text
stringlengths 10
379k
|
|---|
The two major forms of motoneuron disease, proximal spinal muscular atrophy (SMA) and amyotrophic lateral sclerosis, are caused by selective cell death of motoneurons. Among the mechanisms that are thought to play a central role are cell- autonomous mechanisms like oxidative stress and mitochondrial dysfunction (), but also nonautonomous processes such as dysregulated signaling from neighboring glial cells and contacting neurons (; ). Such mechanisms have been studied in great detail for amyotrophic lateral sclerosis (). In contrast, much less is known for proximal SMA, the most common form of motoneuron disease in children and young adults (; ; ; ). This disease is caused by homozygous loss or mutations in the telomeric copy () of the survival of motor neuron gene (; ) on human chromosome 5q13. Whereas the gene allows expression of a functionally intact full- length protein, most of the transcripts from the gene code for a truncated protein lacking the functionally important domains at the C terminus that are encoded by exon 7 (; ). Nevertheless, low expression of full-length Smn protein from the gene occurs, but this is not sufficient for compensating the defect caused by loss, thus leading to motoneuron disease in humans. In contrast to humans, mice carry only one gene, and the homozygous knockout of the gene in mice is lethal in early development, even before blastocysts are formed ().
The gene is ubiquitously expressed, thus raising the question of how reduced levels of this protein lead to specific motoneuron disease. Smn plays a role in the assembly and in recycling of spliceosomal uridine-rich small nuclear RNPs (; ). Dysfunction of such processes should lead to severe defects in all cell types. The clinical phenotype of patients with SMA indicates that low levels of SMN protein, both in the full-length and the truncated form lacking the exon 7–encoded domains, are sufficient for development, survival, and function of most types of cells, but not for motoneurons. It has therefore been hypothesized that motoneurons are more vulnerable, possibly because they are among the biggest cells in the body and their need for proper mRNA expression, processing, and translation is probably higher than in other cell types (). This hypothesis is supported by the observation that injection of assembled small nuclear RNP complexes into early Smn-deficient zebrafish embryos can rescue defects in motoneurons ().
A mouse model for SMA has been generated by introducing the human into a mouse null background (). The phenotype of these mice closely resembles that of humans. These mice develop severe paralysis within a few days after birth and normally die between postnatal day 1 and 5. Surprisingly, the loss of motoneuron cell bodies at late stages of the disease does not exceed 20%, suggesting that most motoneurons develop normally during embryogenesis and that disease becomes apparent before the majority of motoneurons are lost.
embryos does not differ from control motoneurons (
; ). However, they exhibit a specific defect in axon elongation that correlates with a defect in β-actin mRNA translocation to distal axons ().
To study the underlying pathomechanism in Smn-deficient motoneurons, we have analyzed the functional consequences of Smn deficiency in growth cones. Smn-deficient motoneurons show defects in spontaneous excitability, and these defects correlate with reduced integration of voltage-gated Ca channels (VGCCs) into axonal growth cones.
motoneurons and leads to restoration of the morphological and functional deficits in axons and axon terminals. These findings indicate that reduced excitability in growth cones contributes to the disease phenotype. This defect could lead to disturbances of active zones in the presynapse, causing reduced transmitter release at the motor endplate that, in turn, could contribute to motoneuron malfunction and degeneration in SMA.
motoneurons, translocation of β-actin mRNA to distal axons and growth cones is disturbed ().
mutant mouse embryos, and cultured them on laminin-111. The cultured neurons were loaded with Fura-2, a Ca-binding fluorescent dye (, top left) and analyzed over periods of 7.5 min at 3, 4, 5, and 7 d in culture (, bottom left). In parallel to these measurements, motoneuron survival in the presence of brain-derived and ciliary neurotrophic factor (10 ng/ml each) was determined. No difference was observed between Smn-deficient and control motoneurons ().
motoneurons that were cultured on laminin-111, spontaneous spikelike Ca transients were detectable that appeared synchronized in the cell body, dendrites, axons, and axonal growth cones, thus reflecting global spontaneous depolarization as described previously for developing neurons (; ). These transients could be blocked by 1 μM tetrodotoxin (TTX), indicating that they are triggered by opening of voltage-gated Na channels (, top right). The addition of 0.3 and 1.0 μM ω-conotoxin MVIIA (CTX) inhibited these Ca transients, suggesting that N-type Ca channels are responsible for the Ca transients measured in these cultures (, bottom right; and S1 A, available at ).
and control motoneurons ().
motoneurons, confirming previous analyses (), perhaps because the resting potential becomes more negative in embryonic motoneurons that are cultured for longer periods ().
motoneurons ().
We then investigated the frequency of such spikelike spontaneous transients in axons and growth cones ().
motoneurons, indicating that not every transient spreads from the cell body and proximal axon to distal axons at this stage. At day 4, spikelike transient frequency was similar in cell bodies and distal axons in control motoneurons.
motoneurons (0.39 ± 0.11 min) in comparison to control
cells (0.83 ± 0.18 min; P < 0.05).
motoneurons (). TTX and CTX also inhibited these spontaneous Ca transients on day 5 in culture (Fig. S1 A). The reduction of Ca transients was >50% at 0.3 μM CTX and increased to ∼80% at 1 μM CTX, indicating that influx through VGCCs is the predominant source of these fast transients of cytosolic Ca (Fig. S1 A).
The time course of reduced spontaneous activity in distal axons and growth cones resembles that of axonal growth defects in these cultures.
motoneurons are shorter at day 5. Surprisingly, significant differences in axon elongation (P < 0.05) are not detectable at earlier stages (). Normally, between days 3 and 4, a doubling of axon length is observed in cultures of both control and Smn-deficient motoneurons, but the difference between Smn-deficient and control cells was not significant (; P > 0.05).
motoneurons ().
motoneurons. However, these defects occur late, starting at E14 plus four additional days in culture (E14 + 4), at a developmental stage when maximal axon elongation has already occurred and motor endplate differentiation progresses in vivo.
To follow the idea that axonal defects become apparent when Smn-deficient motoneurons get in contact with skeletal muscle, we investigated motoneurons on motor endplate–specific forms of laminin (laminin-211/221). Previous papers have shown that neurite growth of motoneurons is reduced on this substrate (; ). Furthermore, it has been shown that the β2 chain in laminin-221 interacts with the pore-forming (Ca) subunit of the N-type VGCC (Ca2.2; ). Thus, presynaptic differentiation appears mediated through the direct interaction of laminin-221 with Ca2.2 channels. We therefore measured Ca transients that only occurred in growth cones (). Such local transients have been identified in many types of developing neurons, including motoneurons, and some of these transients are caused by Ca influx through other channels than VGCCs (). Evidence has been presented that the transient receptor potential ion channel 5 (TRPC5) is a channel that can regulate neurite growth and growth cone morphology, at least in hippocampal neurons ().
nor in control motoneurons at day 5 in culture (). They were much more frequent on laminin-211/221, both in control (0.22 ± 0.06 min on laminin-211/221 vs. 0.04 ± 0.03 min on laminin-111) and
motoneurons (0.07 ± 0.03 min on laminin-211/221 vs. 0.02 ± 0.01 min on laminin-111; ). In control cultures, ∼50% of these local transients in growth cones could be blocked with CTX (Fig. S1, B and C), indicating that VGCCs are also responsible for some but not all fast local transients in growth cones of isolated embryonic mouse motoneurons.
motoneurons on laminin-211/221 at day 5 in culture (, P < 0.05).
In embryonic motoneurons, the N-type VGCCs are predominantly expressed (; ). These channels are located in axon terminals of motoneurons where they act as receptors for motor endplate–specific forms of laminin ().
motoneurons showed reduced spontaneous Ca transients in growth cones, we investigated expression and cellular distribution of Ca2.2 in control and Smn-deficient motoneurons () using polyclonal antibodies against the α2 chain of this channel. The Ca2.2 signal intensity was quantified as arbitrary units based on quantum levels per pixel per area in cell body, proximal axon, and growth cones. A significantly reduced signal intensity (P < 0.001) was found in growth cones of Smn-deficient motoneurons (8.7 ± 1.1) versus controls (21.9 ± 2.1; ). Applying a fixation protocol without detergent and shortened exposure to 4% PFA to increase the staining intensity of cell surface–exposed versus intracellular Ca2.2, showed this channel to be highly concentrated in protrusions of control growth cones () but not in
growth cones (). Furthermore, Ca2.2 channels colocalize with the active zone protein Piccolo, indicating that clusters of Ca2.2 have formed active zone-like structures in the growth cone protrusions in control motoneurons (Fig. S2, A, C, E, and G, available at ). The colocalization of Ca2.2 with Piccolo was highly reduced in Smn-deficient growth cones (Fig. S2, B, D, F, and H). In control motoneurons, these active zone-like structures cover ∼20% of the whole growth cone area. In Smn-deficient motoneurons <5% of the active zone-like structures are detectable (Fig. S2 I). These structures are only found in protrusions but not the core or the proximal parts of growth cones and axons (Fig. S2, J–L).
and
motoneurons was not different ().
To analyze whether the reduced Ca2.2 expression in distal axons is caused by disturbed subcellular distribution of the corresponding mRNA, in situ hybridization was performed. This experiment did not reveal any difference in cellular distribution or rough differences in expression levels between control and Smn-deficient motoneurons (). The specificity of the in situ hybridization was controlled with a Ca2.2 sense probe ().
We also applied stimulated emission depletion (STED) fluorescence microscopy (; ; ) to investigate whether the reduced accumulation of Ca2.2 reflects a defect in cluster formation of this channel. This method enhances the resolution of confocal microscopy in the xy axis, so that structures <200 nm that normally cannot be resolved by classical light microscopy become detectable. Intracellular vesicles containing Ca2.2 channels are much smaller than the Ca2.2 clusters that form on the cell surface. When we compared the size of the Ca2.2 immunoreactive areas in control and
growth cones, it became apparent that in Smn-deficient growth cones the relative density of large clusters covering an area of at least 0.01 μm versus small vesicles is reduced compared with control growth cones (). Collectively, these data suggest that a defect in Ca2.2 transfer into the cell membrane and active zone-like structures in Smn-deficient growth cones is responsible for reduced frequency of Ca transients.
In parallel to Ca transients, we also analyzed axon elongation on laminin-211/221. Mean axon length of control motoneurons was 264.1 ± 11.2 μm on laminin-211/221 in comparison to 335.2 ± 19.0 μm on laminin-111 after 7 d in culture (). Surprisingly, Smn-deficient motoneurons did not show such a reduction of axon growth on laminin-211/221. In contrast, they exhibited a slight but significant (P < 0.05) increase in axon extension (309.2 ± 12.5 μm) in comparison with control motoneurons (259.6 ± 10.3 μm) on laminin-211/221 ().
motoneurons that were cultured on laminin-111 (). The growth cone area of Smn-deficient motoneurons on laminin-211/221 did not differ from the area on laminin-111 ().
motoneurons were smaller than those of
motoneurons ().
Inhibition of Ca2.2 with CTX blocks global Ca transients in control motoneurons by >80% (Fig. S1 A). In addition, local transients that only occur in axonal growth cones are reduced by >50%, both at 1 and 0.3 μM CTX, which is considered to be highly specific for N-type VGCCs (Fig. S1, A–C). Interestingly, the reduction of local transients in growth cones is smaller, thus confirming earlier observations with motoneurons that other Ca channels contribute to rapid local Ca transients in growth cones (). To determine the role of classical VGCCs for axon growth, we tested whether specific blockade of Ca2.2 with CTX affects axon growth of motoneurons in vitro. CTX was applied at 1 μM and a lower concentration (0.3 μM) that is considered highly specific for N-type VGCCs ( and S1). After 7 d in culture, motoneurons grown on laminin-111 or laminin-211/221 were fixed and stained against tau and microtubule-associated protein 2 to distinguish dendrites and axons (). 1 μM CTX specifically reduces axon growth of control motoneurons on laminin-111 (335.2 ± 19.0 vs. 240.7 ± 8.5 μm), whereas 0.3 μM CTX was less efficient (335.2 ± 19.0 vs. 288.2 ± 11.8 μm; ). Axon growth of Smn-deficient neurons on laminin-111 (233.4 ± 13.8 μm), which was already disturbed, was not further reduced both by 1 and 0.3 μM CTX ().
We then investigated the effect of 0.3 and 1 μM CTX on motoneurons grown on laminin-211/221. Inhibition of Ca2.2 with CTX led to an increase in axon growth in control motoneurons () at both concentrations. Axon elongation reached comparable levels to those seen in motoneurons on laminin-111 (). Smn-deficient motoneurons did not show increased axon elongation after 0.3 and 1 μM CTX treatments (), indicating that the response of axonal growth to blockade of N-type Ca channels is lacking in Smn-deficient motoneurons. Low concentrations (0.3 μM) of CTX showed similar effects as 1 μM CTX on laminin-211/221, thus providing an argument for the specificity of the effect for Ca2.2 channels.
In other types of neurons, TRPC3 and 6 have been shown to mediate Ca fluxes that modulate axon guidance and neuronal survival in response to brain-derived neurotrophic factor (; ). In addition, TRPC5 activation promotes axon guidance and neurite growth in hippocampal neurons (). Therefore, the possibility exists that TRPCs are also involved in the pathological alterations observed in Smn-deficient motoneurons. Among these candidates, TRPC5 is highly expressed in embryonic motoneurons. TRPC6 is also found but at relatively lower levels, whereas TRPC3 is barely detectable (unpublished data). When we analyzed the distribution of TRPC5 and 6 immunoreactivity in growth cones of control and Smn-deficient motoneurons, no difference was observed, both with respect to subcellular distribution (Fig. S3, A–D, available at ) and signal intensity (Fig. S3, E and F). This led us to conclude that alteration in Ca2.2 distribution is responsible for the differences in Ca transients observed in growth cones of Smn-deficient motoneurons.
We further tested whether cAMP, which has previously been described to enhance the frequency of spontaneous Ca transients in developing motoneurons (), could enhance the frequency of Ca transients in growth cones of control and Smn-deficient mouse motoneurons. Surprisingly, there was only little effect (P > 0.05) in cultured control motoneurons ().
axonal growth cones react with a more than twofold increase in the frequency of Ca transients on day 5 in axon terminals after treatment with 100 μM 8-CPT-cAMP. To test whether 8-CPT-cAMP has a direct effect on Ca transients, we added CTX together with 8-CPT-cAMP (Fig. S1, C and D). Enhanced cAMP does not rescue the reduction of Ca transients that is caused by CTX (Fig. S1 C), indicating that there are no CTX-insensitive channels present that open in response to 8-CPT-cAMP.
motoneurons with CTX, there was little further reduction of local Ca transients, but treatment with 8-CPT-cAMP increases Ca transient frequency (Fig. S1 D).
but not in control motoneurons, leading to similar levels of Ca transients than those observed with CTX and 8-CPT-cAMP in control motoneurons (Fig. S1 C).
motoneurons.
We then analyzed whether this increase of Ca transients in 8-CPT-cAMP–treated Smn-deficient motoneurons correlates with increased Ca2.2 expression in Smn-deficient growth cones. Both the signal intensity of the Ca2.2 staining () and β-actin in growth cones () was increased to almost normalized levels by this treatment.
The promoter region contains a CreII-binding element () that mediates cAMP effects on increased transcription in mouse hepatocytes. We therefore tested whether cAMP up-regulates Smn expression and thus restores the morphological and functional alterations in Smn-deficient motoneurons. For this purpose we analyzed Smn protein levels and distribution in cultured embryonic motoneurons. Because the number of motoneurons that can be isolated from one Smn-deficient embryo is not sufficient for quantitative RT-PCR or Western blot analysis, we prepared protein and RNA extracts from cultures of E11.5 forebrain neuronal precursor cells from control and Smn-deficient mice. 100 μM 8-CPT-cAMP increased mRNA (Fig. S4, A and B, available at ) and protein levels by ∼40–100% (Fig. S4 C) in these cells.
and
embryos were treated with 100 μM 8-CPT-cAMP. Both in cell bodies, axons, and axonal growth cones, Smn-specific fluorescence signal intensity was enhanced in Smn-deficient neurons but not fully restored to control levels ().
motoneurons (). 8-CPT-cAMP treatment normalized growth cone size in
motoneurons to control levels and CTX did not abolish the rescue effect, indicating that it does not involve enhanced Ca transients (). We then investigated whether this effect is caused by a normalization of local β-actin levels in distal axons.
motoneurons (), and this ratio was normalized by 8-CPT-cAMP (). The altered β-actin ratio is based on increased actin mRNA in the growth cone of 8-CPT-cAMP–stimulated Smn-deficient motoneurons (). Thus, elevated cAMP increases distal actin mRNA and protein levels in axons, leading to augmented Ca2.2 levels in growth cones and normalization of Ca transient frequency.
motoneurons.
neurons on laminin-211/221 when the cells are treated with 8-CPT-cAMP, indicating that the responsiveness to motor endplate-specific forms of laminin is restored by elevated cAMP levels.
