluster criteria (mean, SD)
Total 41.8 (16.2) 26.6 (11.9) 8.1 (9.2) < .001 A+B vs C vs Axis I
Cluster A 11.3 (8.2) 5.6 (4.6) 1.6 (2.0) < .001 A+B vs C, Axis I
Cluster B 14.7 (10.9) 4.8 (5.4) 2.6 (3.7) < .001 A+B vs C, Axis I
Cluster C 15.8 (10.0) 16.2 (5.7) 3.9 (5.4) < .001 A+B, C vs Axis I
Comorbid Axis I disorder (n, % .60
No 10 (26) 8 (24) -
Yes 29 (74) 25 (76)
GAF (mean, SD) 41.5 (9.1) 51.0 (7.4) 55.4 (9.3) < .001 A+B vs C, Axis I
PSI (mean, SD) 1.80 (.78) 1.24 (.71) 1.24 (.81) .005 A+B vs C, axis I
SAS (mean, SD) Overall adjustment 2.70 (.66) 2.59 (.59) 2.16 (.52) .007 A+B, C vs Axis I
Work 2.86 (1.62) 3.01 (1.80) 2.07 (.50) .94
Social and leisure 3.18 (1.18) 2.96 (1.06) 2.44 (.91) .06
Extended family 2.29 (.60) 2.00 (.51) 1.95 (.68) .05
Marital, partnership 2.21 (.56) 2.59 (.78) 2.14 (.63) .10
Parental 2.00 (.66) 1.91 (.58) 1.52 (.67) .14
Family unit 2.17 (.96) 2.21 (.95) 2.10 (.81) .94
As to psychopathology, the Cluster A and B PDs criteria sum scores were significantly higher in the Cluster A+B group compared to the two other groups, and both cluster A+B and cluster C group had significantly higher scores on cluster C criteria sum score than the Axis I group (Table 1). The Cluster A+B group had significantly lower GAF-score and higher PSI score than the two other groups which did not differ from each other. Mental problems had started significantly earlier in the PD groups compared to the Axis I-group. The SAS Overall adjustment was significantly poorer in the cluster A+B and cluster C groups, compared to the Axis I group.
The diagnostic distribution of PDs and Axis I disorders in the three groups at baseline and follow-up are given in Table 2. The mean number of PDs diagnosed in cluster A+B was 2.2, and in cluster C 1.3 per patient, and the most common PDs were avoidant, borderline, and dependent. One PD was observed in 37 patients (51%), two PDs in 20 (28%), and three PDs in 15 patients (21%). Fifty-four (75%) of the PD patient had at least one co-morbid Axis I-disorder.
Table 2 Diagnostic distribution of the baseline and follow-up samples.
Baseline Follow-up
Cluster A+B (n = 39) Cluster C (n = 33) Axis I (n = 19) Cluster A+B (n = 18) Cluster C (n = 21) Axis I (n = 11)
Personality disorders (DSM-III-R) N (%) N (%) N (%) N (%) N (%) N (%)
Paranoid 14 (36) - - 7 (39) 2 (10) 1 (9)
Schizotyp 6 (15) - - 2 (11) 1 (5) -
Schizoid 4 (10) - - 4 (22) - -
Antisocial 4 (10) - - 1 (6) - -
Narcissistic 0 (0) - - 0 (0) - -
Histrionic 4 (10) - - 0 (0) - -
Borderline 19 (49) - - 2 (11) - -
Dependent 5 (13) 7 (21) - 1 (6) 3 (14) -
Avoidant 15 (39) 25 (76) - 11 (61) 15 (71) 3 (27)
Obsessive-compulsive 5 (13) 8 (24) - 1 (6) 2 (10) -
Passive-aggressive 8 (21) 2 (6) - 1 (6) 2 (10) -
Mean no of PDs 2.2 1.3 - 1.7 1.2 .4
Axis I disorders (DSM-IV)
Major depression 6 (15) 9 (27) 3 (16) 0 (0) 1 (5) 1 (9)
Dysthymia 5 (13) 11 (33) 5 (26) 4 (22) 4 (19) 1 (9)
Panic disorder 7 (18) 7 (21) 2 (11) 0 (0) 2 (10) 1 (9)
Agoraphobia 5 (13) 7 (21) 2 (11) 2 (11) 2 (10) 2 (18)
Social phobia 9 (23) 8 (24) 2 (11) 2 (11) 4 (19) 1 (9)
GAD* 1 (3) 0 (0) 2 (11) 0 (0) 0 (0) 0 (0)
OCD* 4 (10) 2 (6) 2 (11) 0 (0) 0 (0) 1 (9)
Alcohol dependence 10 (26) 2 (6) 3 (16) 1 (6) 0 (0) 2 (18)
Substance dependence 7 (18) 0 (0) 2 (6) 0 (0) 1 (5) 0 (0)
Bulimia nervosa 4 (10) 1 (3) 0 (0) 0 (0) 1 (5) 0 (0)
PTSD* 0 (0) 1 (3) 0 (0) 0 (0) 0 (0) 0 (0)
Mean no of Axis I disorders 1.5 1.5 1.2 .5 0.7 0.8
*GAD: Generalized anxiety disorders, OCD: Obsessive-compulsive disorder, PTSD: Post-traumatic stress disorder
Axis-I disorders were equally common in the two PD groups (mean 1.5 disorder) and with slightly lower mean (1.2) in the Axis I-disorder group. Depressions, anxiety disorders, and alcohol dependence were the most frequent Axis I diagnoses in all groups.
QoL in PD patients
Figure 1. shows that the mean scores on the eight dimensions of the SF-36 of the PD patients at baseline are significantly lower (p < .001 for all) than those of the age- and gender-adjusted norms. The mean difference was least (13 points) for PF, and highest for RF and RE (54 points and 49 points, respectively).
Figure 1 SF-36 mean dimensional scores in personality disorder sample (N = 72) and the age- and gender-adjusted norm sample.
Linear regression analyses showed that control for co-morbid Axis I disorders reduced the PF, GH, VT, SF, and MH scores of the total PD group significantly. Controlling for comorbid cluster C PDs did not influence the SF-36 scores of the cluster A+B PDs to any significant extent, while control for Axis disorders significantly reduced the MH scores in the cluster A+B and cluster C groups.
No significant differences were found between genders on any of the SF-36 dimensions among the PD patients (data not shown). Both the cluster A+B, the cluster C, and the Axis I group differed significantly from their norms on all eight SF-36 dimensions (data not shown). No significant differences were observed on the eight SF-36 dimensions between the three diagnostic groups (Figure 2).
Figure 2 SF-36 mean dimensional scores at baseline for Cluster A+B, Cluster C, and Axis I-disorders groups.
When we compared the patients with one (n = 37), two (n = 20), and three or more (n = 15) PDs, we did not observe any significant differences in mean MCS and PCS scores.
Correlation between SF-36 dimensions and other measures
The eight dimensions of SF-36 are regularly divided into the four physical: PF, RP, BP, and GH, and the four mental ones: VT, SF, RE, and MH. In our PD sample the SF-36 mental dimensions had most significant correlations with the psychiatric measures of the GAF, the SCL-90-R PSI, the sum of positive PDs diagnostic criteria, and the dimensions of the SAS (Table 3). The SF-36 MH had a significant correlation to most of these measures, followed by VT and SF. The physical dimensions of the SF-36 had less frequently a significant correlation to the psychiatric measures.
Table 3 Correlation of SF-36 dimensions with Global Assessment of Functioning (GAF), Personality Severity Index (PSI), and dimensions of Social Adjustment Scale (SAS).
PF RP BP GH VT SF RE MH
GAF .12 .32** .02 .32** .26* .36** .30** .40**
Total no of positive PD criteria .03 -.11 .02 -.27** -.14 -.18 -.05 -.26**
SCL-90-R PSI -.22* -.17 -.12 -.26* -.25* -.34** -.19 -.38**
SAS Overall -.38** -.31** -.23* -.34** -.42** -.45** -.30** -.45**
SAS Work -.17 -.22* -.10 -.16 -.44** -.31* -.26* -.41**
SAS Social and leisure -.30** -.35** -.24* -.32** -.43** -.39** -.29** -.37**
SAS Extended family -.20 -.18 -.23* -.21 -.26* -.32* -.28* -.43**
SAS Marital/partner -.18 .01 -.04 -.18 -.24 -.08 -.05 -.10
SAS Parental -.04 -.06 .06 .02 -.07 -.09 -.07 -.04
SAS Family unit -.38** -.22 -.26* -.20 -.37** -.31* -.27* -.35**
Sum significant correlation 4 4 4 4 6 7 6 8
* Correlation is significant at the .05 level (two-tailed), ** Correlation is significant at the .01 level (two-tailed)
The SAS overall adjustment and the SAS social and leisure functioning had significant correlations to all the SF-36 dimensions, while the SAS marital/ partnership and the SAS parental functioning had none. The SAS work, extended family, and family unit fell in between.
What changes in the QoL of PD patients can be observed from baseline to follow-up two years later?
Although quite intensive search for patients was done for the follow-up examination, only 50 patients (53%) of the 91 patients who rated themselves on the SF-36 complied. The distribution of patients were cluster A+B group (n = 18), cluster C (n = 21), and Axis I-disorder group (n = 11). Due to small sample sizes, the Axis I disorder was dropped from further analysis and the two PD groups were pooled as to the study of changes after treatment. The 39 PD patient with SF-36 ratings both at baseline and follow-up were compared to the 33 PD patients only seen at baseline. The diagnoses at follow-up are shown in Table 2, and among the non-compliant patients those with borderline PD, and alcohol and substance dependence were over-represented.
Few significant differences were observed between compliant and non-compliant PD patients at follow-up (Table 4). In particular, no significant differences of the eight SF-36 dimensions were observed between the compliers and non-compliers at baseline. The compliers had significantly more depressive disorders and cluster C PDs at baseline. All of those who terminated treatment without the consent of their therapist (N = 22) were in the non-compliant group. The non-compliant patients also had a significantly longer mean duration of treatment. The mean treatment time for the PD patients attending follow-up was 16.6 months (SD 5.9), median 18 months, and range 4 to 24 months, and the mean follow-up time since treatment termination was 9.8 months (SD 6.4), median 10.4 months, and range 0 to 26 months. The majority of the patients had weekly individual psychotherapy, although a small proportion also had group psychotherapy in addition. Drug treatment was given to 20 patients of the 39 patients, and to 20 of the 33 non-compliers (ns). Among those seen at follow-up, had 15 got antidepressive and 5 antipsychotic medication, in addition to psychotherapy.
Table 4 Demographic, psychopathological, and treatment features at baseline for patients with personality disorders with (N = 39) and without (n = 33) follow-up examination.
Variable Follow-up + (n = 39) Follow-up - (n = 33) p
Age (mean, SD) 37.9 (11.6) 33.3 (9.9) .08
Gender (n, %) .50
Male 17 (44) 17 (52)
Female 22 (56) 16 (48)
Relationship (n, %) .56
Paired 18 (46) 13 (39)
Non-paired 21 (54) 20 (61)
Basic education level (n, %) .64
≤ 9 years 11 (29) 13 (40)
10 – 12 years 13 (34) 10 (30)
≥ 13 years 14 (37) 10 (30)
≥ 1 cluster A PDs 13 (33) 6 (18) .15
≥ 1 cluster B PDs 10 (26) 15 (45) .08
≥ 1 cluster C PDs 35 (90) 22 (67) .02
Mean (SD) of PD cluster criteria
Total 34.8 (17.9) 34.8 (14.3) .99
Cluster A 9.5 (8.7) 7.7 (5.1) .28
Cluster B 7.6 (9.3) 13.1 (10.3) .02
Cluster C 17.6 (8.9) 14.0 (7.2) .07
≥ 1 depressive disorder 19 (49) 7 (21) .02
≥ 1 anxiety disorder 17 (44) 16 (49) .68
≥ 1 substance use disorder 9 (23) 11 (33) .33
Comorbid Axis I disorder (n, %) .79
No 9 (23) 9 (27)
Yes 30 (77) 24 (33)
GAF (mean, SD) 46.0 (9.4) 45.7 (9.9) .91
PSI (mean, SD) 1.5 (.9) 1.6 (.7) .85
SAS Overall (mean, SD) 2.7 (.6) 2.6 (.6) .86
SF-36 (mean, SD)*
Physical Functioning 79.4 (19.1) 76.7 (22.5) .72
Role Functioning 31.4 (34.3) 25.8 (36.2) .27
Bodily Pain 47.6 (28.9) 48.1 (26.7) .71
General Health 51.4 (23.5) 50.5 (20.39 .99
Vitality 35.0 (19.6) 29.1 (18.6) .16
Social Functioning 45.2 (28.5) 48.9 (21.5) .60
Role Emotional 42.7 (39.7) 32.3 (31.7) .35
Mental Health 42.5 (23.2) 38.3 (19.5) .52
No of sessions (mean, SD)* 16.6 (5.9) 18.8 (26.9) .01
Termination without consensus (n, %) 0 (0.0) 22 (67) < . 001
Treated by specialist (n, %) 14 (36) 16 (49) .28
Additional drug treatment (n, %) 20 (51) 20 (61) .43
*Mann-Whitney tests
In the 39 PD patients who complied at both baseline and follow-up, significant improvement was seen in the RF and MH dimensions of the SF-36, while considerable, but non-significant changes were observed for BP and SF (Table 5).
Table 5 Changes from baseline to follow-up in patients with personality disorders (n = 39).
Measure Baseline Mean (SD) Follow-up Mean (SD) P
SF-36
Physical Functioning 79.4 (19.2) 76.8 (24.6) .95
Role Physical 31.4 (34.3) 51.3 (38.5) .01
Bodily Pain 47.6 (28.9) 57.5 (25.6) .06
General Health 51.4 (23.5) 56.0 (26.6) .22
Vitality 35.0 (19.6) 36.3 (21.4) .70
Social functioning 45.2 (28.5) 53.5 (28.7) .09
Role-emotional 42.7 (39.7) 41.9 (38.0) .89
Mental Health 42.5 (23.2) 50.1 (22.3) .03
Global Assessment of functioning 46.0 (9.4) 54.6 (9.6) < .001
Total no of PD criteria 34.8 (17.9) 25.7 (11.5) < .001
SCL-90-R PSI 1.52 (.86) 1.30 (.80) .035
Social Adjustment Scale (SAS)
Overall adjustment 2.66 (.63) 2.42 (.62) .007
Work 2.76 (1.49) 2.36 (1.39) .20
Social and leisure 3.17 (1.22) 2.87 (1.11) .045
Extended family 2.05 (.52) 1.97 (.50) .34
Both professional-rated measures the GAF, and the mean total number of PD criteria, showed significant improvement. Among the patient-rated measures significant better results at follow-up were found for the SCL-90-R PSI, the SAS overall adjustment, and the SAS social and leisure scales.
Discussion
The main findings of this study of mainly co-morbid PD patients treated at an ordinary POC, was that the QoL on both the physical and mental SF-36 dimensions was significantly lower than that of an age- and gender-adjusted general population sample. According to our knowledge, ours is the first report on QoL-data in such PD patients at a POC. This finding is in accordance with QoL studies of PD patients in the general population [11,13], and correspond to findings of clinical studies of patients with anxiety disorders, depression, schizophrenia, and substance dependence [27-30]. However, the SF-36 dimension mean scores of our PD sample are lower than those reported for these diagnoses, and for co-morbid disorders [31]. In our sample we did not find any significant differences between the SF-36 dimension scores of the cluster A+B, cluster C, or Axis I groups, and all groups had significantly lower scores on all dimensions than their age- and gender-adjusted norm groups.
In contrast to the epidemiological study from Australia [12] we did not find worsening of MCS and PCS with increasing number of PDs present in our sample. This could be due to our small samples, but also due to the fact that our patients with 1 PD had considerably lower QoL than in the Australian survey [MCS: 33.7 (SD 10.6) versus 44.4 (SD 12.0), p < .001, and PCS: 43.8 (SD 8.6) versus 46.9 (SD 11.0), p = .03].
Comorbid Axis I disorders explained a significant part of scores of PF, GH, VT, SF, and MH scores of the total PD group. This is in accordance with the findings of the Australian study [12].
We found that the SF-36 dimensions had variable associations with established psychiatric measures. As expected the SF-36 MH was most strongly associated with the psychiatric measures, but so were also SF and VT. For the SAS we found that overall adjustment and social and leisure activities were significantly correlated to all the SF-36 dimensions. In our PD sample we observed a somewhat different pattern of significant correlations between the GAF and the SF-36 dimensions than reported by Meijer et al. [32] in patients with schizophrenia. Small sample sizes and different diagnostic classes could be the explanation. However, in sum the SF-36 had a considerable association with established psychiatric measures in our PD sample.
For both the patient- and professional-rated psychiatric measures significant changes at follow-up after treatment was observed in the 39 patients who also scored themselves on the SF-36. We cannot say if these changes were related to treatment, and ours is not an outcome study. We wanted to examine if changes in established psychiatric measures were associated with changes in the QoL measured by the SF-36 in the PD patients seen at a POC.
Significant changes at follow-up were found for only two of the SF-36 dimensions, however, one physical (RP) and one mental (MH). While the finding for MH was expected, the change in RP which covered problems with work or other daily activities as a result of physical health was more difficult to explain. The score on that dimension was extraordinarily low at baseline (mean 31.4), and regression towards mean could be a likely explanation. It seemed that only MH of the SF-36 changed in the same way as established psychiatric measures in our study. The SF-36 MH correlated significantly with most of such psychiatric measures, and MH is currently used as a valid measure for mental health in several studies [33].
This result could indicate that the other dimensions of the SF-36 are less valid as measures of changes in mental health of PD patients, or alternatively that most aspects of QoL measured by the SF-36 do not change in PD patients even if established psychiatric measures do.
The main strength of our study was that we were able examine systematically various aspects of the QoL measured by the SF-36 in a clinically relevant sample of PD patients at a POC which is a common setting for such patients in psychiatry.
Our study had a number of weaknesses. The study groups were small with limited statistical power, and there was a considerable risk of type II errors. More significant differences as to the SF-36 dimensions could turn up in larger samples. Although we put considerable efforts into location of patients, we had a lower follow-up rate than we had expected. However, the PD patients who did not show up at follow-up did not differ much from those who did. We cannot, therefore, generalize the discrepancy observed between significant changes in established psychiatric measures and lack of such changes in most of the SF-36 dimensions of PD patients treated at a POC to widely.
The same experienced psychiatrists did all the interviews at baseline and follow-up. Although she was not involved in any treatments, we cannot exclude an expectation bias from her side.
We think that our study has to be considered an exploratory one. Our finding of a generally strongly reduced QoL should be replicated in a PD sample with less comorbid Axis I disorders, although their influence was limited. The same is true for QoL as a valid measure for change in PD patient, since it was not recommended as part of a standard outcome battery and was not used by major treatment studies of PD patients. However, our study confirmed that the SF-36 MH dimension seemed to be a valid psychiatric measure in our PD patient sample.
Conclusion
In this study of the QoL in PD patients seen at an ordinary POC, we found that the PD patients had significantly lower mean scores on all the SF-36 dimensions compared to age- and gender-adjusted norm data. This is in accordance with the SF-36 measurements of other major diagnostic groups of mental disorders. Although the SF-36 dimensions correlated considerably with established psychiatric measures in our PD patients, they did not show the same significant changes over time as the established measures. The use of QoL measures like the SF-36 as an outcome measure in PD patients is in need of further investigation.
List of abbreviations
BP: SF-36 Bodily pain
GAF: The Global Assessment of Functioning
GH: SF-36 General health
MH: SF-36 Mental health
PD: Personality disorder
PDs: Personality disorders
PF: SF-36 Physical functioning
POC: Psychiatric outpatient clinic
PSI: Personality severity index of SCL-90-R
QOL: Health-related quality of life
RE: SF-36 Emotional role functioning
RP: SF-36 Physical role
SAS: The Social Adjustment Scale
SCL-90-R: The Symptom Checklist 90-Revised
SF-36: MOS Short Form 36
SF: SF-36 Social functioning
VT: SF-36 Vitality
Competing interests
The author(s) declare that the have no competing interests.
Authors' contributions
KN conceived and planned the study, prepared the therapists at Furuset Outpatient Department, did all the psychiatric interviews at baseline and follow-up, and drafted the manuscript. AM helped designing of the study, supervised the statistic calculations, and drafted the manuscript. AAD participated in the design and coordination of the study, and drafted 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
Kjersti Narud, MD got research grants from Sommer's Legacy, Andresen's Legacy, Nilsen's Legacy, and the Department of Psychiatry, Aker University Hospital.
==== Refs
American Psychiatric Association Diagnostic and statistical manual of mental disorders 1994 4 Washington, DC: American Psychiatric Press
Beals J Novins DK Spicer P Orton HD Mitchell CM Barón AE Manson SM the AI-SUPERPFP Team Challenges in operationalizing the DSM-IV clinical significance criterion Arch Gen Psychiatry 2004 61 1197 1207 15583111 10.1001/archpsyc.61.12.1197
Kessler RC Barker PR Colpe LJ Epstein JF Gfroerer JC Hiripi E Howes MJ Normand SL Manderscheid RW Walters EE Zaslawsky AM Screening for serious mental illness in the general population Arch Gen Psychiatry 2003 60 184 189 12578436 10.1001/archpsyc.60.2.184
Fayers PM Machin D Quality of life Assessment, analysis and interpretation 2000 Chicester: Wiley
Ware JE Snow KK Kosinski M SF-36® health survey: manual and interpretation guide 2000 Lincoln, RI: Quality Metric Inc
Ware J JrKosinski M Keller SD A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity Med Care 1996 34 220 233 8628042 10.1097/00005650-199603000-00003
Gunderson JG Shea MT Skodol AE McGlashan TH Morey LC Stout RL Zanarini MC Grilo CM Oldham JM Keller MB The Collaborative Longitudinal Personality Disorder Study: development, aims, design, and sample characteristics J Personal Disord 2000 14 300 315 11213788
Karterud S Pedersen G Bjordal E Brabrand J Friis S Haaseth Ø Haavaldsen G Irion T Leirvåg H Tørum E Urnes Ø Day treatment of patients with personality disorders: experiences from a Norwegian treatmentresearch network J Person Disord 2003 17 243 262 10.1521/pedi.17.3.243.22151
Chiesa M Fonagy P Psychosocial treatment for severe personality disorder Br J Psychiatry 2003 183 356 362 14519615 10.1192/bjp.183.4.356
Strupp HH Horowitz LM Lambert MJ Eds Measuring patient changes in mood, anxiety, and personality disorders Towards a core battery 1997 Washington, DC: American Psychological Association
Jackson HJ Burgess PM Personality disorders in the community: a report from the Australian National Survey of Mental Health and Wellbeing Soc Psychiatry Psychiatr Epidemiol 2000 35 531 538 11213842 10.1007/s001270050276
Jackson HJ Burgess PM Personality disorders in the community: results from the Australian National Survey of Mental Health and Wellbeing. Part II. Relationship between personality disorder, Axis I mental disorders and physical conditions with disability and health consultations Soc Psychiatry Psychiatr Epidemiol 2002 37 251 260 12111029 10.1007/s001270200017
Grant BF Hasin DS Stinson FS Dawson DA Chou SP Ruan WJ Pickering RP Prevalence, correlates, and disability of personality disorders in the United States: results from the national epidemiologic survey on alcohol and related conditions J Clin Psychiatry 2004 65 948 958 15291684
Abrams RC Alexopoulos GS Spielman LA Klausner E Kakuma T Personality disorder symptoms predict declines in global functioning and quality of life in elderly depressed patients Am J Geriatr Psychiatry 2001 9 67 71 11156754 10.1176/appi.ajgp.9.1.67
Swinton M Oliver J Carlisle J Measuring quality of life in secure care: comparison of mentally ill and personality disordered patients Int J Soc Psychiatary 1999 45 284 291
Hueston WJ Mainous AG IIISchilling R Patients with personality disorders: functional status, health care utilization, and satisfaction with care J Fam Pract 1996 42 54 60 8537806
Nakao K Gunderson JG Phillips KA Tanak N Yorifuji K Takaish J Nishimura T Functional impairment in personality disorders J Person Disord 1992 6 24 33
Heine Strand B Dalgard OS Tambs K Rognerud M Measuring the mental health status of the Norwegian population: a comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36) Nord J Psychiat 2003 57 113 118 10.1080/08039480310000932
Loranger AW Sartorius N Andreoli A Berger P Buchheim P Channabasavanna SM Coid B Dahl A Diekstra RF Ferguson B The International Personality Disorder Examination. The WHO/ADAMHA International Pilot Study of Personality Disorders Arch Gen Psychiatry 1994 51 215 224 8122958
Lecrubier Y Sheehan DV MINI International Neuro-psychiatric Interview Version 40, Norwegian edition 1996 Oslo: Department of Psychosomatics, National Hospital
Hilsenroth MJ Ackerman SJ Blagys MD Baumann BD Baity MR Smith SR Price JL Smith CL Heindselman TL Mount MK Holdwick DJ Jr Reliability and validity of DSM-IV Axis V Am J Psychiatry 2000 157 1858 1863 11058486 10.1176/appi.ajp.157.11.1858
Moos RH Nichol AC Moos BS Global assessment of functioning ratings and the allocation and outcomes of mental health services Psychiatr Serv 2002 53 730 737 12045311 10.1176/appi.ps.53.6.730
Loge JH Kaasa S Jensen Hjermstad M Kvien TK Translation and performance of the Norwegian SF-36 health survey in patients with rheumatoid arthritis. I. Data quality, scaling assumptions, reliability, and construct validity J Clin Epidemiol 1998 51 1069 1076 9817124 10.1016/S0895-4356(98)00098-5
Weissmann MM Sholomkas D John K The assessment of social adjustment: an update Arch Gen Psychiatry 1981 38 1250 1258 7305605
Karterud S Friis S Irion T Vaglum P A SCL-90-R derived index of the severity of personality disorders J Person Disord 1995 9 112 123
Simon NM Otto MW Korbly NB Peters PM Nicolaou DC Pollack MH Quality of life in social anxiety disorder compared with panic disorder and the general population Psychiatr Serv 2002 53 714 718 12045308 10.1176/appi.ps.53.6.714
Papakostas GI Petersen T Mahal Y Mischoulon D Nierenberg AA Fava M Quality of life assessments in major depressive disorder: a review of the literature Gen Hosp Psychiatry 2004 26 13 17 14757297 10.1016/j.genhosppsych.2003.07.004
Tunis SL Croghan TW Heilman DK Johnstone BM Obenchain RL Reliability, validity and application of the Medical Outcome Study 36-item short-form health survey (SF-36) in schizophrenic patients treated with olanzapine versus haloperidol Med Care 1999 37 678 691 10424639 10.1097/00005650-199907000-00008
Richter D Eikelmann B Berger K Use of the SF-36 in the evaluation of a drug detoxification program Qual Life Res 2004 13 907 914 15233504 10.1023/B:QURE.0000025589.07313.46
Bijl RV Ravelli A Current and residual functional disability associated with psychopathology: findings from the Netherlands Mental Health Survey and Incidence Study (NEMESIS) Psychol Med 2000 30 657 668 10883720 10.1017/S0033291799001841
Meijer CJ Schene AH Koeter MWJ Quality of life in schizophrenia measured by the MOS SF-36 and the Lancashire quality of life profile: a comparison Acta Psychiatr Scand 2002 105 293 300 11942934 10.1034/j.1600-0447.2002.1198.x
Edwards VJ Holden GW Felitti VJ Anda RF Relationship between multiple forms of childhood maltreatment and adult mental health in community respondents: results from the Adverse Childhood Experiences Study Am J Psychiatry 2003 160 1453 1460 12900308 10.1176/appi.ajp.160.8.1453
| 15720726 | PMC552308 | CC BY | 2021-01-04 16:33:03 | no | BMC Psychiatry. 2005 Feb 20; 5:10 | utf-8 | BMC Psychiatry | 2,005 | 10.1186/1471-244X-5-10 | oa_comm |
==== Front
BMC Pulm MedBMC Pulmonary Medicine1471-2466BioMed Central London 1471-2466-5-31570748410.1186/1471-2466-5-3Research ArticleEffects of inhaled corticosteroids on sputum cell counts in stable chronic obstructive pulmonary disease: a systematic review and a meta-analysis Gan Wen Qi [email protected] SF Paul [email protected] Don D [email protected] From the James Hogg iCAPTURE Center for Cardiovascular and Pulmonary Research, St. Paul's Hospital and the Department of Medicine (Pulmonary Division), University of British Columbia, Vancouver, B.C., Canada2005 11 2 2005 5 3 3 13 9 2004 11 2 2005 Copyright © 2005 Qi Gan et al; licensee BioMed Central Ltd.2005Qi Gan 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
Whether inhaled corticosteroids suppress airway inflammation in chronic obstructive pulmonary disease (COPD) remains controversial. We sought to determine the effects of inhaled corticosteroids on sputum indices of inflammation in stable COPD.
Methods
We searched MEDLINE, EMBASE, CINAHL, and the Cochrane Databases for randomized, controlled clinical trials that used induced sputum to evaluate the effect of inhaled corticosteroids in stable COPD. For each chosen study, we calculated the mean differences in the concentrations of sputum cells before and after treatment in both intervention and control groups. These values were then converted into standardized mean differences to accommodate the differences in patient selection, clinical treatment, and biochemical procedures that were employed across original studies. If significant heterogeneity was present (p < 0.10), then a random effects model was used to pool the original data. In the absence of significant heterogeneity, a fixed effects model was used.
Results
We identified six original studies that met the inclusion criteria (N = 162 participants). In studies with higher cumulative dose (≥ 60 mg) or longer duration of therapy (≥ 6 weeks), inhaled corticosteroids were uniformly effective in reducing the total cell, neutrophil, and lymphocyte counts. In contrast, studies with lower cumulative dose (< 60 mg) or shorter duration of therapy (< 6 weeks) did not demonstrate a favorable effect of inhaled corticosteroids on these sputum indices.
Conclusions
Our study suggests that prolonged therapy with inhaled corticosteroids is effective in reducing airway inflammation in stable COPD.
==== Body
Background
Chronic obstructive pulmonary disease (COPD) is characterized by prominent airway inflammation [1,2]. The intensity of the inflammation strongly correlates with disease severity [3,4] and increases even further during exacerbations [5]. Moreover, increased expression of inflammatory markers in the sputum is associated with increased risk of exacerbations [6]. The attenuation of the inflammatory process, on the other hand, is associated with improvements in lung function and airway hyperresponsiveness in COPD [7]. It is possible therefore that the inflammatory process is an integral component in COPD pathogenesis and may represent an important therapeutic target in improving the health status and outcomes of COPD patients [1,2].
One potential therapy for down-regulating the inflammatory process in the airways is through the use of corticosteroids, which are potent but non-specific anti-inflammatory agents. Some in vitro studies have demonstrated that inhaled corticosteroids can modulate certain aspects of the inflammatory cascade in COPD [8,9]; however, other studies have shown less favorable results [10,11]. Despite this uncertainty, large clinical trials have shown that these medications reduce clinically relevant exacerbations by ~30% and improve health status of patients with moderate to severe disease [12]; their withdrawal, on the other hand, leads to increased risk of exacerbations and worsening of health status [13]. Since airway inflammation is associated with exacerbations [6] and since inhaled corticosteroids reduce exacerbations [12], they may also have salutary effect on airway inflammation in COPD. However, to date, the clinical studies, which have addressed this issue, have been small in size and scope and may not have had sufficient statistical power (on their own) to detect subtle but important effect of these medications on inflammatory indices in the airways. Additionally, there may be important methodologic differences between the positive and negative studies that could potentially explain the discrepancy. We, therefore, conducted a systematic review and a meta-analysis to determine whether inhaled corticosteroids do or do not suppress airway inflammation in patients with stable COPD and to explore the potential causes for the heterogeneity in reports.
Methods
Search for relevant studies
MEDLINE (1966–2004), EMBASE (1980–2004), CINAHL (1982–2004), and the Cochrane Databases were searched for randomized, controlled clinical trials that used induced sputum to evaluate the effect of inhaled steroids on airway inflammation in stable COPD. The search was restricted on articles published in the English language, using human participants. Subject headings included disease-specific search terms (COPD, lung diseases, pulmonary diseases, airway obstruction, obstructive pulmonary disease, chronic obstructive pulmonary disease, bronchitis, emphysema, pulmonary emphysema, or mediastinal emphysema), drug-specific search terms (glucocorticosteroids, corticosteroids, beclomethasone, budesonide, fluticasone, or triamcinolone), and laboratory method-specific search terms (biopsy, bronchoalveolar lavage, or sputum). We also scanned the bibliographies and reference lists of retrieved articles to supplement the electronic searches. We contacted the primary authors for additional data and/or clarification of data.
Study selection and data abstraction
The primary objective of this meta-analysis was to compare the changes in sputum inflammatory indices among stable COPD patients before and after treatment with inhaled corticosteroids, using the control group in each individual studies as the referent. We chose sputum as the primary source of the analysis because there was a marked scarcity of quality studies which had evaluated the effect of inhaled corticosteroids from bronchoalveolar lavage fluid or tissue biopsy specimens. The inflammatory indices included total cell, neutrophil, macrophage, eosinophil, lymphocyte, and epithelial cell counts and interleukin (IL)-8 levels. Since the actions of oral corticosteroids may differ from those of inhaled corticosteroids, we excluded studies that evaluated the effects of oral corticosteroids on sputum inflammatory indices. From each selected article, two investigators (WQG, DDS) abstracted the following baseline information: the source of data, study design, inclusion and exclusion criteria, concomitant drugs, demographics of study participants including sample size, age, sex, current smoking status, pack-years of smoking history, predicted forced expiratory volume in one second (FEV1), the ratio of FEV1 to forced vital capacity (FVC), percent predicted reversibility with inhaled bronchodilator, the specific brand of inhaled corticosteroids and the dose as well as the duration of therapy. Cumulative dose of inhaled corticosteroids was calculated by multiplying the average daily dose by the total days of treatment. All formulations were converted to beclomethasone equivalent based on the recommendations from the Canadian Asthma Consensus Report [14]. Any questions or discrepancies were resolved through iteration and consensus.
Statistical methods
To accommodate any differences in patient selection, clinical treatment, and biochemical procedures that were employed across the original studies, we converted the absolute mean differences in the concentrations of the inflammatory cells between the intervention and control groups into standardized mean differences. For each study, standardized mean difference was derived by dividing the mean change in the inflammatory cell concentration at follow-up visit from the baseline visit between intervention and control groups by a pooled standard deviation of the mean change [15,16]. A negative standardized mean difference indicated that the participants assigned to inhaled corticosteroids had lower cell counts compared with placebo at the end of the study phase; whereas a positive number denoted increased cell count relative to the control group. For each inflammatory cell, we tested the heterogeneity of results across the studies, using a Cochran Q test. If significant heterogeneity was present (p < 0.10), then a random effects model was used. In the absence of significant heterogeneity, a fixed effects model was used [16]. We also evaluated the potential modifying effect of cumulative dose and the duration of therapy of the trials. We reasoned that trials that had higher cumulative dose (or longer duration of therapy) defined as greater or equal to the median cumulative dose (or duration of therapy) of all the trials included in this meta-analysis would be more "positive" than those that used lower doses (or were shorter in duration). All analyses were conducted using Review Manager version 4.2 (Revman; The Cochrane Collaboration, Oxford, England) and were two-tailed in nature.
Results
A summary of the search strategy is shown in Figure 1. The original search yielded 155 and 63 citations in MEDLINE and EMBASE, respectively. CINAHL and the Cochrane Databases did not contribute to the search results. The abstracts of these articles were selected and reviewed. Of these, 21 articles were retrieved for a detailed review. We excluded the study from Loppow and colleagues [17] because it included 6 patients with a positive skin prick test against at least one common airborne allergen and 4 patients who had FEV1/FVC > 0.7. We excluded additional 14 articles because of other reasons (Figure 1). This process left 6 original studies meeting the inclusion and exclusion criteria, which were used for the analyses [7,18-22]. The baseline information concerning the study designs is summarized in Table 1. The relevant demographic data are summarized in Table 2. All 162 patients were current or ex-smokers with post-bronchodilator FEV1 <70 % predicted, FEV1 to FVC ratio <0.7, and reversibility with bronchodilator of <15%. The medications used included budesonide, beclomethasone dipropionate, and fluticasone propionate. The study period of these trials ranged from 2 to 12 weeks.
Figure 1 Study selection process
Table 1 Baseline information on original studies included in the meta-analysis.
Source Setting Design Inclusion Criteria Exclusion Criteria Concomitant drugs Withdrawal Sputum specimen
Sugiura et al 2003 [7] NR Randomized, placebo-controlled parallel design. FEV1/FVC < 0.7; all patients wereex-smokers who had stopped smoking for at least 1 year beforethe study. A history of perennial allergic rhinitis; positiveallergen skin prick tests and RAST assay; a history of periodicwheezing; an improvement in FEV1 of more than 12 % predicted oran absolute increase of 200 ml after inhalation of 200 μg salbutamol; had bronchial or respiratory tract infectionsin the month preceding the study; had taken systemic steroids in the 2 monthsbefore the study or inhaled steroids in the month beforethe study. NR None NR
Keatings et al 1997 [18] Outpatient clinics in different hospitals Randomized, single-blind, crossover design with 3–7 day run in period. The clinical part of the study was single-blind, but all differential cell counting and assayswere carried out in a double blind fashion. FEV1/FVC < 0.7; FEV1 < 70% predicted; reversibility with inhaled albuterol of <10% of predicted FEV1; smoking history of at least 10 pack-years; negative results on skin prick testing to four common aeroallergens. Patients who had taken inhaled or oral steroids or who had suffered an exacerbation of their airway disease in the previous 6 weeks, or patients with any history of asthma or variability in symptoms were excluded. Albuterol was allowed. 2 subjects NR
Culpitt et al 1999 [19] Outpatient clinic Randomized, double-blind, placebo-controlled crossover design with a run-in period of 2 weeks. FEV1/FVC < 0.7; postbronchodilat or FEV1 <85% predicted; reversibility with inhaled β2-agonist of <15% of predicted FEV1; smoking history of at least 20 pack-years. Patients who had taken inhaled or oral steroids or who had suffered an exacerbation of their airway disease in the previous 6 weeks, or patients with any history of asthma or atopy or variability in symptoms were excluded. Three subjects had concomitant treatment with albuterol (200 μg twice a day) and ipratropium bromide (40 μg twice a day), one subject with albuterol (200 μg as needed) alone. 12 subjects Samples were considered adequate for analysis if there was < 50% squamous cell contamination.
Confalonieri 1998 [20] Outpatient clinic Randomised, controlled, open study. The clinical parts of the study was open, but all differential cell counting was carried out in a double blind fashion. FEV1/FVC <88% of predicted in men and <89% in women; all patients were current smokers. Patients who had taken inhaled or oral steroids or had suffered a respiratory tract infection in the previous three months were excluded. None of the patients was taking theophyllines or long acting β2 agonists. None Samples were discarded if viability levels were 50% or less, or squamous contamination was 20% or more. An overall differential cell count on 500 nucleated non-squamous cells was performed by two examiners and results reported as mean of the two counts.
Mirici et al 2001 [21] Outpatient clinic Randomized, double-blind, placebo-controlled parallel design. FEV1 < 70% predicted; no self-reported asthma; reversibility with inhaled terbutaline of <15% of predicted FEV1; current smokers. Long-term treatment with oral or inhaled steroids within 6 months of study entry; A respiratory tract infection in previous 3 months; pregnancy or lactation, or presence of other serious systemic diseases. β2 – agonists of all kinds, theophylline, and mucolytics were allowed. 10 subjects Samples were discarded if viabilitylevels were 50% or less, or squamous contamination was 20% or more
Yildiz et al 2000 [22] Outpatient clinic Randomized, placebo-controlled parallel design with a run-in period of 2 weeks. FEV1/FVC < 0.7; FEV1 < 70% predicted; reversibility with inhaled albuterol of <10% of predicted; smoking history of at least 10 pack-years. Patients with any history of asthma or variability in symptoms, and patients who had taken inhaled or oral steroids or had suffered a respiratory tract infection or exacerbation in the previous 6 weeks were excluded. All of the patients continued to inhale both salbutamol and ipatropium bromide. In 9 patients, sustained release theophylline was also administered. None NR
Abbreviations: FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; NR, not reported.
Table 2 The characteristics of COPD patients at baseline.
Source Number of Patients Age (year) Men (%) Current Smokers (%) Pack-years FEV1 (% predicted) Ratio (%) Reversibility (% predicted) Drug Dose (mg/day) Duration (weeks) Cumulative dose (mg) #
Sugiura [7] 18‡ 70(7) 89 0* NR 1.2(0.4)† <70 <12 Beclomethasone 0.8 4 22.4
Keatings [18] 26 45–78 60 46 >10 35.1(4.7) <70 <10 Budesonide 1.6 2 28.0
Culpitt [19] 26 43–73 62 69 >20 49.5(16.6) <70 <15 Fluticasone 1.0 4 56.0
Confalonieri [20] 34 58 (5) 59 100 NR 59.7(37.1) 67 (5) NR Beclomethasone 1.5 8 84.0
Mirici [21] 40 53(10) 75 100 26.5 (16.1) 62.0(7.4) NR <15 Budesonide 0.8 12 84.0
Yildiz [22] 18 64(7) 78 89 52.0 (23.4) 44.5(2.7) 57 (3) <10 Fluticasone 1.5 8 168.0
† FEV1, liter;
‡ 6 patients in control group
* All subjects were ex-smokers and stopped smoking for at least 1 year.
# Cumulative dose = daily dose × days × adjusted factor for beclomethasone equivalence [14].
Continuous variables are presented as mean (SD)
Abbreviations: FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; ratio, the ratio of FEV1 to FVC; NR, not reported/not calculable.
After treatment with inhaled corticosteroids, the total cell counts decreased. Overall, the standardized mean difference between steroid and control groups was -0.43 units (95% confidence interval, CI, -0.75 to -0.11), indicating that inhaled corticosteroids had a favorable effect in reducing total count compared with controls (test for heterogeneity, p = 0.35) (Figure 2). Importantly, the total cumulative dose of inhaled corticosteroids calculated on the basis of mean daily dose and duration of therapy made a material difference to the results. In the studies in which patients were exposed to 60 mg or greater of beclomethasone or its equivalent for the duration of the trial, inhaled corticosteroids were effective in reducing the total sputum cell count (-0.68 units; 95% CI, -1.11 to -0.26). In contrast, trials with cumulative dose of < 60 mg did not demonstrate a favorable effect of inhaled corticosteroids on this sputum index (-0.11 units; 95% CI, -0.58 to 0.37). All of the trials with the higher cumulative dose had exposed the trial participants to inhaled corticosteroids for at least 6 weeks; whereas, the trials with the lower cumulative dose was uniformly less than 6 weeks in duration.
Figure 2 Effect of inhaled corticosteroids on total cell counts in the sputum of stable COPD patients
Inhaled corticosteroids had a salutary effect on neutrophil counts in the sputum. As compared with the control group, the standardized mean difference in those treated with inhaled corticosteroids was -2.16 units (95% CI, -3.81 to -0.50; test for heterogeneity, p < 0.001) (Figure 3). Similar to the findings on the total cell count, trials with a cumulative dose of ≥ 60 mg of beclomethasone (or at least 6 weeks of therapy) demonstrated a significant effect of these medications on sputum neutrophil count (-4.27 units; 95% CI, -6.87 to -1.66); whereas, trials with cumulative dose of < 60 mg (or less than 6 weeks of therapy) failed to demonstrate a beneficial effect (-0.26 units; 95% CI, -0.74 to 0.22).
Figure 3 Effect of inhaled corticosteroids on neutrophils in the sputum of stable COPD patients
Inhaled corticosteroids also reduced the lymphocyte counts in the sputum (standardized mean difference, -0.39 units, 95% CI, -0.74 to -0.05; test for heterogeneity, p = 0.58) (Figure 4). Trials with cumulative dose of ≥ 60 mg (or at least 6 weeks of therapy) demonstrated a significant effect (standardized mean difference, -0.59 units; 95% CI, -1.01 to -0.17); whereas, trials with cumulative dose < 60 mg (or less than 6 weeks of therapy) failed to demonstrate a salutary effect on this endpoint (standardized mean difference, 0.02; 95% CI, -0.59 to 0.62).
Figure 4 Effect of inhaled corticosteroids on lymphocytes in the sputum of stable COPD patients
These medications were also effective in reducing epithelial cell counts compared with the controls (standardized mean difference, -0.51 units, 95% CI, -0.98 to -0.05; test for heterogeneity, p = 0.20) (Figure 5). There was an insignificant trend towards reducing eosinophil counts in the sputum with inhaled corticosteroid therapy (standardized mean difference, -0.28 units, 95% CI, -0.62 to 0.07; test for heterogeneity, p = 0.22) (Figure 6). Inhaled corticosteroids did not appear to have any significant effect on macrophage concentrations in the sputum (standardized mean difference, -0.02 units, 95% CI, -0.34 to 0.29; test for heterogeneity, p = 0.65) (Figure 7). Inhaled corticosteroids did not have significant effect on sputum IL-8 levels (standardized mean difference, -0.22 units; 95% CI, -0.77 to 0.32; test for heterogeneity, p = 0.84).
Figure 5 Effect of inhaled corticosteroids on epithelial cells in the sputum of stable COPD patients
Figure 6 Effect of inhaled corticosteroids on eosinophils in the sputum of stable COPD patients
Figure 7 Effect of inhaled corticosteroids on macrophages in the sputum of stable COPD patients
To evaluate whether the magnitude of the reduction in the inflammatory cells was modified by the absolute levels of the inflammatory cells in the sputum at baseline, we performed a stratified analysis based on the total cell counts at baseline (see Table 3). No significant patterns were observed with any of the cell lines suggesting that baseline "cell load" in the sputum was not a predictor of response to inhaled corticosteroids.
Table 3 Total and differential cell counts at baseline and the standard mean difference (SMD) in cell counts between intervention group and placebo group after treatment.
Source Total cells Neutrophils Lymphocytes Eosinophils Macrophages
Number (× 104/mL) SMD (95% CI) Number (× 104/mL) SMD (95% CI) Number (× 104/mL) SMD (95% CI) Number (× 104/mL) SMD (95% CI) Number (× 104/mL) SMD (95% CI)
Yildiz [22] 350.0 -0.6 (-1.6 to 0.4) 260.0 -2.2 (-3.4 to -1.0) 3.5 -0.5 (-1.4 to 0.5) 7.0 -1.1 (-2.1 to -0.1) 80.0 0.2 (-0.5 to 0.9)
Confalonieri [20] 219.0 -0.4 (-1.1 to 0.3) 158.8 -3.4 (-4.5 to -2.3) 6.6 -0.5 (-1.2 to 0.2) 6.2 -0.6 (-1.3 to 0.1) 45.0 -0.3 (-0.9 to 0.3)
Mirici [21] 196.5 -1.0 (-1.7 to -0.3) 146.5 -7.5 (-9.3 to -5.6) 1.6 -0.7 (-1.4 to -0.1) 1.6 0.2 (-0.4 to 0.8) 38.2 0.5 (-0.5 to 1.5)
Sugiura [7] 165.7 0.2 (-0.8 to 1.2) 102.9 0.1 (-0.9 to 1.1) 6.1 0.04 (-0.9 to 1.0) 4.5 -0.2 (-1.2 to 0.8) 52.0 -0.3 (-1.1 to 0.5)
Culpitt [19] 165.0 -0.3 (-1.1 to 0.5) 145.0 -0.4 (-1.2 to 0.4) NR NR NR NR 25.0 -0.2 (-0.9 to 0.6)
Keatings [18] 6.3* -0.1 (-0.9 to 0.7) 4.3* -0.4 (-1.1 to 0.4) 6.0* 0.0 (-0.7 to 0.8) 0.2* -0.2 (-1.0 to 0.6) 1.8* 0.2 (-0.7 to 1.2)
Pooled Summary -0.4 (-0.8 to -0.1) -2.2 (-3.8 to -0.5) -0.4 (-0.7 to -0.1) -0.3 (-0.6 to 0.1) -0.02 (-0.3 to 0.3)
* cell count/ml
Abbreviations: NR, not reported/not calculable.
After treatment with inhaled steroids, lung function improved slightly but neither the improvement in FEV1 nor FVC reached statistical significance. For predicted FEV1, the overall standardized mean difference was 0.26 units, 95% CI, -0.06 to 0.57 (test for heterogeneity, p = 0.62) (Figure 8). For predicted FVC, the overall standardized mean difference was 0.31 units; 95% CI, -0.09 to 0.70 (test for heterogeneity, p = 0.23).
Figure 8 Effect of inhaled corticosteroids on FEV11% predicted in stable COPD patients. Abbreviation: FEV1, forced expiratory volume in one second
Discussion
By combining data across the clinical studies, we increased statistical power to demonstrate a salutary effect of moderate to high doses of inhaled corticosteroids on some inflammatory indices in the sputum of patients with stable COPD. Over a short term, these medications reduced neutrophil, lymphocyte and epithelial cell counts in the sputum of stable COPD patients. They had smaller (and insignificant) effect on sputum eosinophils and IL-8. They had little effect on sputum macrophages. Although the magnitudes of these reductions were relatively small, they may explain why inhaled corticosteroids decrease cough and sputum production [23], reduce exacerbations [24], and hospitalizations [25].
We also found that duration of therapy and total cumulative dose, which are related constructs, made a material difference to the overall results. Short trials (less than 6 weeks in duration) were uniformly "negative"; while longer term trials (at least 6 weeks of therapy) were mostly positive. Similarly, trials that exposed the patients to higher cumulative dose were more "positive" than those that exposed patients to lower dose. This suggests that duration of therapy and total cumulative doses may be important determinants of the effect of inhaled corticosteroids on airway inflammation.
Although corticosteroids delay neutrophil apoptosis and may increase neutrophil survival [11,26], they also have significant inhibitory action on neutrophil performance. Likely through the annexin-I (lipocortin-1) pathways, for instance, corticosteroids interfere with neutrophil chemotaxis, adhesion, transmigration, oxidative bursts, and phagocytosis, thereby down-regulating the overall inflammatory cascade [9,27]. Indeed, Llewellyn-Jones and co-workers [28] showed that 4 weeks of inhaled fluticasone therapy can significantly reduce sputum chemotactic activity for neutrophils and increase its elastase inhibitory capacity in patients with well-characterized COPD. These data suggest that inhaled corticosteroids can reduce recruitment and/or adhesion of neutrophils to the airways of COPD patients, thereby lowering the overall concentration of these cells in COPD airways.
Superficially, the present data on sputum eosinophils appear to be inconsistent with the known effect of corticosteroids in general on eosinophils. Many experiments have shown that eosinophils are exquisitely sensitive to corticosteroids [29,30]. The current data, however, suggest otherwise. Several studies have demonstrated that among COPD patients with irreversible airflow obstruction (as was the case for a majority of patients enrolled in the original studies contained in this meta-analysis), eosinophils are present in relatively small quantities in the sputum of such patients [10,31]. In most COPD patients, eosinophils account for less than 2% of the total cells in the sputum. This could have introduced a "floor" bias wherein the overall signal to the noise ratio for eosinophils may have been too small to detect subtle but important effect of inhaled corticosteroids on these cells. Although by combining data from these published studies we increased the power of the present analysis to detect salient changes in the inflammatory indices of the sputum, we may still have had insufficient power for analyses of cells with a relatively small signal. Our analysis may also have had insufficient power to assess the effects of inhaled corticosteroids on FEV1. Although there was a trend towards improvement, we did not find a statistically significant effect of inhaled corticosteroids on FEV1. Larger randomized trials have demonstrated, however, that inhaled corticosteroids significantly improve FEV1 over the first three to six months of therapy [25,32-34], suggesting that for certain endpoints our present analysis still lacked sufficient power. Therefore, the "negative" associations must be interpreted cautiously. It is also important to note that none of the studies included in the present review evaluated the effects of inhaled corticosteroids on the function or performance of inflammatory cells in the airway. Thus, we can not discount the possibility that these medications could have salutary effects on the functional performance of these cells.
In the present review, we did not include randomized studies that used bronchoalveolar lavage (BAL) or bronchial biopsies to measure inflammatory cells in the airways. However, in one study, Balbi and colleagues [35] observed significant reductions in the total number of cells, neutrophil counts, IL-8, and myeloperoxidase levels in the BAL fluid of COPD patients after 6 weeks of inhaled beclomethasone therapy. A similar finding was observed and reported by Thompson and coworkers [36]. In another experiment, Hattotuwa at al [23] randomly treated a group of COPD patients with 3 months of inhaled fluticasone propionate (1 mg/d) or placebo. The group that received fluticasone had significantly fewer mast cells in the subepithelial layer as well as a reduced ratio of CD8 to CD4 positive cells in the epithelial layer than those treated with placebo. Most importantly, the fluticasone group had significant improvements in cough and sputum scores and decreased use of reliever medications and experienced fewer exacerbations than did the placebo group [23]. Verhoeven et al [37] evaluated 23 patients with COPD and randomly treated 10 patients to fluticasone (1 mg/d) and the remainder to placebo. After 6 months, fluticasone treatment resulted in a significant reduction in the number of MBP and CD68 positive cells in the lamina propria and reduced tryptase levels in the epithelium. In addition, there was a trend towards fewer CD3, CD4 CD68 positive cells in epithelium of the group treated with fluticasone compared with the group treated with placebo [37]. The results from the BAL and bronchial biopsy studies largely support data from the sputum studies and are consistent with the notion that inhaled corticosteroids reduce airway inflammation in COPD.
We also did not include studies that used systemic corticosteroids. Barcyk and colleagues [38] have reported that oral prednisone therapy (0.5 mg/kg/d) for 2 weeks significantly reduced myeloperoxidase levels in the sputum of COPD patients. Brightling and colleagues [39] showed that 2 weeks of oral prednisone therapy resulted in fewer eosinophils in the sputum of COPD patients. Similar findings were reported by Fujimoto and colleagues [40]. These data suggest that oral prednisone can reduce certain components of airway inflammation (e.g. eosinophils) in COPD; however, most of the studies were very short in duration, which makes it difficult to compare these data against those studies that used inhaled corticosteroids.
Although in the present review, we could not adequately determine the effects of tobacco smoke exposure on the relationship between inhaled corticosteroids and airway inflammation, there is a growing body of evidence to suggest that active smoking may attenuate the effectiveness of corticosteroids in suppressing airway inflammation. Active smoking increases oxidative stress and up-regulates the production of various pro-inflammatory cytokines including Il-6, IL-8, IL-1β and monocyte chemoattractant protein-1 in airways, which may through a series of complex pathways lead to a state of steroid resistance [41]. Additionally, cigarette smoke may reduce histone deacetylase activity and its expression in alveolar macrophages, making these cells relatively resistant to corticosteroids since one of the principal targets of corticosteroid action is by switching off gene expression of inflammatory genes through the recruitment of histone deacetylases [41]. Therefore smoking cessation remains the single most important intervention in COPD management. Inhaled corticosteroids should be considered as a possible adjunctive therapy in patients who remain symptomatic despite smoking cessation.