We have investigated the correlation between defective axon elongation and spontaneous excitability in motoneurons from a mouse model of SMA. We observed that the reduction of spontaneous Ca transients in distal axons and growth cones is caused by defective Ca2.2 accumulation and clustering in the axonal growth cones, thus influencing axon growth in the Smn-deficient motoneurons. These defects can be at least partially compensated by 8-CPT-cAMP treatment. Motoneuron disease in SMA, both in humans and mouse models, becomes apparent after motoneurons have made contact with skeletal muscle. In a mouse model of SMA type I, motoneuron loss is not enhanced during a critical period of development when motoneurons depend on trophic support from target tissues (). Motoneuron numbers are normal at birth, but decrease at postnatal days 3–5. During this period, differentiation of motor endplates takes place, and we therefore investigated neurons that were cultured on motor endplate–specific forms of laminin (laminin-211/221). Previous papers have shown that neurite growth of motoneurons is impaired on this substrate (). Furthermore, it has been shown that the β2 chain in laminin-221 interacts with the pore-forming (Ca) subunit of the N- and P/Q-type–specific VGCC (Ca2.2 and Ca2.1; ), indicating that presynaptic differentiation is mediated through the direct interaction of laminin-221 with Ca2.2 channels. Smn-deficient motoneurons exhibit a reduced accumulation of Ca2.2 channels in growth cones. This finding correlates with reduced Ca influx and reduced spontaneous electrical activity in this part of the neuron.
Gene knockout mice for the laminin β2 chain (
) or Ca2.1 (
) exhibit strong synaptic maturation defects. Lamb2-deficient mice develop neuromuscular junction (NMJ) degeneration, which is characterized by disturbed active zones just after birth (; ). 2.2 knockout mice do not show any signs of motoneuron disease, as the defect can most probably be compensated by Ca2.1 expression. Ca2.1-deficient mice develop normally until the third postnatal week ().
mice, the NMJs degenerate and exhibit a decrease of active zone proteins (; ). The delayed disease onset in Ca2.1-deficient mice can be explained by a compensatory effect of residual Ca2.2, which is substituted postnatally by Ca2.1. Altogether, these data indicate that the β2 chain interaction of laminin-221 with Ca2.1 and Ca2.2 supports postnatal development and maintenance of NMJs (). Our data suggest that the reduced Ca2.2 accumulation in the axonal growth cone protrusions of Smn-deficient motoneurons is responsible for reduced axon elongation in cell culture and reduced responsiveness to synapse-specific laminin isoforms.
The ventrolateral part of the lumbar spinal cord of E14 embryos was dissected and transferred to HBSS. After 15 min of treatment with 0.05% trypsin, cells were triturated and cultured after enrichment by panning with antibodies against the mouse p75 neurotrophin receptor (Abcam). Cells were plated at a density of 2,000 cells/cm in 4-well dishes (Greiner Bio-One) and cultured as described previously (). The culture dishes were precoated with polyornithine and laminin-111 or laminin-211/221 (Invitrogen), respectively. 100 μM 8-CPT-cAMP (dissolved in HBSS; Calbiochem), 1 μM TTX (Sigma-Aldrich), and 1 or 0.3 μM CTX (Sigma-Aldrich), respectively, was added by changing the medium every second day.
Immunocytochemistry was performed as described previously (). For analysis of membrane-exposed N-type Ca channels, we fixed the cells only for 2 min with 4% PFA in 1× TBS without acetone. In addition, Tween 20 (Sigma-Aldrich) was omitted from all buffers for this set of experiments (). The following primary antibodies were used: rabbit polyclonal antibodies against tau at 1 μg/ml (1:1,000; Sigma-Aldrich), an N-type Ca channel (1:200; Sigma-Aldrich), TRPC5 (1:200, Sigma- Aldrich), and TRPC6 (1:200, Chemicon), mouse monoclonal antibodies against 1 μg/ml β-actin (Abcam), 1 μg/ml microtubule-associated protein 2 (Sigma-Aldrich), and 2 μg/ml Smn (BD Biosciences). Cells were then washed three times with 1× TBS-T (20 mM Tris-HCl, pH 7.6, 137 mM NaCl, and 0.1% Tween 20) and incubated for 1 h at room temperature with Cy2 (1:200)- and Cy3 (1:300)-conjugated secondary antibodies (Dianova). The Atto 647N–conjugated secondary antibody (Atto Technology) was used for STED microscopy. Confocal images were obtained either with a microscope (TCS 4D; Leica) with a 20× 0.5 objective (PL FLUOTAR) or a microscope (SP2; Leica) with a 100× 1.4 oil-immersion objective (HCX PL APO CS), with identical settings for pinhole and voltage for control and Smn-deficient motoneurons. For high-resolution microscopy, a STED setup mounted to a microscope (SP5; Leica) with a 100× 1.4 oil immersion objective was applied.
Forebrain neuronal precursor cells grown on laminin-111 for 72 h were collected from the dishes, and protein extraction for Western Blotting was performed as described previously (). Primary antibodies, 1 μg/ml anti–mouse β-actin antibody (Abcam), 1 μg/ml anti–mouse β-III tubulin antibody (RDI), and 2 μg/ml Smn IgG1 (BD Biosciences) were used.
For Ca imaging analysis, cultured motoneurons were grown on glass coverslips. After gently washing in phenol red–free HBSS and permeabilization with 0.25% pluronic F-127 (Sigma-Aldrich) over 5 min, cells were loaded with 2 μM FURA-2 AM (Invitrogen) at 37°C for 45 min in Ca and Mg containing HBSS. After washing, coverslips were mounted on a heated microscope stage (37°C) and constantly superfused with Ca/Mg-HBSS under linear flow conditions.
Images were acquired at a frequency of 1 Hz and exposure time was 10 ms. The measurements in proximal axons were taken at a distance of 10–30 μm from the cell body. Spontaneous Ca transients (340:380) were recorded separately in cell bodies, dendrites, axons, and growth cones over a time period of 7.5 min.
Permeabilizing the motoneurons with 20 μM digitonin allowed background signal detection after cytosolic dye release for background correction. Cells were constantly superfused with Ca/Mg-HBSS over the registration time.
Hybridization solution containing 3′ biotinylated sense or antisense actin or N-type Ca channel oligonucleotides (200 ng/ml; GeneDetect) was applied to the coverslips. Hybridization was performed as described previously (). Images were acquired using a microscope (Axiophot; Carl Zeiss MicroImaging, Inc.) equipped with a charge-coupled device camera using Axioplan 2 software (Carl Zeiss MicroImaging, Inc.).
Total RNA from neuronal precursor cells was extracted by Trizol (Invitrogen) according to the manufacturer's protocol, and 1 μg of total RNA was used for cDNA amplification. Amplification of the cDNA was performed with Ex 5f (5′-CCACTTACTATCATGCT-3′) and Ex 8r (5′-CTACAACACCCTTCTCACAG-3′) primers under the following PCR conditions: 3 min at 94°C (1 cycle), 30 s at 94°C, 30 s at 56°C, 45 s at 72°C (30 cycles), and 5 min at 72°C (1 cycle).
For the quantification of β-actin and N-type Ca channel distribution within the different cellular compartments, the staining intensity in the cell body, the proximal and the distal third of the axon, and the growth cone were analyzed with AIDA software (Raytest). Background intensity was measured for every single picture. The intensity for β-actin and N-type Ca channels was measured as arbitrary units per area, based on quantum levels per pixel, according to the manufacturer's instructions.
The final processing of all images was performed with Photoshop 7.0 (Adobe) and Illustrator 10 (Adobe). Linear contrast enhancement was applied to , , and , and all individual panels contained in these figures were treated similarly.
Fig. S1 shows that Ca transient frequency is reduced by TTX and CTX in cultured motoneurons from 14-d-old mouse embryos. Fig. S2 shows colocalization of Ca2.2 and the presynaptic protein Piccolo in protrusions of axonal growth cones in control and
motoneurons. Fig. S3 shows distribution and semiquantitative analysis of TRPC5 and 6 immunoreactivity in the growth cones of control and Smn-deficient motoneurons. Fig. S4 shows that 8-CPT-cAMP stimulates and transcription and increases Smn protein levels in Smn-deficient forebrain neuronal precursor cells. Online supplemental material is available at . |
Polarity is a physical attribute of most eukaryotic cells that is indispensable for their function. Generation and maintenance of cell polarity requires the active segregation of molecular components and imparts distinct properties to subcellular domains. Polarization occurs along a major axis in epithelia, neurons, and asymmetrically dividing cells. In epithelia, polarization relative to the tight junction functionally separates the apical and basolateral domains; neuronal polarization allows for differential development of dendrites and axons. During asymmetric cell division, a polarity cue in the mother cell directs distribution of cell fate determinants in daughter cells. In all these cases, failure to establish cell polarity compromises tissue differentiation and function.
Sustained polarization of epithelial cells lining renal tubules is essential for kidney development and function (). Tubules in the permanent kidney develop through reciprocal interactions between the ureteric bud and metanephric mesenchyme (). The ureteric bud invades the metanephros and undergoes a series of branching events that give rise to the collecting duct system. In response to signals released from ureteric bud tips, mesenchymal cells transform into polarized epithelia, which differentiate into the tubule cells along the remainder of the nephron. Epithelial cell polarization relies on tight junctions, which connect tubule cells, provide paracellular barriers to ion and fluid movement, and organize the membrane into apical and basolateral domains (; ; ). Polarized expression of channels and transporters allows for vectorial transport of solutes along the nephron, resulting in physiological urine formation. Aberrations in epithelial cell polarity are implicated in the pathogenesis of renal cysts, renal fibrosis, and renal failure (; D.B. ); however, the molecular mechanisms underlying these complex diseases remain poorly understood.
Genetic studies of invertebrates have identified two conserved tight junction PDZ protein complexes, the Crumbs (CRB) and partition-defective-3 (PAR-3) complexes, which contribute to cell polarity (; ; ; ; ; ; ). CRB is a transmembrane protein (), whose C-terminus binds to the PDZ domain of PALS () or PATJ (). In turn, PALS and PATJ bind one another through heterodimerization of their L27 domains (). The CRB complex is an apical membrane determinant that stabilizes the apical membrane cytoskeleton through interaction with β-spectrin and D-moesin (). The second complex includes the PDZ protein PAR-3, which binds to atypical protein kinase C (aPKC) (), and the PDZ protein PAR-6 (; ). This PAR-3 complex excludes certain basolateral proteins from the apical domain via aPKC phosphorylation (; ). The CRB and PAR-3 polarity complexes were originally thought to act independently, but physical and functional interactions between these complexes establish cell polarity ().
Genetic analyses of invertebrates also identified several basolateral proteins required for epithelial cell polarization. Discs large () (), lethal giant larvae () (), and scribble () (; ) were originally identified as tumor suppressor genes in . Mutations in , , or disrupt epithelial cell polarity and cause neoplastic overgrowth of tissues. DLG genetically interacts with the multi-PDZ protein SCRIB and the WD40 motif protein LGL (). Interplay between these basolateral polarity proteins and the apical polarity complexes establish and maintain cellular polarity. For example, LGL is phosphorylated by aPKC, thereby excluding LGL from the apical compartment (; ).
The mammalian homologue of LIN-7 (MALS) can bind components from both the apical () and basolateral (; ) polarity complexes, suggesting an additional mode for interplay. Three MALS genes occur in mammals, and each is a small protein comprising an L27 and a PDZ domain (; ; ). The L27 domain links MALS to either the CRB () or DLG () complex, but a role for MALS in these polarity complexes is unknown. Biochemical and cell biological experiments suggest a variety of possible roles for MALS in mammalian epithelial cell lines. MALS localizes to the basolateral membrane of Madin-Darby canine kidney (MDCK) cells (; ). By analogy to the role of LIN-7 in , the PDZ domain of MALS may anchor receptors to the basolateral domain of MDCK cells (; ). Alternatively, MALS may stabilize PALS and mediate tight junction formation (). Other studies suggest that MALS interacts with β-catenin and mediates organization of adherens junctions (). Finally, MALS can regulate endocytosis or endosomal sorting in epithelial cell lines (; ). Despite these numerous suggestions, genetic analyses have yet to identify functions for MALS in mammalian epithelia.
We characterize knockout (−/−) mice and identify an essential role for MALS-3 in defining polarity of renal epithelial cells. mice have hypomorphic kidneys characterized by numerous cysts and fibrosis. These developmental defects owe to a loss of polarity specifically in epithelia derived from the metanephric mesenchyme. Immunoprecipitation proteomics analysis reveals MALS-3 binds to the CRB tight junction and DLG basolateral complexes. Biochemical studies show the L27 domain of MALS-3 assembles and stabilizes these complexes. In proximal tubule cells from MALS-3 mutant mice, the CRB complex is lost from tight junctions and DLG mislocalizes to tight junctions. These studies demonstrate that MALS-3 organizes these two discrete polarity complexes and that disruption of epithelial cell polarity can result in renal agenesis, cysts, and fibrosis.
Kidneys from neonatal and adult mice are noticeably smaller than those from heterozygous (+/−) littermates (), and ∼15% of display unilateral renal agenesis (unpublished data). Furthermore, kidneys are studded with numerous cysts (; and Fig. S1 A, available at ). Most cysts (8 of 11) fail to express tubule segment-specific markers, only 1 of 11 cysts stained with a lectin that labels proximal tubules (Fig. S1 B). Kidneys from are dramatically reduced in weight relative to (). Whereas the kidney is reduced by nearly 45%, only a ∼10% reduction in the size of heart, spleen, and overall body weight is observed. Renal hypoplasia occurs in mice backcrossed to either the 129/Sv or C57BL/6 strain and is specific to , as kidneys from mice lacking both MALS-1 and MALS-2 are normal in size ().
Microscopic analysis of adult and newborn kidneys reveals dramatic cellular abnormalities in kidney architecture (; Fig. S1 C). kidneys display marked tubulointerstitial changes including tubular dilatation and dedifferentiation/simplification (). Trichrome staining reveals interstitial fibrosis (; Fig. S1 D), a condition characterized by the accumulation of extracellular matrix proteins and fibroblasts, which is a hallmark of end-stage renal disease (; J.M. ). Renal cysts and fibrosis in are accompanied by a loss of epithelial cell polarity. The Na/K ATPase, which is normally localized basolaterally, becomes diffusely localized in dedifferentiated tubular epithelia ().
mice also manifest defects in renal function (). Compared with , mice exhibit a considerable increase in urine output and sodium excretion. Creatine clearance, a measure of glomerular function, was modestly decreased whereas blood urea nitrogen (BUN) was elevated (). This is likely a consequence of extracellular fluid volume contraction resulting from the polyuria. Urine concentrating ability, measured as the response to 12-h water deprivation and vasopressin injection after a high water load, was also impaired; urine osmolarity only increased 3 ± 0.9-fold (1,024 ± 490 mOsm/l change) in mice compared with 7.2 ± 0.9-fold (2,764 mOsm/l + 332 mOsm/l) in (Fig. S1 E). Together with the histopathological observations, these results are reminiscent of nephronophthsis, the most common genetic disorder of progressive renal failure in children.
To determine the pathogenesis of these kidney defects, the cellular distribution of MALS-3 protein was examined in kidney sections. MALS-3 is present in epithelial cells of renal tubules but is absent from glomeruli (). In the kidney cortex, MALS-3 occurs in collecting duct epithelia that label for aquaporin-2 () and proximal tubule epithelia that stain intensely for phalloidin (). In proximal tubule cells, MALS-3 localizes to the basolateral membrane and the tight junction (; Fig. S2 A, available at ). In the inner medulla, MALS-3 is present in tubule cells of the loop of Henle and collecting duct (). In collecting duct epithelia, MALS-3 is restricted to the basolateral membrane and is not detected at the tight junction (). As expected, MALS-3 staining is absent in mice (; and Fig. S2 A).
To define the proteins associated with MALS-3, we conducted immunoprecipitation proteomics. Immunoprecipitation of MALS-3 from adult kidney homogenates shows a series of specific protein bands that are absent from immunoprecipitates (). As determined by mass spectroscopy, MALS-3 interacts with several proteins found in complexes that determine cell polarity. MALS-3 immunoprecipitates contain DLG and CASK, proteins that localize to the basolateral surface, and PALS-1, PALS-2, PALS-4 (Mpp7), and PATJ, proteins that localize to tight junctions. Western blotting confirmed all of these interactions and demonstrated coimmunoprecipitation of CRB-3, the prototypical tight junction polarity protein, with similar recovery ().
To characterize the role of MALS-3 in these polarity complexes, we first assessed protein levels in kidneys. Strikingly, levels of all components of the CRB-3 tight junction complex are drastically diminished in kidneys (). CRB-3 itself is decreased by 60% and PALS-1 and PATJ are each decreased by ∼80% (). In contrast, aPKCζ and PAR-3, members of a distinct tight junction polarity complex, were unchanged in . Levels of basolateral proteins, LGL, CASK, and DLG are also significantly decreased (). Structural proteins of the tight junction, claudin-7 and -8, are unaffected in , while ZO-2 is modestly decreased. E-cadherin and β-catenin of the adherens junctions are also unchanged (). MALS-2 protein is up-regulated in kidneys of , which has also been noted in brains of MALS mutants ().
Through L27 domain interactions, MALS-3 associates directly with CASK and PALS (; ; ), which in turn bind to DLG and PATJ (; ), respectively. Interestingly, levels of DLG and PATJ, which do not directly bind to MALS-3, are reduced to a greater or similar extent as CASK and PALS (). This suggests that MALS-3 may mediate cooperative assembly of these L27 domain complexes. To explore this biochemically, the L27 domains of CASK and DLG were expressed in the absence of the MALS L27 domain in bacteria. Under these conditions, the CASK/DLG binary complex is substantially degraded upon isolation and further deteriorated within 24 h at 4°C (). Inclusion of the MALS L27 domain promotes formation of a MALS/CASK/DLG L27 complex that is stable for several days at 35°C (). Furthermore, the MALS/CASK/DLG ternary complex was more resistant to urea denaturation than either the MALS/CASK or DLG/CASK binary complexes (). MALS also stabilized the MALS/PALS/PATJ L27 complex (unpublished data). The oligomeric nature of protein complexes containing four L27 domains has not been determined. NMR showed that the isolated CASK and DLG L27 domains form a “dimer of dimers” structure (), which might imply high order oligomerization of L27 domain complexes containing three proteins. However, gel filtration analysis shows that the MALS/CASK/DLG complex elutes as a single peak corresponding to ∼38 kD (), implying a 1:1:1 stoichiometry. Similarly, the tandem PALS L27 domains elute with MALS L27 and PATJ L27 as a single peak corresponding to a 1:1:1 stoichiometry (unpublished data).