There are certain limitations with the present analysis. First, although we used stringent entry criteria in order to minimize the heterogeneity in the research methods employed by each of the selected study, there were still some variations in the study design, the exposure medications, and the target population across the original studies. However, the differences in the characteristics of the studies were relatively small and unlikely to have materially affected the overall findings of the current review. We also contacted the primary authors to clarify any ambiguities or to obtain additional data, where necessary, to further minimize the "noise" inherent to meta-analyses. Moreover, to accommodate various differences in the methodology of data collection and laboratory techniques employed across the original studies, we converted the individual data into standardized mean estimates, which enhanced the comparability of data across the original studies. Second, it is possible that corticosteroid therapy could have affected the volume of sputum recovery, decreasing the total sputum cell counts in those patients exposed to this therapy. To mitigate this possibility, the cell counts were expressed as cells per volume of sputum recovered.
Conclusions
In summary, the present meta-analysis suggests that inhaled corticosteroids when used for longer than 6 weeks can significantly reduce neutrophil counts and other inflammatory indices in the sputum of patients with stable COPD. Large randomized controlled trials are needed in the future to confirm these early findings and to determine whether these salutary effects persist longer than 3 to 4 months of therapy.
Abbreviations
COPD chronic obstructive pulmonary disease
FEV1 forced expiratory volume in 1 second
FVC forced vital capacity
SD standard deviationIL-8 interleukin-8
Competing interests
DDS and SFP have received honoraria for speaking engagements from GlaxoSmithKline (GSK) & AstraZeneca, and have received consultation fees and research funding from GSK. However, no part of this work was financed by these companies. This work was funded by Canada Research Chair and a Michael Smith/St. Paul's Hospital Foundation Professorship in COPD.
Authors' contributions
All the authors contributed to the design and implementation of the study. Data analyses were performed by WQG and DDS. All authors contributed to the write-up of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors thank Dr. Füsu Yildiz, Dr. Hisatoshi Sugiura, and Dr. Arzu Mirici for providing additional data for this study.
DDS is supported by a Canada Research Chair (Respiration) and a Michael Smith/St. Paul's Hospital Foundation Professorship in COPD.
==== Refs
Calverley PM Walker P Chronic obstructive pulmonary disease Lancet 2003 362 1053 1061 14522537 10.1016/S0140-6736(03)14416-9
Barnes PJ Shapiro SD Pauwels RA Chronic obstructive pulmonary disease: molecular and cellular mechanisms Eur Respir J 2003 22 672 688 14582923
Di Stefano A Capelli A Lusuardi M Balbo P Vecchio C Maestrelli P Mapp CE Fabbri LM Donner CF Saetta M Severity of airflow limitation is associated with severity of airway inflammation in smokers Am J Respir Crit Care Med 1998 158 1277 1285 9769292
Saetta M Turato G Facchini FM Corbino L Lucchini RE Casoni G Maestrelli P Mapp CE Ciaccia A Fabbri LM Inflammatory cells in the bronchial glands of smokers with chronic bronchitis Am J Respir Crit Care Med 1997 156 1633 1639 9372687
Roland M Bhowmik A Sapsford RJ Seemungal TA Jeffries DJ Warner TD Wedzicha JA Sputum and plasma endothelin-1 levels in exacerbations of chronic obstructive pulmonary disease Thorax 2001 56 30 35 11120901 10.1136/thorax.56.1.30
Bhowmik A Seemungal TA Sapsford RJ Wedzicha JA Relation of sputum inflammatory markers to symptoms and lung function changes in COPD exacerbations Thorax 2000 55 114 120 10639527 10.1136/thorax.55.2.114
Sugiura H Ichinose M Yamagata S Koarai A Shirato K Hattori T Correlation between change in pulmonary function and suppression of reactive nitrogen species production following steroid treatment in COPD Thorax 2003 58 299 305 12668791 10.1136/thorax.58.4.299
Patel IS Roberts NJ Lloyd-Owen SJ Sapsford RJ Wedzicha JA Airway epithelial inflammatory responses and clinical parameters in COPD Eur Respir J 2003 22 94 99 12882457 10.1183/09031936.03.00093703
Llewellyn-Jones CG Hill SL Stockley RA Effect of fluticasone propionate on neutrophil chemotaxis, superoxide generation, and extracellular proteolytic activity in vitro Thorax 1994 49 207 212 8202875
Keatings VM Collins PD Scott DM Barnes PJ ifferences 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
Zhang X Moilanen E Kankaanranta H Beclomethasone, budesonide and fluticasone propionate inhibit human neutrophil apoptosis Eur J Pharmacol 2001 431 365 371 11730731 10.1016/S0014-2999(01)01437-6
Sin DD McAlister FA Man SF Anthonisen NR Contemporary management of chronic obstructive pulmonary disease: scientific review JAMA 2003 290 2301 2312 14600189 10.1001/jama.290.17.2301
van der Valk P Monninkhof E van der Palen J Zielhuis G van Herwaarden C Effect of discontinuation of inhaled corticosteroids in patients with chronic obstructive pulmonary disease: the COPE study Am J Respir Crit Care Med 2002 166 1358 1363 12406823 10.1164/rccm.200206-512OC
Boulet LP Becker A Berube D Beveridge R Ernst P Canadian Asthma Consensus Report, 1999. Canadian Asthma Consensus Group CMAJ 1999 161 S1 61 10906907
Curtin F Altman DG Elbourne D Meta-analysis combining parallel and cross-over clinical trials. I: Continuous outcomes Stat Med 2002 21 2131 2144 12210629 10.1002/sim.1205
Sutton AJ Abrams KR Methods for meta-analysis in medical research 2000 England: John Wiley 57 86
Loppow D Schleiss MB Kanniess F Taube C Jorres RA Magnussen H In patients with chronic bronchitis a four week trial with inhaled steroids does not attenuate airway inflammation Respir Med 2001 95 115 121 11217907 10.1053/rmed.2000.0960
Keatings VM Jatakanon A Worsdell YM Barnes PJ Effects of inhaled and oral glucocorticoids on inflammatory indices in asthma and COPD Am J Respir Crit Care Med 1997 155 542 548 9032192
Culpitt SV Maziak W Loukidis S Nightingale JA Matthews JL Barnes PJ Effect of high dose inhaled steroid on cells, cytokines, and proteases in induced sputum in chronic obstructive pulmonary disease Am J Respir Crit Care Med 1999 160 1635 1639 10556133
Confalonieri M Mainardi E Della Porta R Bernorio S Gandola L Beghe B Spanevello A Inhaled corticosteroids reduce neutrophilic bronchial inflammation in patients with chronic obstructive pulmonary disease Thorax 1998 53 583 585 9797758
Mirici A Bektas Y Ozbakis G Effect of inhaled corticosteroids on respiratory function tests and airway inflammation in stable chronic obstructive pulmonary disease: A randomised, double-blind, controlled clinical trial Clinical Drug Investigation 2001 21 835 842
Yildiz F Kaur AC Ilgazli A Celikoglu M Kacar Ozkara S Paksoy N Ozkarakas O Inhaled corticosteroids may reduce neutrophilic inflammation in patients with stable chronic obstructive pulmonary disease Respiration 2000 67 71 76 10705266 10.1159/000029466
Hattotuwa KL Gizycki MJ Ansari TW Jeffery PK Barnes NC The effects of inhaled fluticasone on airway inflammation in chronic obstructive pulmonary disease: a double-blind, placebo-controlled biopsy study Am J Respir Crit Care Med 2002 165 1592 1596 12070058 10.1164/rccm.2105025
Alsaeedi A Sin DD McAlister FA The effects of inhaled corticosteroids in chronic obstructive pulmonary disease: a systematic review of randomized placebo-controlled trials Am J Med 2002 113 59 65 12106623 10.1016/S0002-9343(02)01143-9
Lung Health Study Research Group Effect of inhaled triamcinolone on the decline in pulmonary function in chronic obstructive pulmonary disease N Engl J Med 2000 343 1902 1909 11136260 10.1056/NEJM200012283432601
Heasman SJ Giles KM Ward C Rossi AG Haslett C Dransfield I Glucocorticoid-mediated regulation of granulocyte apoptosis and macrophage phagocytosis of apoptotic cells: implications for the resolution of inflammation J Endocrinol 2003 178 29 36 12844333 10.1677/joe.0.1780029
Goulding NJ Euzger HS Butt SK Perretti M Novel pathways for glucocorticoid effects on neutrophils in chronic inflammation Inflamm Res 1998 47 158 165 9831319 10.1007/s000110050310
Llewellyn-Jones CG Harris TA Stockley RA Effect of fluticasone propionate on sputum of patients with chronic bronchitis and emphysema Am J Respir Crit Care Med 1996 153 616 621 8564107
Little SA Chalmers GW MacLeod KJ McSharry C Thomson NC Non-invasive markers of airway inflammation as predictors of oral steroid responsiveness in asthma Thorax 2000 55 232 234 10679543 10.1136/thorax.55.3.232
Schleimer RP Bochner BS The effects of glucocorticoids on human eosinophils J Allergy Clin Immunol 1994 94 1202 1213 7798561 10.1016/0091-6749(94)90333-6
Papi A Romagnoli M Baraldo S Braccioni F Guzzinati I Saetta M Ciaccia A Fabbri LM Partial reversibility of airflow limitation and increased exhaled NO and sputum eosinophilia in chronic obstructive pulmonary disease Am J Respir Crit Care Med 2000 162 1773 1777 11069811
Paggiaro PL Dahle R Bakran I Frith L Hollingworth K Efthimiou J Multicentre randomised placebo-controlled trial of inhaled fluticasone propionate in patients with chronic obstructive pulmonary disease: International COPD Study Group Lancet 1998 351 773 780 9519948 10.1016/S0140-6736(97)03471-5
Vestbo J Sorensen T Lange P Brix A Torre P Viskum K Long-term effect of inhaled budesonide in mild and moderate chronic obstructive pulmonary disease: a randomised controlled trial Lancet 1999 353 1819 1823 10359405 10.1016/S0140-6736(98)10019-3
Pauwels RA Lofdahl CG Laitinen LA Schouten JP Postma DS Pride NB Ohlsson SV Long-term treatment with inhaled budesonide in persons with mild chronic obstructive pulmonary disease who continue smoking: European Respiratory Society Study on Chronic Obstructive Pulmonary Disease N Engl J Med 1999 340 1948 1953 10379018 10.1056/NEJM199906243402503
Balbi B Majori M Bertacco S Convertino G Cuomo A Donner CF Pesci A Inhaled corticosteroids in stable COPD patients: do they have effects on cells and molecular mediators of airway inflammation? Chest 2000 117 1633 1637 10858395 10.1378/chest.117.6.1633
Thompson AB Mueller MB Heires AJ Bohling TL Daughton D Yancey SW Sykes RS Rennard SI Aerosolized beclomethasone in chronic bronchitis. Improved pulmonary function and diminished airway inflammation Am Rev Respir Dis 1992 146 389 395 1489129
Verhoeven GT Wijkhuijs AJ Hooijkaas H Hoogsteden HC Sluiter W Effect of an inhaled glucocorticoid on reactive oxygen species production by bronchoalveolar lavage cells from smoking COPD patients Mediators Inflamm 2000 9 109 113 10958384 10.1080/096293500411578
Barczyk A Sozanska E Trzaska M Pierzchala W Decreased levels of myeloperoxidase in induced sputum of patients with COPD after treatment with oral glucocorticoids Chest 2004 126 389 393 15302722 10.1378/chest.126.2.389
Brightling CE Monteiro W Ward R Parker D Morgan MD Wardlaw AJ Pavord ID Brightling CE Monteiro W Ward R Parker D Morgan MD Wardlaw AJ Pavord ID Sputum eosinophilia and short-term response to prednisolone in chronic obstructive pulmonary disease: a randomised controlled trial Lancet 2000 356 1480 1485 11081531 10.1016/S0140-6736(00)02872-5
Fujimoto K Kubo K Yamamoto H Yamaguchi S Matsuzawa Y Eosinophilic inflammation in the airway is related to glucocorticoid reversibility in patients with pulmonary emphysema Chest 1999 115 697 702 10084478 10.1378/chest.115.3.697
Barnes PJ Ito K Adcock IM Corticosteroid resistance in chronic obstructive pulmonary disease: inactivation of histone deacetylase Lancet 2004 363 731 733 15001333 10.1016/S0140-6736(04)15650-X
| 15707484 | PMC552309 | CC BY | 2021-01-04 16:30:12 | no | BMC Pulm Med. 2005 Feb 11; 5:3 | utf-8 | BMC Pulm Med | 2,005 | 10.1186/1471-2466-5-3 | oa_comm |
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-101572072510.1186/1471-2474-6-10Research ArticleOptimal sampling of MRI slices for the assessment of knee cartilage volume for cross-sectional and longitudinal studies Zhai Guangju [email protected] Changhai [email protected] Flavia [email protected] Graeme [email protected] Menzies Research Institute, University of Tasmania, Hobart, Australia2 Department of Epidemiology and Preventive Medicine, Monash University, Alfred Hospital, Prahran, Vic, Australia2005 20 2 2005 6 10 10 10 9 2004 20 2 2005 Copyright © 2005 Zhai et al; licensee BioMed Central Ltd.2005Zhai 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
MRI slices of 1.5 mm thickness have been used in both cross sectional and longitudinal studies of osteoarthritis, but is difficult to apply to large studies as most techniques used in measuring knee cartilage volumes require substantial post-image processing. The aim of this study was to determine the optimal sampling of 1.5 mm thick slices of MRI scans to estimate knee cartilage volume in males and females for cross-sectional and longitudinal studies.
Methods
A total of 150 subjects had a sagittal T1-weighted fat-suppressed MRI scan of the right knee at a partition thickness of 1.5 mm to determine their cartilage volume. Fifty subjects had both baseline and 2-year follow up MRI scans. Lateral, medial tibial and patellar cartilage volumes were calculated with different samples from 1.5 mm thick slices by extracting one in two, one in three, and one in four to compare to cartilage volume and its rate of change. Agreement was assessed by means of intraclass correlation coefficient (ICC) and Bland & Altman plots.
Results
Compared to the whole sample of 1.5 mm thick slices, measuring every second to fourth slice led to very little under or over estimation in cartilage volume and its annual change. At all sites and subgroups, measuring every second slice had less than 1% mean difference in cartilage volume and its annual rate of change with all ICCs ≥ 0.98.
Conclusion
Sampling alternate 1.5 mm thick MRI slices is sufficient for knee cartilage volume measurement in cross-sectional and longitudinal epidemiological studies with little increase in measurement error. This approach will lead to a substantial decrease in post-scan processing time.
==== Body
Background
Osteoarthritis (OA) is the most common form of arthritis and a leading cause of musculoskeletal disability in most developed countries [1]. The knee is one of the most frequently affected joints with a prevalence of 30% in people older than 65 years [2] and high resultant disability [3]. Defects in cartilage are widely considered to be the initial problem in OA [4,5], although this viewpoint is not shared by all investigators [6]. Detection of cartilage morphological change is critical in the evaluation, diagnosis, and monitoring of OA. Conventional radiography is used in evaluating the progression of OA but is limited by its inability to directly visualise cartilage. Magnetic resonance imaging (MRI) offers the distinct advantage of detecting morphologic changes in articular cartilage and is a sensitive and accurate test for evaluating articular cartilage non-invasively [7-11]. The correlation coefficient is 0.99 between knee cartilage volumes measured by MRI and the true volumes by means of water displacement [9]. This method uses 1.5 mm thick MRI slices and has high reproducibility with coefficients of variation of 2–3% [12] and has been used in both cross sectional and longitudinal studies of OA [12-15]. However, the method is difficult to apply to large studies as most techniques used in measuring knee cartilage volumes require substantial post-image processing [12] and the process has not yet been automated. One possible solution is to select a sample from within the 1.5 mm thick slices to reduce the post-image processing time, as has been reported for the estimation of brain compartment volume [16] and fetal volume[17]. The aim of the study, therefore, was to determine the optimal sampling of 1.5 mm thick MRI slices required to estimate the volumes of and rate of change in lateral, medial tibial and patellar cartilage with minimal increase in measurement error.
Methods
Subjects
The present study consisted of two datasets; one was part of the Tasmanian Older Adult Cohort Study (TASOAC), an ongoing prospective population-based study aimed at identifying the environmental, genetic, and biochemical factors associated with the development and progression of OA at multiple sites (hand, knee, hip, and spine), which commenced in 2002. Subjects aged between 50 and 79 years were selected randomly from the electoral roll of Southern Tasmania, with an equal number of males and females. Another dataset was a younger adult sample from the Knee Cartilage Volume study (KCV) as previously reported [15]. Both studies were approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee and all subjects provided informed written consent.
MRI
An MRI scan of the right knee was performed on all subjects. Knee cartilage volume was determined by means of image processing on an independent work station using the software program Osiris as previously described [12,15]. Two observers were utilised. Knees were imaged in the sagittal plane on a 1.5-T whole body magnetic resonance unit (Picker) with use of a commercial transmit-receive extremity coil. The same machine scanned all knees and Philips Quality Procedure (Philips ACR Support Program, XJR153-2922.03) was utilised for MRI slice thickness quality assurance. The following image sequence was used: a T1-weighted fat saturation 3D gradient recall acquisition in the steady state; flip angle 55 degrees; repetition time 58 msecs; echo time 12 msec; field of view 16 cm; 60 partitions; 512 × 512 matrix; acquisition time 11 min 56 sec; one acquisition. Sagittal images were obtained at a partition thickness of 1.5 mm and an in-plane resolution of 0.31 × 0.31 (512 × 512 pixels). The image data were transferred to the workstation. The volumes of individual cartilage plates (medial tibial, lateral tibial and patella) were isolated from the total volume by manually drawing disarticulation contours around the cartilage boundaries on a slice-by-slice basis. All individual slice areas for each cartilage site and each subject were subsequently transferred to and recorded on a spreadsheet. The total area of each individual cartilage was then multiplied by the slice thickness to produce a volume estimate. This "all slice" estimate of cartilage volume (based on slice thickness of 1.5 mm) was used as the gold standard for other comparisons.
Then, the volumes of all individual cartilage plates were recalculated based on different sampling intervals from 1.5 mm thick slices by extracting one in two, one in three, and one in four slice areas from the individual data file. These were then summed and the total was multiplied by the corresponding slice distance.
Femoral cartilage volume was not assessed in this study as it is strongly correlated with tibial cartilage volume and thus adds little extra information [18], tibial cartilage volume is the parameter that is most frequently examined in the literature [12,19-23], and femoral cartilage volume has worse reproducibility than tibial cartilage volume [11].
Other measurements
Weight was measured to the nearest 0.1 kg (with shoes, socks and bulky clothing removed) using a single pair of electronic scales (Seca Delta Model 707) which were calibrated using a known weight at the beginning of each clinic. Height was measured to the nearest 0.1 cm (with shoes and socks removed) using a stadiometer. Body Mass Index (BMI) was calculated as weight (kg) / height (m2).
A standing AP semi-flexed view of the right knee was performed in all subjects. Radiographs were then assessed utilising the Altman atlas[24]. Each of the following was assessed: medial joint space narrowing (0–3), lateral joint space narrowing (0–3), medial osteophytes (femoral and tibial combined) (0–3) and lateral osteophytes (femoral and tibial combined) (0–3). Each score was arrived at by consensus with two readers simultaneously assessing the radiograph with immediate reference to the atlas. Any knee ROA was defined as total score ≥ 1. The total score could vary from 0–12. This method had high reproducibility in our hands with ICCs >0.98 [25].
Statistics
Descriptive statistics of the characteristics of the study subjects were tabulated. The annual change in knee cartilage volume was calculated as percent change by means of dividing absolute volume change by baseline cartilage volume. Intraclass correlation coefficient was utilized to assess the measurement agreement. The difference in cartilage volume measured with different samples extracting one in two, one in three, and one in four 1.5 mm thick slices of MR image compared to that measured using 1.5 mm thickness was calculated and expressed as percent absolute difference. Desirable agreement was defined as an ICC ≥ 0.98 with ≤ 1% difference between two measurements. In addition, Bland & Altman plots [26] were also utilized. Desirable agreement was defined as the mean difference between two measurements close to zero with 95% of individual differences being within 2 SD. All analyses were performed using the SPSS statistical package (version 12.1, SPSS, Chicago, IL).
Results
A total of 150 subjects took part in this study: 100 subjects with cross-sectional data (female: 48, male: 52) were from the TASOAC study and 50 subjects with longitudinal data (female: 31, male: 19) were from the KCV study. Characteristics of the study sample are presented in Table 1. Subjects from TASOAC were older, heavier and had a higher prevalence of ROA than those from KCV. Most of participants with ROA were mild with a total ROA score ≤ 3 out of 12. Lateral and medial tibial cartilage volumes were lower in subjects from KCV than those from TASOAC.
Table 1 Characteristics of the study population*
TASOAC dataset N = 100 KCV dataset N = 50
Age (year) 62.3(7.6) 42.8(6.1)
Sex (female %)† 48 62
Height (cm) 167.4(8.7) 168.6(7.9)
Weight (kg) 76.0(15.0) 73.9(13.7)
BMI (kg/m2) 27.1(4.3) 25.9(4.1)
Any knee ROA (%)† 62 18
Knee ROA total score (0-12) 1.3 (1.7) 0.2(0.7)
Lateral tibial cartilage volume (ml)‡ 3.0(0.7) 2.6(0.5)
Medial tibial cartilage volume (ml)‡ 2.7(0.5) 2.2(0.5)
Patellar tibial cartilage volume (ml)‡ 3.5(1.0) 3.5(0.9)
Lateral tibial cartilage volume change (%) per year‡ - -1.2(3.4)
Medial tibial cartilage volume change (%) per year‡ - -2.9(3.9)
Patellar cartilage volume change (%) per year‡ - -3.8(3.4)
*Values are mean (SD) except for indicated. BMI: body mass index. ROA: radiographic osteoarthritis. † Percentage. ‡ Measured with the whole sample of 1.5 mm thick MRI slices.
In cross-sectional analysis, compared to the cartilage volume measured using 1.5 mm thickness, decreasing the number of the slices by extracting one in two to one in four led to a very little underestimation in the magnitude of the average cartilage volume at lateral, medial tibial and patellar sites with ICCs of 0.98–1.00 (Table 2). The maximum underestimation was 3.3% at the medial tibial site with one in four slices (Table 2). Similar results were obtained when the analysis was done separately for people with and without ROA (Table 3) although the differences tended to be larger in the ROA group. The difference also tended to be larger for medial tibial cartilage in the TASOAC sample and lateral tibial cartilage for the KCV sample (Table 2). At all sites and subgroups, cartilage volume measured with one in two slices had less than 1% difference compared to that measured with all 1.5 mm slices with an ICC of 1.0 (Table 2 &3). Bland & Altman plots showed that the mean difference was zero for lateral tibial cartilage and -0.01 ml for medial tibial and patellar cartilage with 95% of individual differences within ± 2SD. The variability was random and uniform throughout the range of cartilage volume (Figure 1).
Table 2 Agreement analysis of knee cartilage volume measured with different samples of 1.5 mm thick MRI slices*
Whole sample (n = 150) TASOAC sample (n = 100) KCV sample (n = 50)
%Difference (SD) ICC† %Difference (SD) ICC† %Difference (SD) ICC†
Lateral tibial cartilage
The whole sample Reference Reference Reference Reference Reference Reference
1/2 whole sample‡ -0.04(1.5) 1.00 0.35(1.4) 1.00 -0.84(1.4) 1.00
1/3 whole sample‡ -0.61(2.3) 1.00 0.11(2.1) 1.00 -2.09(1.8) 1.00
1/4 whole sample‡ -1.12(3.4) 1.00 -0.11(3.0) 1.00 -3.18(3.3) 0.99
Medial tibial cartilage
The whole sample Reference Reference Reference Reference Reference Reference
1/2 whole sample‡ -0.50(1.7) 1.00 -0.98(1.4) 1.00 0.46(1.7) 1.00
1/3 whole sample‡ -1.70(3.3) 0.99 -2.97(2.8) 0.99 0.83(2.9) 1.00
1/4 whole sample‡ -3.27(5.0) 0.98 -5.09(3.9) 0.97 0.38(4.9) 0.99
Patellar cartilage
The whole sample Reference Reference Reference Reference Reference Reference
1/2 whole sample‡ -0.36(1.2) 1.00 -0.40(1.2) 1.00 -0.29(1.3) 1.00
1/3 whole sample‡ -0.91(2.0) 1.00 -0.93(2.0) 1.00 -0.86(1.9) 1.00
1/4 whole sample‡ -2.24(3.0) 1.00 -2.12(2.9) 1.00 -2.50(3.3) 1.00
* SD: standard deviation. ICC: intraclass correlation coefficient. † All P < 0.001. ‡ Derived by extracting one in two, one in three, or one in four of the 1.5 mm thick MRI slices.
Table 3 Agreement analysis of cartilage volume measured with different samples of 1.5 mm thick MRI slices in people with and without ROA*
ROA absent (n = 76) ROA present (n = 68)
Difference (SD) ICC† Difference (SD) ICC†
Lateral tibial cartilage
The whole sample Reference Reference Reference Reference
1/2 whole sample‡ -0.30(1.4) 1.00 0.24(1.6) 1.00
1/3 whole sample‡ -1.14(2.3) 1.00 -0.01(2.1) 1.00
1/4 whole sample‡ -1.85(3.4) 0.99 -0.29(3.4) 1.00
Medial tibial cartilage
The whole sample Reference Reference Reference Reference
1/2 whole sample‡ -0.39(1.7) 1.00 -0.77(2.2) 1.00
1/3 whole sample‡ -1.20(3.3) 0.99 -2.13(3.4) 0.99
1/4 whole sample‡ -2.56(5.3) 0.98 -3.77(4.5) 0.98
Patellar cartilage
The whole sample Reference Reference Reference Reference
1/2 whole sample‡ -0.38(1.2) 1.00 -0.40(1.2) 1.00
1/3 whole sample‡ -0.87(1.9) 1.00 -1.10(2.0) 1.00
1/4 whole sample‡ -2.02(2.9) 1.00 -2.50(3.2) 1.00
*Six subjects had missing values for ROA. Difference in cartilage volume measured with different thick slices of MR images is expressed as percentage. ICC: intraclass correlation coefficient. ROA: radiographic osteoarthritis. SD: standard deviation. † All P < 0.001. ‡ Derived by extracting one in two, one in three, or one in four 1.5 mm thick MRI slices.
Figure 1 Bland & Altman plots of cartilage volume measured by every second 1.5 mm thick MRI slice compared to that measured by the total sample at lateral (a), medial tibial (b), and patellar (c) sites. The x-axis represents average values of two measurements while the y-axis represents the individual difference between two measurements, and the three horizontal lines stand for mean individual difference ± 2 SD.
Similarly, in longitudinal analysis, compared to the cartilage volume change using 1.5 mm thick slices, decreasing the number of the slices by extracting one in two to one in four slices led to very little over or under estimation of the mean changes in cartilage volume at lateral, medial tibial and patellar sites (Table 4). The mean difference ranged from -0.05% to 0.14% with the maximum difference at the patellar site. ICCs ranged from 0.85 to 0.99 (Table 4). The difference became larger but all were ≤ 1% in subjects with and without ROA (Table 4). At all sites, the annual change in cartilage volume measured with one in two slices had an ICC ≥ 0.98 with less than 0.3% difference compared to that measured using all the slices. Bland & Altman plots showed that 95% of the individual differences were within ± 2 SD and the variability was random and uniform throughout the range of cartilage volume (Figure 2).
Table 4 Agreement analysis of the annual change in knee cartilage volume measured with different samples of 1.5 mm thick MRI slices*
Whole sample (n = 50) ROA present (n = 9) ROA absent (n = 41)
Difference (SD) ICC† Difference (SD) ICC† Difference (SD) ICC†
Lateral tibial cartilage
The whole sample Reference Reference Reference Reference Reference Reference
1/2 whole sample‡ 0.06(0.9) 0.99 0.23(1.1) 0.99 0.02(0.9) 0.98
1/3 whole sample‡ 0.05(1.5) 0.96 -0.65(1.4) 0.98 0.20(1.5) 0.95
1/4 whole sample‡ -0.03(2.2) 0.92 -0.04(2.4) 0.95 -0.02(2.2) 0.91
Medial tibial cartilage
The whole sample Reference Reference Reference Reference Reference Reference
1/2 whole sample‡ -0.05(1.1) 0.98 -0.29(1.0) 0.99 0.00(1.1) 0.98
1/3 whole sample‡ -0.03(1.8) 0.95 0.24(1.8) 0.97 -0.10(1.8) 0.95
1/4 whole sample‡ 0.02(3.0) 0.85 -1.04(2.7) 0.92 0.25(3.1) 0.83
Patellar cartilage
The whole sample Reference Reference Reference Reference Reference Reference
1/2 whole sample‡ 0.10(0.8) 0.99 -0.07(0.7) 1.00 0.13(0.8) 0.99
1/3 whole sample‡ 0.10(1.5) 0.96 -0.18(1.4) 0.98 0.16(1.5) 0.95
1/4 whole sample‡ 0.14(1.8) 0.93 0.61(1.5) 0.97 0.03(1.9) 0.92
* Difference in the annual change in cartilage volume was expressed in percentage. SD: standard deviation. ROA: radiographic osteoarthritis. ICC: intraclass correlation coefficient. † All P < 0.001. ‡ Derived by extracting one in two, one in three, or one in four 1.5 mm thick MRI slices.
Figure 2 Bland & Altman plots of the annual change in cartilage volume measured by every second 1.5 mm thick MRI slice compared to that measured by the total sample at lateral (a), medial tibial (b), and patellar (c) sites. The annual change in cartilage volume was expressed as a percentage. The x-axis represents average values of two measurements while the y-axis represents the individual difference between two measurements, and the three horizontal lines stand for mean individual difference ± 2 SD.
Discussion
This study suggests that lateral, medial tibial and patellar cartilage volumes measured from up to one in four 1.5 mm thick slices are quite comparable to those obtained from 1.5 mm thick slices. If the agreement is defined at high levels expected to lead to minimal measurement error, then knee cartilage volume can be measured sufficiently and accurately with one in two slices both cross-sectionally and longitudinally regardless of ROA status and./or reader. This approach will lead to a substantial decrease in post-scan processing time and make large-scale studies of knee cartilage volume more feasible.
Currently, there is no reported information on the number of the slices of MRI scans to measure cartilage volume apart from a recent paper from our own group which had similar findings to this study with different readers and geographic location [27]. In a study estimating fetal volume by MRI, Roberts et al reported that using the same thickness of MRI slices (10 mm), volume measured from the low sampling intensity (the distance between scan section midplanes T = 4.5 cm) was virtually identical to those obtained with the high sampling intensity (T = 1.5 cm) with a coefficient of error (CE) < 5% [17]. In the study estimating brain compartment volume from MR Cavalieri slices [16], irrespective of slice thickness, a minimum of 3, 5, and 10 slices provided estimates of the true total volume of grey matter and white matter in the cerebrum with coefficients of error (CEs) of 10, 5, and 3%. For a given number of slices CE decreases rapidly when the slices are thicker than the gaps between them; when the slices are thinner than the gaps, then CE is similar to that in the situation when the slice thickness is zero. The current study demonstrates similar results for knee cartilage. Decreasing the number of slices by extracting up to one in four 1.5 mm slices resulted in a very little underestimation in average volume of lateral, medial tibial and patellar cartilage. The maximum mean difference in cartilage volume obtained from one in four slices to that obtained from all slices was 3.3%, which is substantially smaller than the difference of 9% between cartilage volume obtained from 1.5 mm thick slices of MR image and that measured by means of water displacement [9,19,28,29]. The difference increased slightly when we analysed the data separately for people with and without ROA, but the results were similar for both groups, suggesting ROA within the range we report has very limited effect on the cartilage volume measured with subsamples of MRI slices. If we arbitrarily define an ICC ≥ 0.98 with ≤ 1% difference as optimal as it is expected to minimise the measurement error and only slightly increase the variance, then cartilage volume and its rate of change can be measured accurately with one in two 1.5 mm thick slices for lateral, medial tibial and patellar cartilage. Bland & Altman plots confirmed this with a random scatter about zero as would be expected if there is no difference between two measurements and uniform variability throughout the range of measurements. Of note, for longitudinal data even decreasing the number of slices by extracting up to one in four resulted in a maximum difference of 0.14% in mean annual change in cartilage volume which is very small when compared to the 5% cartilage loss annually we have reported in patients with OA [30]. Thus, a subsample of MRI slices could also be utilised with marked decreases in processing time allowing greater numbers of subjects to be studied offsetting the accompanying increase in measurement error.
Ideally, the more slices used, the more accurate the estimation of the object's volume, as they may contain more information. However, for a completely regular structure, such as a cylinder, the area of a single slice with length gives an exact volume. It is therefore reassuring but not surprising that the current study demonstrates a minimum reduction in the knee cartilage volume and volume change over time as tibial and patellar cartilages have a relatively regular structure. A different interpretation may apply to femoral cartilage and we do not have data on this imaging site.
The current study simply examined the effect of decreasing the number of slices on the estimation of knee cartilage volume and volume change while all other variables were kept constant. We did not re-scan the study subjects but simply estimated the cartilage volume by using one in two, one in three, or one in four slices. This has an advantage of allowing us to examine the single effect of sampling intensity in the situation where all other variables such as re-positioning the subject and measurement were kept constant. The effect of these errors on measurement have been well-documented [9,31]. For longitudinal analysis, all the MR images were processed by a single observer. For cross sectional analysis, two observers processed the MR images, one for TASOAC data, and another for the KCV study. However, the difference was even smaller in the whole sample than in the two separate samples providing reassurance that our results may be generalisable to different observers as documented with different readers and machines in Melbourne [27].
The current study has a number of potential limitations. Firstly, which sampling intensity should be used in the MRI scan of knee cartilage depends on the purpose of the measurement. Our results cannot be applied to individual cartilage volume, but only for mean cartilage volume in groups as the individual difference in cartilage volume increases with decreasing sampling intensity. Secondly, decreasing sampling intensity will increase measurement error as the remaining slices focus on different portions of the irregularly shaped cartilage. Depending on what particular surfaces remain, however, the overall volume may be increased or decreased. If this is random, then the mean will remain the same as demonstrated in the current study. Thirdly, the ICC can be influenced by traits in the sample in which it is assessed. Age, sex and BMI have been reported to be associated with knee cartilage volume [32]. These may result in a higher ICC in the current study, as between-subject variance would become larger. However, subgroup analyses by sex, BMI (< 25, >= 25), and age (<50, >= 50 yr) did not change the results (data not shown). Further analysis using the Bland & Altman method confirmed the good agreement and interchangability between thick and thin slices, indicating that the result of the current study should be applicable to other populations regardless of the demographic factors related to cartilage volume. Fourthly, the participants in the study had only mild ROA, and these conclusions may not apply to subjects with more advanced OA. Lastly, the annual change in cartilage volume in our sample can not be generalized to other populations as half of our longitudinal study sample had a higher genetic susceptibility to OA [23,33].
Conclusion
Knee cartilage volume and its rate of change can be accurately measured with every second 1.5 mm thick MR slice. This approach will lead to a substantial decrease in post-scan processing time and make large-scale studies of knee cartilage volume more feasible.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GZ designed and carried out the study planning, data collection, analysis and interpretation of the analysis, and preparation of the manuscript. CD participated in data collection and critical revision of the manuscript. FC participated in the study planning and critical revision of the manuscript. GJ designed the study, participated in analysis and interpretation of the analysis, and critical revision of 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
We would like to thank Martin Rush who performed the MRI scans, and Lesa Hornsey who conducted the X-ray measures. A special thanks goes to Drs Velandai Srikanth, Helen Cooley, and Fiona Scott for reading the radiographs and Jayne Fryer for statistical support. The role of Ms Catrina Boon in coordinating both studies is gratefully acknowledged. We would also like to thank the participants who make this study possible. We thank Dr. Philippa Taplin for comments on the revision.
Sources of support: National Health and Medical Research Council, Masonic Centenary Medical Research Foundation, Tasmanian Community Fund, Tasmanian Government Icon scheme
==== Refs
Reginster JY The prevalence and burden of arthritis Rheumatology (Oxford) 2002 41 3 6 12173279
Felson DT Naimark A Anderson J Kazis L Castelli W Meenan RF The prevalence of knee osteoarthritis in the elderly. The Framingham Osteoarthritis Study Arthritis Rheum 1987 30 914 918 3632732
Spector TD Hart DJ How serious is knee osteoarthritis? Ann Rheum Dis 1992 51 1105 1106 1444621
Venn M Maroudas A Chemical composition and swelling of normal and osteoarthrotic femoral head cartilage. I. Chemical composition Ann Rheum Dis 1977 36 121 129 856064
Maroudas A Venn M Chemical composition and swelling of normal and osteoarthrotic femoral head cartilage. II. Swelling Ann Rheum Dis 1977 36 399 406 200188
Dieppe P Osteoarthritis: time to shift the paradigm. This includes distinguishing between severe disease and common minor disability Bmj 1999 318 1299 1300 10323792
Recht MP Piraino DW Paletta GA Schils JP Belhobek GH Accuracy of fat-suppressed three-dimensional spoiled gradient-echo FLASH MR imaging in the detection of patellofemoral articular cartilage abnormalities Radiology 1996 198 209 212 8539380
Waterton JC Solloway S Foster JE Keen MC Gandy S Middleton BJ Maciewicz RA Watt I Dieppe PA Taylor CJ Diurnal variation in the femoral articular cartilage of the knee in young adult humans Magn Reson Med 2000 43 126 132 10642739 10.1002/(SICI)1522-2594(200001)43:1<126::AID-MRM15>3.0.CO;2-#
Peterfy CG van Dijke CF Janzen DL Gluer CC Namba R Majumdar S Lang P Genant HK Quantification of articular cartilage in the knee with pulsed saturation transfer subtraction and fat-suppressed MR imaging: optimization and validation Radiology 1994 192 485 491 8029420
Morgan SR Waterton JC Maciewicz RA Leadbetter JE Gandy SJ Moots RJ Creamer P Nash AF Magnetic resonance imaging measurement of knee cartilage volume in a multicentre study Rheumatology (Oxford) 2004 43 19 21 12923282
Eckstein F Heudorfer L Faber SC Burgkart R Englmeier KH Reiser M Long-term and resegmentation precision of quantitative cartilage MR imaging (qMRI) Osteoarthritis Cartilage 2002 10 922 928 12464552 10.1053/joca.2002.0844
Jones G Glisson M Hynes K Cicuttini F Sex and site differences in cartilage development: a possible explanation for variations in knee osteoarthritis in later life Arthritis Rheum 2000 43 2543 2549 11083279 10.1002/1529-0131(200011)43:11<2543::AID-ANR23>3.0.CO;2-K
Cicuttini F Wluka A Wang Y Stuckey S The determinants of change in patella cartilage volume in osteoarthritic knees J Rheumatol 2002 29 2615 2619 12465162
Jones G Ding C Glisson M Hynes K Ma D Cicuttini F Knee Articular Cartilage Development in Children: A Longitudinal Study of the Effect of Sex, Growth, Body Composition, and Physical Activity Pediatr Res 2003 54 230 6 12736391 10.1203/01.PDR.0000072781.93856.E6
Ding C Cicuttini F Scott F Glisson M Jones G Sex differences in knee cartilage volume in adults: role of body and bone size, age and physical activity Rheumatology (Oxford) 2003 42 1317 23 12810930
McNulty V Cruz-Orive LM Roberts N Holmes CJ Gual-Arnau X Estimation of brain compartment volume from MR Cavalieri slices J Comput Assist Tomogr 2000 24 466 477 10864088 10.1097/00004728-200005000-00021
Roberts N Garden AS Cruz-Orive LM Whitehouse GH Edwards RH Estimation of fetal volume by magnetic resonance imaging and stereology Br J Radiol 1994 67 1067 1077 7820398
Cicuttini FM Wluka AE Stuckey SL Tibial and femoral cartilage changes in knee osteoarthritis Ann Rheum Dis 2001 60 977 980 11557657 10.1136/ard.60.10.977
Graichen H von Eisenhart-Rothe R Vogl T Englmeier KH Eckstein F Quantitative assessment of cartilage status in osteoarthritis by quantitative magnetic resonance imaging: technical validation for use in analysis of cartilage volume and further morphologic parameters Arthritis Rheum 2004 50 811 816 15022323 10.1002/art.20191
Eckstein F Muller S Faber SC Englmeier KH Reiser M Putz R Side differences of knee joint cartilage volume, thickness, and surface area, and correlation with lower limb dominance--an MRI-based study Osteoarthritis Cartilage 2002 10 914 921 12464551 10.1053/joca.2002.0843
Gandy SJ Dieppe PA Keen MC Maciewicz RA Watt I Waterton JC No loss of cartilage volume over three years in patients with knee osteoarthritis as assessed by magnetic resonance imaging Osteoarthritis Cartilage 2002 10 929 937 12464553 10.1053/joca.2002.0849
Cicuttini F Wluka A Davis S Strauss BJ Yeung S Ebeling PR Association between knee cartilage volume and bone mineral density in older adults without osteoarthritis Rheumatology (Oxford) 2004 43 765 769 15039496
Zhai G Stankovich J Ding C Scott F Cicuttini F Jones G The genetic contribution to muscle strength, knee pain, cartilage volume, bone size, and radiographic osteoarthritis: a sibpair study Arthritis Rheum 2004 50 805 810 15022322 10.1002/art.20108
Altman RD Hochberg M Murphy WAJ Wolfe F Lequesne M Atlas of individual radiographic features in osteoarthritis Osteoarthritis Cartilage 1995 3 3 70 8581752
Jones G Ding C Scott F Glisson M Cicuttini F Early radiographic osteoarthritis is associated with substantial changes in cartilage volume and tibial bone surface area in both males and females Osteoarthritis Cartilage 2004 12 169 174 14723876 10.1016/j.joca.2003.08.010
Bland JM Altman DG Statistical methods for assessing agreement between two methods of clinical measurement Lancet 1986 1 307 310 2868172
Cicuttini F Morris KF Glisson M Wluka AE Slice thickness in the assessment of medial and lateral tibial cartilage volume and accuracy for the measurement of change in a longitudinal study J Rheumatol 2004 31 2444 2448 15570649
Cicuttini F Forbes A Morris K Darling S Bailey M Stuckey S Gender differences in knee cartilage volume as measured by magnetic resonance imaging Osteoarthritis Cartilage 1999 7 265 271 10329301 10.1053/joca.1998.0200
Burgkart R Glaser C Hyhlik-Durr A Englmeier KH Reiser M Eckstein F Magnetic resonance imaging-based assessment of cartilage loss in severe osteoarthritis: accuracy, precision, and diagnostic value Arthritis Rheum 2001 44 2072 2077 11592369 10.1002/1529-0131(200109)44:9<2072::AID-ART357>3.0.CO;2-3
Wluka AE Stuckey S Snaddon J Cicuttini FM The determinants of change in tibial cartilage volume in osteoarthritic knees Arthritis Rheum 2002 46 2065 2072 12209510 10.1002/art.10460
Eckstein F Westhoff J Sittek H Maag KP Haubner M Faber S Englmeier KH Reiser M In vivo reproducibility of three-dimensional cartilage volume and thickness measurements with MR imaging AJR Am J Roentgenol 1998 170 593 597 9490936
Cicuttini FM Wluka A Bailey M O'Sullivan R Poon C Yeung S Ebeling PR Factors affecting knee cartilage volume in healthy men Rheumatology (Oxford) 2003 42 258 262 12595619
Jones G Ding C Scott FS Cicuttini FM Genetic mechanisms of knee osteoarthritis: a population-based case control study Ann Rheum Dis 2004 63 1255 9 15361382 10.1136/ard.2003.015875
| 15720725 | PMC552310 | CC BY | 2021-01-04 16:32:04 | no | BMC Musculoskelet Disord. 2005 Feb 20; 6:10 | utf-8 | BMC Musculoskelet Disord | 2,005 | 10.1186/1471-2474-6-10 | oa_comm |
==== Front
BMC SurgBMC Surgery1471-2482BioMed Central London 1471-2482-5-11572370410.1186/1471-2482-5-1Study ProtocolResorbable screws versus pins for optimal transplant fixation (SPOT) in anterior cruciate ligament replacement with autologous hamstring grafts: rationale and design of a randomized, controlled, patient and investigator blinded trial [ISRCTN17384369] Stengel Dirk [email protected] Gerrit [email protected] Julia [email protected] Volker [email protected] Sven [email protected] Grit [email protected] Axel [email protected] Kai [email protected] Michael [email protected] Dirk [email protected] SPOT Group [email protected] Department of Orthopaedic and Trauma Surgery, Unfallkrankenhaus Berlin Trauma Center, Warener Str 7, 12683 Berlin, Germany2 Department of Orthopaedic and Trauma Surgery, Sauerbruchstr., University Hospital of Greifswald, 17487 Greifswald, Germany3 Department of Clinical Epidemiology, Unfallkrankenhaus Berlin Trauma Center, Warener Str 7, 12683 Berlin, Germany4 Institute of Radiology, Unfallkrankenhaus Berlin Trauma Center, Warener Str 7, 12683 Berlin, Germany2005 21 2 2005 5 1 1 2 2 2005 21 2 2005 Copyright © 2005 Stengel et al; licensee BioMed Central Ltd.2005Stengel 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
Ruptures of the anterior cruciate ligament (ACL) are common injuries to the knee joint. Arthroscopic ACL replacement by autologous tendon grafts has established itself as a standard of care.
Data from both experimental and observational studies suggest that surgical reconstruction does not fully restore knee stability. Persisting anterior laxity may lead to recurrent episodes of giving-way and cartilage damage. This might at least in part depend on the method of graft fixation in the bony tunnels. Whereas resorbable screws are easy to handle, pins may better preserve graft tension. The objective of this study is to determine whether pinning of ACL grafts reduces residual anterior laxity six months after surgery as compared to screw fixation.
Design/ Methods
SPOT is a randomised, controlled, patient and investigator blinded trial conducted at a single academic institution. Eligible patients are scheduled to arthroscopic ACL repair with triple-stranded hamstring grafts, conducted by a single, experienced surgeon. Intraoperatively, subjects willing to engage in this study will be randomised to transplant tethering with either resorbable screws or resorbable pins. No other changes apply to locally established treatment protocols. Patients and clinical investigators will remain blinded to the assigned fixation method until the six-month follow-up examination.
The primary outcome is the side-to-side (repaired to healthy knee) difference in anterior translation as measured by the KT-1000 arthrometer at a defined load (89 N) six months after surgery. A sample size of 54 patients will yield a power of 80% to detect a difference of 1.0 mm ± standard deviation 1.2 mm at a two-sided alpha of 5% with a t-test for independent samples.
Secondary outcomes (generic and disease-specific measures of quality of life, magnetic resonance imaging morphology of transplants and devices) will be handled in an exploratory fashion.
Conclusion
SPOT aims at showing a reduction in anterior knee laxity after fixing ACL grafts by pins compared to screws.
==== Body
Background
Anterior cruciate ligament (ACL) rupture belongs to the most common musculoskeletal injuries in the western world. In the United States, 100,000 new cases occur each year, with 10% of all injuries leading to occupational disability [1]. In Germany, the prevalence of torn ACL among subjects between 20 and 35 years averages 0.4%. In the general population, the yearly incidence of ACL rupture reaches 32/100,000, but peaks to 70/100,000 in athletes [2].
An ACL deficient knee is at risk of developing secondary damage to the cartilage and is liable to undergo progressive intra-articular worsening. Roughly half of all acute ACL disruptions attend meniscus damage. Arthroscopic reconstruction surgery by autologous grafting emerged as the therapy of choice. Predictions call for 175,000 ACL replacements performed yearly in the United States. In Germany, around 50,000 patients undergo ACL repair each year.
Selecting the ideal graft remains an issue of debate. Randomized controlled trials suggest a lower degree of persistent laxity with bone-patellar-tendon-bone (BPTB) comparing with two-, three- or four-bundle hamstring (that is, semitendinosus and gracilis tendon) transplants (HT).
However, the biomechanical advantage does not frame higher patient satisfaction, or differences in scoring after long-term follow-up [3]. In contrast, harvesting BPTB grafts often produces notable donor site morbidity, and refractory kneeling pain [4,5].
Available data from randomized, quasi-randomized and uncontrolled trials signal a weighted mean difference of 2.28 mm (95% confidence interval [CI] 1.83 – 2.73 mm) in anterior laxity between the injured and healthy knee with HT reconstruction (see Figure 1) [3,6-16].
Figure 1 Persisting instability following ACL repair with HT autografts (KT-1000 measurements). Individual study results were weighted by their inverse variance to derive a common point estimate with 95% confidence interval (diamond).
Many features contribute to an unsatisfactory or failed ACL replacement, for example, imprecise tunnel positioning, the presence of degenerative changes, or the onset of arthrofibrosis.
The choice of tibial graft fixation affects later stability. The intact ACL has a tensile strength around 2200 N. To avoid loosening, the graft must be fixed under firm traction (around 40 N), with the knee in a smoothly flexed position. A common way to anchor the tibial end of the graft is by titanium or biodegradable interference screws, for example, the BioCryl® (DePuy Mitek) screw that contains both resorbable poly-L-lactid and osteoconductive tricalcium phosphate (see Figure 2).
Figure 2 Appearance of the BioCryl® screws (left, courtesy of DePuy Mitek), and their positioning (right).
However, in a recent biomechanical study, extracortical fixation devices like the EndoButton® (Smith & Nephew) or RigidFix® (DePuy Mitek) provided better strength than did the interference screws [17]. The possible advantage of RigidFix® over other tethering methods is a splicing of strands, tightening the contact between the tendon surface and the bony tunnel over the entire graft circumference (see Figure 3).
Figure 3 Left: Positioning of RigidFix® pins. Right: splicing of graft bundles leading to close adherence to the surrounding bone.
Methods/ Design
Objectives
The present study aims at comparing later laxity in subjects undergoing arthroscopic anterior cruciate ligament replacement with either RigidFix® pinning or BioCryl® screwing of HT grafts. Both implants are CE approved, and were introduced to ACL-repair in Germany in 2002.
We have secondary objectives in imagining graft incorporation by MRI-scanning, functional results, residual pain, resumption of occupational and leisure activity, and quality of life by generic and disease-specific questionnaires.
Primary endpoint
We pose the primary hypothesis that the RigidFix® system preserves graft tension gained during surgery, and leads to lower KT-1000 arthrometer side-to-side differences than the BioCryl® screw after six months of follow-up.
Specifically, we will test the hypothesis that RigidFix® decreases the average difference gained with interference screws by 1.0 ± standard deviation 1.2 mm. The investigators consider this difference clinically sound, important, and measurable by KT-1000 arthrometer testing. Twenty-four subjects a treatment arm will allow for detecting this difference with an 80% chance at a two-sided alpha-level of 5%. Assuming a drop-out rate of 10%, 54 patients will be enrolled in this study.
Secondary endpoints
As secondary endpoints, we consider functional outcomes by means of the Lysholm scale, the Tegner score, and the International Knee Documentation Committee evaluation form (IKDC) in its German translation, 2000 revision [18]. Besides disease-specific items, this questionnaire also contains the Short-Form 36 (SF-36) generic health assessment tool. The noted instruments have proven reliability, validity, and responsiveness for use in clinical research.
Confirmatory testing will apply for the primary endpoint only. All secondary endpoints will be addressed in an exploratory fashion.
Design
SPOT is a patient and investigator blinded, randomised controlled trial conducted at a single academic institution.
Randomisation is carried out in the operating theatre shortly before transplant fixation, with random codes drawn from sealed envelopes. We use block-randomisation with five subjects a block following a computer-generated random list [19].
Inclusion criteria
Men and women (providing that they are not pregnant) being at least 18 years old are recruited to this trial. Subjects may engage in this study if they
- faced a first one-sided total or subtotal rupture of the anterior cruciate ligament, proven either by arthroscopy or MRI-scanning
- had met an acute knee distorsion event likely to have caused the index injury at least six weeks before scheduled repair
- have been physically examined in the ambulatory of the study hospital before assigning an admission date, and were screened and considered suitable to enter this trial by one of the investigators
Also, patients must be able to give voluntary written informed consent, and to comply with the post operative follow up regime
Exclusion criteria
We exclude patients
- with related lower limb fractures
- with active infection affecting the limb subject to needed treatment
- who have previously took part in this investigation or who are taking part in another clinical investigation
- with contraindications for MRI-scanning (that is, large indwelling orthopaedic implants made of metals others than titanium, or pacemakers)
Ethical considerations
This protocol and all accompanying documents were approved by the local Institutional Review Board (IRB).
According to IRB recommendations and requirements, information leaflets explicitly note that "a benefit from participation in this trial cannot be guaranteed."
We will stress the principle of randomisation as "treatment assignment by chance, without the possibility of the investigator, other health care professionals involved in this study, or the patient influencing the choice of treatment." We also tell patients that, as long as they keep agreement in participation, they will not know their assigned treatment until the six-month follow-up visit.
We will notify the IRB of any significant changes to the protocol. Also, we will notify the IRB within ten working days of the discovery of any severe adverse events which occur during this investigation.
Confidentiality of subject data will always be maintained. Subject anonymity will be guaranteed and all documentation about a subject (including radiographs) will be kept in secure location.
This investigation strictly adheres to the relevant articles of the Declaration of Helsinki as adopted by the 18th World Medical Assembly in 1964 and its later revisions, as well as to principles of GCP, developed within the Expert Working Group (Efficacy) of the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH).
Surgery and rehabilitation
All devices and instrumentations used in this clinical investigation bear the CE mark. They belong to the regular implants used for ACL repair at the study hospital since 2002. Except different graft anchoring, similar treatments apply to patients in both study groups. All participants undergo internationally accepted surgically procedures by a single surgeon (D.C.) with extensive experience in ACL repair using both the BioCryl® screw and RigidFix® cross pins. Also, postoperative care and rehabilitation programs do not differ from those employed outside a clinical trial.
All repairs are carried out under general anaesthesia, with the patient in a supine position. Perioperative antibiotic prophylaxis comprises 2 g of cefotiam. Patients receive 40 mg of enoxaparin daily for prophylaxis of thromboembolic events until full weight bearing.
The knee joint is accessed through two to three standard portals. Meniscal injuries are addressed with partial resection or repair.
Hamstring tendons are harvested via a small incision over the insertion of the pes anserinus at the anterior medial tibia by a closed tendon stripper, and prepared as triple-stranded grafts.
Tibial and femoral tunnels are drilled to approximate graft thickness (usually 8 to 9 mm) with the use of a guiding wire. Grafts are fixed with the knee in 30° flexion to achieve firm tension.
Postoperatively, the knee is stabilized for three days by a zero-degree splint. Afterwards, flexion is limited to 90° by a Secutec® orthosis for six weeks. Patients are allowed partial weight bearing with walking crutches. Subjects are prescribed intense physical therapy for motion exercise, and to strengthen thigh muscles. Normally, full range of motion and weight bearing is achieved until week 12 after surgery.
Patients in the experimental group have their grafts secured by tibial and femoral RigidFix® pinning. Patient in the control group receive tibial and femoral BioCryl® screws.