We next asked whether loss of MALS-3 affects the apico-basal polarization of its interaction partners. PALS, PATJ, and CRB-3 are enriched at tight junctions of proximal tubule epithelia (). Consistent with Western blotting results, PALS and PATJ staining is lost from tight junctions in (); CRB-3 staining is reduced, and the remaining protein concentrates abnormally in punctate vesicles in the subapical region (). These major defects are not observed in collecting duct. In collecting duct epithelia, PALS and PATJ staining is reduced, but not lost, at the tight junction of the knockout, and CRB-3 localization appears normal (Fig. S3, A–D; available at ). Molecular compensation does not account for the localization of the CRB complex in collecting duct epithelia, as MALS-1 and - 2 are not detected (Fig. S3 E).
Loss of MALS-3 specifically disrupts the CRB complex. In proximal tubule and collecting duct epithelia, aPKCζ, PAR-3, and ZO-1 remain at the tight junction (Fig. S3 H, Fig. S4, A, B, and D; and unpublished data). Other junction-associated proteins are also unaltered in renal epithelia from (Fig. S4, E and F).
The distribution DLG undergoes complex changes in (). Normally, DLG is predominantly distributed basolaterally in proximal tubule epithelia () and in more distal tubular segments of the nephron (; Fig. S3 F). Weaker tight junction and cytoplasmic staining is also observed in control animals (). In , DLG is lost from the basolateral surface and becomes abnormally concentrated at tight junctions of epithelia in proximal tubules (; and Fig. S2 B) and in more distal tubules (; Fig. S3 F). Other basolateral polarity proteins, scribble and LGL, remain concentrated along the basolateral membrane (Fig. S4 C, Fig. S3 G; and unpublished data).
The resemblance of the null phenotype to human nephronophthisis raises the possibility that MALS-3 may be involved in a common pathway with the six known nephronophthisis genes (NPHP1–6), which affect the organization and/or development of primary cilia (). Consistent with this idea, CRB-3 localizes to cilia and lacking , a CRB-3 homologue, exhibit renal hypoplasia, cysts, and tubular dilatation. Renal abnormalities in mutants are associated with malformation and dysfunction of primary cilia (). A conserved role for CRB-3 in cilia formation in mammalian epithelia has been demonstrated (). Although CRB-3 is lost from the tight junction in renal epithelia lacking MALS-3, we found that CRB-3 localizes properly to primary cilia (), suggesting an alternative mechanism for targeting CRB-3 to cilia. Consistent with this hypothesis, PALS and PATJ are not detected in cilia (unpublished data) and the morphology of primary cilia appears unaltered in mice (). Analysis of EST databases () revealed an alternatively spliced variant of CRB-3 (). The resulting protein product, CRB-3b, shares identical extracellular and transmembrane domains with canonical CRB-3, but has a divergent cytoplasmic tail. Importantly, the splicing variant results in a longer C-terminal tail that lacks the PDZ ligand () required for PALS/PATJ interaction (; ). Interestingly, the CRB-3 antibody recognizes a doublet in kidney homogenates, with the larger product, likely CRB-3b, being more moderately reduced in (). Normalcy of ciliary structure and CRB localization do not exclude that more specific aspects of ciliary function are compromised in these animals ().
Dysfunction of primary cilia and associated centrosomes can impair cell proliferation () and thereby reduce kidney size in mutants. To determine whether cell proliferation is altered in mice, mitotic cells in kidneys from E14.5 embryos were labeled using anti-phospho-histone3 antibody. We found no difference in cell proliferation between (0.388 ± 0.013) and (0.375 ± 0.013; P = 0.61) littermates (), excluding reduced cell division as the cause for the smaller kidney size.
Next, we examined whether changes in apoptosis/cell death contribute to this phenotype. Apoptotic cells are found in nephrogenic regions () of E14.5 kidneys from both control and littermates (). However, abnormal cell death was detected in cells underlying the renal capsule in kidneys (), resulting in a 10-fold increase in apoptotic signal from (5.80 ± 0.76) as compared with kidneys (0.42 ± 0.19; P = 0.007). These changes are detected as early as E12.5 (unpublished data), indicating that increased apoptosis causes reduced kidney size in mice.
The profound disruption of the CRB-3 complex in proximal tubule cells and the normalcy of PALS and CRB-3 localization in the collecting duct suggest that MALS-3 serves distinct functions in these two epithelial cell types. Proximal tubule and collecting duct cells derive from different tissue types (). During embryonic development, the ureteric bud extends from the Wolffian duct, invades the metanephric mesenchyme, and induces mesenchymal to epithelial cell transition. Tubules derived from metanephric mesenchyme extend from the glomerulus to the distal tubule, whereas the collecting duct derives from the ureteric bud ().
To assess the role of MALS-3 in each of these embryonic tissues, a new allele was created (Fig. S5, A–C, available at ). Breeding these mice to transgenics that express cre recombinase under the control of the HoxB7 or Pax3 promoter () allows deletion of specifically from ureteric bud () [UBKO] () or metanephric mesenchyme () [MMKO] (; ) (Fig. S5, D–F).
Levels of MALS-3 are differentially reduced in UBKO and MMKO mice (; Fig. S5, D–F). Importantly, gross abnormalities in kidney structure are only noted in the MMKO mice. Kidneys from MMKO mice are reduced in size () with renal pathologies identical to mice (). These abnormalities are not observed in the UBKO mice (). Finally, in MMKO proximal tubules, but not UBKO, the CRB-3 and DLG complexes are affected ( and unpublished data) in a manner similar to .
Kidneys from adult MMKO mice phenocopy those of mice, indicating a shared developmental defect. Like the , kidneys from MMKO embryos (E14.5) show extensive apoptosis underlying the renal capsule (). To determine whether defects in the CRB-3 and DLG polarity complexes underlie these developmental abnormalities, we examined the expression levels of these protein complexes in kidneys from MMKO embryos. Western blotting showed that both the CRB and DLG complexes were dramatically reduced (). These results show that MALS-3 is a critical component of the CRB and DLG complexes and that disruption of these complexes in the metanephric mesenchyme is associated with the renal defects in mice.
#text
Isoform specific and pan antibodies against MALS-1, -2, and -3 were generated in rabbits as described previously (). Anti-SAP-97 antibody was generated in rabbit (); for immunohistochemistry an anti-SAP-97 antibody from Morgan Sheng (Massachusetts Institute of Technology, Cambridge, MA) was used. Rabbit anti-JAMS-A and anti-claudin antibodies were purchased from Invitrogen. PALS, PATJ, and Crumbs-3 antibodies were gifts from Ben Margolis (University of Michigan, Ann Arbor, MI) and rabbit anti-LGL antibody was provided by Valeri Vasioukhin (Fred Hutchinson Cancer Research Center, Seattle, WA). Rabbit anti-PAX2 was from BAbCO. Goat anti-aquaporin-2 antibody and rabbit anti-aPKCζ were from Santa Cruz Biotechnology, Inc. Rabbit anti-PAR-3 and anti-phospho-histone3 antibodies were from Upstate Biotechnology and goat anti-scribble antibody was from GeneTex, Inc. Mouse anti-CASK, anti-ZO-2, anti-β-catenin, anti-E-cadherin, and anti-N-cadherin were all purchased from Transduction Laboratories. Alexa-conjugated Phalloidin and secondary antibodies were purchased from Molecular Probes and mouse anti-pancytokeratin was from Sigma-Aldrich.
The following mouse lines were used in this study: STOCK Gt(ROSA)26Sortm1(Smo/EYFP)Amc/J, STOCK Tg(HoxB7-cre)13Amc/J, 129S4/SvJaeSor-Gt(ROSA)26Sortm1(FLP1)Dym/J, C57BL/6, and 129/Sv were purchased from The Jackson Laboratory. Pax3-Pro-Cre mice were a gift from Jonathan Epstein (University of Pennsylvania, Philadelphia, PA). MALS-1, -2, and -3 have been described previously (; ). For characterization of mice lacking MALS-1 and -2, double-knockout mice were compared with wild-type mice of a similar age and genetic background. Generation of mice is described below.
Isolation of MALS-3 genomic DNA (BAC clone, mCG15974) has been described previously (). For construction of the targeting vector, a 1.9-kb region encoding the targeted exons (the fourth and fifth exon) of MALS-3 was PCR-amplified, digested with ClaI, and subcloned into the ClaI site of pks2loxPFRTNT (a gift from Shinya Yamanaka, Kyoto University, Kyoto, Japan). A 1-kb region downstream from the targeted exons was PCR-amplified, digested with BamHI and EcoRI, and subcloned into the BamHI-EcoRI sites of pks2loxPFRTNT. Finally, a 5.8-kb genomic region upstream to the targeted exons was PCR-amplified, digested with KpnI and BamHI, and inserted into the KpnI-SalI sites of the pks2loxPFRTNT vector. Two loxP sites flank targeted exons and a neomycin cassette. Two FRT sites flank the neomycin cassette for its removal.
The targeting vector was linearized and electroporated into R-1 ES cells. Clones resistant to G418 and gancyclovir were analyzed for recombination by PCR. To ensure proper homologous recombination, PCR-positive clones were further analyzed by Southern blotting using probes containing genomic sequences outside of the targeting vector and with a neo probe. Properly targeted clones were injected into blastocysts from C57BL6 mice and transferred to surrogate mothers (Transgenic Facility, Stanford University, Stanford, CA). Male chimeras were mated with 129S4/SvJaeSor-Gt(ROSA)26Sortm1(FLP1)Dym/J (The Jackson Laboratory) females for transmission of the mutated allele through the germ line and for removal of the neomycin cassette from the targeted allele. Genotypes for mice lacking the neomycin cassette were determined by Southern blotting or by PCR using the primers: 5′-GAAAATGCTTCTGTCCGTTTGC-3′ and 5′-ATTGCTGTCACTTGGTCGTGGC-3′, which yields a 280-bp product for the wild-type allele and a 350-bp product for the targeted allele. Presence of the cre-recombinase and EYFP transgenes was determined using the following primer pairs: 5′-GAAAATGCTTCTGTCCGTTTGC-3′ and 5′-ATTGCTGTCACTTGGTCTGGC-3′ for cre and 5′-CCCTGAAGTTCATCTGCACCACC-3′ and 5′-GGACTTGTACAGCTCGTCCATGCC-3′ for EYFP.
Genes corresponding to the L27 domain of rat DLG (L27, residues 1–65), the L27 domain of mouse MALS (L27, residues 2–78), and the tandem L27 domain of rat CASK (L27NL27C, residues 329–460) were PCR amplified from the respective full-length cDNAs. The single-chain fusion protein, containing L27, L27NL27C, and L27 connected with a thrombin-cleavable segment (Leu-Val-Pro-Arg-Gly-Ser-Ser-Gly), was cloned into a modified version of the pET32a vector in which the S-tag and the thrombin recognition site were replaced by a sequence encoding a protease 3C cleavage site (Leu-Glu-Val-Leu-Phe-Gln-Gly-Pro). Similarly, the single chain fusion protein, containing L27 and L27NL27C connected with a thrombin-cleavable segment, was PCR-amplified and inserted into the modified pET32a vector. Bacterial cells harboring the fusion protein expression plasmid were grown at 37°C, and protein expression was induced by IPTG at the same temperature for 3 h. The His-tagged, thioredoxin-containing protein was purified under native conditions using Ni-NTA agarose (QIAGEN) affinity chromatography. After protease 3C digestion the N-terminal His-tag and thioredoxin were removed by passing the digestion mixture through an S-200 gel filtration column, then the single-chain protein was digested by thrombin and the L27 domain complex protein was further passed through the same size-exclusion column. The expression and purification of L27/L27C and L27N/L27 complex proteins were described previously (, ).
Adult mice were anesthetized and perfused with 2% paraformaldehyde in PBS. Kidneys were removed, immersed in the same fixative for 2 h at 4°C, and then cryoprotected in PBS containing 30% sucrose overnight at 4°C. Frozen sections (10 μm) were rehydrated with PBS containing 0.1% Triton X-100 (PBS-X) for 20 min and incubated for 1 h in blocking solution (PBS-X containing 1% BSA). Primary and secondary antibodies were diluted in blocking solution and incubated overnight at 4°C and 2 h at room temperature, respectively. Antibody incubations were followed by three washes in PBS-X. Sections were mounted with coverslips in Flouromount-G (Southern Biotechnology Associates, Inc.). Images were taken with a confocal microscope (LSM 5 Pascal Axioplan 2; Carl Zeiss MicroImaging, Inc.) under Zeiss 63× 1.4 or 40× 1.3 oil-immersion objectives or a Zeiss 10× 0.3 objective at room temperature and acquired using LSM 5 Pascal (v.3.2) software. Images were processed using Adobe Photoshop. To study primary molecular defects in mice, only kidneys that showed normal arrangement of tubules with mild anatomical abnormalities were used for localizing aPKC/Crumbs/DLG complexes. More than three sections from six KO and six control littermates were examined. Proximal tubules were identified by phalloidin staining of apical brush border, whereas collecting ducts were stained using aquaporin-2 antibody.
Mice were housed in metabolic cages on a 12-h light-dark cycle and fed ad libitum on normal chow. After 24 h, urine was collected under oil, animals were anesthetized with Inactin (100 mg/kg, i.p.), and blood was collected by cardiac puncture. Blood and urine chemistries were measured by automated methods (IDEXX, West Sacramento, CA). For measurements of urine diluting and concentrating ability, urine osmolarity was measured using a vapor pressure osmometer (Wescor) after animals were fed a high water content diet (, ) for 48 h and then again after mice (12 h) were injected with deamino-Cys, D-Arg9-vasopressin (DDAVP, 1 μg/Kg, i.p.) and feeding water was withdrawn.
Adult mouse kidneys from or mice were homogenized in three volumes of STE buffer (320 mM sucrose, 20 mM Tris, pH 7.4, and 2 mM EDTA) containing 10 μg/ml leupeptin, 10 μg/ml aprotinin and 200 μg/ml PMSF. Homogenates were centrifuged at 20,000 for 1 h and pellets were resuspended in TET buffer (20 mM Tris, pH 8.0), 1 mM EDTA, and 1.3% Triton X-100) containing 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 50 μg/ml PMSF. Lysates were pelleted at 100,000 for 1 h. Precleared lysates were immunoprecipitated with 5 μg of MALS-3 antibody or control rabbit IgG overnight at 4°C. To collect immunoprecipitated protein complexes, 80 μl of a 50% protein A–Sepharose slurry was added to the lysates and incubated for 1 h at 4°C. Immunoprecipitates were washed extensively and loaded onto SDS-PAGE to separate the proteins. Gels were either silver stained or transferred to nitrocellulose for Western blotting.
Gel bands were reduced with 10 mM dithiothreitol (DTT) followed by alkylation with 55 mM iodoacetamide. Proteins were digested with trypsin and extracted with a 50% acetonitrile/5% formic acid solution. The peptides were dried down and resuspended in 0.1% formic acid then separated via HPLC using a 75 μM × 15 cm reverse-phase C-18 column (LC Packings) running a 3–32% acetonitrile gradient in 0.1% formic acid on an Agilent 1100 series HPLC. The LC eluent was coupled to a micro-ionspray source attached to a QSTAR Pulsar mass spectrometer (MDS Sciex). Peptides were analyzed in positive ion mode.
Size-exclusion chromatography was performed on an AKTA FPLC system using a Superose 12 10/30 column (GE Healthcare). Protein samples were dissolved in 100 mM potassium phosphate buffer containing 1 mM DTT. The column was calibrated with the low molecular mass column calibration kit from GE Healthcare.
Urea denaturation of L27 domain complexes was monitored by acquiring circular dichroism spectra of protein samples at each urea concentration. Data were collected on a JASCO J-720 spectropolarimeter at room temperature. The sample contained 15 μM protein in buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 1 mM DTT).
To detect dying cells, LysoTracker (Invitrogen) labeling of embryonic kidneys was performed as described previously (). In brief, isolated kidneys (six kidneys each for control and ) were incubated with 5 μl/ml LysoTracker solution in HBSS at 37°C for 30 min, rinsed in HBSS, fixed in 4% PFA in PBS, and stored at −20°C in 100% methanol. After rehydration into PBS, E14.5 kidneys were sectioned (100 μm), permeabilized in PBS contaning 0.1% Tween 20, and stained with anti-cytokeratin (Sigma-Aldrich). After extensive washing, cytokeratin antibody was detected using Alexa488-conjugated donkey anti–mouse antibody (Invitrogen) and nuclei were labeled with DAPI. All images were taken with a confocal microscope (LSM 5 Pascal Axioplan 2; Carl Zeiss MicroImaging, Inc.) under Zeiss 25× 0.8 objective (zoom setting 1 or 2) at room temperature and acquired using LSM 5 Pascal (v.3.2) software. Total fluorescent intensity was analyzed using MetaMorph software and images were processed using Adobe Photoshop. The average intensity of three individual optical sections from each stack was calculated. Five stacks of individual kidneys were analyzed for each condition. For apoptosis in E12.5 whole mounts, kidneys were rehydrated into PBS containing Tween 20 (0.1%) and counterstained with anti-PAX2 antibody. The rest of the staining procedure was performed as described above. To detect cells in mitosis, isolated kidneys (13 sections from 5 control kidneys and 14 from 5 kidneys) were fixed in 4% PFA, washed, sectioned (100 μm), and incubated with rabbit anti-phospho-histone H3 antibody (1:200; Upstate Biotechnology) and co-stained with anti-cytokeratin as above. Alexa546-conjugated donkey anti–rabbit antibody was used for labeling phospho-histone antibody. All images were obtained with a confocal microscope and >20 optical sections were analyzed for each kidney.