Baseline assessment
Each subject considered eligible for entry into this investigation has the following information and procedures recorded at the pre-investigational examination:
- Demographic details including date of birth and gender
- Medical history, coexisting diseases, and accompanying medication
- Physical examination, including circumferential measurement of both legs at defined landmarks, Lachman and pivot shift tests, KT-1000 arthrometer objectifying of instability, one-legged hop test, Lysholm, Tegner, and IKDC scores, knee and kneeling pain measured by visual analogue scales
- Radiographic examination, including a conventional roentgenogram of the injured knee in anteroposterior and lateral projection, and a preoperative MRI scan according to local standards
Intraoperative assessments
During surgery, we record procedure details in the electronic CRF. We assess the duration of surgery (from cut to skin closure), and operating theatre time (from induction of anaesthesia to arrival at the recovery room). A clinical knee examination is performed and documented with the subjects under general anaesthesia and relaxation. Arthroscopy findings (accompanying injuries to or degenerative changes of the cruciate or collateral ligaments, menisci, or cartilage) are recorded by video and/ or hard copy images.
We document eventual blood loss, and any other adverse event occurring during surgery.
The responsible surgeon judges the handling of implants and his overall satisfaction with the intraoperative result using five-point Likert scales. We detail any problems or complications on an Adverse Events form.
Follow-up assessments
Patients are appointed outpatient visits as part of the clinical investigation at 3, 6, and 12 months postoperatively. For study purposes, except quality of life measurements, patients do not undergo any diagnostic or other procedure not belonging to the common repertoire of assessments carried out after ACL repair. Specifically, we avoid invasive procedures, blood sampling, or imaging tests exposing subjects to radiation or contrast agents. The investigators consider the possible burden caused by extra clinical tests negligible.
We assess the following items at the scheduled visits:
Three months postoperatively
- Physical examination, including circumferential measurement of both legs at defined landmarks, Lachman and pivot shift tests, KT-1000 arthrometer objectifying of instability, one-legged hop test, Lysholm, Tegner, and IKDC scores, knee and kneeling pain measured by visual analogue scales
- Any complications or complaints raised by the patient, GP/ ambulatory surgeon, or Clinical Investigators
- Resumption of work
- MRI scan according to local standards
Six months postoperatively
- Physical examination, including circumferential measurement of both legs at defined landmarks, Lachman and pivot shift tests, KT-1000 arthrometer objectifying of instability, one-legged hop test, Lysholm, Tegner, and IKDC scores, knee and kneeling pain measured by visual analogue scales
- Any complications or complaints raised by the patient, GP/ ambulatory surgeon, or Clinical Investigators
- Resumption of work and leisure activities
- MRI scan according to local standards
Patients and Investigators responsible for follow-up examinations may learn about the assigned treatment after completing the six-month CRF.
Twelve months postoperatively
- Physical examination, including circumferential measurement of both legs at defined landmarks, Lachman and pivot shift tests, KT-1000 arthrometer objectifying of instability, one-legged hop test, Lysholm, Tegner, and IKDC scores, knee and kneeling pain measured by visual analogue scales
- Any complications or complaints raised by the patient, GP/ ambulatory surgeon, or Clinical Investigators
- Resumption of work and leisure activities
- MRI scan according to local standards
MRI studies
Radiological evaluation comprises
- Tunnel widening
- Impingement
- Transplant sufficiency, morphology of transitional areas and tendon-bone-interfaces
- Degree of degradation of screws and pins
- Degree of inflammation (synovitis) and effusion
- Presence or progression of arthrofibrosis
- Changes in cartilage and meniscal morphology
Physical examination
Before physical examination, both knees are prepared by applying opaque dressings to hide scars and to blind the examining doctors for the managed side.
Examination is performed independently by two of three board-certified surgeons (D.S., K.B., V.T.) who had not operated on any of the patients in the study. The responsible surgeon (D.C.) conducts a third clinical examination after completion of the case report forms. We document his examination findings separately and consider them as the diagnostic reference standard. We assess both interobserver agreement by kappa statistics and the accuracy of measurements taken by independent observers comparing with those of the responsible surgeon.
KT-1000 arthrometer testing
Objective translation measurements comprise a defined load (89 N). We measure the anterior translation of the injured and healthy side (in mm), as well as the difference between both knees.
Safety assessment and reporting of adverse event
We define an adverse event as 'any undesirable clinical instance in a subject whether it is considered treatment related or not'. In addition, an adverse device effect, undesirable side effect, is defined as 'a device related adverse event'.
A record of all adverse events, including details of the nature, onset, duration, severity, relationship to the device, relationship to the operative procedure and outcome, will be made on the relevant section of the subject's CRF. The subject will be questioned about any adverse event at each later follow-up assessment visit.
An adverse event or an adverse device effect may be mild, moderate or severe and are usually unexpected.
A severe adverse event or adverse device effect is defined as any experience that
- is fatal or life-threatening
- is permanently disabling
- needs or prolongs in-patient hospitalization because of a potential disability, danger to life or forces an intervention
All severe adverse events or adverse effects which occur during this investigation must be and will be reported immediately by telephone or facsimile to
Bundesinstitut für Arzneimittel und Medizinprodukte, Kurt-Georg-Kiesinger-Allee 3, 53175 Bonn, Phone: ++49 30 228 207 30, Fax: ++49 30 228 207 5207
Data management
For data collection, we set up a Microsoft XP Professional Access® Database, run on a mobile computer separately from the hospital documentation and the intranet. Data collection and storage comply with orders fixed by the data safety board of the Unfallkrankenhaus Berlin, and follow German laws for data safeguard and protection (Bundesgesetz über den Schutz personenbezogener Daten [Datenschutzgesetz 2000 – DSG 2000], 17. August 1999, BGBl. I Nr. 1999/165). We ensure data storage for five years.
For study documentation, we assign patients an identification number. Electronic sources do not contain names or addresses of participants. Linking lists are stored in a study folder with copies of adverse events forms.
Since this study runs at a single centre, we do not appoint an external monitor for data handling and management. We regularly (at least twice a month) check datasets for consistency, completeness and plausibility.
Statistical analysis
We will conduct all analyses following the intention to treat principle (that is, patients will be evaluated as randomized).
We will express measurements as means, medians or proportions with their proper distribution indices (that is, standard deviations, ranges, and interquartile ranges). In case of skewed distributions, normalizing will be achieved by logarithmic transformation, where necessary.
As pointed out earlier, we will address only the primary hypothesis in a confirmatory fashion, whereas all other results will be evaluated in a plain exploratory intent.
We will employ the student's t-test for independent samples to test for the difference in anterior laxity between both fixation methods at six months of follow-up.
For secondary endpoint analysis, we will calculate cross tables, 95% confidence intervals for normally or binomially distributed data, and differences in means, proportions, and ratios. In case of obvious benefits or harms with either device in a certain subgroup of patients, we will use stratified analyses (for example, according to Mantel-Haenszel).
Where statistically and/ or clinically sound, we will consider linear and logistic regression analyses, or more sophisticated regression models for correlated data (for example, generalized estimating equations). We will, however, respect the small sample size of this study, and limit statistical analyses to the necessary minimum.
In case of missing data, we will use both a last observation carried forward approach, and imputation methods by regression or semi-Bayesian modelling. Separate analyses will be performed for raw and modelled data.
Discussion
After ACL repair, most patients rarely recognize slightly weakened anterior knee stability in everyday life. However, subjects with a high recreational and sporting activity and physically strenuous professions often suffer from recurrent events of giving-way, especially on hastened movements. This poses a high risk for secondary knee injury. Of note, muscular training cannot compensate for residual laxity, outbalancing the anticipated benefit from surgical repair.
Thus, attempts to optimize the surgical technique may be valuable. Currently, surgeons performing ACL reconstruction use screws, pins, buttons, and cramps for graft fixation because of individual preference, or institutional orders. The latter are chiefly driven by cost considerations. For example, the purchase price of a RigidFix® tray is ten times higher than that of BioCryl® screws. Obviously, the more expensive implant must prove a distinct clinical advantage over the common one to justify its further use. Unfortunately, there is lack of robust evidence on the effectiveness of either fixation method beyond laboratory and animal experiments.
Although conceptually impressive, there is no comparative study that proved a clinically measurable advantage of RigidFix® over screws. The investigators consider the equipoise principle fulfilled, since it is unclear whether screw or pins lead to better long-term stability, or show any measurable differences at all. We hope that the results from this pragmatic study can clarify this issue.
Competing interests
MW, DS and AE have worked as independent scientific consultants for DePuy® International, and received project-related funding that does not apply to this work. No support in any form was provided, or will be provided by third parties to set up the trial protocol, or to conduct this study. This study aims at investigating biomechanical principles, not certain implants.
The members of the SPOT Group have no financial or non-financial competing interests in this work.
Authors' contributions
DS drafted the manuscript. KB and GM edited the manuscript.
DC is in charge of all surgical procedures. DS is responsible for statistical analyses.
All authors participated in the design of this study, and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Freedman KB D'Amato MJ Nedeff DD Kaz A Bach BR Arthroscopic anterior cruciate ligament reconstruction: a metaanalysis comparing patellar tendon and hamstring tendon autografts Am J Sports Med 2003 31 2 11 12531750
Lobenhoffer P Kniebandverletzungen. I. Anatomie, Biomechanik, Diagnostik, Indikationsstellung Chirurg 1999 70 219 230 10097870 10.1007/s001040050075
Beynnon BD Johnson RJ Fleming BC Kannus P Kaplan M Samani J Renstrom P Anterior cruciate ligament replacement: comparison of bone-patellar tendon-bone grafts with two-strand hamstring grafts. A prospective, randomized study J Bone Joint Surg 2002 84-A 1503 1513 12208905
Weiler A Anatomische „Hamstringsehnen" Verankerung mit Interferenzschrauben beim Kreuzbandersatz. Biomechanische und tierexperimentelle Untersuchungen PhD Thesis 2002 Charité Campus Virchow-Klinikum, Berlin, Germany
Kousa P Graft fixation in ACL reconstruction Academic Dissertation 2004 University of Tampere Medical School, Tampere, Finland
Jansson KA Linko E Sandelin J Harilainen A A prospective randomized study of patellar versus hamstring tendon autografts for anterior cruciate ligament reconstruction Am J Sports Med 2003 31 12 18 12531751
Ejerhed L Kartus J Sernert N Köhler K Karlsson J Patellar tendon or semitendinosus tendon autografts for anterior cruciate ligament reconstruction? Am J Sports Med 2003 31 19 25 12531752
Haieb MD Kan DM Chang SK Marumoto JM Richardson AB A prospective randomized comparison of patellar tendon versus semitendinosus and gracilis tendon autografts for anterior cruciate ligament reconstruction Am J Sports Med 2002 30 214 220 11912091
Corry IS Webb JM Clingeleffer AJ Pinczewski LA Arthroscopic reconstruction of the anterior cruciate ligament Am J Sports Med 1999 27 444 454 10424213
Potel JF Boussation M Djian P Franceschi JP Reparation arthroscopique du ligament crois anterieur comparaison tendon rotulien versus tendon de la patte d'oie. Etude rétrospective multicentrique de la Société Française d'Arthroscopie Annales de SFA 1999
Marder RA Raskind JR Carroll M Prospective evaluation of arthroscopically assisted anterior cruciate ligament reconstruction. Patellar tendon versus semitendinosus and gracilis tendons Am J Sports Med 1991 19 478 484 1962713
Maeda A Shino K Horibe S Nakata K Buccafusca G Anterior cruciate ligament reconstruction with multistranded autogenous semitendinosus tendon Am J Sports Med 1996 24 504 509 8827311
Yasuda K Tsujino J Tanabe Y Kaneda K Effects of initial graft tension on clinical outcome after anterior cruciate ligament reconstruction. Autogenous doubled hamstring tendons connected in series with polyester tapes Am J Sports Med 1997 25 99 106 9006702
Röpke M Becker R Urbach D Nebelung W Semitendinosussehne vs. Ligamentum patellae Unfallchirurg 2001 104 312 316 11357697 10.1007/s001130050733
Karlson JA Steiner ME Brown CH Johnston J Anterior cruciate ligament reconstruction using gracilis and semitendinosus tendons. Comparison of through-the-condyle and over-the-top graft placements Am J Sports Med 1994 22 659 666 7810790
Howell SM Taylor MA Brace-free rehabilitation, with early return to activity, for knees reconstructed with a double-looped semitendinosus and gracilis graft J Bone Joint Surg 1996 78-A 814 825 8666598
Kousa P Jarvinen TL Vihavainen M Kannus P Jarvinen M The fixation strength of six hamstring tendon graft fixation devices in anterior cruciate ligament reconstruction. Part I: femoral site Am J Sports Med 2003 31 174 181 12642249
Irrgang JJ Anderson AF Boland AL Harner CD Kurosaka M Neyret P Richmond JC Shelborne KD Development and validation of the international knee documentation committee subjective knee form Am J Sports Med 2001 29 600 613 11573919
randomization.com
| 15723704 | PMC552311 | CC BY | 2021-01-04 16:28:03 | no | BMC Surg. 2005 Feb 21; 5:1 | utf-8 | BMC Surg | 2,005 | 10.1186/1471-2482-5-1 | oa_comm |
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-141571591710.1186/1472-6963-5-14Research ArticleIntegrating quantitative and qualitative methodologies for the assessment of health care systems: emergency medicine in post-conflict Serbia Nelson Brett D [email protected] Kerry [email protected]Šćepanović Milena [email protected]ć Mihajlo [email protected]ć Miloš [email protected]ć Ljiljana [email protected] Michael J [email protected] Center for International Emergency, Disaster and Refugee Studies Johns Hopkins Bloomberg School of Public Health: Department of International Health, Johns Hopkins University School of Medicine: Department of Emergency Medicine, 615 N. Wolfe Street, Baltimore, Maryland 21205, USA2 Emergency Center of the Clinical Center of Serbia, School of Medicine, Belgrade University, Pasterova 2, 11000 Belgrade, Republic of Serbia3 Division of International Health and Humanitarian Programs, Brigham and Women's Hospital: Department of Emergency Medicine, Harvard Medical School, 75 Francis Street Boston, Massachusetts 02115, USA2005 17 2 2005 5 14 14 21 4 2004 17 2 2005 Copyright © 2005 Nelson et al; licensee BioMed Central Ltd.2005Nelson 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
Due to the complexity of health system reform in the post-conflict, post-disaster, and development settings, attempts to restructure health services are fraught with pitfalls that are often unanticipated because of inadequate preliminary assessments. Our proposed Integrated Multimodal Assessment – combining quantitative and qualitative methodologies – may provide a more robust mechanism for identifying programmatic priorities and critical barriers for appropriate and sustainable health system interventions. The purpose of this study is to describe this novel multimodal assessment using emergency medicine in post-conflict Serbia as a model.
Methods
Integrated quantitative and qualitative methodologies – system characterization and observation, focus group discussions, free-response questionnaires, and by-person factor analysis – were used to identify needs, problems, and potential barriers to the development of emergency medicine in Serbia. Participants included emergency and pre-hospital personnel from all emergency medical institutions in Belgrade.
Results
Demographic data indicate a loosely ordered network of part-time emergency departments supported by 24-hour pre-hospital services and an academic emergency center. Focus groups and questionnaires reveal significant impediments to delivery of care and suggest development priorities. By-person factor analysis subsequently divides respondents into distinctive attitudinal types, compares participant opinions, and identifies programmatic priorities.
Conclusions
By combining quantitative and qualitative methodologies, our Integrated Multimodal Assessment identified critical needs and barriers to emergency medicine development in Serbia and may serve as a model for future health system assessments in post-conflict, post-disaster, and development settings.
==== Body
Background
In June 1991, a series of civil wars began as the Socialist Federal Republic of Yugoslavia dissolved violently into the four independent republics of Slovenia, Croatia, Bosnia-Herzegovina, the Former Yugoslav Republic of Macedonia, and the semi-autonomous nations of Serbia and Montenegro. Kosovo remains a province of Serbia but under the administration of the international community. The events of the last decade in the Balkans have significantly impacted the health care system of Serbia. The nation and its health care system were devastated by international sanctions, soaring unemployment, political instability, near economic collapse, and the North Atlantic Treaty Organization (NATO) air campaign [1]. Consequently, the health care budget was cut dramatically, further limiting the system's capabilities of providing adequate health care. These difficult years resulted in a substantial fall in major health indices and left a crippled health system struggling to provide for the needs of its citizens [2].
A country's emergency medical services are a particularly important component of the health care system for they provide an essential front-line resource for trauma, medical emergencies, and those without other access to health care. Unfortunately, the system of emergency services in Serbia did not escape the devastation of the 1990s and continues to suffer from problems common throughout the health care system. Limited health care funding has resulted in a lack of necessary equipment, supplies, medications, personnel wages, and an economic infrastructure unable to support necessary health care reforms in the public and private sectors. Furthermore, the country has received little external support for health system reform, underscoring the need for focused assessments to best determine priorities for health care development.
There is a desperate need for reorganizing and restructuring emergency medical services throughout Serbia. To better understand the problems, needs, and obstacles to development, a team of investigators from Johns Hopkins University Schools of Medicine and Public Health collaborated with the Belgrade University Emergency Center to perform a multimodal assessment of emergency medicine in Serbia. Our goal in this study was to identify the most pressing needs and achievable goals, as perceived by emergency health care personnel, in order to ensure that development programs and available funds are appropriately directed [3].
Evaluating health care systems through the use of demographic data collection, surveys, or focus groups alone – as is often done in health system assessments – cannot adequately elucidate the intricacies of a country's health care system. In order to fully appreciate the complex dynamics of the health care system, it is necessary to perform a complete initial assessment; include strong local participation; identify and address significant barriers to change; and identify unique local needs and other cultural dimensions. To do this effectively, it is important to avoid simply introducing a "carbon copy" of a Western health care system that would further burden the system with inappropriate and unsustainable programs. We propose for the rapid assessment of health care services an Integrated Multimodal Assessment that uniquely combines diverse methodologies, whose individual effectiveness has been well described in the literature.
Methods
This study used a multimodal approach to assess the strengths, needs, problems, and obstacles related to the development of emergency medical services in Belgrade. The four study modalities were 1) demographic information and observational data; 2) focus group discussions with 68 emergency medicine personnel; 3) individual, free-response questionnaires with 39 emergency medicine personnel; and 4) by-person factor analysis (or Q-methodology) of the attitudes and opinions of 36 emergency medicine personnel regarding the development of emergency medical services. Participants in the study were a non-probabilistic sampling of emergency medicine physicians, nurses, and department directors from all pre-hospital and hospital-based emergency medicine institutions in Belgrade. Many participants took part in more than one study modality. Data collection occurred during July-August 2002. The authors received institutional review board approval and the written informed consent of all participants.
Characterization of emergency medical services
The capacity of emergency medical services in Serbia to provide for the needs of the country's citizens was characterized through health system data collection. Health system information was obtained through meetings with key officials in established health surveillance institutions such as the Emergency Center of the Clinical Center of Serbia; World Health Organization Belgrade office; Public Health Institute of Belgrade; Republic of Serbia Ministry of Health; European Agency for Reconstruction; Belgrade University, Institute of Social Medicine, Statistics and Health Research; as well as several non-governmental organizations with offices in Belgrade. Key officials were asked a series of quantitative, closed-response questions about the Serbian health care system and emergency medical services. Developed from a pilot health care assessment in Serbia the previous year, the questions focused on health system organization; patient population size and demographics; health care personnel; medical education and training; materials, supplies, and medications; and provider aptitude and morale [4].
Focus group discussions
The two-fold purpose of the focus group discussions was 1) to better understand the needs, problems, and obstacles to system development as perceived by those involved directly in providing emergency medical services and 2) to collect a wide variety of opinions and attitudes for subsequent by-person factor analysis. The focus group discussions were held with 68 emergency medicine physicians, nurses, and directors through a non-probabilistic sample taken from the institute for pre-hospital emergency services, the Emergency Center of the Clinical Center of Serbia, and four of Belgrade's six community emergency departments. The discussions were conducted by the primary author in the Serbian language. The audio recordings of the 30- to 60-minute discussions were transcribed and translated into English by a native-English speaker fluent in Serbian (BDN) and a native-Serbian speaker fluent in English (MŠ). The translation was later verified by additional native-Serbian speaking co-authors fluent in English (MV, LM).
The investigative field team (BDN, KD, MŠ, MV, LM) summarized the final transcripts and extracted common themes and divisive statements for by-person factor analysis.
Emergency medical provider free-response questionnaire
Free-response questionnaires enabled assessment of emergency health care needs and priorities for change as perceived by those most directly involved in patient care: the emergency health care providers. A non-probabilistic sample of 39 participants was obtained by open invitation, without compensation, through the director's office of each institution. As with the focus group discussions, the institute for pre-hospital emergency services, the Emergency Center of the Clinical Center of Serbia, and four of the city's six community emergency departments were represented. All subjects were administered a written Serbian-language questionnaire comprised of 10 free-response items (Table 1). The questions were developed through discussion with emergency medicine leaders at the University of Belgrade as well as through adaptation of earlier health care assessments conducted by the authors in the region. Each questionnaire was translated from Serbian to English by two bilingual translators (BDN, MŠ) and verified by additional authors (MV, LM) to ensure accuracy. Translated responses were coded and assessed for content by a researcher (KD) blinded to the demographics and specialty training of the respondents. More than one response was coded for each subject when necessary. Duplicate answers were only coded once. Illegible, blank, and off-subject answers were coded as missing data. Data were analyzed using SPSS 11.0 for Windows.
Table 1 Free-response questionnaire questions. Investigators developed and administered to participants a Serbian-language questionnaire containing the following open-ended questions.
Free-response questionnaire questions
• What functions best in emergency medicine in your country and why?
• How is the status of physicians in emergency medicine in comparison to that of physicians in other specialties?
• With regard to education and training, what are the obstacles to becoming a physician in the field of emergency medicine?
• What are the current problems in the system of emergency medicine? And what is being done to solve these problems?
• What are the priorities for improving the system of emergency medicine?
• What are the barriers to the future development of the system of emergency medicine?
• What types of training should be offered for the improvement of the system of emergency medicine?
• What are the strong and weak points of the training of physicians in the system of emergency medicine?
• What do you think would be the reaction of the health care community (physicians, nurses, administrators) to development of the system of emergency medicine?
• What do you think would be the response of the public and government to development of the system of emergency medicine?
By-person factor analysis
By-person factor analysis, or Q-methodology, avoids many of the limitations of other modalities by allowing the grouping of participants based on their subjective responses to an issue while preventing investigator preconceptions from influencing the grouping structure. In a letter to Nature in 1935, physicist and psychologist William Stephenson introduced by-person factor analysis as a means for the scientific study of subjectivity [5]. The methodology uses a unique combination of qualitative and quantitative methods to subdivide a study population, evaluate the degree of consensus among the participants, and identify any discordant opinions. The process begins by creating a concourse, or collection, of opinions and perceptions toward the subject of interest – in this case, emergency medicine in Serbia. From this concourse, the investigators develop statements representing the spectrum of opinion and request that respondents assign each statement to a position within a quasi-normal grid distribution as to whether they completely disagree with, feel indifferently or ambivalently towards, or completely agree with the statement [6,7]. In our present study, a concourse was developed through the aforementioned focus groups and free-response questionnaires. Directly from this concourse, the field investigators (BDN, KD, MŠ, MV, LM) produced a Q-sample of 23 statements (Table 2), which was administered to 36 Belgrade emergency medicine physicians, nurses, and department directors. The participants were attained by open invitation through the director's office of each institution. Collected demographic information of participants included age, gender, and specialty training. Respondents were asked to sort the Q-sample using the condition of instruction, "Please sort these statements with respect to your opinion of the current system of emergency medicine in Serbia". Sorting involved the forced ranking of all statements within the grid of quasi-normal distribution. Statements with which the participant most strongly disagreed were placed toward one end of the grid (-3 agreement score). Toward the grid's other extreme (+3 agreement score) were placed statements with which the participant most strongly agreed. The participant placed statements that evoked ambivalence or neutrality near the center (0 agreement score) of the quasi-normal distribution. Factor analysis was then performed on the responses using PQMethod 2.10 [8] followed by manual factor rotation. This analysis leads to the subdivision of the study population into distinct "respondent types" (i.e. factors), which are clusters of participants grouped by their common opinions and perceptions. To define and characterize the resultant respondent types, all participants who loaded heavily on a respondent type (>60% concordance) were selected as respondent type "loaders". The participants who loaded heavily and specifically for a single respondent type (i.e. >60% concordance on the type of interest and <30% concordance on remaining types) were designated respondent type "definers" and closely reviewed to assist in characterizing each respondent type.
Table 2 Responses of emergency medicine physicians and administrators to Q-statements regarding emergency medicine in Serbia. Statements are listed from greatest to least participant consensus. With each statement, an averaged agreement score is calculated for all participants and for each identified respondent type. Scores represent the spectrum of participant agreement/disagreement (i.e. "strongly disagree" (-3), "ambivalent/neutral" (0), or "strongly agree" (+3) with the statement). To aid discussion of respondent types, summary labels ("Utilize", "Develop", and "Invest") characterize unique qualities of each respondent type. "Utilize" respondent type is most concerned with the poor utilization of emergency services. "Develop" respondent type advocates the further development of emergency medicine. "Invest" respondent type emphasizes the need for greater investment in emergency medicine.
Q-statements (listed in order of greatest consensus to least consensus) Averaged participant agreement score Agreement scores of identified respondent types
"Utilize" (5 loaders, 4 definers) "Develop" (6 loaders, 3 definers) "Invest" (4 loaders, 2 definers)
It is NOT necessary for patients to be seen by physicians in the field, but rather patients should be brought immediately to an emergency department for care. -2.0 -2 -1 -3
Health management training for health care leaders is essential for the improvement of emergency medical services. -0.1 -1 -1 -2
The public overuses ambulance services because there is no charge for the use of these services. 0.6 1 0 1
The first priority for the development of emergency medicine should be to improve the organization of emergency services. 0.4 2 1 0
Patients arriving to the emergency facility should be taken, according to their illness, directly to a specific specialty department. -0.4 -1 -1 -1
Protocols should be developed to standardize the treatment of patients throughout Serbia. 1.2 0 3 -1
Emergency medicine should be taught as a required course during the last year of medical school. 0.8 -2 2 0
Emergency medicine in Serbia would function better if it were financed by the federal budget rather than by the social health insurance fund. 0.0 1 0 -1
Emergency medicine should be a separate specialty in which physicians are trained to exclusively practice emergency medicine. 0.0 1 2 -2
Primary health care providers in the health houses are sufficiently trained in the triage of emergent and non-emergent patients. -1.8 -1 -2 -3
All institutions that provide emergency medical services should be open 24 hours a day. 1.4 0 2 1
A priority in the development of emergency medicine is to increase the number of appropriately equipped ambulances. -0.6 -2 -3 1
There should be national guidelines to determine which illnesses/injuries should be treated at each type of health care facility. 0.2 2 0 1
Much of the burden on emergency health care providers is due to the time spent caring for non-emergent patients. 1.3 3 0 0
Continuing medical education should be required by law of all physicians working in emergency medicine. 1.0 0 1 2
There is poor coordination among the various specialties that provide emergency medical services. -0.3 -1 1 -1
The public should be better educated about the level of care that each health care institution provides in order to properly use the available health care services. 0.3 3 0 0
The medical school is currently playing a sufficient role in the development of emergency medicine in Serbia. -1.6 -3 -1 -1
There is poor cooperation between the emergency centers, clinical-hospital centers, pre-hospital emergency services, and health houses. 0.6 1 1 3
Physicians working in emergency medicine in Serbia need greater expertise and technical skills to provide an appropriate level of care. 0.8 -2 3 3
The problems in emergency medicine would be solved if there were money and equipment with which to work. 0.4 2 -2 2
There is an appropriate balance of theoretical and practical training for physicians in emergency medicine. -2.0 -3 -3 -2
Radio communication does not function effectively between the ambulances and the medical institutions. -0.1 0 -2 2
Results
Characterization of emergency medical services
Emergency medicine is not a recognized health care specialty in Serbia. In smaller communities, emergency services are provided by primary health stations (zdravstvene stanice) and primary health centers (domovi zdravlja). In the capital city of Belgrade, there exists a network of 6 part-time, hospital-based (kliničko-bolnički centri) emergency departments that are supported by around-the-clock pre-hospital emergency services (hitna pomoć) and the academic Emergency Center (Urgentni centar) at the Clinical Center of Serbia. Like many European countries, Serbia has adopted a largely Franco-German model of emergency medicine in which pre-hospital emergency services are provided in the field by physician-staffed ambulances [9]. Care is subsequently provided in-hospital by physicians of multiple medical specialties. Patients access emergency services through referral from Belgrade's 16 primary health centers, telephoning "94" for pre-hospital emergency services, or by personal referral to the emergency institutions. The academic Emergency Center at the Clinical Center of Serbia receives patients directly as well as by referral from community emergency departments that are closed or overwhelmed. During 2001, the Emergency Center received 130,877 patients.