Figure S1 shows renal cysts in kidney sections from adult and embryonic mice as well as urine concentrating ability in mutant mice. Figure S2 shows the colocalization of MALS and DLG with the tight junction. Figure S3 shows partial disruption of polarity complexes in collecting ducts of . Figure S4 shows that localization of PAR-3, aPKC, and SCRIB does not require MALS-3. Figure S5 depicts the generation and characterization of the allele. Online supplemental material is available at . |
Both of my parents and others in my family were involved in science—they were doctors and engineers. But that career option was not a possibility for me when I was growing up in China. That was during the Cultural Revolution, so essentially all the universities were closed. I was considering whether I wanted to be a farmer or a factory worker; those were the options.
But I was a good student, and I had a high school teacher, Mr. Lu, who believed that I should go to college somehow. Normally, someone from a family like mine—intellectuals—would have been sent to a remote farm to work, but he helped me stay in Shanghai, where I worked in a factory on heavy machines.
Fortunately, at that time, the Cultural Revolution was ending, and China reinstituted college entry exams. Mr. Lu asked me to study the old high school curriculum, from before the Cultural Revolution, because my education had been really basic and elementary. You know, every semester English started with, “Long live Chairman Mao” [].
So I studied for four months really, really hard. They estimated that probably more than 100,000 people took the exam, and I actually scored the highest. And I went to Fudan University in Shanghai.
When I needed to decide what to study, my grandfather, who was a professor of organic chemistry, told me, “Study biology, it's good for girls. And chemistry is also very interesting.” He said the combination of chemistry and biology, meaning biochemistry, would be the thing of the future.
That was also a coincidence, because when I was in college, it wasn't really possible to leave the country to study. But my mother was a professor at a medical school, and she told me that she heard rumors that the best student in her college would be sent abroad to study. So I applied to graduate school in her medical college, and I again scored best on the entrance exam.
That qualified me to take another national test sponsored by Harvard. They had organized a program to bring Chinese students to top graduate schools in America. That's how I got to Harvard.
One of the courses I took at Harvard was called Neurological Diseases, and they actually brought in patients who had diseases like Alzheimer's and Huntington's. That left a really deep impression on me. Even today, I can remember exactly what that Huntington's patient looked like.
I also took a class on developmental neurobiology. At the time, classical work done by Ron Oppenheim and others had shown that up to 50% of neurons die during development. I was intrigued that cell death, something associated with disease, was also associated with a beautiful thing like development.
At the end of the courses, I concluded two things. The first was that all of these neurodegenerative diseases are caused by the selective death of specific populations of neurons that should not have died. I thought it was striking that early in embryonic development you can have very orderly, selective neuronal death. And later on in life, you can also have very selective neuronal deaths. It was a first-year student's fantasy, but I thought, “There may be some similarities between them.”
I even talked to some professors about it. But their response was always negative, because nobody believed that cell death was something interesting or regulated. They just thought neurons died in Alzheimer's because they were poisoned, and neurons died in development because they were starved to death by a lack of growth factors.
The second thing I concluded was that, although we were hearing lectures from eminent physicians, they really didn't know anything about these diseases in terms of mechanism. All they could tell the patient was which way they were going to die. And if I were the patient, I really wouldn't care which way it was; they were all pretty bad. That made me interested in cell death as a biological mechanism.
Nobody at Harvard then was studying cell death. So I went to our graduate director, Ed Kravitz, and said, “I cannot find a lab that I want to work in here.” Thinking back, anybody else would probably have thought I was so arrogant—I'd just recently arrived from China, and I definitely didn't speak English well. “What do you mean? None of the hundreds of labs here would be good enough for you?” Fortunately, Ed was really nice to students, and he said, “Oh, if you can't find a lab here, you can work in Cambridge,” meaning on the Harvard main campus because I was at the medical school, “or you can even go to MIT.”
That made me really happy because I had just heard Bob Horwitz from MIT talk about cell death in . He believed that there were cells that were determined to die, just like there are cells determined to be a neuron. It was striking to me that his and mutants had such specific defects in cell death, and nothing else seemed to be wrong with them. I thought those were the most interesting mutants I'd ever heard of. So I ended up in Bob Horwitz's lab, but with a degree from Harvard.
From all this work, I've realized that even though apoptosis and caspases are contributing to many neurodegenerative diseases—Aβ can induce apoptosis in Alzheimer's, PolyQ can induce apoptosis in Huntington's—for most of these diseases, inhibiting cell death will be too late for the patient. The disease actually starts with neuronal dysfunction caused by these proteins. In many cases, the mutant proteins manifest as an accumulation of misfolded proteins. So I think it's very important to learn how cells normally remove misfolded proteins.
Maybe the best example is Huntington's disease. The penetrance is 100% when they have PolyQ repeats longer than 40-something. The key, though, is why does this person live fine for 40 or 50 years, and then all of a sudden have Huntington's disease? It's not because these patients don't express PolyQ proteins when they're young. It's pretty much always around.
So we're hoping to understand what tips the threshold. Because if we can push that back another 10 or 20 years, it will mean a cure for many of these people.
I don't want to give away too much of our progress yet. But the key I think is accumulation. And what leads to accumulation must be the degradation mechanism—it's somehow defective.
I think they are very good. What it does most for me is really free me from writing grants all the time instead of doing more productive work. I can follow the work in the lab better, I can really think more in depth about the projects. It gives me more time with my students and postdocs.
The Pioneer people are looking for someone to do things that are different from what they originally studied, but the hope is to spark creativities in the differences. The other thing they look for is a good track record of creativity, a willingness to stick your neck out and do something different. That pushed me toward working on a subject that I honestly wouldn't have worked on otherwise. |
DNA double-strand breaks (DSBs) can be accidentally introduced in cells by the action of ionizing radiation or certain reactive radicals. These agents have the ability to initiate a series of chemical reactions that ultimately sever the DNA backbone, resulting in chromosome breakage and fragmentation of genes (). Because such corruption of genetic material inevitably leads to problems with replication and cell division, it is of the utmost importance that cells have a mechanism to counteract DSBs. In addition, DSBs are generated in developing B and T cells during normal V(D)J recombination, implying that a working DSB repair system is not only necessary for an effective defense against DNA-modifying agents but also for a functional immune system in higher organisms (for review see ). As a result, two highly efficient DSB repair pathways have evolved in eukaryotic cells: homologous recombination (HR) and nonhomologous end joining (NHEJ).
The HR process mediates DSB repair by using a homologous DNA sequence as a template to guide proper restoration of the break. Because homologous templates are found on sister chromosomes, HR is thought to be active during the S and G2 cell cycle phases. Contrasting, NHEJ is characterized by its ability to directly ligate the two ends of the broken DNA molecule. This process does not have the need for a homologous template and is therefore theoretically not restricted to a certain phase of the cell cycle.
The NHEJ pathway utilizes several enzymes that capture both DNA ends, bring them together in a synaptic complex, and facilitate direct ligation of the DNA break (). The process is initiated by the association of DNA ends with the Ku 70/80 heterodimer, a protein with a ring-shaped structure that displays an extraordinary affinity for open DNA ends (). The DNA–Ku 70/80 complex then functions as a scaffold to assemble the other key NHEJ proteins at the DNA termini.
One of the first enzymes to be attracted to the DNA–Ku 70/80 scaffold is a large, 460-kD serine/threonine kinase called DNA-dependent protein kinase catalytic subunit (DNA-PK). The protein complex that is formed after the association of both Ku 70/80 and DNA-PK at the DNA ends is generally referred to as DNA-dependent protein kinase (DNA-PK). The large DNA-PK molecule forms a distinct structure at the DNA termini, which is likely to play an active role in the formation of a synaptic complex that holds the two ends of the broken DNA molecule together (; for review see ).
The positioning of the two DNA ends in a synaptic complex sets the stage for religation of the broken DNA molecule. Before this can take place effectively, it is necessary that noncomplementary ends are processed. In the case of single-stranded overhangs, DNA termini can be made ligatable by either filling of the missing nucleotides or by resection of the overhang. Polymerases μ and λ, terminal deoxynucleotidyltransferase, polynucleotide kinase, and several nucleases have been shown to play a role in this processing. The best-characterized processing nuclease is the endonuclease Artemis, whose activities include the removal of single-strand overhangs (). After correct processing, the two tethered DNA termini can finally be ligated. Ligation is mediated by the ligase IV–XRCC4 complex, possibly in conjunction with the recently discovered XLF (XRCC4-like factor)/Cernunnos protein (), although at this point, little is known about the mechanism by which XLF stimulates ligation.
After the introduction of a DSB, the DNA-PK enzyme is quickly recruited to the DNA–Ku scaffold. The serine/threonine kinase activity of DNA-PK is not activated until the DNA-PK molecule is associated with both the Ku 70/80 heterodimer and a DNA terminus. This latter requirement makes the DNA-PK protein kinase truly DNA dependent. Many targets for the DNA-PK kinase have been identified in vitro, including XRCC4, Ku 70/80, Artemis, p53, and even DNA-PK itself (autophosphorylation). This plethora of possible targets suggests that DNA-PK may be involved in several aspects of the NHEJ process, including activation of components of the synaptic complex and signal transduction to cell cycle regulators (). Surprisingly, however, little evidence has been found for the biological relevance of these activities. To the contrary, much more convincing data are available that argue for a regulatory role for DNA-PK at the synapsis of the two DNA ends during NHEJ.
Information on the shape and conformation of the DNA–Ku–DNA-PK complex is scarce and provides little detail as a result of technical difficulties in the production of DNA-PK crystals for crystalographic structure studies. The best available information on the three-dimensional structure of the DNA–Ku–DNA-PK heterotrimer was obtained by single- particle electron microscopy at a 25-Å resolution (). These images clearly showed dimeric structures in which DNA ends were brought into close proximity by the formation of a synaptic complex that consisted of two DNA ends, two Ku 70/80 molecules, and two DNA-PK molecules (). In addition, both atomic force microscopy studies and biochemical experiments have shown that DNA-PK not only assembles at DNA ends but also facilitates the tethering of multiple DNA ends (for review see ). Collectively, these findings suggest that DNA-PK is responsible for the formation of a synaptic bridge between the two termini of the broken DNA molecule.
It has been well established that the presence of the large DNA-PK molecule at DNA ends effectively blocks access of either processing nucleases or ligases (; ). Although this capping of DNA ends may have an important function in protection of the DNA termini against degradation or premature and incorrect ligation, it is clear that the DNA-PK cap has to be removed or altered before religation of the DNA ends and repair of the DSB can take place. Several authors have demonstrated that autophosphorylation of DNA-PK results in release of the cap and accessibility of the DNA termini for either processing enzymes or ligases (; ; ). Recent research has demonstrated that DNA-PK autophosphorylation can occur in trans during the synapsis of two DNA-bound DNA-PK molecules (). These findings clearly argue for a model in which DNA-PK protects the termini of a broken DNA molecule by capping it until the DNA ends find each other and are properly aligned in a synaptic complex. DNA-PK trans-autophosphorylation then liberates the DNA ends for proper processing and ligation by other NHEJ factors like Artemis () and ligase IV–XRCC4 (). In this model, DNA-PK functions as a gatekeeper of the NHEJ process, regulating access to the DNA ends by autophosphorylation ().
At present, it is not clear how we should envision the conformational changes in the DNA–DNA-PK complex that take place during autophosphorylation. It is also not unequivocally known which autophosphorylation sites need to be phosphorylated to facilitate the accessibility of DNA termini. Thus far, 16 amino acid residues that can be phosphorylated by the action of the DNA-PK kinase have been identified within the DNA-PK molecule (). The importance of DNA-PK autophosphorylation during NHEJ is demonstrated by numerous observations, showing that mutation of phosphorylation sites causes increased radiosensitivity and less efficient DSB repair (; ; ; ) and that an active DNA-PK kinase is required for NHEJ ().
Inhibiting phosphorylation of the entire 2609 cluster () leads to a severe increase in radiosensitivity and diminished processing of DNA ends, whereas mutation of single residues within this cluster has a less severe (but still present) effect on radiosensitivity (; ). In addition, we found that mutation of either the DNA-PK kinase domain or the 2609 cluster and the 2056 residue results in a rigid binding of DNA-PK to DNA ends in vivo, which most likely interferes with the NHEJ process (). Collectively, these results demonstrate that autophosphorylation of the 2609 cluster influences the DNA–DNA-PK interaction in a manner that facilitates end joining.
In contrast, the 2056 cluster () has been reported to inhibit DNA end processing upon phosphorylation, suggesting that the 2609 and 2056 clusters work in an opposite direction (). The latter finding has at present not been confirmed, and it is not entirely transparent how this double regulation mechanism cooperates during DSB repair.
Phosphorylation of the recently discovered 3950 residue, which is located in the C-terminal kinase domain, most likely plays a role in regulation of the DNA-PK kinase activity. Mutation of this site with phosphomimic aspartic acid results in deficient V(D)J recombination and increased radiation sensitivity (). This finding suggests that the 3950 residue may be involved in the regulation of DNA-PK autophosphorylation by mediating kinase activity. Three other autophosphorylation sites have been identified in the C-terminal region of DNA-PK (3821, 4026, and 4102), although no in vivo functionality has been reported for these sites at present ().
Recent experiments have shown that in vivo phosphorylation of the DNA-PK 2609 and 2647 residues can still occur at DSB sites in the absence of DNA-PK kinase activity (; ). Clearly, in these cases, the DNA-PK molecule is not responsible for the phosphorylation events, and another kinase (or kinases) must be involved. These studies raise doubts as to whether all phosphorylation sites within the DNA-PK molecule are genuine autophosphorylation sites or whether they can be targets for other kinases as well.
One study reports that phosphorylation of the 2609 and 2647 residues (but not the 2056 residue) of DNA-PK is dependent on the ataxia telangiectasia mutated (ATM) protein (). ATM is a damage-responsive protein kinase and a member of the phosphatidylinositol 3-kinase–like kinase group (PIKK), to which DNA-PK also belongs. Another study finds no absolute requirement for the presence of ATM but does acknowledge that several DNA-PK residues can be phosphorylated in the absence of a functional DNA-PK kinase, leaving open the possibility that ATM contributes to the phosphorylation of DNA-PK residues in vivo (). A third study suggests that yet another member of the PIKK group, ataxia telangiectasia related protein (ATR), is involved in the regulation of DNA-PK phosphorylation at the 2609 and 2647 residues after the onset of UV damage (). At present, it is not clear whether the ATR-mediated phosphorylation of DNA-PK is relevant for DSB repair via the NHEJ pathway.
At this point, it is not known whether ATM (or ATR)- mediated phosphorylation of the DNA-PK 2609 and 2647 residues is redundant with autophosphorylation, but it is tempting to speculate that both routes of DNA-PK phosphorylation have their own biological significance. ATM is activated at an early stage after the introduction of a DSB, is present at DSB sites, and is involved in signal transduction and phosphorylation of a plethora of proteins that mediate cell cycle arrest during DSB repair (). It is not unlikely that the ATM kinase, with its central role in the regulation of DSB repair and its physical presence at DSB sites, would be involved in regulation of the DNA-PK phosphorylation status and, thus, in regulation of the accessibility of DNA ends during the repair process.
It is becoming increasingly clear that DNA-PK plays a central role during the NHEJ process. This enzyme not only captures and tethers the two ends of a broken DNA molecule but also regulates the access of modifying enzymes and ligases to the DNA termini. A body of evidence supports a model in which regulation of the accessibility of DNA ends is mediated by autophosphorylation of DNA-PK in trans over the synaptic cleft between the two tethered DNA molecules ().
The latest findings that suggest the involvement of ATM in the phosphorylation of at least two DNA-PK residues open a new perspective on the regulation of DNA-PK in the synaptic complex. The codependence of DNA-PK phosphorylation on ATM would ensure that at least two signals need to be present before DNA-PK alters its conformation at the DNA ends: phosphorylation of the ATM-responsive sites within the DNA-PK molecule and autophosphorylation in trans over the synaptic cleft. Such a mechanism would tightly control DNA-PK activity, not allowing access to the DNA termini until two criteria have been met: a defined DNA damage response (as indicated by activation of the ATM kinase) and a proper alignment of the two DNA ends. Until these criteria are met, the DNA termini will be protected by the DNA-PK molecule: a lock with multiple keys. |
Mitotic chromosome segregation requires the coordination of both regulatory and mechanical molecular machines and culminates in the delivery of two complete sets of chromosomes to two daughter cells. Chromosomes contain long, continuous strands of DNA that are folded and assembled into higher order structures, which, in human cells, results in a 10–20,000-fold linear compaction of DNA (). Besides the core histones, many nonhistone chromosomal proteins have been identified (), but a full identification and functional characterization of chromosomal proteins has so far been unavailable.
Chromosomes assemble specific structures called kinetochores that serve as the molecular machines to mediate attachment, checkpoint signaling, and force generation at the ends of spindle microtubules (; ). Kinetochores are built either at the primary constriction of centric chromosomes or along the whole length of holocentric chromosomes. The molecular components of kinetochores are best characterized in , and many of the components of yeast kinetochores are highly conserved (; ; ; ). Nonetheless, a full inventory of the components of the animal cell kinetochore is still lacking.
Cell-free cytoplasmic extracts from eggs have previously been used for functional studies of chromosomes and kinetochores (; ; ; ). This system targets many chromosome and kinetochore proteins to chromatin in a cell cycle–dependent fashion and has the advantage of providing a method of preparing chromatin and chromosomes that are largely free of cytoplasmic contaminants. We have previously developed methods for preparing a soluble fraction of chromatin and chromosome-associated proteins () and have used two-dimensional gel electrophoresis of these preparations to reveal >350 distinct polypeptides associated with in vitro–assembled mitotic chromosomes, although the exact number depended on the resolution of the gel system ().
We have subsequently used liquid chromatography tandem mass spectrometry to characterize our preparations of solubilized mitotic chromosome proteins. In this study, we have selected four of the unknown chromosome proteins identified in this primary proteomics screen for further characterization. We have investigated the function of their human homologues using a secondary screen based on time-lapse fluorescence imaging of mitotic progression after RNAi-mediated depletion of each unknown. This analysis has identified Bod1, a novel vertebrate centrosomal and outer kinetochore protein that is required for proper chromosome biorientation.
Mass spectrometric analysis identified >250 proteins that associate with chromosomes assembled in metaphase egg extracts. Experimental details and results of the chromatin proteomic data are deposited at ( chromatin proteome survey). We chose four of these proteins that were novel, uncharacterized, and had well-conserved orthologues in other species (FLJ13263, ABCF, NPL4, and FAM44B; Fig. S1 B, available at ). To determine whether these proteins were involved in the generation of condensed chromosomes or in chromosome segregation, we constructed the pU6YH vector that expresses histone H2B-YFP () and a short hairpin RNA (shRNA) against the target protein (Fig. S1, A and C). The expression of histone H2B-YFP allows the visualization of chromosomes and simultaneously marks cells that are transfected with the shRNA-containing vector. In control experiments with pU6YH coding for shRNA targeting Aurora B, cells expressing histone H2B-YFP always showed knockdown of the target protein, but the amount of histone H2B-YFP detected was poorly correlated with the level of Aurora B knockdown (unpublished data). Regardless, the knockdown was efficient enough to allow us to screen for mitotic phenotypes by monitoring chromosome dynamics by time-lapse imaging of living cells.
and S1 (D–G) show maximum intensity projections of selected time points from time-lapse videos for each targeted protein. Cells expressing shRNA to FLJ3263 or ABCF generally proceeded through mitosis with no obvious phenotypes, similar to control cells expressing scrambled shRNA () or shRNA targeting paraspeckle component 1 (PSP1; Fig. S1 D), a nuclear protein that shows no mitotic phenotype upon depletion by siRNA (). In contrast, cells expressing shRNA targeting Npl4 or Fam44B showed marked mitotic defects. Cells depleted of Npl4 either failed to form a defined metaphase plate (Fig. S1 G) or persisted in metaphase to the end of the experiment (not depicted). Upon entry into anaphase, we often observed the apparent formation of multipolar spindles resulting in cut phenotypes. Cells depleted of Fam44B generally failed to form an organized metaphase plate and maintained this state for extended periods of time before entering an aberrant anaphase with a classic cut phenotype ().
The frequency of aberrant anaphase events is quantified in Fig. S1 H. Transfection of plasmids bearing cassettes coding for scrambled shRNA or shRNAs targeting PSPC1, FLJ13263, or ABCF1 caused a low level of aberrant anaphases, whereas the depletion of Npl4 or Fam44B resulted in anaphase defects in 75–80% of cell divisions. After this work was completed, a paper was published describing the function of Npl4 in the regulation of mitotic spindle assembly (). This gave us confidence that our screens are identifying proteins with important roles in chromosome segregation. The defects observed upon Fam44B depletion strongly suggested that this novel protein is required for proper function of the mitotic spindle and possibly for the interactions between kinetochores and microtubules. Because our further characterization showed that Fam44B (provisionally named as a member of a protein family of unknown function) is required for chromosome biorientation, we have named this protein Biorientation Defective 1 (Bod1).
To characterize endogenous Bod1, we generated a Bod1 polyclonal antibody using a recombinant protein antigen. This antibody recognized a 22-kD protein on immunoblots of HeLa cell lysates (Fig. S2 A, available at ), Bod1-GFP in human cells (Fig. S2 A), and recombinant Bod1 expressed in (not depicted). We were unable to detect Bod1 in HeLa cells by standard immunofluorescence protocols but did detect Bod1 at kinetochores of nocodazole-arrested cells subjected to swelling and spreading using an antibody raised against recombinant Bod1 (). This suggests that Bod1 is a component of mitotic kinetochores.
To more fully characterize the properties of Bod1 through the cell cycle, we constructed a Bod1-GFP fusion and used fluorescence microscopy to localize Bod1-GFP. shows representative fixed cell images of Bod1-GFP throughout interphase and mitosis 48 h after transfection. Bod1-GFP localized strongly to centrosomes throughout the cell cycle, only dissociating during cytokinesis (). Bod1-GFP also localized at kinetochores from prometaphase until anaphase (, arrowheads). During metaphase and anaphase, levels of centrosome-bound Bod1 decreased, whereas levels increased on spindle microtubules. To further refine the kinetochore localization of Bod1, we compared the localization of Bod1-GFP with Aurora B (to mark the inner centromere) or anticentromere antibody (ACA; to mark the inner kinetochore; ). The localization of Bod1-GFP was separate from Aurora B and adjacent to, but not overlapping, ACA, suggesting that Bod1 is a component of the outer kinetochore.
The kinetochore of is well characterized; upwards of 70 proteins have been identified as components of separable subcomplexes (; ; ). Therefore, we looked for potential orthologues of Bod1 in . BLAST searches of the genome using the entire Bod1 sequence failed to find any matches. Splitting the Bod1 sequence into 4–10 amino acid segments and using the patmatch function () also failed to find any direct orthologues. However, Bod1 is highly conserved throughout metazoans (Fig. S3 A, available at ), with clear orthologues in mouse, rat, , , and . No apparent orthologue could be found in .
Bod1 is one of three related proteins that comprise the Fam44 protein family in the human genome. These proteins are encoded by genes on three different chromosomal loci (4p16.1, 5q35.2, and 18q21.31 for Fam44A, Bod1, and Fam44C, respectively), suggesting that the three genes have arisen from a gene duplication event. Fig. S3 B demonstrates that Bod1 and Fam44C are most closely related. The N terminus of Fam44A is very similar to Bod1 except that it contains a long string of proline residues at the extreme N terminus. Fam44A is approximately twice as large as the other family members with a large C-terminal extension, which does not appear to relate to any other known protein. Fam44A is conserved among vertebrates, but, given the similarity of Fam44C to Bod1, it is was difficult to distinguish whether there were Fam44C-specific genes in species other than (GenBank/EMBL/DDBJ accession no. ).
To determine whether Bod1 is associated with any other proteins, we examined its hydrodynamic properties. Lysates of nocodazole-arrested HeLa cells were fractionated by gel filtration and sedimentation through glycerol gradients and were analyzed with a polyclonal antibody that recognizes Bod1. Both analyses showed that Bod1 exists in two forms: one that is most likely a monomeric, unbound form and one in a complex of ∼490 kD (). To date, we have not identified the components of this larger complex but note that Ndc80/Hec1, a fundamental component of the kinetochore (), does not comigrate with this complex. Further characterization will be necessary to determine whether the large Bod1 complex is a component of the kinetochore or a spindle pole–associated complex.
Bod1 was depleted from HeLa cells by targeting three different sequences by siRNA (see Materials and methods), resulting in efficient depletion of the protein ( and not depicted). This depletion was specific, as we observed no change in the levels of Fam44A or Fam44C by RT-PCR ().
FACS analysis showed that Bod1 treatment reduced the proportion of G1 cells and caused a substantial increase in apoptotic cells with a sub-G1 DNA content (). This increase in the apoptotic population was also confirmed by counting the number of apoptotic cells in fixed DAPI-stained samples. Only 0.47 ± 0.01% of control cells were apoptotic compared with 3.9 ± 0.4% of Bod1cells.
To assess any effects on the mitotic spindle checkpoint, we treated Bod1 HeLa cells with either nocodazole or taxol and counted phospho-H3–positive cells. In both cases, we observed a robust mitotic arrest, suggesting that the depletion of Bod1 does not impair the spindle checkpoint (). In addition, immunostaining Bod1 cells with anti-Bub1 (), anti-BubR1, or anti-Mad2 antisera (not depicted) demonstrated that the majority of unaligned kinetochores stained strongly for spindle assembly checkpoint proteins.
Detailed examination of Bod1 HeLa cells by immunofluorescence revealed the presence of somewhat elongated disorganized bipolar mitotic spindles with a mean pole–pole distance of 12.5 μm compared with 9.2 μm in control cells (). Given the disorganization of the mitotic spindle, we examined the localization of Eg5 and Aurora A (). Although Eg5 localized properly to spindles, which is consistent with the formation of a bipolar spindle, Aurora A staining was much less focused in Bod1 cells, suggesting that Bod1 may play a role in organization of the spindle pole.
Bod1 cells contained many unaligned chromosomes (). All kinetochores of unaligned chromosomes formed either end-on or lateral attachments with spindle microtubules (, inset), with frequent syntelic microtubule–kinetochore attachments (). In addition, all unaligned chromosomes still had robust Hec1 and Mis12 staining at kinetochores, suggesting that after the depletion of Bod1, at least two critical kinetochore subcomplexes were still targeted properly (unpublished data). These results suggest that after Bod1 depletion, kinetochores can form end-on and lateral attachments to microtubules but, in many cases, cannot achieve correct alignment on the metaphase plate.
Unaligned kinetochores in Bod1 cells might be unable to congress to the metaphase plate because of defects in microtubule attachments or microtubule plus end dynamics. Alternatively, the microtubule attachments might be functional, but syntelic attachments might be inappropriately stabilized. To discern between these possibilities, we performed a cold stable kinetochore fiber assay. shows that although the overall density of microtubules was reduced, all kinetochores in Bod1 cells were attached to microtubules after cold treatment, as in control cells. To assess the function of these attachments, we cotransfected Bod1 cells with plasmids expressing GFP–centromere protein B (CENP-B) and mCherry-tubulin (see Materials and methods). shows stills from time-lapse videos of projections of control and Bod1 cells. Control cells formed normal mitotic spindles, achieved normal chromosome alignment, and progressed through mitosis ( and Video 1, available at ). In contrast, Bod1 cells aligned some chromosomes but contained many misaligned chromosomes ( and Videos 2–4). The severity of this phenotype varied, with some cells showing a small number of misaligned chromosomes and others showing many chromosomes on the distal side of spindle poles (Video 2). Most misaligned chromosomes were associated with end-on attachments to microtubule bundles and underwent oscillatory movements (; and Videos 3 and 4).
Sister centromeres normally only separate when under tension in a bioriented state (). In Bod1 cells expressing GFP–CENP-B and mCherry-tubulin, we observed the oscillatory separation of unaligned sister centromeres. These centromeres appeared to be syntelically attached to a bundle of microtubules, and sister centromere separation was not aligned with the pole–pole axis (, “un” arrowhead; and Video 4). This observation was confirmed by time-lapse analysis by measuring distances between two bioriented sister centromeres (used as an internal control) and an unaligned pair of sister centromeres in a Bod1 cell (). We conclude that kinetochores on misaligned sister chromosomes were attached syntelically to microtubules from the neighboring pole and that the separation of misaligned sister centromeres reflects asynchronous microtubule end dynamics on a centromere pair, resulting in force across the pair. Kinetochores in Bod1 cells can therefore attach to microtubule ends and generate force, and the major defect in Bod1 cells appears to be an inability to detect or resolve syntelic microtubule attachments.
The mitotic profile of Bod1 cells was very similar to control cells except for a marked increase in cells with major biorientation defects and a corresponding decrease in normal metaphase cells (). Given the severity of the biorientation defect, we were surprised not to see an increase in aberrant anaphases. Long-term time-lapse imaging of cells cotransfected with CENP-B– GFP and control or Bod1 siRNA () revealed that the biorientation defect in Bod1 cells would persist for up to 12 h before cells either directly entered apoptosis or exited mitosis and apoptosis several hours later, explaining why a high rate of aberrant anaphases were not observed.
To determine whether the biorientation defects in Bod1 cells were caused by a failure to resolve syntelic attachments, we artificially increased the occurrence of these attachments by the addition of the drug monastrol, which causes monopolar spindles and results in the majority of kinetochore microtubule attachments being syntelic (). shows that Bod1 cells arrest as efficiently as control cells with monopolar spindles when treated with monastrol. 1 h after release from monastrol into media containing MG132, to prevent cells going into anaphase, 60% of control cells formed a fully aligned metaphase plate with virtually no biorientation defects. In contrast, only 37% of mitotic Bod1 cells had properly aligned their chromosomes, and 26% had severe biorientation defects. Detailed inspection by immunofluorescence demonstrated that these cells contained syntelic attachments (). Therefore, by artificially increasing the number of syntelic attachments in Bod1-depleted cells, we increased the frequency of biorientation defects from 15% in unperturbed cells to 26% in cells released from the monastrol block (). We conclude that the depletion of Bod1 compromises the efficient resolution of syntelic attachments.
The destabilization of syntelic attachments allowing subsequent correction requires the Aurora B protein kinase (; ). Aurora B phosphorylates mitotic centromere-associated kinesin (MCAK; and possibly Kif2), and this phosphorylation appears to be required for this destabilization (; ; ). Therefore, we analyzed the localization and phosphorylation of MCAK in Bod1-depleted cells. After Bod1 depletion, Aurora B still localized to mitotic chromosomes, and we detected no difference in the amount of Aurora B in unaligned and apparently aligned chromosomes (). We detected no change in chromosome staining with anti–phosphohistone H3 () or anti-phospho–CENP-A (not depicted) after Bod1 depletion. Because both are markers of Aurora B activity (), these results suggest that Aurora B kinase activity was not dramatically impaired by the loss of Bod1. To further assay the function of Aurora B, we determined the localization of MCAK, which localizes to the inner centromere in its phosphorylated form but concentrates at kinetochores in its dephosphorylated state (). At unaligned sister kinetochores or in kinetochore pairs not yet fully under tension, MCAK is predominantly located at the inner centromere (; ). In Bod1 cells, we observed that although total MCAK present at unaligned centromeres was similar to control cells (), its precise localization was abnormal, forming multiple foci stretching out to one or both sister kinetochores.
Because MCAK localization to centromeres and kinetochores depends on the state of MCAK phosphorylation, we examined the levels of phosphorylated MCAK using an anti–phospho-Ser92 MCAK antibody (). Phosphorylation of MCAK was substantially reduced at the inner centromere of unaligned chromosomes in Bod1 cells compared with the control cells (). These results suggest that Bod1 depletion impairs the formation of bioriented attachments across sister kinetochores, possibly by impairing the correct localization of MCAK at centromeres and, thereby, preventing its phosphorylation and timely correction of syntelic attachments. We have not detected any effect of Bod1 on the in vitro phosphorylation of MCAK by Aurora B (unpublished data), so Bod1 may modulate MCAK phosphorylation by interacting with other proteins. Aurora B activity and kinetochore oscillations are necessary for syntelic correction (), and our data further suggest that syntelic correction may require MCAK phosphorylation. Whether there is any subtle perturbation in kinetochore oscillations in Bod1-depleted cells is not yet known and will require much higher resolution live cell imaging.
In summary, by using a cell cycle–dependent analysis of the chromatin proteome, we have identified a novel protein required for proper chromosome biorientation called Bod1. Bod1 is a member of the FAM44 protein family and is highly conserved throughout metazoans. Depletion of Bod1 in human cells causes severe biorientation defects, although kinetochores appear to generate force and oscillate. Bod1 is not required for the spindle assembly checkpoint but appears to be required either for the efficient detection or removal of syntelic attachments. Thus, it plays a critical role in defining and monitoring the proper attachment of microtubules to the kinetochore.
HeLa S3 cells were grown in DME supplemented with 10% FCS, 2 mM -glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen) at 37°C with 5% CO in a humidified incubator.
HeLa cells were transfected with pU6YH plasmid (Fig. S1) encoding shRNAi to Bod1 or other target proteins. 48 h after transfection, cells were split onto 40-mm-diameter glass coverslips (Bioptechs), cultured overnight, and transferred to CO independent media (Invitrogen) with supplements as above. Cells were maintained at 37°C using an FCS2 chamber in conjunction with an objective heater (Bioptechs). Images were acquired on a restoration microscope (DeltaVision Spectris; Applied Precision) with a 100× 1.35 NA objective and a cooled charge-coupled device camera (CoolSNAP HQ; Roper Scientific). SoftWorx software (Applied Precision) was used for image analysis. Datasets were deconvolved using the constrained iterative algorithm (; ) using SoftWorx software. Time courses were presented as maximum intensity projections of deconvolved three-dimentional datasets. Images were loaded into Photoshop (Adobe) or OMERO () and adjusted for display.
To deplete Bod1, HeLa cells were transfected with siRNA duplexes targeting the sequence UUCAUGAGUUCCUGGCGGCTT (MWG Biotech) or STEALTH siRNA duplexes (Invitrogen) targeting the sequences GCCACAAAUAGAACGAGCAAUUCAU or GGAAUGGAAUCCUACGAUGAACAAA for 48 or 72 h. Scrambled siRNA duplexes were used as controls. RT-PCR analysis was performed using a One Step RT-PCR kit (QIAGEN) according to the manufacturer's instructions.