Focus group discussions
Through focus group discussions, the investigators were able to determine the opinions and perceptions of emergency service providers and administrators on the current emergency medical system in Serbia. Sixty-eight physicians participated in one of six focus group discussions (female: 39.7%; average work experience: 19.3 years; specialties: surgery (34%), internal medicine (29%), anesthesia (12%), general medicine (7%), emergency medicine (3%), other (15%)). Discussions were held at the institute for pre-hospital emergency services, the Emergency Center, and four hospital emergency departments.
The participants' comments from the focus group discussions are collectively summarized in Table 3. Most providers believe the lack of sufficient funding for emergency medical services is one of the greatest problems affecting emergency medicine in Serbia. Financial support for medications, medical supplies, modern equipment, employee salaries, and facility maintenance is extremely limited and restricts the capabilities of health care providers. Poor organization of emergency medical services, including the lack of government regulation, absence of uniform treatment protocols, improper system management, and poor triaging and routing of patients between facilities, also dominated the focus group discussions.
Table 3 Foremost problems of emergency medicine in Serbia and priorities for system development. A summary of the findings from focus group discussions with providers of emergency medical services.
Results of focus group discussions with providers and administrators of emergency medical services
CITED PROBLEMS IN THE SYSTEM OF EMERGENCY MEDICINE IN SERBIA
FINANCE:
≺ Inadequate financial resources for essential equipment, supplies, and medications
≺ Discouraged emergency medical service personnel due to meager salaries, difficult work conditions, and large workloads
≺ Inadequate number of properly equipped ambulances and functioning radio equipment
≺ Very few computers and no health information systems to track patient health records
ORGANIZATION:
≺ Need for federal regulation of emergency medical services
≺ Lack of sufficient protocols for the standardization of triage and treatment
≺ Inadequate coordination between the institutions providing emergency medical services
≺ Need for further development of emergency medicine as its own specialty
EDUCATION:
≺ Inadequate training in emergency medicine during medical school
≺ Insufficient practical training of emergency medical service providers
≺ Few opportunities for professional development and continuing education of emergency service providers
≺ Limited access to medical innovations through the internet, foreign professional journals, conferences, courses, and seminars
≺ Lack of health management training for leaders of health care institutions
≺ Need for public education about the emergency medical services system and how to properly utilize them
SUGGESTED PRIORITIES FOR THE DEVELOPMENT OF EMERGENCY MEDICINE IN SERBIA
FINANCE:
≺ Secure funding for essential medications, supplies, equipment, employee salaries, and maintenance of health care facilities
≺ Consider long-term sources of continuous funding for emergency services such as the government budget instead of the social health insurance fund
ORGANIZATION:
≺ Develop national protocols for the standardization of emergency triage and treatment
≺ Further develop emergency medicine as its own specialty
≺ Clearly define the responsibilities and emergency services of physicians in each health care facility
≺ Institute a system to promote better coordination between the primary health centers, the hospital emergency departments, and the Emergency Center
≺ Implement quality control measures for the delivery of emergency medical services
≺ Establish a health information system to facilitate the tracking of patients
EDUCATION:
≺ Introduce required continuing medical education supported by legislation that would provide health care professionals leave from work to attend this periodic training
≺ Provide health care professionals with access to continuing medical education through the internet, professional journals, conferences, seminars, and practical training
≺ Develop a fellowship program for emergency medicine physicians
≺ Increase the level of practical emergency medical experience provided in medical school and postgraduate training
≺ Implement training in BLS, ALS, and emergency triage for all health care providers
≺ Institute a health management training courses for leaders of health care institutions
≺ Educate the public regarding the level of emergent care that each health care institution provides and how to properly utilize the available health care services
In addition, physicians and medical directors report that their emergency medical system is in significant need of improved education and training programs. Suggested improvements include further training in emergency medicine during medical school, bedside training to develop practical skills, development of standardized treatment protocols, and access to continuing professional education.
Emergency medical provider free-response questionnaire
Questionnaires were completed by 39 physicians at the institute for pre-hospital emergency services, the Emergency Center, and four hospital emergency departments (female: 41.0%; average work experience: 20.8 years; specialties: surgery (31%), internal medicine (31%), anesthesia (13%), general medicine (10%), emergency medicine (5%), other (10%)).
Respondents show several areas of high agreement – particularly remarkable considering the questionnaire's free-response format. The most frequently reported strength of Serbia's current health care system are the health care workers (69%). Worker enthusiasm (33%) and the collaboration of various medical specialties providing emergency services (21%) are most frequently cited as the greatest positive attributes of emergency service providers. In addition, 46% of respondents report that the various emergency medical service providers (especially the Emergency Center and the pre-hospital emergency services) are other strengths of emergency medical services in Serbia.
Thirty percent of the participants surveyed believe that lack of incentives to enter the specialty, difficulty of the work, and poor financial compensation are important barriers to becoming a physician in emergency medicine in Serbia. The poor reputation of physicians in emergency medicine relative to other physicians (reported by 46% of participants) may contribute to lower interest in specializing in emergency medicine. Poor organization (26%) and insufficient opportunities for professional development (15%) were also reported as major impediments to being an emergency service provider.
Participants were asked what they believed were major problems of emergency medical services in Serbia. Inadequate finances, medications, medical supplies, and modern medical equipment were cited by 54% of the emergency medicine personnel as being a major problem (Table 4). Of these respondents, 62% feel that they have sufficient technical skills but that insufficient medical equipment significantly restricts their capabilities as health care providers. Another 24% believe that the absence of adequate and continuous funding for emergency medicine is a critical problem. With improved funding they could obtain the necessary supplies, equipment, and medication they need, as well as improve employee salaries and facility maintenance. As a way to ensure a more reliable source of funding, 78% of the emergency medicine personnel surveyed emphasized the need for government financing of emergency medical services through the national budget rather than through the unpredictable social health insurance fund.
Table 4 Quantitative results of emergency medical provider questionnaires. Positive responses were calculated as a percentage of the number of providers that included the statement in their free response. More than one response was coded per subject when applicable. Italicized responses denote breakdown of individuals with the above response.
Select questionnaire responses of emergency medical providers (n = 39) Positive responses (%)
CURRENT PROBLEMS IN EMERGENCY MEDICINE
Organization 59
An organized system doesn't exist 35
No government regulation 22
Lack of treatment/triage protocols 22
Poor coordination between health care facilities / lack of team work 17
Poor division of labor of those providing emergency services 17
Other 26
Lack of supplies, equipment, and medications 54
Insufficient equipment 62
Inadequate funding 24
Lack of optimal ambulances 5
Poor diagnostic & therapeutic procedures 5
Training and education 36
Lack of incentives / difficult field of work / poor compensation 23
No answer / there is no system of emergency medicine 18
Pre-hospital emergency services / ambulance services 5
PRIORITIES FOR REFORM
Organization 77
Treatment protocols/guidelines 27
Improve coordination between health care facilities (team work) and within hospitals (between specialties) 27
Increase the number of physicians (personnel) trained in EM 17
Government regulation and organization 13
Legal regulation 10
Increase the number of beds 10
Establishment of EM as a separate specialty 7
Develop a computer database for tracking patients 3
Improve efficiency of system 3
Supplies, equipment, and medication (improved diagnostics) 54
Training and education 36
Financing 33
Incentives / work conditions / compensation 21
BARRIERS TO FUTURE DEVELOPMENT
Economics/resources 69
Organization 36
Political/government 26
Inadequate education and training 18
Reorganization of emergency medical services is a priority for development according to 77% of participants. Of these responses, poor coordination and team work between health care facilities (27%), the absence of treatment protocols (27%), insufficient numbers of emergency medical physicians (17%), and the lack of government regulation (13%) were the most frequently cited priorities for improving the organization of emergency medical services. Other concerns include lack of quality control measures, burden of caring for non-emergent cases, inadequate admission and triage areas, and insufficient hours of operation of hospital-based emergency departments.
Many emergency medicine personnel are also critical of the education and training of physicians in emergency medicine. Although 58% of those surveyed feel that physicians receive good theoretical training, 77% report that their educational system is in significant need of improved medical school and postgraduate training programs. Of the individuals advocating for improved education and training, 27% feel that further training in emergency medicine should be included in the medical school curriculum, and 53% believe that additional bedside training to develop practical skills is imperative. Furthermore, 56% of participants in this study express a need for better access to continuing professional education, including shared practical experiences and training between health care institutions. A large number of emergency medicine personnel (36%) stress the need for greater access to international journals, conferences, seminars, and the internet. Many (15%) also communicate a need for international collaboration and training programs between Serbia and foreign medical institutions.
Survey participants cite numerous obstacles to the development of emergency medical services in Serbia, including insufficient funding and resources (69%), poor organization of emergency services (36%), lack of governmental support (26%), and inadequate education and training (18%). Also vital to development efforts are the perceptions and attitudes of the people impacted by and involved in these efforts – the public, health care community, and government officials. A slight majority (62%) of individuals surveyed believe that the health care community will support further development of emergency medical services. However, only 46% and 41% of participants feel that the public and government, respectively, will respond positively to development.
By-person factor analysis
Of the 36 emergency medicine personnel invited to participate in factor analysis, 33 individuals (91.7%) completed the exercise correctly (female: 45.4%; average work experience: 16.8 years; specialties: surgery (21%), internal medicine (27%), anesthesia (27%), general medicine (6%), emergency medicine (6%), other (12%)). The institute for pre-hospital emergency services, the Emergency Center, and two hospital emergency departments participated in this component of the study.
Factor analysis followed by manual factor rotation determined levels of agreement among the 33 participants and identified three unique types of respondents. Table 2 displays the averaged level of agreement and the level of agreement for each respondent type (the spectrum of agreement includes "I strongly disagree" (-3), "I feel ambivalent/neutral" (0), or "I strongly agree" (+3)). With statements written in the form of agreement, a majority of participants believe patients should be seen by physicians in the field (+2.0), protocols should be developed to standardize the treatment of patients throughout Serbia (+1.2), primary health care providers are not sufficiently trained in the triage of emergent and non-emergent patients (+1.8), emergency medical institutions should be open 24 hours a day (+1.4), much of the burden on emergency health care providers is due to time spent caring for non-emergent patients (+1.3), continuing medical education should be required by law of all physicians working in emergency medicine (+1.0), the medical school is not playing a significant role in developing emergency medicine in Serbia (+1.6), and there is not an appropriate balance of theoretical and practical training for physicians in emergency medicine (+2.0).
In addition to assessing levels of agreement among participants, by-person factor analysis allowed us to identify three different respondent types, each involving multiple respondents with diverse demographics but who share common and unique perspectives relative to the remaining participants. To aid discussion of the three respondent types, each was assigned a brief label: "Utilize", "Develop", and "Invest".
Respondent type "Utilize" was heavily loaded by a total of 5 respondents (15.2% of participants) and defined by 4 (12.1%). The principal concerns of the "Utilize" respondent type include the poor utilization of emergency medical services. These individuals believe that emergency personnel are well trained, educated, and prepared to address the emergent needs of the community. However, they consider their role ill-defined and misunderstood, leading to the ineffective use of their services by the public.
Respondent type "Develop" was characterized by 6 loaders (18.2%) and 3 definers (9.1%). These individuals support the further development of emergency medicine as an independent specialty in Serbia. They do not consider inadequate finances, supplies, or equipment to be the chief concern of the system of emergency medicine. Instead, the "Develop" respondent type encourages the development of standardized treatment guidelines, improved coordination between the specialties providing emergency care, and additional education and training in the specialty of emergency medicine.
The third respondent type, "Invest", included 4 loaders (12.1%) and 2 definers (6.1%). These individuals consider the system of emergency medicine appropriately organized but lacking adequate investment. They believe greater resources should be applied towards equipment and professional training and development. They are also concerned with an apparent lack of coordination between institutions providing primary health care and emergency services.
Additional respondent types were defined by single respondents, which prevented adequate characterization. However, the demographic data on each of these isolated respondent types were closely examined to assure that the individual was not in a unique position of authority that could disproportionately influence future health system reform. The specific responses of respondent types "Utilize", "Develop", and "Invest" to the 23 Q-statements are listed in Table 2.
Discussion
The purpose of a rapid assessment is to provide organizations with timely and reliable information to aid targeted interventions. Many rapid assessment methodologies have been utilized, each with its own strengths and weaknesses. One of the most commonly used methodologies is the KAP (Knowledge, Attitudes, and Practices) survey, which is traditionally a standard questionnaire with pre-developed, closed-response questions [10]. Because of its ease in study development, administration, and analysis, a KAP survey can provide very rapid results for truly emergent situations. However, KAP surveys are limited in flexibility, community involvement, and internal validity. An alternative assessment method, Rapid Assessment Procedures (RAP), successfully addresses some of KAP's limitations through an ethnographic, participatory problem-solving process [11]. Nevertheless, while these methods may provide limited quantitative information for use in health sector assessment, they can lack sufficient detail to identify major barriers to system improvements.
The use of more detailed assessment tools – employing a combination of qualitative and quantitative methodologies – can more accurately characterize the country's health care needs and the significant political and cultural barriers to sustainable health care reform. Our proposed Integrated Multimodal Assessment utilizes a diverse approach for strong internal validity and by-person factor analysis to uniquely identify critical subpopulations. By understanding the concerns of subpopulations, more targeted interventions can be developed to directly address these concerns. A trial study evaluating primary health care in Serbia has demonstrated the advantages of using a combination of qualitative and quantitative tools for health system assessment [12]. Each of the integrated methodologies contributes its own strengths to the overall assessment, complementing shortcomings of the other methodologies. For example, while the questionnaires provide concise, quantifiable, and less equivocal responses, the focus group discussions allow investigators to seek clarification and have the unique benefit of participant interaction often leading to entirely novel ideas.
Out of the near collapse of the Serbian health care system comes the opportunity to establish a health system more effective than ever in meeting the needs of its citizens. Our focus group discussions, questionnaires, individual surveys, and by-person factor analysis reveal that most emergency medical physicians believe the greatest problems in their system are poor organization of emergency medical services; lack of essential funding, medical supplies, medication, and technical equipment; and inadequate education and training. Several respondents also share the idea that poor incentives for specializing in emergency medicine, difficulties of emergency service work, and inadequate compensation are significant barriers to advancing the field of emergency medicine in Serbia. Although many of these are problems ubiquitous to developing health systems, this study illuminates specific opportunities for emergency medicine providers, the Serbian government, and international institutions to work together to address these issues.
According to the majority of emergency medicine providers and administrators surveyed, of foremost priority is reorganization of emergency medical services. In order to begin addressing this need, several participants propose the development of triage and treatment protocols. To facilitate improved routing of patients within the hospitals and to the appropriate emergency health care facility, all health care providers, including those within the primary health centers, should receive formal training in triage. Several participants suggest that leadership development and management training should be provided to medical directors and leading health care providers to improve the organization within and between emergency health care centers. Public education programs should also be developed to increase awareness of the level of care that each health care institution provides so that the public can more properly utilize available health care services.
Study participants further reported that the attainment of necessary medications, supplies, and equipment should be a top priority in the development of emergency services. The professionals working in the health care system of Serbia are severely limited by a shortage of these resources and, therefore, are unable to apply their knowledge and skills for proper diagnosis and treatment. Obsolete medical equipment further constrain their diagnostic capabilities, and lack of basic medications and supplies impedes their treatment strategies.
This multimodal assessment also shows that the education and training of physicians in emergency medicine need further development. Elements of the educational system needing improvement include medical school and postgraduate bedside training programs; training between health care institutions; access to continuing professional education materials including international journals, conferences, seminars, and internet access; and international collaboration between Serbia and foreign medical institutions, including opportunities for Serbian physicians to gain international clinical experience. Several participants also suggest that all health care workers, including physicians working in the primary health centers, receive periodic training in Basic Life Support (BLS), Advanced Life Support (ALS), and emergency triage.
Further development of emergency medicine in Serbia will be complex and will face numerous challenges. Reform will require funding, political commitment, supportive legislation, a revised medical curriculum, multi-phasic implementation, and post-interventional evaluation. Respondents believe that the funding needed for development of emergency services represents a significant barrier to change, but that, however, it is not the universal solution.
Although a majority of the emergency medicine personnel surveyed feel that the health care community will support further development of emergency services, just under half of the participants expect similar support from the public and government. Additional studies specifically assessing the needs and concerns of the public and government are recommended.
The use of multiple assessment modalities and the resulting complexity of this study involve several limitations that require consideration. All participants in this study were invited to participate by the director's office at each hospital or clinic. The study participants were aware that their participation was entirely voluntary and that they would not receive compensation. Although the study did not provide an economic incentive for participation, it did provide the opportunity for those providers with strong opinions to openly voice their concerns and suggestions. As a result, the study population may not represent all emergency health care providers. While every emergency medicine institution in Belgrade participated, this study was limited to the pre-hospital and hospital-based institutions within the capital city (covering 14.9% of the population of Serbia (excluding Kosovo)). Nevertheless, the participants represent the breadth of medical specialties in Serbia, and they constitute a large number of the stakeholders involved in further development of emergency medicine.
Conclusions
Despite the challenges, development of emergency services in Serbia can be accomplished through the dedication and commitment of providers. As a result of studies on emergency services in Kosovo, developmental programs similar to those suggested here have been successfully implemented to improve the training in and provision of Kosovo's emergency medical services [13]. Visiting experts currently travel to the province to provide didactic and bed-side training to physicians and other health care providers on the latest advances in emergency medicine, internal medicine, pediatrics, and several other fields of medicine. Similarly, a one-year fellowship program in emergency medicine and a leadership training program have been established at the University of Prishtina.
It is the hope of the authors that similar reforms will also be implemented to address the specific situation in the Republic of Serbia and that this multimodal assessment will assist by identifying the critical needs, barriers, and priorities for sustainable development. By utilizing both quantitative and qualitative methodologies, the authors submit that this Integrated Multimodal Assessment tool offers a more robust alternative to standard surveys, KAP studies, or the use of anecdotal information for quickly identifying priorities in health system reconstruction in the post-conflict, post-disaster, or development setting.
List of abbreviations
KAP: knowledge, attitudes, and practices
RAP: rapid assessment procedures
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
Each author has participated sufficiently in the work being reported to take public responsibility for the content. BDN, KD, and MVR conceived and designed the study and obtained research funding. MVR and MM supervised the completion of the study. BDN, KD, MS, MV, and LM undertook recruitment of participating centers and physicians. BDN, KD, and MS completed the data collection, statistical analysis, and first draft of the manuscript. All authors reviewed and contributed to the revision of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank for their generous assistance Ivana Mišić MD, the physicians and directors of the Emergency Center of the Clinical Center of Serbia, the Belgrade Institute for Emergency Medical Services, and the Belgrade hospital-based emergency departments.
This study has been supported in part by a grant from the Emergency Medicine Foundation and the Society for Academic Emergency Medicine.
==== Refs
Silber L Little A Yugoslavia: Death of a Nation 1997 New York: Penguin Books
Kazic S Health System in Yugoslavia Lancet 2001 357 1369 11347589 10.1016/S0140-6736(00)04497-4
Kazic S Health System in Yugoslavia Lancet 2001 357 1369 11347589 10.1016/S0140-6736(00)04497-4
Nelson BD Simić S Beste L Vuković D Bjegović V VanRooyen MJ A Multimodal Assessment of the Primary Health Care System of Serbia: A Model for Evaluating Post-Conflict Health Systems Prehospital Disaster Med 2003 18 6 13 14694894
Stephenson W Technique of factor analysis Nature 1935 136 297
Barbosa JC Willoughby P Rosenberg CA Mrtek RG Statistical Methodology: VII. Q-Methodology, a Structural Analytic Approach to Medical Subjectivity Acad Emerg Med 1998 5 1032 1040 9862598
Brown SR Q methodology and qualitative research Qualitative Health Research 1996 6 561 567
Schmolck P PQMethod – 2.10 (November 2002)
Dick WF Anglo-American versus Franco-German Emergency Medical Services System Prehospital Disaster Med 2003 18 29 37 14694898
Yolles TK Kelman HR Varma AO Physician knowledge and attitudes toward an emergency medical services system Ann Emerg Med 1981 10 2 10 7458025
Weiss W Bolton P Shankar A Rapid Assessment Procedures (RAP): Addressing the Perceived Needs of Refugees & Internally Displaced Persons Through Participatory Learning and Action Second
Nelson BD Simić S Beste L Vuković D Bjegović V VanRooyen MJ A Multimodal Assessment of the Primary Health Care System of Serbia: A Model for Evaluating Post-Conflict Health Systems Prehospital Disaster Med 2003 18 6 13 14694894
Eliades MJ Lis J Barbosa J VanRooyen M Post-war Kosovo: Part 2, Assessment of emergency medicine leadership development strategy using multimodal assessments and Q-Analysis Prehospital Disaster Med 2001 16 268 274 12090209
| 15715917 | PMC552312 | CC BY | 2021-01-04 16:31:52 | no | BMC Health Serv Res. 2005 Feb 17; 5:14 | utf-8 | BMC Health Serv Res | 2,005 | 10.1186/1472-6963-5-14 | oa_comm |
==== Front
BMC BiolBMC Biology1741-7007BioMed Central London 1741-7007-3-41571004210.1186/1741-7007-3-4Research ArticleSequential super-stereotypy of an instinctive fixed action pattern in hyper-dopaminergic mutant mice: a model of obsessive compulsive disorder and Tourette's Berridge Kent C [email protected] J Wayne [email protected] Kimberly R [email protected] Xiaoxi [email protected] Department of Psychology, University of Michigan, Ann Arbor, USA2 Department of Neurology, University of Michigan, Ann Arbor, USA3 Wayne State University Medical School, Detroit, USA4 Department of Neurobiology, Pharmacology, and Physiology, University of Chicago, Chicago, USA2005 14 2 2005 3 4 4 11 10 2004 14 2 2005 Copyright © 2005 Berridge et al; licensee BioMed Central Ltd.2005Berridge 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
Excessive sequential stereotypy of behavioral patterns (sequential super-stereotypy) in Tourette's syndrome and obsessive compulsive disorder (OCD) is thought to involve dysfunction in nigrostriatal dopamine systems. In sequential super-stereotypy, patients become trapped in overly rigid sequential patterns of action, language, or thought. Some instinctive behavioral patterns of animals, such as the syntactic grooming chain pattern of rodents, have sufficiently complex and stereotyped serial structure to detect potential production of overly-rigid sequential patterns. A syntactic grooming chain is a fixed action pattern that serially links up to 25 grooming movements into 4 predictable phases that follow 1 syntactic rule. New mutant mouse models allow gene-based manipulation of brain function relevant to sequential patterns, but no current animal model of spontaneous OCD-like behaviors has so far been reported to exhibit sequential super-stereotypy in the sense of a whole complex serial pattern that becomes stronger and excessively rigid. Here we used a hyper-dopaminergic mutant mouse to examine whether an OCD-like behavioral sequence in animals shows sequential super-stereotypy. Knockdown mutation of the dopamine transporter gene (DAT) causes extracellular dopamine levels in the neostriatum of these adult mutant mice to rise to 170% of wild-type control levels.
Results
We found that the serial pattern of this instinctive behavioral sequence becomes strengthened as an entire entity in hyper-dopaminergic mutants, and more resistant to interruption. Hyper-dopaminergic mutant mice have stronger and more rigid syntactic grooming chain patterns than wild-type control mice. Mutants showed sequential super-stereotypy in the sense of having more stereotyped and predictable syntactic grooming sequences, and were also more likely to resist disruption of the pattern en route, by returning after a disruption to complete the pattern from the appropriate point in the sequence. By contrast, wild-type mice exhibited weaker forms of the fixed action pattern, and often failed to complete the full sequence.
Conclusions
Sequential super-stereotypy occurs in the complex fixed action patterns of hyper-dopaminergic mutant mice. Elucidation of the basis for sequential super-stereotypy of instinctive behavior in DAT knockdown mutant mice may offer insights into neural mechanisms of overly-rigid sequences of action or thought in human patients with disorders such as Tourette's or OCD.
==== Body
Background
Overly rigid sequential patterns of movement and thought characterize several human brain disorders involving dysfunction in basal ganglia systems (i.e. dopamine nigrostriatal projections to the neostriatum and related brain structures). For example, pathological repetitions of spoken words in Tourette's syndrome, and the tormenting habits and thoughts of obsessive-compulsive disorder (OCD), involve overly rigid sequential patterns of action, language or thought [1-9], which in part may be influenced by genetic factors [10-13].
Normal sequential patterns of action, language and thought also have been suggested to depend on proper basal ganglia function [14,15]. For example, Marsden proposed that "The sequencing of motor action and the sequencing of thought could be a uniform function carried out by the basal ganglia" [15], and a variety of computational models have been proposed to carry out the general sequencing functions of basal ganglia [16-19]. According to this view, basal ganglia systems evolved originally to coordinate syntactic patterns of instinctive movements, and were extended subsequently by natural selection to participate in sequencing cognitive and linguistic functions as well.
Almost all behavior is sequential, so what do we mean by 'syntactic sequence'? In the simplest terms, a syntactic sequence is one that follows normative rules that determine the temporal progression of its elements and impart a lawful predictability to the sequence as a whole [14,20,21]. Human language has real syntax, as the prototypical example, complete with recursive generative rules [14,21,22]. But other behavior can be described as having properties of syntax too, if the behavioral flow is governed by lawful sequential patterns that follow normative rules to produce a complex serial order [14,20,23-26].
Neuroethological studies of natural behavior in animals have shown that neostriatum, substantia nigra, and their connecting dopamine projections are critical to sequential stereotypy for complex serial patterns of instinctive behavior [26-35]. In particular, a complex fixed action pattern displayed spontaneously by rodents during grooming behavior, called a syntactic grooming chain, has been exploited by neuroethological studies that point to basal ganglia systems as the controlling neural mechanisms for the stereotypy of complex sequential patterns [27,28,36]. A syntactic chain is a 4-phase series of up to 25 elements, each phase containing recursive iterations of its characteristic element (Figure 1; see Additional movie file 1). This syntactic sequence occurs spontaneously during grooming behavior of most rodents. Mice, rats, gerbils, hamsters, guinea pigs, ground squirrels and other species all have their own signature patterns of syntactic chains, with different details, but all follow the syntactic 4-phase rule [37]. In one squirrel species (Spermophilus beecheyi), syntactic chains have been even further ritualized into a stereotyped display, and adapted for territorial communicative use [38]. As is typical of fixed action patterns, no two syntactic chains may be absolutely identical, but they are highly similar, stereotyped, and easily recognized, and always follow the same serial patterning rule [39,40]. Thus syntactic grooming chains are complex multi-component patterns that are sequentially stereotyped, and capable of interacting with evolutionary selection pressures that alter the genotype to modulate behavioral patterns. They represent precisely the sort of sequencing function that ancestral basal ganglia systems might originally have evolved to perform [2,9,14,15,24,29-31].