HeLa cells were cotransfected with mCherry1 () fused to human β-tubulin (a gift from A. Straight, Stanford University, Stanford, CA), CENP-B–GFP (a gift from V. Draviam and P. Sorger, Harvard Medical School, Boston, MA), and either control or Bod1 siRNA. 24 h after transfection, cells were trypsinized and seeded onto 35-mm glass-bottom Microwell dishes (MatTek Corp.). Imaging was started 48 or 72 h after transfection. Datasets (512 × 512 pixels with 2 × 2 binning, 0.05-s exposure, and five z sections spaced by 0.5 μm) were acquired every 1 or 2 min on a microscope (DeltaVision Spectris; Applied Precision) fitted with a 37°C environmental chamber (Solent).
Anti–human Bod1 antibody was generated against recombinant GST-Bod1 fusion and used to immunize rabbits (Diagnostics Scotland). The antibody was affinity purified by incubating serum with recombinant myelin basic protein–Bod1 conjugated to Affigel 10 active ester agarose (Bio-Rad Laboratories).
Mitotic cell extracts were prepared from HeLa cells treated with 100 ng/ml nocodazole for 16 h. Size exclusion chromatography and glycerol density gradients were performed as previously described () except that 5-ml gradients and H150 buffer (50 mM Hepes, 150 mM KCl, 1 mM EDTA, and 1 mM MgCl, pH 7.9) were used. The native molecular weight and shape (frictional coefficient) of protein complexes was calculated using established equations (; ; ).
Cells were fixed with 3.7% PFA or in methanol for 2 min at −20°C and processed as described previously . Aqueous chromosome spreads were performed as described previously (). Mouse anti–α-tubulin DM1A (Sigma-Aldrich), rabbit anti-HEC1 antibody (Abcam), mouse anti–Aurora B antibody AIM-1 (BD Biosciences), and mouse anti-Bub1 (Chemicon) were used at 1:500. Rabbit anti–Aurora A (Abcam), mouse anti-Eg5 (BD Biosciences), and human CREST autoantisera (ACA; a gift from W.C. Earnshaw, University of Edinburgh, Edinburgh, Scotland, UK) were used at 1:1,000. Sheep anti-MCAK and anti–phospho-MCAK antibodies () were used at 1 μg/ml. Rabbit anti–phospho-H3 (Ser10; Upstate Biotechnology) was used at 1:200. Fluorescently labeled secondary antibodies were all obtained from Jackson ImmunoResearch Laboratories.
Fig. S1 shows the use of shRNAi and live cell imaging to identify candidate proteins. Fig. S2 shows the relative expression levels of Bod1-GFP. Fig. S3 shows Fam44 protein family sequence alignment. Videos 1–4 are videos relating to stills shown in Online supplemental material is available at . |
In eukaryotic cells, messenger precursor molecules must undergo a series of maturation events that include 5′ capping, splicing, 3′ end cleavage, and polyadenylation. During processing, nascent mRNA assembles together with RNA binding proteins into ribonucleoprotein particles (mRNPs; ; ). Mature particles are exported to the cytoplasm and several lines of evidence indicate that mRNPs move from the sites of transcription to the nuclear pores by random Brownian motion. As diffusion cannot be regulated, traffic control of newly synthesized mRNA molecules is thought to rely on retention at dedicated sites within the nucleus (). According to the current view, any failure compromising the integrity of an mRNA may cause its retention in the nucleus and trigger its degradation. There is evidence suggesting that such a surveillance mechanism operates in close proximity to the gene template () and, at least in yeast, at the nuclear pore ().
A key connection between transcription and mRNP biogenesis is provided by the C-terminal domain (CTD) of the largest subunit of RNA polymerase II (RNA Pol II LS), which binds several proteins essential for pre-mRNA processing (). The CTD of RNA Pol II LS is highly conserved, increasing in length and diversifying in structure with the complexity of organisms (). Contrasting with yeast, which contains 26 repeats of a conserved heptapeptide with the consensus sequence YSPTSPS, the mammalian CTD has 52 repeats, of which 21 obey the conserved consensus while the remainder display a variety of substitutions. Most of these nonconsensus repeats are located in the C-terminal part of the CTD (heptads 27–52; ), and the last repeat (CTD52) is essential for cell viability and Pol II stability (). At the very C terminus, the mammalian CTD further comprises a specific 10-amino acid motif. CTD deletion analysis has shown that heptad repeats 1–15 or 1–25 support capping but not splicing or 3′ end formation, whereas heptads 27–52 plus the C-terminal 10 residues can support efficient capping, splicing, and 3′ end formation (). More recent studies have demonstrated that scrambling the 10 residues that lie C-terminal of heptad 52 impairs efficient release of RNA from the site of transcription (). However, this mutation also reduces splicing and 3′ end cleavage (), arguing that the CTD requirement for RNA release may be a consequence of its role in promoting pre-mRNA processing.
To further investigate the role of the CTD in transcript release, we generated murine erythroleukemia (MEL) cell lines that express α-amanitin–resistant RNA Pol II LS with either full-length or truncated forms of the CTD. Our results reveal that deleting 21 C-terminal heptads of the CTD causes transcript retention at the site of transcription but without inhibiting splicing or 3′ end formation. This implies a previously unsuspected involvement of the CTD in mRNP maturation events that occur after splicing, cleavage, and polyadenylation have taken place.
MEL cells were stably transfected with an α-amanitin–resistant form of the RNA polymerase II largest subunit (RNA Pol II LS; ) containing either wild-type or deletion mutants of the CTD with 31 (Δ31) or 5 heptad repeats (Δ5) (; ; ). Each of these plasmids was cotransfected with a second plasmid containing the human β-globin gene () micro-locus control region (βLCR) and a puromycin resistance gene (; ). Given the tendency of multiple copies of plasmid transgenes to co-integrate as tandem arrays, we reasoned that this cotransfection procedure should, in many cases, result in the positioning of the βLCR-bearing plasmid upstream of the α-amanitin–resistant RNA Pol II LS constructs. As the βLCR is able to activate heterologous, nonerythroid promoters (; ) with the minimum requirements being a CAAT and CACCA or GC-rich (e.g., Sp1) elements (; and unpublished data), this configuration should confer erythroid-specific induced transcription on the human cytomegalovirus (CMV) promoter linked to the α-amanitin–resistant RNA Pol II LS cassettes. Stably transfected clones were selected with puromycin and screened by an S1-nuclease protection assay for expression of the transfected RNA Pol II LS gene. We selected clones that showed low levels of exogenous, transgene-derived α-amanitin–resistant RNA Pol II LS expression in preinduced cells and high levels after 4 d of differentiation. Expression of the exogenous RNA Pol II LS was confirmed by Western blotting analysis with an antibody that recognizes the haemagglutinin (HA) epitope (), as previously described ().
We next confirmed that the exogenous RNA Pol II LS was functional. The endogenous RNA Pol II LS is degraded upon binding of α-amanitin (), but the exogenous transgene-derived protein is resistant due to a single amino acid substitution that decreases its affinity for the toxin (). We therefore determined the ability of the different clones to transcribe endogenous murine β-globin gene () in the presence of α-amanitin. After 17 h of α-amanitin treatment, S1-nuclease protection assays revealed no signal for pre-mRNA in untransfected MEL C88 cells, confirming that transcription by the endogenous RNA Pol II was abolished (, lane C88 +). In marked contrast, pre-mRNA was present in clones expressing the α-amanitin–resistant forms of RNA Pol II LS containing either the full-length CTD (, lane wt +) or the Δ31 truncation (, lane Δ31 +), but not the Δ5 variant (, lane Δ5 +). The finding that none of the clones expressing RNA Pol II Δ5 were able to support transcription of was surprising, taking into account that this mutant was previously shown to transcribe under control of the SV40 promoter () and a rat homeobox reporter gene stably integrated into the genome of HeLa cells (). Additional studies indicated that the truncated variant of the CTD with only 5 heptad repeats did not affect TATA-box–mediated transcription (; ). However, this same form of the CTD in yeast and humans abolished activator- dependent induction of transcription of specific genes (; ; ). Furthermore, nuclear run-on experiments in mammalian cells suggested a global defect in transcription of endogenous genes (). Because endogenous murine is under control of the LCR (), our observation that RNA Pol II Δ5 fails to support LCR-dependent transcription is consistent with previous data indicating a requirement of the CTD for enhancer-driven transcription (; ; ).
Having selected MEL cell clones that express functional α-amanitin–resistant wild-type or Δ31 RNA Pol II LS, we next super-transfected these cells with an transgene that was either wild type (βWT) or a mutant variant possessing a GT to AC mutation at the 5′ splice site of the second intron (βSM) (). As this mutation inhibits splicing and causes retention of the RNA at the transcription site (), we were interested in determining whether the CTD is involved in recognition of the resulting aberrantly processed βSM pre-mRNA. Cells were induced for 3 d and analyzed by FISH. As previously described (), the transcription site of the βWT or βSM transgene is detected as a focus in the nucleus (). The wild-type mRNA is exported from the nucleus and accumulates in the cytoplasm (), in contrast to the mutant RNA, which is not detected in the cytoplasm (). Treatment of cells devoid of exogenous transgene-derived RNA Pol II with α-amanitin results in the disappearance of nuclear foci (). Contrastingly, nuclear foci remain clearly visible in cells transfected with α-amanitin–resistant forms of RNA Pol II LS ().
We have previously shown that treating MEL cells with the transcription inhibitor actinomycin D for a short period of time (5–15 min) causes a dramatic, rapid reduction in the relative number of cells that contain a detectable nuclear focus of βWT transcripts within the nucleus, whereas the percentage of cells harboring βSM RNA foci remained largely unaltered, suggesting that these mutant RNAs were not being released from the site of transcription (). We therefore performed the same assay using cells that express the α-amanitin–resistant forms of RNA Pol II LS in an effort to gain insight into the possible role of the CTD in the process of mRNA release from the transcription site. The results show that actinomycin D treatment of cells expressing either endogenous RNA Pol II LS or the α-amanitin–resistant full-length CTD transgene product resulted in a significant decrease in the percentage of cells with βWT transcription foci (), as well as in the intensity of the remaining signal (). However, when transcription is dependent on the α-amanitin–resistant RNA Pol II LS containing the truncated Δ31 CTD, actinomycin D treatment does not cause any significant reduction in the percentage of cells with a visible focus () or in the mean fluorescence intensity of each focus (). We therefore conclude that the transcripts are not being efficiently released. Parallel experiments performed with cells expressing the mutant transgene showed that treatment with actinomycin D causes no significant change in the percentage of cells with a visible nuclear focus of βSM RNA (), regardless of whether this gene is transcribed by full-length () or truncated () CTD versions of α-amanitin–resistant RNA Pol II LS. Thus, reducing the CTD to 31 heptad repeats is sufficient to prevent release of RNA transcribed from a normal gene while it does not interfere with the ability to retain transcripts derived from a gene with a severe splice mutation.
Previous studies have shown that deletion mutants of the CTD induce defects in splicing and 3′ end cleavage (; ; ). We therefore analyzed wild-type (βWT) transcripts produced by endogenous or exogenous transgene RNA Pol II LS by RNase protection assays to monitor splicing of introns 1 and 2 as well as cleavage at the poly(A) addition site (). As expected, no signal for unspliced (US; , lane 2) and uncleaved (UC; , lane 4) RNA was detected after inactivation of the endogenous RNA Pol II LS with α-amanitin (). However, bands corresponding to spliced (SP) and cleaved (CL) RNA were still present, most probably due to the long half-life of mRNA.
To avoid complications caused by the presence of mRNA synthesized by the endogenous RNA Pol II LS and which had accumulated before α-amanitin treatment, cells were exposed to this toxin immediately after the first day of induced erythroid differentiation. In MEL cells containing the βWT transgene and which rely exclusively on the endogenous RNA Pol II for transcription, this resulted in massive cell death, while there was survival of cells transfected with the α-amanitin–resistant forms of RNA Pol II LS. In these cells, after inactivation of the endogenous RNA Pol II LS, bands corresponding to both unspliced and spliced (, lanes 2 and 3) as well as uncleaved and cleaved (, lanes 5 and 6) products were clearly detected. In a control experiment, we analyzed RNA extracted from MEL cells containing the βWT transgene that had undergone the same period (3 d) of differentiation but not exposed to α-amanitin (, lanes 1 and 4). The results () revealed no substantial reduction in the percentage of spliced and 3′ cleaved βWT mRNA transcribed by RNA Pol II LS Δ31 compared with that synthesized by either RNA Pol II LS wild type or endogenous RNA Pol II. We further observed that poly(A) tail length of mRNA was similar in transcripts synthesized by either endogenous RNA Pol II RNA (, lane 2), RNA Pol II LS wild type (, lane 3), or Pol II LS Δ31 (, lane 4). Thus, a CTD with heptad repeats 1–23, 36–38, and 48–52 followed by the unique terminal 10-amino acid motif is sufficient to support efficient pre-mRNA processing, as predicted from previous studies (; ).
Collectively, our data reveal that βWT transgene mRNA synthesized by the RNA Pol II Δ31 mutant is efficiently spliced, cleaved, and polyadenylated and yet remains in close vicinity to the gene after inhibition of transcription. This strongly suggests that the CTD is required for mRNA release from the site of transcription in a manner independent of splicing and 3′ formation. A possible splicing-independent involvement of the CTD in RNA release was also noted by Bentley and colleagues, who observed that intron-less pre-mRNA synthesized by a terminal 10-amino acid motif mutant CTD remained at the site of transcription, whereas synthesis by a wild-type CTD resulted in RNA release (). However, the use of an intron-less reporter gene in these studies precluded definitive conclusions to be drawn.
We have previously shown that exon junction complex (EJC) proteins and core spliceosome components (U snRNPs) accumulate on nascent wild-type transcripts, but fail to associate with mutant transcripts that are not released from the transcription site (). To assess if the presence of truncation mutants of the RNA Pol II LS CTD affected this process, we conducted double-labeling (FISH plus immunocytochemical staining) of MEL cells that contained the βWT transgene transcribed by either endogenous or exogenous RNA Pol II (full-length or the Δ31 CTD mutant) after treatment with α-amanitin as before (). We used a probe to detect βWT transcripts and antibodies to detect snRNP Sm proteins (), and EJC components SRm160 () and Aly/REF (). Using a previously described quantitative single-cell assay (; ), we detected all three proteins recruited to nascent transcripts irrespective of CTD length (, a′–k′ and d, h, and l). Moreover, we observed that colocalization of EJC proteins with transcripts synthesized by RNA Pol II LS Δ31 CTD persists after inhibition of transcription by actinomycin D, which adds additional evidence that RNA transcribed by RNA Pol II harboring a truncated Δ31 CTD and retained at the transcription site is normally spliced.
In the yeast , retention of defective mRNA at the site of transcription requires Rrp6p and other components of the nuclear exosome, suggesting that this complex is part of a quality control checkpoint that monitors for correct processing of pre-mRNA (; ; ; ). We therefore investigated whether the mammalian orthologue of Rrp6p associates with nascent βWT transcripts by immunofluorescence using a specific antibody (). This protein was readily detectable throughout the nucleoplasm with a high concentration at the site of βWT transgene transcription (). Similar results were observed in cells that express exogenous RNA Pol II LS with either a wild-type (, n′, and p) or truncated Δ31 (, o′, and p) CTD.
Accumulation of nascent mRNA in close proximity to their transcription site is thought to represent a surveillance mechanism that prevents defectively processed transcripts from entering the flow to the cytoplasm (; ). A major player in cotranscriptional pre-mRNA maturation is the CTD of RNA Pol II, which acts by facilitating specific interactions between processing factors while the transcript is still attached to the polymerase (for review see ). Previous work revealed that a mutation of the terminal 10-amino acid motif of the CTD inhibited splicing, 3′ end cleavage (), and RNA release from the site of transcription (). Based on these observations it was proposed that the CTD is required for transcript release as a consequence of its role in splicing and 3′ end cleavage (). However, in this report we show that a partial truncation of the CTD (Δ31) containing heptads 1–23, 36–38, and 48–52 including the terminal 10-amino acid motif is sufficient to support transcription, splicing, 3′ end cleavage, and polyadenylation, but the newly synthesized mRNA fails to be efficiently released ( and ). These novel observations imply that the CTD is involved in processes that control the release of transcripts by a mechanism independent from splicing and cleavage. Collectively with previously reported data (), our results further suggest that different segments of the CTD play distinct roles in pre-mRNA processing and mRNA release from the vicinity of the gene template.
Although the mechanism via which the RNA Pol II LS CTD is involved in the release of mRNA from the transcription site is unknown, we speculate that the Δ31 CTD truncation mutant used in this study fails to bind and therefore recruit protein factors required to complete maturation of spliced and 3′ end cleaved/polyadenylated mRNA into export-competent mRNPs. We further propose that this defect in production of fully mature mRNA activates a quality control checkpoint or surveillance mechanism that prevents diffusion of mRNPs to the nuclear pores by tethering them near the gene template, with stalled mRNA being subsequently degraded by the exosome present at the transcription site. A prediction from this mechanistic model is that most mRNA transcribed by the Δ31 CTD mutant should not translate into protein. In good agreement with that prediction is the observation that the Δ31 CTD mutation cannot support long-term cell viability ().
The maintenance, induction of erythroid differentiation, and stable transfection of the MEL cell line C88 were described elsewhere in detail (). The cells were cotransfected as previously described () with an α-amanitin–resistant RNA Pol II LS gene with either a full-length CTD (52 heptad repeats, wild type) or a CTD with 31 (Δ31) or 5 (Δ5) repeats (; ; ; plasmids provided by W. Schaffner, University of Zürich, Zürich, Switzerland) and a plasmid containing the βLCR () modified to carry a puromycin resistance gene under the control of a phosphoglycerate kinase promoter (). Before transfection the RNA Pol II LS wild-type plasmid was linearized with MluI, the Δ31 and Δ5 plasmids were linearized with ClaI and the micro-βLCR plasmid was linearized with PvuI. Stable transfected clones were obtained by culture in the presence of 2.5 μg/ml puromycin (Sigma-Aldrich). Clones selected for further studies were then super-transfected with either the wild-type (βWT) or mutant (βSM) genes in the micro-βLCR expression vector (; ) with stably transfected cells isolated in the presence of 800 μg/ml G418. The function of endogenous RNA Pol II LS was inhibited by adding α-amanitin (Sigma-Aldrich) to the cell culture medium to a final concentration of 2.5 μg/ml.
transcripts were visualized by FISH () and double labeling for RNA and protein was as previously described (). The probe used for FISH was a 740-bp fragment of the human β-globin gene extending from the SnaBI site at −265 bp from the transcriptional start point to the BamHI site at +475 bp. The fragment was labeled by nick-translation with either digoxigenin-11-dUTP (Roche) or Cy3-AP3-dUTP (GE Healthcare). The following primary antibodies were used for immunofluorescence: rabbit polyclonal directed against SRm160 (1:500; ; provided by B. Blencowe, University of Toronto, Ontario, Canada), mouse monoclonal directed against Aly/REF (1:100; clone 11G5; AbCam), human autoantiserum C45, specific for Sm proteins (1:75; provided by W. van Venrooij, University of Nijmegen, Nijmegen, Netherlands) and rabbit serum against PM/Scl-100 (1:75; provided by Ger Pruijn, University of Nijmegen). The secondary antibodies used were: AlexaFluor 488–conjugated goat anti–rabbit IgG (1:200; Jackson ImmunoResearch Laboratories, Inc.), AlexaFluor 488–conjugated goat anti–mouse IgG (1:200; Jackson ImmunoResearch Laboratories, Inc.), and FITC-conjugated donkey anti–human IgG (1:100; Jackson ImmunoResearch Laboratories, Inc.).
Images were acquired on a laser scanning confocal microscope (LSM 510 or LSM 510 META; Carl Zeiss MicroImaging, Inc.) using the PlanApochromat 63×/1.4 objective. FITC and AlexaFluor 488 fluorescence was detected using the 488-nm line of the argon ion laser. Cy3 was excited with the 543-nm line of the helium-neon laser on the Zeiss LSM 510 and with the DPSS 561-10 laser on the Zeiss LSM 510 META confocal microscope. To quantify the intensity of the nuclear foci, single-cell images were acquired with no saturated pixels, always using the same settings. The mean intensity of fluorescence in the nuclear RNA focus was determined using ImageJ (). Line profiles were obtained from unprocessed images using the LSM 510 software.
Total cell extracts were prepared as described in . Volumes of total extract equivalent to 10 cells were fractionated on a 7% polyacrylamide-SDS gel and proteins were transferred to nitrocellulose in 24 mM Tris, 193 mM glycine, and 20% methanol for 16 h at 30 mA. Western blotting with mouse monoclonal against HA epitope (HA.11; Covance) and mouse monoclonal against α-tubunin (clone B-5-1-2; Sigma-Aldrich) was as previously described ().
The levels of endogenous murine β-globin gene () pre-mRNA were analyzed by an S1-nuclease protection assay as described (). RNase protection assays were performed as described (; ). RNase protection probes () were prepared by in vitro transcription with T7 RNA polymerase in the presence of α-[P]UTP (GE Healthcare). In each reaction nuclear RNA, prepared as described (), was incubated with excess of the antisense RNA probe overnight at 50°C, and the hybridization products were digested with a mixture of RNase T1 and A (Ambion) at 37°C for 1 h. The RNase-protected fragments were resolved on a 6% denaturing polyacrylamide gel and the intensity of the bands was quantified using a PhosphorImager (Molecular Dynamics). After quantification of each gel band, background was subtracted and the values were normalized for different U residue contents of the protected probe fragments.
The poly(A) tail length analysis of transcripts was performed by PCR using the ligase-mediated poly(A) test (LM-PAT) described by . For cDNA synthesis we used 3 μg of total RNA, 50 ng of phosphorylated oligo dT, and 1 μg of oligo(dT)-anchor primer. As a control, RNA was deadenylated by digestion with RNase H in the presence of oligo dT. For PCR amplification, 1 μl of the template LM-PAT cDNA was added to a standard 25-μl PCR reaction containing 12.5 pmol of oligo(dT)-anchor primer and 12.5 pmol of an specific primer that hybridizes 254 nt upstream from the polyadenylation signal (5′-GCAACGTGCTGGTCTGTGTGCTG-3′). The amplified products were resolved by electrophoresis on a 2% agarose gel stained with ethidium bromide. |
xref
sub
#text
We screened a deletion mutant library covering all nonessential genes () and a yeast strain collection harboring essential genes under the control of a regulatable promoter (). We fused the last 36 residues of the mammalian inwardly rectifying potassium channel Kir6.2 to Pmp2, a single-spanning (type I) yeast membrane protein, to obtain Pmp2GFP-LRKR. As shown previously (), the reporter localized to the ER due to the exposure of the well-characterized R-based signal present in the Kir6.2 tail (). We transformed a plasmid encoding the Pmp2GFP-LRKR reporter into both strain collections and analyzed the resulting transformants by light microscopy (Fig. S1 A, available at ). Inactivation of only 13 genes was found to affect the steady-state ER localization of the reporter (). To further characterize the hits, a plasmid encoding Pmp2GFP-KKTN () was introduced into the strains identified in the screen (Fig. S1 A; and unpublished data). In wild-type cells, Pmp2GFP-KKTN localizes to the ER due to the exposure of the C-terminal di-lysine (KKXX) signal (). R-based sorting motifs and KKXX signals fall into distinct classes because the details of how they are recognized are not identical (; ). The ER localization of Pmp2GFP-LRKR and Pmp2GFP-KKTN was compromised in all candidate strains (, Fig. S1 A; and unpublished data). Strikingly, several genes identified in the screen coded for COPI subunits or subunits of the Dsl1/Dsl3 tethering complex that is thought to be part of the ER target site for COPI-coated vesicles (; ). All but one coatomer subunit and the members of the Dsl1/Dsl3 tethering complex are essential. Hence, it is not surprising that the shut-off of these genes affects both types of cargo.
Because COPI functions as a heptameric complex, the genes encoding the α-, ε-, and ζ-COP subunit should have been identified in the screen. We investigated the localization of Pmp2GFP-LRKR and Pmp2GFP-KKTN in the corresponding deletion or shut-off strains (Fig. S1, B and C). The result shows that all COPI subunits but ε-COP (Fig. S2 A, available at ) are required for efficient ER localization of Pmp2GFP-LRKR. This result is consistent with the fact that ε-COP is the only dispensable COPI subunit (). Presumably the genes encoding α- and ζ-COP were missed in the screen because of incomplete down-regulation. Similarly, the gene encoding Dsl1 that associates with Dsl3 to form a tethering complex thought to be involved in the fusion of retrograde vesicles with the ER was not detected in the screen, although the gene encoding Dsl3 was (, Fig. S1 A). Consistent with the proposed role for a complex containing Dsl1 and Dsl3 in COPI-mediated retrograde transport, we found that down-regulation of the gene resulted in the mislocalization of Pmp2GFP-LRKR and Pmp2GFP-KKTN (Fig. S1 B). The accumulation of coatomer subunits in the screen strongly suggests an intimate relationship between the sorting of membrane proteins exposing an R-based signal and coatomer. Therefore, we decided to concentrate on the role of COPI in the recognition of R-based signals.
The two WD40 domains of α- and β′-COP bind distinct but overlapping sets of di-lysine signals, and hence, both contribute to the recycling of proteins with di-lysine signals (). To assess whether coatomer discriminates between Pmp2GFP-LRKR and Pmp2GFP-KKTN, we tested the localization of both reporters in a mutant containing an allele with a mutation in the region of α-COP that encodes the WD40 domain (Fig. S2 B). Confirming previous results (; ), we observed mislocalization of Pmp2GFP-KKTN, whereas Pmp2GFP-LRKR localized robustly to the ER. These experiments support the notion that the binding site for R-based signals present in the COPI complex is distinct from the WD40 domain of α-COP that recognizes the KKTN motif ().
The yeast two-hybrid system has been successfully used to detect the interaction between adaptor subunits and tyrosine-based cargo sorting motifs and between K(X)KXX motifs and COPI subunits (, ; ; ). We used fusions to the Gal4 DNA-activation domain (AD) presenting a C-terminal tail with a very strong R-based signal (KLRRRRI) or one of low efficacy (NVRNRRK). Both synthetic signals were originally identified in a combinatorial screen of signal variants in mammalian cells (). Their differential strength was revealed by quantitative trans-Golgi network processing and cell surface expression assays. Fusions of COPI subunits to the Gal4 DNA-binding domain (BD) () were used to assay the putative interaction between the R-based ER sorting motif and individual COPI subunits (). Two plasmids encoding the Gal4-BD fusions to β-COP or to δ-COP gave rise to specific activation of the reporter gene (, “interaction test”) when tested against the strong variant of the R-based sorting motif.
We used the yeast two-hybrid assay to narrow down the sequence stretch on each COPI subunit that was required to observe a specific interaction with the strong R-based signal. A series of C-terminal deletion mutants of the Gal4-BD fusion to the δ-COP subunit was tested against the strong and the weaker R-based sorting motif (). Two constructs ending with residue 399 or 388 marked the transition from a δ-COP protein that was able to recognize R-based signals to one that was not (). We aligned the surrounding region of bovine, yeast, human, and fly δ-COP orthologues () and found the region to be highly homologous, consistent with the evolutionarily conserved recognition of R-based sorting motifs (). We deleted the conserved region (aa 388 to 413) from the full-length δ-COP Gal4-BD fusion construct and tested whether it had lost its ability to interact with the R-based sorting motif, which was indeed the case (). The internal deletion mutant of δ-COP had retained part of its function because δ-COP (Δ388-413) was still capable of binding β-COP in a two-hybrid assay (unpublished data).
β- and δ-COP are known to interact tightly in the trunk domain of the adaptor-like subcomplex (; ). To determine the region on the β-COP subunit required for the interaction with the R-based signal, we tested a series of constructs encoding N-terminally truncated forms of β-COP. Removing the N-terminal 210 aa had no effect on the specific interaction, whereas the signal was lost upon deletion of further 156 aa (). The responsible region was narrowed down to aa 318–338. A β-COP subunit lacking these residues was incapable of interacting with the strong R-based signal as assayed by the yeast two-hybrid system. As observed for δ-COP, alignment of the corresponding stretch with β-COP orthologues from yeast, human, mouse, and fly revealed high conservation of the putative binding site (). Our yeast two-hybrid analysis implicated the β- and δ-subunit of COPI as candidates harboring the binding site for R-based signals. This result could be explained by different hypotheses: Both COPI subunits contain independent binding sites for the R-based signal; or Both COPI subunits contribute to a common binding site and the affinity of the partial binding pocket is sufficient for a positive signal in the assay.
To distinguish between these possibilities, we created a strain that expresses a variant of the COPI coat specifically incapable of recognizing R-based sorting motifs.
DLD/NAN, (β-COP*) in the critical region and δ-COP against Δ414-435 (δ-COP*). Strains expressing β-COP*, δ-COP*, or both were viable, although the double-mutant strain grew slowly and was temperature sensitive. These strains enabled us to test the localization of different reporters in the absence of the putative binding site(s).
We then assessed the functional consequence of removing the two putative binding sites for R-based signals individually and in combination by determining the subcellular localization of Pmp2YFP reporters presenting different variants of the R-based signal (). In mammalian cells, the two tails containing the sequences KLRRRRI or NVRNRRK were shown to function respectively as a very strong and a weak R-based ER localization signal by two independent quantitative assays (see in ). Although these results show that the CD4-GFP reporter exposing the weak signal was able to leave the ER, most of the protein localized to the ER in the steady state as assessed by the GFP staining pattern in HeLa cells (Fig. S3, available at ). Consistent with the strict conservation of the relative efficacy of R-based signals in yeast (), Pmp2YFP- NVRNRRK was also localized in the ER at steady state in wild-type yeast cells (). In the strains expressing either β-COP* or δ-COP*, Pmp2YFP-NVRNRRK (exposing a weak R-based signal) was no longer retained in the ER and efficiently reached the vacuole (, arrows), whereas Pmp2YFP-KLRRRRI (exposing a strong R-based signal) was still maintained in the ER (completely in the strain expressing δ-COP* and substantially in the strain expressing β-COP*; ). Vacuolar localization is an indication of ER exit and loss of R-based signal recognition because Pmp2 reporters with inactive R-based signals efficiently reach the vacuole (). In the strain coexpressing the two altered COPI-subunits, both R-based signal-exposing reporters strongly accumulated in the vacuole as demonstrated by differential interference contrast and colocalization with the fluorescent dye FM4-64 (). FM4-64 is taken up from the plasma membrane and reaches the vacuole via the endocytic route (). In contrast, no colocalization with the vacuolar marker was observed for Pmp2GFP-KKTN. We tested whether the observed changes in steady-state localization could be explained by a reduced affinity of the double-mutant COPI coat for the respective signal-containing tails. Binding assays with cytosol obtained from wild-type or double-mutant yeast cells revealed that the mutant COPI complex containing both β-COP* and δ-COP* recognized a KKTE signal at the C terminus of Mst27 () almost as efficiently as the wild-type coat, whereas binding to the last 36 aa of Kir6.2 containing the LRKR signal was strongly reduced ().
To address the specificity of the sorting defect () we investigated the steady-state localization of two additional membrane proteins known to be retrieved by COPI. Rer1 is a retrieval receptor for ER membrane proteins that is dynamically localized to the Golgi apparatus by COPI (). Emp47 is a cargo receptor that shuttles between the ER and the Golgi (; ). The C-terminal KTKLL signal of Emp47 is thought to be recognized by the WD40 domain of β′-COP (). Indirect immunofluorescence and subcellular fractionation on sucrose gradients revealed similar steady-state localization patterns for both Rer1 and Emp47 in wild-type and double-mutant cells (). The punctate staining pattern is consistent with their accumulation in Golgi sub-compartments. The vacuolar protein carboxypeptidase Y (CPY) was found to float to the light fractions (labeled 1, 2 in ) of the gradient for both strains, whereas the heavy fractions (labeled 7, 8, 9) of the gradient were enriched in the ER marker Kar2. Detection of Pmp2YFP-NVRNRRK (exposing a weak R-based signal) revealed that the reporter cofractionated with the ER marker Kar2 in the wild-type strain and partially redistributed to the lighter fractions in the strain coexpressing both, β-COP* and δ-COP*. This confirms the mislocalization of this reporter in the double-mutant strain. The most apparent defect in the double-mutant strain consisted of multiple vacuoles as observed by FM4-64 staining () or indirect immunofluorescence of a vacuolar membrane protein ().
We conclude that the COPI coat containing both β-COP* and δ-COP* is specifically incapable of recognizing R-based signals. At the same time this mutant coat is still capable of sorting three other cargo proteins with high fidelity. The two independent or one common binding site(s) formed by β- and δ-COP are necessary and sufficient to recognize R-based peptide sorting motifs. Importantly, the region contributing to the recognition site for R-based signals is highly conserved across eukaryotic species, consistent with the fact that the signals are effective and follow the same consensus in yeast and mammalian cells. The effects of mutating or deleting the critical stretch in either β- or δ-COP were additive, e.g., individual mutations in either β- or δ-COP caused considerable ER exit of the Pmp2YFP reporter exposing a weak R-based signal, whereas the reporter exposing the strong signal remained mostly ER-localized (). When both mutations were combined, the less efficient signal was completely inactive as an ER localization motif and reporters presenting the stronger LRKR or KLRRRRI signals accumulated outside the ER. This additive behavior supports the notion that the regions in the two subunits contribute to a common binding site that is completely distinct from the binding site recognizing KKXX signals.
No high-resolution structure of the COPI coat complex is available, but sequence homology and partial structures suggest that the β-, δ-, γ-, and ζ-COP subcomplex is structurally similar to the clathrin-adaptor core complex, whereas the α-, β'-, and ε-COP subcomplex is thought to be functionally equivalent to clathrin (; ; ). To better understand the structural relationship between the stretches identified in β- and δ-COP, we built a homology model of the adaptor-like COPI subcomplex containing β-, δ-, γ-, and ζ-COP based on the crystal structure of the clathrin adaptor 1 core () (; compare the original structure shown as for reference). Our homology model is consistent with the idea that the relevant regions of β- and δ-COP come into close proximity and with the concept of one binding site for R-based signals at the subunit interface. Strikingly, this binding site in the adaptor-like trunk is structurally comparable to the site where YXXΦ endocytic motifs are recognized by clathrin adaptors (, compare red and blue regions in to black ellipse). This suggests that the general architecture of the trunk structure has evolved to accommodate completely different cargo-sorting signals. It will be interesting to test how the recognition of R-based signals is coupled to the sorting of ion channel and receptor subunits.