Figure 1 Prototypical syntactic grooming chain pattern. Choreograph shows mouse movements of the left/right paws over the face (time proceeds from left to right). Lines deviating above/below the horizontal axis show the trajectory height of left/right paws. Large black box denotes bout of body licking, and placement of asterisk in box shows which left/right side flank was chosen by the mouse to initiate body licking. Phase I: series of ellipse-shaped strokes tightly around the nose. Left and right paws often take alternating turns as the major/minor trajectory. Phase II: series of unilateral strokes, each made by one paw, that reach up the mystacial vibrissae to below the eye. Mice often make hybrid Phase I/II strokes, in that one paw makes a Phase II unilateral stroke while the remaining paw makes a smaller Phase I type ellipse. Phase III: series of bilateral strokes made by both paws simultaneously. Paws reach back and upwards, ascending usually high enough to pass over the ears, before descending together over the front of the face. Phase IV (strong or classic form): sustained bout of body licking, preceded by postural cephalocaudal transition to move mouth and tongue from facial and paw grooming to body grooming. Mouse-typical pattern modified from Berridge (1990). See Additional movie file 1 for examples of syntactic grooming chains by DAT-KD mutant mice.
The firing of some basal ganglia neurons in neostriatum and in substantia nigra codes the serial pattern of syntactic grooming chains as an entire sequence in rats [27,28]. In addition, the integrity of basal ganglia neurons is necessary for normal sequential stereotypy of the instinctive pattern. For example, after lesions of neostriatum, rats lose the ability to complete the 4-phase pattern properly (especially after lesions of anterior dorsolateral neostriatum, which contains the neurons that particularly code the syntactic pattern), even though the lesions do not impair constituent grooming movements [27,36]. Similar deficits in grooming syntax are caused by disruption of dopamine neurotransmission in mice lacking dopamine D1 receptors [41], and in normal rats with neostriatal dopamine depletion caused by 6-hydroxydopamine lesions of nigrostriatal projections [42].
Brain lesions that disrupt behavioral sequences indicate a potential sequencing function for the targeted structures. However, factors besides sequencing loss may contribute to disrupted serial patterns after lesions. An alternative and stronger proof for dopamine mediation of action syntax would be to demonstrate enhanced stereotypy of behavioral sequences, by boosting nigrostriatal dopamine neurotransmission. Enhanced sequential stereotypy would be reflected if the complex serial pattern as a whole entity became more sequentially rigid or persistent. Indeed, in rats, pharmacological boosting by dopamine D1 agonists administered systemically or into brain ventricles produces sequential super-stereotypy of syntactic grooming chains [43-45]. In a state of sequential super-stereotypy, the stereotyped pattern becomes even more predictable than normal, which is evident as an increase in the probability that all four phases will be completed in syntactic order [43,44]. Such rigidity of complex multiple-phase sequences contrasts with simpler repetition stereotypies (e.g., associated with D2 receptor activation), in which the same movement is repeated over and over again [46-50].
In human pathologies such as Tourette's or OCD, complex sequential super-stereotypy often occurs spontaneously in human patients. If sequential super-stereotypy of complex instinctive behavior sequences is to serve as a model of human disorders involving sequential super-stereotypy, it ought to be able to occur spontaneously in some individual animals too. In addition, it should possess features of compulsive behavioral sequences.
Compulsive behavior may have several features, including both perseverative tendencies and more rigid sequences of entire serial patterns. To date, prior genetically-modified mouse models of spontaneous compulsive behavior have successfully captured the perseverative feature, but it is not yet clear whether these animal models also share the exaggerated serial pattern feature of compulsive behavior. For example, the Hoxb8lox mutant model has been reported to exhibit OCD-like increased persistence of self-directed grooming and body-licking, and even mutual grooming of other mice [13,51,52]. Similarly, the D1CT mutant mouse, caused by transgenic potentiation of D1-associated brain circuits, shows OCD-like persistence of grooming, as well as persistence of other behaviors such as digging, climbing, and tics [3,53-57]. However, it is unknown whether these or any other animal models also show excessively rigid sequences, in the sense of a stronger multi-element and rule-governed sequence that becomes more rigid as a single complex pattern. For modeling the serial rigidity feature of OCD or Tourette's, an animal model is needed that spontaneously produces an overly-rigid and serially-complex sequence of behavior, such as a syntactic grooming chain.
Here we show that this serial pattern feature of sequential super-stereotypy indeed appears spontaneously without drugs in DAT-KD mutant mice with genetic knockdown of the dopamine transporter (DAT) [58]. DAT-KD mutant mice have 10% normal DAT expression in dopamine neurons [58], which impairs synaptic re-uptake of dopamine, resulting in elevated (170%) levels of extracellular dopamine in neostriatum (wild-type mice = 100%) [58]. DAT-KD mutant mice show other behavioral evidence for high levels of dopamine activation. They tend to be hyperactive, to walk in perseverative straight paths, and to over-pursue certain incentive stimuli [58-60]. The question asked in the present study was whether these mutant mice would also show sequential super-stereotypy in their syntactic chains – that is, do they have excessively rigid serial patterns of instinctive grooming behavior?
Results
Syntactic chains
Hyper-dopaminergic mutant mice and wild-type control mice each generated syntactic chains of grooming as described above (Figures 1, 2 &3). Syntactic grooming chains by DAT-KD mice had virtually all the typical features of wild-type chains and of syntactic chains previously reported for outbred mice and D1 receptor knockout mice [37,41] (Figures 1 &3; see Additional movie file 1).
Figure 2 Sequential super-stereotypy of syntactic pattern. Cumulative rates of full pattern completion by DAT-KD mutant (dark symbols) and wild-type mice (open symbols) for each type of syntactic chain (Perfect, Insertion of unpredicted component, Phase Reversal, Phase Skip, Substitution of paw lick for Terminal Phase IV component). Choreographs at bottom show example for each type of syntactic chain. Mutant mice have higher rates of syntactic completion for all forms of the chain that terminate in the strong form of Phase IV, body licking, which characterizes the prototypical Phase IV for all rodents. Wild-type mice use a weak form of Phase IV (paw lick substitution) to terminate a substantial proportion of their syntactic chains. All mice show less pattern completion when grooming in the laboratory (top) than when grooming in their home cage (bottom), but mutant mice show more rigid sequential patterns than wild-type mice while grooming in both environments. * p < 0.05; ** p < 0.01.
Figure 3 Sample choreographs of actual syntactic chains. Both mutant mice and wild-type mice emit every type of syntactic chain described in the text (Perfect, Insertion of unpredicted component, Phase Reversal, Phase Skip, and Substitution of paw lick for Terminal Phase IV).
Syntactic grooming chains
The sequential pattern of a syntactic grooming chain contains up to 25 movements serially combined into 4 syntactic or rule-governed phases that form one chain pattern [61] (Figure 1; see Additional movie file 1). Each of the 4 phases contains recursive repetitions of its particular component movement.
Phase I consists of 5–10 rapid elliptical forepaw strokes made with both paws simultaneously over the nose and mystacial vibrissae. In mice, Phase I ellipses are often slightly asymmetrical and alternating, in the sense that the 'major paw' makes a slightly larger stroke than the 'minor paw' [37]. Typically, the major/minor role alternates over successive Phase I strokes between left and right paws. The entire Phase I lasts for about one second.
Phase II is short (0.25 s) and consists of 1–4 unilateral or highly asymmetrical strokes made by one forepaw. The unilateral stroke is typically of small or medium amplitude ascending to about the level of the eye. In mice, the other paw not participating in the Phase II stroke often makes a smaller Phase-I ellipse-type stroke simultaneously [37,41]. Thus, Phase II in mice typically contains several hybrid Phase I-II strokes, in contrast to rats, which move only a single forepaw [37]. Mice generally alternate between left and right paws in making Phase II strokes (though sometimes the same paw repeats a short series of Phase II strokes).
Phase III is highly visually distinctive, and consists of 1–5 large bilateral strokes with both paws. Both paws move very symmetrically almost as mirror images of the other, typically ascending together high up the side of the face, and passing forward synchronously over the ears. Phase III strokes are extremely stereotyped, usually all of the same height, and with both paws traveling back down to the nose between successive Phase III strokes [37]. The entire Phase III lasts 1–3 s.
Phase IV concludes the prototypical chain, and consists of a postural turn to the side and caudally, and lowering of the head to bring the tongue towards the flank or side of the body, followed immediately by a 2–5 s bout of body licking directed to the flank.
Syntactic Initiation: rate of starting chains
In terms of the number of syntactic chains started during a grooming bout, DAT-KD mutant mice initiated marginally more syntactic chains overall than wild-type mice (F(1,86) = 3.592, p = 0.061; Figure 4). The difference in chain initiation was context dependent. All mice were twice as likely to initiate syntactic chains in the laboratory than at home (F(1, 82) = 85.73, p < 0.001), and mutant mice in particular initiated approximately 25% more chains than wild-type mice in the laboratory environment (F(1,86) = 17.315, p < 0.001; Figure 4), compared to only 5% more in the home environment. If the laboratory context was considered more stressful than the home cage environment, then stress dramatically promoted the tendency to begin a highly stereotyped sequence, especially for mutants.
Figure 4 Initiation of stereotyped syntactic chain pattern. Rates of initiation of syntactic chains are slightly higher for mutant mice, especially while grooming in the laboratory, measured cumulatively across the entire observation period (top). In more detail, initiation rates are broken down as occurring either early versus late in grooming bouts (bottom). All mice tend to start the stereotyped sequential pattern more often early in a grooming bout. Mutant mice are even more likely than wild-type mice to start the pattern in an early grooming bout, both in home and laboratory environments. ** p < 0.01.
The nature of the context-dependence of the difference was further clarified by a closer look at the time course of exactly when syntactic chains were begun by mice during a grooming bout. The overwhelming majority of syntactic chains tended to be initiated early in a grooming bout by all mice (Figure 4). Mutants initiated up to twice as many chains as wild-type mice per minute of grooming in the first quarter of a bout, whereas by the last quarter of their grooming bouts mutant and wildtype initiation rates no longer differed significantly (Figure 4). Mutants tended to begin more grooming bouts than wild-type mice especially in the laboratory (described below), which may have facilitated the mutants' greater tendency to initiate the syntactic chain pattern in the laboratory (Figure 4). In short, syntactic chains were initiated early in a grooming bout by all mice, but mutant mice were even more likely than wild-type mice to initiate chains during those early portions of a grooming bout, and the mutant advantage was greatest in the laboratory environment (which might be the most stressful environment).
Syntactic completion
In mice and rats, once a syntactic chain pattern begins with Phase I, each remaining action can be predicted with roughly 80–90% accuracy. The entire syntactic chain occurs with a frequency over ten thousand times greater than could be expected by chance (based upon the relative probabilities of the component actions). However, several types or degrees of chain completion are possible.
Types of syntactic chains
A prototypical or perfect syntactic chain requires Phases I, II, III and IV in order, with no deviations, additions or omissions. Perfect chains were occasionally performed by both mutant and wildtype mice. After performing Phase I, II and III strokes over the face, a mouse performs Phase IV by transitioning to body grooming. For this transition to body licking, the mouse must bend down and backward to bring its mouth toward a side flank, and then begin a vigorous bout of body licking that continues for 1–4 s.
In addition, several types of imperfect syntactic chains were observed in both mutant and wildtype mice. Imperfect sequences proceed from Phase I to IV with some minor deviation from the prototypical pattern along the way. In this study, we recognized three forms of imperfect completion. All involved a minor imperfection, which was either an insertion, reversal or replacement of a component action within the syntactic chain. Imperfect completion forms were: (i) Reversal of Phases II-III, where Phase II unilateral strokes were emitted after Phase III bilateral strokes (instead of before them), but the chain was otherwise syntactically correct; (ii) Insertion of an unexpected movement component in between phases, usually a quick paw lick or several paw licks inserted between Phases III and IV; (iii) Skip or omission of one phase en route to completion, where a chain lacked either any Phase II unilateral face stroke, or any Phase III face strokes (never both), but was otherwise syntactically correct (e.g. an observed order of I-III-IV). Finally, a fourth type of syntactic chain was observed that failed to be completed in the prototypical sense, but where the mice substituted paw licking in the terminal place of Phase IV (which might provisionally be regarded as an "attempt" to complete syntactically). We called this Terminal substitution: the final Phase IV component (body licking) was completely replaced with a different type of licking movement (paw licking), and the chain was otherwise syntactically correct (e.g. I-II-III-paw lick bout).
Terminal substitution never attains a prototypical Phase IV, and so is not really a form of syntactic completion by criteria used in earlier studies. However, the terminal substitution of paw licking might be viewed as an attempt to complete syntactically with a transition from paw strokes to licking, compared to other forms of incompletion such as either simply stopping or immediately launching into a sequentially flexible series of grooming strokes. Thus for the purpose of analysis, we examined the consequences of allowing terminal substitution to count provisionally as a form of "weak" completion.
At the completion of Phase IV (strong or weak), over 93% of syntactic chains led to continued grooming of body or face in sequentially flexible and much less predictable patterns compared to syntactic chains. After 7% of chains, the end of Phase IV terminated the entire grooming bout, and the mouse rested quietly afterwards or began to explore the chamber.
Syntactic rigidity: strength of pattern completion
DAT-KD mutant mice not only started more syntactic chain patterns, they were also more likely than wild-type mice to complete the syntactic chain patterns they started – in both laboratory and home environments (Figures 2 &3). Sequential super-stereotypy (i.e. more predictable and stereotyped completion of entire sequence) of DAT-KD mutant mice was the most consistent and robust finding of our study (F(1,78) = 12.33, p < 0.001; Figure 2).
The higher syntactic rigidity of mutants was visible qualitatively and verified quantitatively (Figures 2 &3), and it interacted with the various types of syntactic completion described above (interaction between mutant/wildtype × perfect/imperfect types: F(4,184) = 5.96, p < 0.001). Hyper-dopaminergic mutant mice nearly always completed their syntactic chains with the strong form of Phase IV (body licking), whereas wild-type mice completed roughly half their chains with only the weaker form of Phase IV (paw licking). Mutant mice completed a higher percentage of insertion, reversal and omission types of syntactic chains than wild-type mice (F(4, 184) = 129.01, p < 0.001; each subtype; Figures 2 &3). These stronger or more rigid chains of mutant mice more closely corresponded to the prototypical 4-phase syntax pattern (including the prototypical terminal Phase IV component: body licking).
In other words, mutant mice were better than wild-type mice at resisting disruption of the pattern by minor flaws that occurred along the way, and mutants more often returned to the full-blown pattern after any distraction. For example, insertion chains included 1 or 2 extraneous movements, such as a nonsyntactic paw lick action inserted between Phases III and IV. After an insertion, mutants were nearly 50% more likely than wild-type mice to reach a strong form of Phase IV completion (mutant vs. wildtype, p < 0.01, Bonferroni). Similarly, a reversal error reversed the serial order of Phases II and III, or followed a Phase II stroke with a late Phase I ellipse stroke, and after a reversal mutant mice were nearly 50% more likely than wild-type mice to go on to complete a strong form of Phase IV. Finally, in an omission chain, a mouse would omit either Phase II or Phase III (never both), and after an omission mutants were again roughly 50% more likely than wild-type mice to successfully return to the full pattern and reach a strong form of terminal Phase IV completion (each p < 0.01, Bonferroni).
In contrast, wild-type mice had a greater proportion of terminal substitution chains that never achieved a full syntactic transition to body grooming. Wild-type mice instead substituted a weaker paw-lick form of Phase IV as terminal component. In terminal substitution, a mouse completely omitted the normal Phase IV shift to body licking, and instead simply continued to lick its paws, never changing posture or moving its head caudally out of the normal facial grooming position (the complete failure of transition to body licking after paw licking marked the difference between Insertion and Terminal Substitution chains). Wild-type mice had nearly twice the proportion of terminal substitutions as mutant mice (F(1,78)= 11.47, p < 0.001).
If terminal substitution is regarded as failure to complete the pattern, then wild-type mice simply failed to complete over half the syntactic chains they began. More leniently, wild-type mice could approach an 80% – 90% rate of syntactic completion – if we took the unprecedented step of allowing Phase IV terminal substitution to count as weak completion (Figure 2). Allowing this weaker criterion was the only way to consider wild-type mice able to achieve the 80%–90% syntactic completion level that mutant mice successfully achieved through the stronger prototypical form of Phase IV.
In summary, DAT-KD mutant mice had more rigid sequential patterns than wild-type controls in several ways. Mutant mice were more likely than wild-type mice to proceed syntactically through Phases I, II and/or III to reach the syntactic final Phase IV (body licking). Even after encountering minor imperfections along the way, mutants persevered in the sequential pattern. Wild-type mice introduced the same imperfections in their syntactic pattern, but did not return to the full pattern or complete Phase IV as strongly, ending their chains without ever reaching the full-blown transition to body grooming that normally terminates a syntactic chain pattern.
Finally, syntactic completion was highest in home environment grooming for all mice (even though more syntactic chains were begun in laboratory) (F(1,78)14.31, p < 0.001). This difference suggests that stress may promote the initiation of stereotyped sequences, but impede their lawful completion, and is consistent with reports that stress disrupts completion of syntactic chain sequences [62]. However, mutant mice were equally more likely than wild-type mice to complete strong patterns in both laboratory and home environments.
Motor control for movement capacity
In order to reject motor confounds that might have provided an alternative explanation of some results, we assessed whether wild-type mice were simply less able to perform body licking movements than mutant mice. If wild-type mice had motor deficits that impaired their ability to perform body-lick posture/movements, then wild-types might have had weakened syntactic chains simply because of their motor incapacity to perform Phase IV movements, rather than because mutants had stronger sequencing tendencies. Therefore we analyzed whether wild-type mice spent a lower proportion of their total grooming behavior time making body licking movements compared to mutant mice. However, wild-type mice did not have significantly lower total cumulative duration scores for body licking overall than mutant mice (F(1,78) = 0.56, n.s.), indicating there was no motor impairment of Phase IV movements. That suggests the difference in tendency to complete syntactic chains represents a true difference in sequence rigidity or pattern strength, and not in simple motor capacity.
Overall grooming behavior: amount, bout number, and bout duration
All mice groomed twice as much in their home cages than in the laboratory environment, suggesting that the relatively novel laboratory environment might have acted to suppress spontaneous grooming behavior (F(1,82) = 1.773, p < 0.001; Figure 5). Grooming behavior in the laboratory was less than half that of the home cage for both mutants and wildtypes (in terms of cumulative grooming duration per hour of observation). DAT-KD mutant mice spent 10%–50% more time than wild-type control mice in grooming behavior overall (F(1,86) = 3.949, p < 0.05), and the mutant propensity to groom more was most visible in the home environment (p < 0.1).
Figure 5 General amount and bout features of grooming behavior. Cumulative time spent in grooming behavior during observation (total duration), Duration of individual bouts of grooming, and the Number of bouts of grooming emitted during observation session. Mutant mice tend to spend more time in grooming than wild-type mice, and to have longer grooming bouts, in the home environment. Mutant mice tend to emit a greater number of fragmented bouts when grooming in the laboratory environment. These general features of grooming enhancement in mutant mice are flexible and context-dependent, in contrast to the greater mutant rigidity of sequential pattern that is constant across both environments (shown in Figure 2). * p < 0.05; ** p < 0.01.
Closer analysis of grooming compared the relative contributions to increased grooming time of greater bout numbers versus longer bout durations (Figure 2). The increased time spent grooming by mutant mice in the home cage was due to longer grooming bouts (but not to a greater number of bouts) compared to wild-type mice. In their home cages, grooming bouts in mutant mice were 80% longer than in wild-type mice (F(1,86) = 4.083, p = 0.008), while bout numbers did not differ.
Although mutants emitted only marginally more body licking than wild-type mice in an analysis that combined data from both home and laboratory environments (F(1,60) = 3.403, p < 0.07), a separate analysis of grooming specifically in the home cage showed that mutants at home had longer cumulative durations of body licking (p < 0.05, Bonferroni), consistent with prolongation of the later components of cephalocaudal grooming bouts in that home environment [63]. However, as a percentage of total grooming, the proportion of mutant body licking to facial stroke components was not higher than for wild-type mice, either overall (F(1,60) = 0.58, n.s.), or even in the home cage (p = 0.32), which suggests that the mutants' longer grooming bouts in the home cage may also have included more facial strokes than wild-type mice. Thus, longer mutant grooming bouts in the home cage likely involved expansion of several components of grooming, including longer body licking bouts and facial strokes. These perseverative features of DAT-KD mutant grooming in the home cage therefore may overlap with perseverative body grooming tendencies reported for other genetic animal models of compulsive behavior, such as Hoxb8lox and D1CT mutant mice [13,51-53,55,57].
Conversely, in the laboratory environment, DAT-KD mutants' higher grooming was chiefly due to a greater number of grooming bouts (but not longer bouts). In the laboratory environment, mutants began more grooming bouts than wild-type mice (F(1,86) = 3.478, p = 0.026), but their duration of bouts did not differ. Thus, different features of grooming bouts (length versus number) were enhanced in mutant mice depending on their environmental context of the moment. However, as described above, in both home and laboratory the hyper-dopaminergic mutants were always more likely than wild-type mice to perform more rigid and strongly stereotyped syntactic chain sequences.
Discussion
Sequential super-stereotypy: pattern completion
Our results reveal that hyper-dopaminergic mutant mice show excessively strong and rigid manifestations of a complex fixed action pattern compared to wild-type mice. Their sequential super-stereotypy was produced by DAT knockdown mutation, which reduces DAT to 10% of wild-type levels and causes extracellular dopamine elevation to 170% in neostriatum [58]. Mutant mice showed more stereotyped and predictable syntactic grooming chains, the instinctive fixed action pattern that serially links up to 25 movements into 4 predictable phases that follow 1 syntactic rule. That entire pattern became even more stereotyped and resistant to disruption in hyper-dopaminergic mutant mice.
The stronger pattern was evident in several ways. First, DAT-KD mutant mice were more likely to begin a syntactic chain pattern than wild-type mice, especially during the early minutes of a grooming bout (when the highly stereotyped serial pattern is most likely to be produced), and especially in the novel laboratory environment (a potential stressor). Further, once the complex sequence began, DAT-KD mutant mice went on to execute chains that were more stereotyped and rigid, both qualitatively and quantitatively. Qualitatively, mutant mice almost always achieved the strongest form of the terminal phase (Phase IV), successfully making a transition from head grooming to body grooming. By comparison, wild-type mice ended far more of their chains with a weaker terminal substitution for Phase IV, which left them stuck in head grooming without ever making a transition to body grooming. Quantitatively, DAT-KD mutant mice returned more often to the prototypical pattern after minor mistakes, whereas wild-type mice failed to reach full Phase IV after such mistakes. Mutant mice returned more often to the full pattern after extraneous component insertion, phase omission, or serial reversal of phases. The mutants' elevated pattern strength for this complex sequence was evident in both home and laboratory environmental contexts.
If the less-stereotyped sequential patterns of wild-type mice are viewed as the norm (and not as a sequential deficit), then the mutant tendency to complete stronger syntactic patterns must be viewed as sequential super-stereotypy, representing the exaggerated serial rigidity feature of compulsive behavior. Here sequential super-stereotypy is manifest in a complex behavioral sequence that is instinctive and naturally stereotyped to begin with, but becomes even more stereotyped or excessively rigid as a consequence of the DAT mutation.
It may be important that the mutant pattern strength is revealed not as an elimination of errors, but rather primarily as a resistance to disruption by errors. In other words, mutants did not have more frequent perfect chains than wild-types: both generated similar moderate rates of minor errors (e.g., inserting extra actions, omitting one syntactic phase from where it ought to be, or reversing the order of 2 phases in the 4-phase pattern). Instead, the mutants' stronger syntactic pattern was like a tightened elastic band, pulling them back after such errors to finish the prototypical pattern. Stronger return to the pattern could only be possible if DAT knockdown strengthened the entire pattern as a global whole, facilitating the mutants' ability to maintain a neural representation of the pattern during an error and to resume the remaining pattern after the error. That suggests that neural mechanisms of pattern coordination were better able to persist in rule maintenance in the face of disruption, and to successfully compete to re-establish control of the behavioral stream after the disruption.
Thus, stronger patterns were not simply the results of strengthened Markov sequential transitions among individual pairs of actions, producing a stimulus-response (S-R) reflex chain. If sequence composition was simply a probabilistic construction based only on the frequency of transitions between individual pairs of actions, then stronger perfect completion might have been expected in mutants, but not stronger return after an error. Errors would still terminate or weaken the pattern. Instead, we observed the opposite result: mutants kept errors but recovered better after them, and took the full pattern up again where it had left off.
Relation to other nigrostriatal manipulations and behaviors
These results are the first demonstration to our knowledge of sequential super-stereotypy of a complex behavioral pattern, occurring spontaneously without drugs. In previous studies, dopamine D1 agonists were needed to cause sequential super-stereotypy of syntactic grooming chains, whereas D2 agonists in contrast reduced initiation and completion of syntactic grooming chains (even though D2 agonists can cause simple repetitive movement stereotypies) [43-45,64]. Future studies will be needed to confirm whether the sequential super-stereotypy of DAT knockdown mutant mice depends specifically on increased D1 receptor activation. However, it is notable that there is a consistent trend of D1 circuit-activation inducing OCD-like behavioral persistence in D1 agonist-treated rodents, D1-circuit potentiated D1CT mice, and hyperdopaminergic DAT-KD mice [43-45,53,55,57,64]. This suggests that the D1 circuit may play an important role in features of compulsive behavior related to perseveration and sequential rigidity. It also would be of interest for future studies to examine if other animal models of perseverant grooming behavior, such as Hoxb8lox and D1CT mutant mice, also show any exaggerated serial rigidity features in their fixed action patterns similar to those found here [13,51,53-55,57,65]. Finally, it would clearly be of interest to examine whether any other instinctive fixed action patterns belonging to those of DAT-KD mutant mice show sequential super-stereotypy similar to syntactic grooming chains.
We should note that although our study is the first to produce spontaneous sequential super-stereotypy, several previous studies reported weakening of the syntactic chain pattern by other genetic manipulations. For example, the ability to complete syntactic grooming chains is impaired in several types of mutant mouse, caused by either a knockout of D1 dopamine receptors [41], or by a Weaver gene mutation that alters the nigrostriatal dopamine system [65,66]. In the D1 knockout study, mutant D1 mice were less able than wild-type mice to complete the full grooming pattern of syntactic chains they started [41]. Our DAT-KD findings provide an opposite demonstration to complement that D1 knockout study: DAT knockdown strengthens the same pattern presumably by elevating extra-cellular dopamine. Both results might therefore reflect essentially linear effects on the sequential stereotypy of this complex behavior pattern, mirroring up or down changes in basal ganglia dopamine neurotransmission.
Evolution co-opts sequential super-stereotypy
We acknowledge that there is one other known form of genetically-related sequential super-stereotypy for syntactic grooming chains. However, that sequential super-stereotypy is not caused by a single targeted gene mutation but rather is a naturally evolved adaptation of the fixed action pattern in a species of ground squirrel, Spermophilus beecheyi [38], which is probably polygenic in origin. California ground squirrels defend their individual mating territories in the Sierra mountains against other same-sex ground squirrels (especially males against other males). One of their behavioral territory displays is a specialized exapted form of the syntactic grooming chain [38].
Display forms of Spermophilus beecheyi syntactic chains are ritualized, more sequentially rigid and predictable than normal self-grooming chains, and occur as a single grooming chain with no other grooming before or after [38]. Phase III elements are amplified and made more visually distinctive, and an extra Phase V component is appended to the end of the pattern (the squirrel seizes and licks its tail, which is also visually distinctive). Syntactic grooming chains are usually performed at the boundary where two adjacent territories meet. Syntactic grooming chain displays appear to be communicative, in that they are emitted in conjunction with other territorial displays, such as scent-marking of objects, and have the social consequence of subsequently reducing the likelihood of a physical fight between the two adversaries [38].
Thus the evolution of Spermophilus beecheyi ground squirrels appears to have exapted the pre-existing pattern of a syntactic grooming chain, which likely evolved in ancestral rodents over 60 million years ago, and co-opted it into a sequentially super-stereotyped form for specific communicative use [38]. It may have been selected because of the same feature that led us to study syntactic chains, namely its recognizable sequential stereotypy. Also, the sequential pattern appears highly sensitive to the underlying genotype; for example, the detailed 'signature patterns' of the syntactic grooming chains that distinguish mice from squirrels, rats, guinea pigs and other rodents can be used to construct taxonomic trees of relatedness for them (similar to taxonomies based on differences in skull structure or in DNA sequences) [37]. The genetic sensitivity of the pattern may explain why evolutionary selection exploited it for use by California ground squirrels, and also explain why knockdown of a single gene can change the strength of the entire complex sequential pattern in studies such as ours [41,65,66].
Neural systems and clinical implications of sequential super-stereotypy
Altered neurochemical signaling within basal ganglia neural circuits may be the mechanism by which DAT knockdown produces sequential super-stereotypy of grooming syntax. Electrophysiological studies have shown that neurons in neostriatum and in substantia nigra pars reticulata code the sequential pattern of syntactic grooming chains and other natural sequences of behavior [24,27,28]. For example, 40% of neostriatal neurons in rats code sequential aspects of the syntactic chain pattern, especially in anterior dorsolateral neostriatum [24,27]. Neurochemical boosting of dopamine signalling from substantia nigra pars compacta on to neostriatal neurons might be one candidate mechanism to modulate sequential super-stereotypy of the pattern in DAT-KD mutants. Similarly, neurons in the substantia nigra pars reticulata appear especially to code initiations of the complex behavioral sequence, and so modulated input to them might be more relevant to the elevated mutant tendency to begin the syntactic pattern [24,28].
Nigrostriatal mechanisms for sequencing instinctive action may have been co-opted in subsequent mammalian and human evolution into use in sequencing learned and cognitive psychological elements [67-69]. In that way, the same basal ganglia mechanisms used for movement syntax may participate in sequential habits that result from learning [20,29,70-72]. A view of basal ganglia as a general purpose sequencing mechanism is compatible also with computational sequencing models of basal ganglia [16-19]. Beyond the basal ganglia, DAT-KD mutant mice might also have elevated extra-cellular dopamine concentrations in other target structures, including prefrontal cortex and amygdala. Such systems might also contribute to OCD and Tourette's syndromes in humans and to some aspects of compulsive-like behavior in mutant mice. Elaborated applications of dopamine-related circuits for sequencing may thus extend from instinctive animal behavior to abstract human cognition and behavior, including syntactic sequencing of action plans, linguistic syntax, and the serial order of streams of thought [14,73].
A clinical implication of the embeddedness of basal ganglia in sequencing function may be a vulnerability to sequential dysfunction in some human disorders involving nigrostriatal systems [74,75]. Both Tourette's syndrome and obsessive-compulsive disorder show symptoms of sequential super-stereotypy, in the form of overly rigid patterns of action, language or thought [76,77]. Basal ganglia are believed to be involved in generating such pathologically-strong and complex sequential stereotypies [1,2,8,9,74,78-85]. Hyper-dopaminergic function in nigrostriatal and related neural systems might thus play a role in causing the excessive rigidity of behavioral tics, repetitive language utterances, and obsessive chains of thought [2,74,79,81,86,87].
Finally, while highly speculative, it is at least conceivable that an evolutionary specialization of dopamine-related neural mechanisms for self-grooming sequences, suggested by our current results, might also influence the theme or content, as well as the syntactic stereotypy, of some human super-stereotypies involving washing or purifying compulsions [74].
Pathologically-intense rituals of cleanliness, security behavior, or concerns with contamination, all share a focus that might relate to grooming of oneself [74]. Conceivably, excessive activation in brain circuits linked by evolution to self-grooming behavior might tip the thematic focus of some human stereotyped sequences toward rituals of cleanliness or reaction to perceived contamination, in addition to strengthening their syntactic rigidity. Whether or not such a direct overlap exists between human pathology and animal instinctive behavior, our results indicate that DAT-KD mutant mice show sequential super-stereotypy in a complex instinctive fixed action pattern.
Methods
Subjects
DAT-KD mutant mice (n = 12 male) and wild-type control mice (n = 12 male) were generated at the University of Chicago by breeding heterozygous mutants on a 129 Sv/J genetic background as described earlier [58]. Such a design minimizes any contribution to behavioral phenotype by genetic background difference or by differences in genetic modifiers that are linked to the Slc6a3 locus. DAT knockdown was achieved by insertion of the tetracycline regulatable system into the 5' untranslated region in the second exon of the DAT gene (Slc6a3). Such an insertion reduced the DAT promoter strength without affecting its expression pattern. It also allows regulation of DAT expression by dietary tetracycline, although that feature was not used in this study. DAT knockdown reduces adult DAT expression to 10% of wild-type levels and raises extracellular dopamine levels in neostriatum to 170% (wild-type control = 100%) [58]. Once housed at the University of Michigan, mutant and wild-type mice (age 2–4 months) were allowed to habituate to their new surroundings for two weeks before any behavioral testing. Mice were housed at ~21°C on a 12 h light/dark cycle with lights on at 7 a.m., in groups of two to three same-type individuals during the laboratory environment testing phase. During the home cage testing phase of the experiments, mice were housed individually to facilitate videotaping. Food (Purina Rat Chow; St. Louis, MO) and water (tap water) were always available.
Behavioral testing
It was important to determine whether any sequential stereotypy difference between mutant and wild-type mice in grooming behavior was a stable difference in action syntax strength, and not merely an artifact of testing conditions. Grooming behavior of rodents is sensitive to environmental contexts, both in quantity and in fine structure, and stressors in particular can either suppress or increase grooming behavior depending on type [88]. All mice were therefore tested for grooming behavior in 2 environmental contexts: 1) a standard behavioral neuroscience laboratory chamber, and 2) their own home cages (a relatively stress-free environment).
Laboratory environment
Immediately prior to testing, mice were transported in their home cage on a cart down a 30 m hallway to a laboratory testing room with standard white fluorescent lighting, and placed individually in a test chamber (light intensity 550–650 lux; sound intensity 65–70 decibels measured within chamber). The laboratory test chamber consisted of a transparent cylinder (19 cm high, 12.5 cm diameter) suspended over a tilted mirror. A camera lens focused on this mirror gave a close-up view of the mouse's face, forepaws, and upper body. For behavioral testing, each mouse was placed individually in a test chamber and videotaped for 30 minutes. Each mouse received 3 habituation days in the laboratory test procedure before grooming behavior data were collected over the next 2 consecutive days in 30 min sessions.
Home environment
Testing in the home environment took place during the dark phase under dim red light conditions. Mice were housed singly in transparent rectangular cages (12 cm high × 19 cm long × 10 cm wide). Videotaping of grooming sequences took place from the side of the transparent home cage, for 30 min each day on 2 consecutive days, with the camera focused closely on the mouse.
Behavioral video analysis
Videotaped grooming behavior was scored in slow motion (frame-by-frame to 1/10th actual speed; scorer blind to genotype) for grooming amount (cumulative durations), grooming bout number and bout length, and occurrence of syntactic chains. Syntactic grooming chains were identified and classified in frame-by-frame analysis as either Perfect, Imperfect but completed by full Phase IV (omission, insertion, or reversal types), Terminal substitution of paw lick for Phase IV body licking, or Incomplete (grooming stops before Phase IV, or reverts to sequentially flexible facial grooming and paw strokes) [37,41], [43,44]. We also made choreograph diagrams of syntactic chains from each mouse to compare details of their form and sequential pattern [61]. Behavioral data were statistically analyzed by 3-factor, 2-factor, or 1-factor ANOVA as indicated above. When significant results were obtained, post hoc paired comparisons were subsequently performed using Bonferroni or Tukey tests (alpha set equal to original 0.05 level).
Authors' contributions
KCB conceived and supervised the study and drafted the manuscript; JWA co-conceived the study and participated in interpretation and writing; KRH carried out behavioral testing, videoanalysis, and statistics; XZ developed and generated the mutant mice, and participated in writing the manuscript.
Supplementary Material
Additional File 1
Movie: Sequential super-stereotypy of an instinctive fixed action pattern in hyper-dopaminergic mutant mice. Windows Media Player movie file (.avi): DAT Knockdown grooming fixed action pattern.aviExamples of syntactic grooming chains performed by three hyperdopaminergic mutant mice are shown in the accompanying movie file. Choreograph diagrams of component movements' form and sequence are displayed for each syntactic chain, and strokes are illuminated sequentially in synchrony with their corresponding movements. Note that the first two syntactic chains contain insertion or reversal errors (Mutant mouse 1: paw lick insertions in Phase II, between Phases II and III, and between Phases III and IV; also reversal insertion of a Phase I ellipse stroke within Phase II. Mutant mouse 2: paw lick insertions within Phase I, within Phase III, and between Phases III and IV). However, the syntactic chains are not disrupted by these errors, and the mutants continue on with the sequential pattern to successfully complete Phase IV (body licking). Mutant mouse 3 also shows the ventral view that permits the viewer to see both forepaws simultaneously, which was used to score all syntactic chains in the laboratory.
Click here for file
Acknowledgments
We thank Dr. Roger Albin for helpful discussion of issues presented here, Dr. Susana Peciña for assistance in preparing the figures, Ben Long for assistance with data scoring, and anonymous reviewers for helpful suggestions. This research was supported by National Science Foundation Grant IBN 0091661, National Institutes of Health Grants MH63649 and DA015188, the National Alliance for Research on Schizophrenia and Depression, and the Tourette Syndrome Association.
==== Refs
Albin RL Koeppe RA Bohnen NI Nichols TE Meyer P Wernette K Minoshima S Kilbourn MR Frey KA Increased ventral striatal monoaminergic innervation in Tourette syndrome Neurology 2003 61 310 315 12913189
Albin RL Walkup J Neurobiology of Basal Ganglia and Tourette Syndrome: Striatal and Dopamine Function Tourette Syndrome New York: Raven Press
Nordstrom EJ Burton FH A transgenic model of comorbid Tourette's syndrome and obsessive-compulsive disorder circuitry Mol Psychiatry 2002 7 617 625 524. 12140785 10.1038/sj.mp.4001144
Kim CH Koo MS Cheon KA Ryu YH Lee JD Lee HS Dopamine transporter density of basal ganglia assessed with [123I]IPT SPET in obsessive-compulsive disorder Eur J Nucl Med Mol Imaging 2003 30 1637 1643 14513291 10.1007/s00259-003-1245-7
Rauch SL Whalen PJ Curran T Shin LM Coffey BJ Savage CR McInerney SC Baer L Jenike MA Probing striato-thalamic function in obsessive-compulsive disorder and Tourette syndrome using neuroimaging methods Adv Neurol 2001 85 207 224 11530429
Peterson BS Thomas P Kane MJ Scahill L Zhang H Bronen R King RA Leckman JF Staib L Basal Ganglia volumes in patients with Gilles de la Tourette syndrome Arch Gen Psychiatry 2003 60 415 424 12695320 10.1001/archpsyc.60.4.415
Krause KH Dresel S Krause J Kung HF Tatsch K Lochmuller H Elevated striatal dopamine transporter in a drug naive patient with Tourette syndrome and attention deficit/ hyperactivity disorder: positive effect of methylphenidate J Neurol 2002 249 1116 1118 12420715 10.1007/s00415-002-0746-9
Saka E Graybiel AM Pathophysiology of Tourette's syndrome: striatal pathways revisited Brain Dev 2003 25 S15 19 14980366 10.1016/S0387-7604(03)90002-7
Graybiel AM Rauch SL Toward a neurobiology of obsessive-compulsive disorder Neuron 2000 28 343 347 11144344 10.1016/S0896-6273(00)00113-6
Grados MA Walkup J Walford S Genetics of obsessive-compulsive disorders: new findings and challenges Brain Dev 2003 25 S55 61 14980374 10.1016/S0387-7604(03)90010-6
Pato MT Pato CN Pauls DL Recent findings in the genetics of OCD J Clin Psychiatry 2002 63 30 33 12027117
Diaz-Anzaldua A Joober R Riviere JB Dion Y Lesperance P Richer F Chouinard S Rouleau GA Tourette syndrome and dopaminergic genes: a family-based association study in the French Canadian founder population Mol Psychiatry 2004 9 272 277 15094788 10.1038/sj.mp.4001411
Graybiel AM Saka E A genetic basis for obsessive grooming Neuron 2002 33 1 2 11779470 10.1016/S0896-6273(01)00575-X
Lieberman P Human language and our reptilian brain: the subcortical bases of speech, syntax, and thought 2000 Cambridge, Mass.: Harvard University Press
Marsden CD Evered D, O'Conner M Which motor disorder in Parkinson's disease indicates the true motor function of the basal ganglia? Functions of the Basal Ganglia 1984 London: Pitman 225 237
Berns GS Sejnowski TJ A computational model of how the basal ganglia produce sequences J Cogn Neurosci 1998 10 108 121 9526086 10.1162/089892998563815
Joel D Niv Y Ruppin E Actor-critic models of the basal ganglia: new anatomical and computational perspectives Neural Netw 2002 15 535 547 12371510 10.1016/S0893-6080(02)00047-3
Ivry RB The representation of temporal information in perception and motor control Cur Opin Neurobio 1996 6 851 857 10.1016/S0959-4388(96)80037-7
Beiser DG Houk JC Model of cortical-basal ganglionic processing: encoding the serial order of sensory events J Neurophysiol 1998 79 3168 3188 9636117
Lashley KS Jefferies LA The problem of serial order in behavior Cerebral mechanisms of behavior 1951 New York: Wiley 112 136
Hauser MD Chomsky N Fitch WT The faculty of language: what is it, who has it, and how did it evolve? Science 2002 298 1569 1579 12446899 10.1126/science.298.5598.1569
Premack D Psychology. Is language the key to human intelligence? Science 2004 303 318 320 14726578 10.1126/science.1093993
Fentress JC Stilwell FP Grammar of a movement sequence in inbred mice Nature 1973 244 52 53 4582491
Aldridge JW Berridge KC Rosen AR Basal ganglia neural mechanisms of natural movement sequences Canadian Journal of Physiology and Pharmacology 2004 82 732 739 15523530 10.1139/y04-061
Hamilton DA Rosenfelt CS Whishaw IQ Sequential control of navigation by locale and taxon cues in the Morris water task Behav Brain Res 2004 154 385 397 15313026 10.1016/j.bbr.2004.03.005
Aldridge JW Whishaw IQ, Kolb B Grooming The behavior of the laboratory rat: a handbook with tests 2005 New York: Oxford University Press 141 149
Aldridge JW Berridge KC Coding of serial order by neostriatal neurons: A "natural action" approach to movement sequence J Neurosci 1998 18 2777 2787 9502834
Meyer-Luehmann M Thompson JF Berridge KC Aldridge JW Substantia nigra pars reticulata neurons code initiation of a serial pattern: implications for natural action sequences and sequential disorders Eur J Neurosci 2002 16 1599 1608 12405974 10.1046/j.1460-9568.2002.02210.x
Baxter LR Jr Basal ganglia systems in ritualistic social displays: reptiles and humans; function and illness Physiol Behav 2003 79 451 460 12954439 10.1016/S0031-9384(03)00164-1
Kermadi I Joseph JP Activity in the caudate nucleus of monkey during spatial sequencing J Neurophysiol 1995 74 911 933 7500161
Graybiel AM Building action repertoires: memory and learning functions of the basal ganglia Cur Opin Neurobio 1995 5 733 741 10.1016/0959-4388(95)80100-6
Schmitzer-Torbert N Redish AD Neuronal activity in the rodent dorsal striatum in sequential navigation: separation of spatial and reward responses on the multiple T task J Neurophysiol 2004 91 2259 2272 14736863 10.1152/jn.00687.2003
Kimura M Matsumoto N Okahashi K Ueda Y Satoh T Minamimoto T Sakamoto M Yamada H Goal-directed, serial and synchronous activation of neurons in the primate striatum Neuroreport 2003 14 799 802 12858035 10.1097/00001756-200305060-00004
Cenci MA Whishaw IQ Schallert T Animal models of neurological deficits: how relevant is the rat? Nat Rev Neurosci 2002 3 574 579 12094213 10.1038/nrn877
Berridge KC Whishaw IQ Cortex, striatum and cerebellum: Control of serial order in a grooming sequence Exp Brain Res 1992 90 275 290 1397142 10.1007/BF00227239
Cromwell HC Berridge KC Implementation of action sequences by a neostriatal site: A lesion mapping study of grooming syntax J Neurosci 1996 16 3444 3458 8627378
Berridge KC Comparative fine structure of action: Rules of form and sequence in the grooming patterns of six rodent species Behav 1990 113 21 56
Bursten SN Berridge KC Owings DH Do California ground squirrels (Spermophilus beecheyi) use ritualized syntactic cephalocaudal grooming as an agonistic signal? Journal of Comparative Psychology 2000 114 281 290 10994844 10.1037//0735-7036.114.3.281
Barlow GB Ethological units of behavior Foundations of animal behavior: Classic papers with commentaries 1996 Chicago: University of Chicago Press 138 153
Hinde RA Animal Behaviour: a synthesis of ethology and comparative psychology, 1970 2 London: McGraw-Hill
Cromwell HC Berridge KC Drago J Levine MS Action sequencing is impaired in D1A-deficient mutant mice Euro J Neurosci 1998 10 2426 2432 10.1046/j.1460-9568.1998.00250.x
Berridge KC Substantia nigra 6-OHDA lesions mimic striatopallidal disruption of syntactic grooming chains: A neural systems analysis of sequence control Psychobiol 1989 17 377 385
Berridge KC Aldridge JW Super-stereotypy I: Enhancement of a complex movement sequence by systemic dopamine D1 agonists Synapse 2000 37 194 204 10881041 10.1002/1098-2396(20000901)37:3<194::AID-SYN3>3.0.CO;2-A
Berridge KC Aldridge JW Super-stereotypy II: Enhancement of a complex movement sequence by intraventricular dopamine D1 agonists Synapse 2000 37 205 215 10881042 10.1002/1098-2396(20000901)37:3<205::AID-SYN4>3.0.CO;2-A
Deveney AM Waddington JL Psychopharmacological distinction between novel full-efficacy "D-1-like" dopamine receptor agonists Pharmacol Biochem Behav 1997 58 551 558 9300618 10.1016/S0091-3057(97)00248-7
Delfs JM Kelley AE The role of D1 and D2 dopamine receptors in oral stereotypy induced by dopaminergic stimulation of the ventrolateral striatum Neurosci 1990 39 59 67 10.1016/0306-4522(90)90221-O
Kuczenski R Segal DS Sensitization of amphetamine-induced stereotyped behaviors during the acute response: role of D1 and D2 dopamine receptors Brain Res 1999 822 164 174 10082894 10.1016/S0006-8993(99)01149-X
Mittleman G LeDuc PA Whishaw IQ The role of D1 and D2 receptors in the heightened locomotion induced by direct and indirect dopamine agonists in rats with hippocampal damage: an animal analogue of schizophrenia Behav Brain Res 1993 55 253 267 8102851 10.1016/0166-4328(93)90121-6
Capper-Loup C Canales JJ Kadaba N Graybiel AM Concurrent activation of dopamine D1 and D2 receptors is required to evoke neural and behavioral phenotypes of cocaine sensitization J Neurosci 2002 22 6218 6227 12122080
Saka E Goodrich C Harlan P Madras BK Graybiel AM Repetitive behaviors in monkeys are linked to specific striatal activation patterns J Neurosci 2004 24 7557 7565 15329403 10.1523/JNEUROSCI.1072-04.2004
Greer JM Capecchi MR Hoxb8 is required for normal grooming behavior in mice Neuron 2002 33 23 34 11779477 10.1016/S0896-6273(01)00564-5
Reilly CE Disruption of Hoxb8 gene leads to obsessive grooming behavior J Neurol 2002 249 499 501 11967666 10.1007/s004150200052
Campbell KM de Lecea L Severynse DM Caron MG McGrath MJ Sparber SB Sun LY Burton FH OCD-Like behaviors caused by a neuropotentiating transgene targeted to cortical and limbic D1+ neurons J Neurosci 1999 19 5044 5053 10366637
Campbell KM McGrath MJ Burton FH Behavioral effects of cocaine on a transgenic mouse model of cortical-limbic compulsion Brain Res 1999 833 216 224 10375697 10.1016/S0006-8993(99)01544-9
McGrath MJ Campbell KM Burton FH The role of cognitive and affective processing in a transgenic mouse model of cortical-limbic neuropotentiated compulsive behavior Behav Neurosci 1999 113 1249 1256 10636303 10.1037//0735-7044.113.6.1249
McGrath MJ Campbell KM Veldman MB Burton FH Anxiety in a transgenic mouse model of cortical-limbic neuro-potentiated compulsive behavior Behav Pharmacol 1999 10 435 443 10780249
McGrath MJ Campbell KM Parks CR Burton FH Glutamatergic drugs exacerbate symptomatic behavior in a transgenic model of comorbid Tourette's syndrome and obsessive-compulsive disorder Brain Res 2000 877 23 30 10980239 10.1016/S0006-8993(00)02646-9
Zhuang X Oosting RS Jones SR Gainetdinov PR Miller GW Caron MG Hen R Hyperactivity and impaired response habituation in hyperdopaminergic mice Proc Natl Acad Sci USA 2001 98 1982 1987 11172062 10.1073/pnas.98.4.1982
Ralph-Williams RJ Paulus MP Zhuang X Hen R Geyer MA Valproate attenuates hyperactive and perseverative behaviors in mutant mice with a dysregulated dopamine system Biol Psychiatry 2003 53 352 359 12586455 10.1016/S0006-3223(02)01489-0
Pecina S Cagniard B Berridge KC Aldridge JW Zhuang X Hyperdopaminergic mutant mice have higher "wanting" but not "liking" for sweet rewards J Neurosci 2003 23 9395 9402 14561867
Berridge KC Fentress JC Parr H Natural syntax rules control action sequence of rats Behav Brain Res 1987 23 59 68 3828046 10.1016/0166-4328(87)90242-7
Kalueff AV Tuohimaa P Grooming analysis algorithm for neurobehavioural stress research Brain Res Brain Res Protoc 2004 13 151 158 15296852 10.1016/j.brainresprot.2004.04.002
Richmond G Sachs BD Grooming in Norway rats: The development and adult expression of a complex motor pattern Behav 1978 75 82 96
McNamara FN Clifford JJ Tighe O Kinsella A Drago J Fuchs S Croke DT Waddington JL Phenotypic, ethologically based resolution of spontaneous and D(2)-like vs D(1)-like agonist-induced behavioural topography in mice with congenic D(3) dopamine receptor "knockout" Synapse 2002 46 19 31 12211095 10.1002/syn.10108
Coscia EM Fentress JC Neurological dysfunction expressed in the grooming behavior of developing Weaver mutant mice Behavior Genetics 1993 23 533 541 8129695 10.1007/BF01068144
Bolivar VJ Danilchuk W Fentress JC Separation of activation and pattern in grooming development of weaver mice Behav Brain Res 1996 75 49 58 8800659 10.1016/0166-4328(96)00156-8
Packard MG Knowlton BJ Learning and memory functions of the basal ganglia Annu Rev Neurosci 2002 25 563 593 12052921 10.1146/annurev.neuro.25.112701.142937
Doyon J Laforce R JrBouchard G Gaudreau D Roy J Poirier M Bedard PJ Bedard F Bouchard JP Role of the striatum, cerebellum and frontal lobes in the automatization of a repeated visuomotor sequence of movements Neuropsychologia 1998 36 625 641 9723934 10.1016/S0028-3932(97)00168-1
Saint-Cyr JA Frontal-striatal circuit functions: Context, sequence, and consequence J Int Neuropsychol Soc 2003 9 103 127 12570364 10.1017/S1355617703910125
Graybiel AM The basal ganglia and chunking of action repertoires Neurobiology of Learning and Memory 1998 70 119 136 9753592 10.1006/nlme.1998.3843
Greenberg N MacLean PD Ferguson JL Role of the paleostriatum in species-typical display behavior of the lizard (Anolis carolinensis) Brain Res 1979 172 229 241 466473 10.1016/0006-8993(79)90535-3
Brown LL Schneider JS Lidsky TI Sensory and cognitive functions of the basal ganglia Cur Opin Neurobiol 1997 7 157 163 10.1016/S0959-4388(97)80003-7
Knowlton BJ Mangels JA Squire LR A neostriatal habit learning system in humans Science 1996 273 1399 1402 8703077
Szechtman H Woody E Obsessive-compulsive disorder as a disturbance of security motivation Psychol Rev 2004 111 111 127 14756589 10.1037/0033-295X.111.1.111
Ho AK Bradshaw JL Cunnington R Phillips JG Iansek R Sequence heterogeneity in Parkinsonian speech Brain & Language 1998 64 122 145 9675046 10.1006/brln.1998.1959
Toates F Coschug-Toates O Obsessive Compulsive Disorder 2002 London: Class publishing
Rapoport JL Wise SP Obsessive-compulsive disorder: Evidence for basal ganglia dysfunction Psychopharmacol Bull 1988 24 380 384 3153497
Hollander E Liebowitz MR DeCaria CM Conceptual and methodological issues in studies of obsessive-compulsive and Tourette's disorders Psychiat Dev 1989 7 267 296
Goodman WK McDougle CJ Price LH Riddle MA Pauls DL Leckman JF Beyond the serotonin hypothesis: a role for dopamine in some forms of obsessive compulsive disorder? J Clin Psychiat 1990 51 36 43 discussion 55-38.
Ridley RM The psychology of perseverative and stereotyped behaviour Prog Neurobiol 1994 44 221 231 7831478 10.1016/0301-0082(94)90039-6
McDougle CJ Update on pharmacologic management of OCD: agents and augmentation J Clin Psychiat 1997 58 11 17
Rapoport JL Le "spectre obsessionnel-compulsif": Un concept utile? Encephale 1994 20 Spec No 4 677 680 7895635
Albin RL Young AB Penney JB The functional anatomy of disorders of the basal ganglia Trends Neurosci 1995 18 63 64 7537410 10.1016/0166-2236(95)93872-U
Luthi-Carter R Strand A Peters NL Solano SM Hollingsworth ZR Menon AS Frey AS Spektor BS Penney EB Schilling G Decreased expression of striatal signaling genes in a mouse model of Huntington's disease Hum Mol Genet 2000 9 1259 1271 10814708 10.1093/hmg/9.9.1259
Spektor BS Miller DW Hollingsworth ZR Kaneko YA Solano SM Johnson JM Penney JB JrYoung AB Luthi-Carter R Differential D1 and D2 receptor-mediated effects on immediate early gene induction in a transgenic mouse model of Huntington's disease Brain Res Mol Brain Res 2002 102 118 128 12191502 10.1016/S0169-328X(02)00216-4
McDougle CJ Goodman WK Price LH The pharmacotherapy of obsessive-compulsive disorder Pharmacopsychiatry 1993 26 24 29 8378419
Poyurovsky M Bergman Y Shoshani D Schneidman M Weizman A Emergence of obsessive – compulsive symptoms and tics during clozapine withdrawal Clin Neuropharm 1998 21 97 100
Fentress JC Expressive contexts, fine structure, and central mediation of rodent grooming Ann N Y Acad Sci 1988 525 18 26 3291664
| 15710042 | PMC552313 | CC BY | 2021-01-04 16:39:16 | no | BMC Biol. 2005 Feb 14; 3:4 | utf-8 | BMC Biol | 2,005 | 10.1186/1741-7007-3-4 | oa_comm |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.