Molecular cloning followed standard procedures as described in . All plasmids are listed in Table S1 (available at ). All constructs were verified by sequencing. Pmp2GFP and YFP fusions were as described in . Two additional constructs were prepared where Pmp2 was fused to YFP followed by a unique NotI site. The NotI site encoded three alanines linking YFP and the tail. Oligonucleotides coding for the different tails were cloned into NotI and XhoI sites, resulting in constructs pR889 –AAATLASKLRRRRISLS and pR890 –AAATLASNVRNRRKSLS (compare Table S1).
The GST fusions are derivatives of the GST-MST27-KKXX construct (). The construct is a fusion of the C-terminal tail of Mst27 (codons 159–234 including stop codon) to GST. The different R-based signals were introduced as oligonucleotides using a ClaI site (codon 226 of ) and an XhoI site present in the pGex6P vector. The sequences of the different tails are as follows in the indicated constructs (compare Table S1): pO732 -DALLKKTE and pF291 -DLLDALTLASSRGPLRKRSVAVAKAKPKFSISPDSLSGSRSHHHHHH. Bold-print letters indicate the peptide-sorting motif.
The yeast strains used in this study are listed in Table S2 (available at ). A diploid heterozygous Δ strain (Y25865) was transformed with an -containing plasmid (pQ849) carrying the temperature-sensitive allele. The strains were then sporulated and spores were selected for their capacity to grow on geneticin (the Δ genotype was marked with a selection marker in strain Y25865) and for 5-fluororotic acid (5′FOA) sensitivity to ensure loss of the wild-type allele present in the diploid strain. Plasmids carrying either wild-type (Y25865-KM1) or Δ414-435 (Y25865-KM2) under the control of the promoter were then integrated into the locus of the resulting haploid strains. These strains were selected on 5-FOA for loss of the -carrying plasmid. In Y25865-KM2, the presence of only the altered variant after gene swapping was confirmed by the polymerase chain reaction and by Western blot analysis of total cell lysates using an anti-COPI antiserum.
To test whether different mutant alleles could substitute for the wild-type gene, we used a strain (RDY122) with an integrated copy of under the control of the promoter (). The strain was transformed with a plasmid carrying the respective variant of (
DLD/NAN or DILR/NAAA or Δ318-338) under the control of a promoter (p894-p897) and a marker. Serial dilutions of individual transformants were spotted on selective medium lacking leucine and containing glucose to suppress transcription from the promoter. All transformants that had received wild- type or mutant alleles but not those that had received only the empty plasmid were able to grow in the presence of glucose indicating that all Sec26-derived proteins were functional components of coatomer.
Strain Y25865-KM2 was transformed with a -marked 2μ plasmid carrying the wild-type gene. Next, a cassette disrupting the coding region of was introduced by homologous recombination and the resulting transformants were selected for growth on nourseothricin-containing medium. This strain was then transformed with a -marked plasmid carrying the different alleles (
DLD/NAN or DILR/NAAA or Δ318-338). Transformants were selected on 5′FOA- containing medium to select against the plasmid carrying the wild-type gene yielding strains Y25865-KM3 and Y25865-KM4.
Large-scale transformations were performed in 96-well plates using the lithium acetate method () and transformants were plated on synthetic complete (SC) medium lacking uracil or leucine. Liquid cultures were inoculated in 96-well plates containing SC medium lacking uracil or leucine and viewed by epifluorescence after overnight growth at 30°C. For screening of the promoter allele library 40 μg/ml doxycycline was added to the medium during the overnight growth period before microscopy. Automated microscopy of the two yeast libraries was performed on an Olympus ScanR screening system (40 × 0.95 objective) and 96-well glass-bottom plates (MMI).
Double transformants of strain AH109 containing Gal4-BD and -AD plasmid were selected on SC medium lacking tryptophane and leucine. In contrast to , we replaced the auto-activating Gal4-BD fusions of bovine β- and ζ-COP with the respective fusions to the corresponding yeast COP subunits (encoded by and ) and performed the two-hybrid assays at low stringency (e.g., dilution series of double transformants grown in SD lacking tryptophane and leucine were spotted on SD medium lacking tryptophane, leucine, and histidine to test for the activation of the reporter gene). The plasmid encoding the Gal4-BD fusion to Ret3 proved toxic to the yeast and was thus omitted from further experiments. The deletion analysis was performed on bovine δ-COP and yeast β-COP.
Crude cell extracts were prepared essentially as described by . In brief, yeast cells were grown to logarithmic phase, harvested, and resuspended in cold wash buffer (20 mM Hepes, pH 7.4, 0.7 M sorbitol, and 1 mM PMSF). Cells in wash buffer were transferred to a syringe and pelleted for 5 min at 2,000 rpm. Supernatant was removed and the paste was extruded from the syringe into liquid nitrogen. Frozen cells were ruptured with a pestle and mortar filled with liquid nitrogen. The resulting white powder was transferred to a reaction tube and an equal volume of binding buffer (20 mM Hepes, pH 6.8, 2% glycerol, 150 mM KAc, 5 mM Mg(Ac), 1 mM EDTA, 1 mM DTT, and 0.1% Triton-X 100 containing protease inhibitor mix: 2.5 μg/ml leupeptin, 1.5 μg/ml antipain, 0.5 μg/ml chymostatin, and 1 μg/ml pepstatin A) was added. After thawing, the crude extract was cleared for 5 min at 2,000 rpm to remove unbroken cells.
GST fusion proteins were induced in BL21 (DE3 star) carrying pRosetta by adding 0.2 mM IPTG for 2 h at 37°C. Cells were harvested and sonified in breaking buffer (20 mM Hepes, pH 6.8, 2% glycerol, 150 mM KAc, 5 mM Mg(Ac), 1 mM EDTA, 1 mM DTT, 1 mM PMSF, and 1× Complete protease inhibitor cocktail [Roche]). The lysate was cleared for 5 min at 2,000 rpm. The resulting supernatant was subjected to ultra-centrifugation for 30 min at 100,000 . The GST fusion proteins were purified from the supernatant by incubating with GSH-agarose. The agarose column was washed with breaking buffer and proteins were eluted by adding elution buffer (20 mM Hepes, pH 9.5, 10 mM GSH, 5% glycerol, 150 mM KAc, 5 mM Mg(Ac), 1 mM EDTA, and 1 mM DTT). For purification of the GST fusion protein with −LRKR-HexaHis C-terminus a Protino-2000 Ni-column (Macherey) was used as per the manufacturer's instructions. Appropriate fractions were dialyzed overnight in binding buffer and 0.5 μg dialyzed protein bound to 1 μl GSH-agarose (Sigma-Aldrich).
For binding experiments, crude extract was cleared of all insoluble material for 30 min at 100,000 and the supernatant was used for binding experiments. 100 μg total protein in 500 μl binding buffer (containing protease inhibitors, see above) was used for a binding assay with 2.5 μg bait on 5 μl GSH-agarose and incubated 2 h at 4°C. Beads were washed five times in binding buffer. Bound proteins were eluted with 1× SDS loading buffer containing 1 mM DTT, resolved by SDS PAGE, and detected by Western blotting using rabbit anti-COPI and rabbit anti-GST (Sigma- Aldrich) primary antibodies and an anti–rabbit AlexaFluor 680 (Invitrogen) secondary antibody. Blots were scanned (700 nm) and quantified with the Li-COR Odyssey system.
Separation of organelles was performed according to . In brief, cells were grown to mid-logarithmic phase, converted to spheroplasts, and lysed by douncing in a hypotonic buffer (0.3 M sorbitol and 50 mM triethanol amine, pH 8.9). The homogenate was layered on freshly prepared ten-step sucrose gradients and centrifuged for 12 h at 23,500 rpm in an SW28 rotor (Beckman Coulter). Membranes from the different fractions were pelleted by centrifugation (SW28, for 30 min at 23,500 rpm).
Cells were grown to early logarithmic phase and fixed at room temperature in freshly prepared fixative for at least 8 h and processed as described by . In brief, spheroblasts were permeabilized by a short incubation in 1% SDS in 1.2 M sorbitol and then stained with the indicated primary and Alexa594-conjugated secondary antibodies (Invitrogen).
Microscopy was performed using a microscope (DM IRE2; Leica) controlled by OpenLab software (Improvision) with a 100×/1.4–0.7 HCX PL APO CS oil immersion objective. Images were captured by an Orca-ER CCD camera (Hamamatsu) with excitation at 470/40 nm and emission at 525/50 nm (GFP, YFP), or excitation at 510/40 nm and emission at 610 nm/longpass (FM4-64, Alexa594). All images shown are representative of several independent experiments. Steady-state staining with FM4-64 was performed in SC medium for 1 h at room temperature. For live-cell imaging, yeast were incubated in SC medium at room temperature. Fixed yeast cells were mounted in ProLong Gold AntiFade reagent with DAPI (Invitrogen). Live cell images of transiently transfected HeLa cells were acquired using a 63×/1.4–1.6 HCX PL APO oil immersion objective in L-15 medium (Leibovitz; Sigma-Aldrich) at 37°C. Images were transferred to Adobe Photoshop CS2 for slight gamma adjustments.
A comparative three-dimensional model for the tetrameric subcomplex of yeast COPI was built based on the Crystal Structure of the clathrin adaptor protein core AP-1 (PDB: 1w63; ). Template identification for all four chains of the yeast COPI trunk lacking the appendage domains was performed by searching the PDB database () for suitable template structures in SwissModel Workspace (). Target-template alignments were generated by structure-guided multiple sequence alignment using 3DCoffee () for aligning γ-COP with template chain 1w63:A, β-COP with 1w63:B, and ζ-COP with 1w63:S. The alignment between δ-COP with 1w63:M was generated by HMM-HMM alignment (). Model coordinates for the tetrameric complex were generated using satisfaction of spatial restraints () after visual assessment of placements of insertions and deletions in the alignments.
Figure S1 shows microscopic images of selected hits in the reverse genetic screen expressing Pmp2YFP-LRKR and Pmp2YFP-KKTN. Figure S2 shows the localization of different Pmp2-GFP reporters in strains lacking ε-COP or expressing mutant α-COP. Figure S3 shows the steady-state localization of CD4GFP reporters exposing a very strong and a weak R-based signal in HeLa cells. Table S1 provides information about the plasmids and Table S2 about the yeast strains used in this study. Online supplemental material is available at . |
In , the larval somatic (skeletal) musculature arises from the fusion of two distinct types of myoblasts, the founders and fusion-competent cells (for review see ). Subsequent differentiation programs, including activation of muscle-specific gene expression and asymmetrical cell fusion between the differentially marked founders and fusion-competent myoblasts, are required for the generation of syncytial muscle fibers. Maturation of these syncytia into functional muscle fibers involve additional events, including pathfinding processes and the formation of attachments to the tendon cells, as well as the establishment of neuromuscular junctions (for review see ; ). The functional characterization of integrins and downstream effectors of integrin signals has underscored the importance of this pathway in establishing muscle attachment sites (for review see ). However, the molecular basis for many other aspects of morphogenesis and maintenance of the mature muscles is still poorly defined.
Herein, we present a functional characterization of (), which shares structural similarities with its paralogue Unlike , is prominently expressed in muscle progenitors and differentiated musculatures. We show that loss of activity leads to muscle detachment and massive muscle degeneration. We also demonstrate that functions in a novel integrin- and Notch-independent manner to maintain the integrity of the mature somatic musculature.
() encodes a 1,050-amino acid protein with several notable features (). A ZZ zinc finger domain within the N-terminal portion is flanked by two regions that share homology with HERC2, a protein that may function in protein trafficking and degradation pathways (; ). The ZZ domain, characterized by Cys-X-Cys and Asp-Tyr-Asp-Leu motifs, is found in a small number of proteins, including some transcriptional adaptor proteins and Dystrophin/Utrophin, and is implicated in protein–protein interactions (). Following the ZZ and HERC2-like domains is a repeated sequence that is specific to Mib proteins. Eight Ankyrin repeats are in the middle portion of the protein and two RING fingers at the C-terminal end. Ankyrin repeats are present in a vast number of proteins and their role in protein–protein interactions is well documented (), while RING fingers proteins are known to participate in protein–protein interactions in the ubiquitination pathway (). The presence of these various domains suggests that Mib2 functions as an adaptor-type of protein and/or as a component of a ubiquitination pathway.
The Mib2 protein is conserved during evolution. Mib2 and its murine orthologue display a similar structural organization and considerable degree of amino acid conservation within all the aforementioned domains (; and Fig. S1, available at ). When compared with Mib2 proteins across species, Mib1, an E3 ubiquitin ligase that has been shown to be important in Notch signaling (; ; ; ; ), shows a lower level of homology in most of these domains, indicating that Mib2 is a paralogue of Mib1. In addition, the Mib2 proteins have only two RING finger domains while the Mib1 proteins have three.
Maternally derived transcripts are detected prominently in the fertilized egg (). Zygotic expression is first observed at low levels panmesodermally, and beginning at stage 11, high levels of expression appear in progenitors of somatic and visceral muscles () and persist in the differentiated muscles of late stage embryos (; and unpublished data). is not detectable in cardiomyocytes. Co-localization of RNA (cytoplasmic) and LacZ protein (nuclear) in embryos derived from the rP298 enhancer trap line (), which carries a insertion within the () gene that is active in all founders (), confirmed that expression is specific for founder myoblasts (). Accordingly, is not detected in Lame duck (Lmd)–positive fusion-competent cells (; ). Mib2 protein expression is identical to that of mRNA and appears to be in the cytoplasm of founder cells (). In contrast, expression is not detectable in mesodermal cells (unpublished data).
Genetic and molecular analysis in the vicinity of the 37B10 locus identified the lethal complementation group as a likely candidate for (; see Materials and methods). We obtained the two extant alleles, and , for further analysis. Sequence analysis of the protein-coding exons showed that the gene on the mutant chromosome contains a nucleotide change (C to T) that converts Gln to a nonsense codon ( and Fig. S1). On the chromosome, a two-base pair deletion converts Asn to a Thr, which is then followed by a nonsense codon. As shown below, expression of wild-type in mutant embryos can rescue the observed muscle phenotype. We conclude that the and alleles correspond to bona fide mutations and henceforth designate these alleles as and , respectively. Based upon our analysis, the mutant Mib2 protein lacks all Ankyrin repeats and RING fingers while the mutant Mib2 protein lacks the RING fingers but retains four out of the predicted eight Ankyrin repeats.
To assess the consequence of loss of function on muscle development, we stained wild-type and mutant embryos with an antibody against Myosin to visualize the muscle pattern. We focused more on the allele because the molecular nature of this mutation suggests that it is a stronger mutant allele. As compared with wild-type embryos, stage 15 mutant embryos (derived from germline clones and zygotically /, termed “ m&z”), which lack both maternal and zygotic activity, have a well-developed somatic musculature, although a very limited number of detached muscles can already be detected (compare with ). At stage 16, the mutant embryos exhibit a highly deranged muscle pattern that is characterized by a massive number of detached muscles (). Many of the rounded muscles have become smaller, followed by rapid muscle degeneration. Consequently, in stage 17 mutant embryos, normal somatic muscles are absent and the size of the rounded muscles decreases dramatically (). We observed the same types of muscle deterioration with mutant embryos that are null for both maternal and zygotic activity (unpublished data). Unlike the somatic muscles, the midgut muscles do not disintegrate in mutant embryos; however, the incompletely constricted midgut of these embryos suggests that also plays a role in visceral muscles (Fig. S3, A and B; available at ). Cardiac morphology is not affected, as predicted from the absence of expression in myocardial cells (see ). Embryos that lack only zygotic function and homozygous deficiency embryos show similar somatic muscle and gut defects as those derived from germ line clones, although the defects are delayed and less severe (see ; and unpublished data). Previous knockdown by RNAi injections also caused some muscle detachments ().
To analyze the temporal progression and cause of the observed muscle phenotype, we recombined , which is a common marker for muscle 12 (or VL1) development (; ), and onto the chromosome. Wild-type and mutant embryos were double-labeled for Tropomyosin and LacZ expression. At late stage 14, the somatic musculature of mutant (m&z) embryos, including muscle 12, which is in the final stages of establishing its normal attachments, looks normal (). A low degree of muscle detachment becomes detectable at stage 15, although muscle 12 does not seem to be affected immediately, suggesting some differences in susceptibility to loss of function among the various muscles (). However, massive muscle detachments and degeneration, which include muscle 12, occur by stage 16 (). In the aggregate, this analysis indicates that function is not needed for the formation of somatic muscles and their initial attachment to tendon cells, but rather it is required at late embryogenesis for maintaining the attachments and the integrity of the mature musculature.
Because of the muscle detachment phenotype, we examined whether loss of function disrupts the localization of integrin signaling components that are known to establish stable muscle/tendon attachments (for review see ). In (m&z) embryos, α integrin localizes normally to the attachment sites within the tips of muscle 12 before and during the early stages of their detachment (compare with ). Hence, the gradual disappearance of localized α integrin during later stages () is presumably a consequence of the muscle detachments and deterioration rather than a cause of the detachment. Likewise, all other integrin pathway components examined, including Talin (), Pinch (not depicted), ILK (), and Tyr-phosphorylated FAK (Fig. S3, C and D) are initially localized normally within the muscle ends near the attachment sites in the absence of activity. Several components on the epidermal side of the attachments were also unaffected (unpublished data). These observations argue against a function of in establishing stable muscle attachments via integrin signaling components unless there is a yet unknown parallel pathway to ILK that is affected.
As shown in , Mef2-positive nuclei are still present in the large rounded muscles of zygotic (z) mutant embryos at stage 16, whereas they are absent in the rounded muscles that are severely decreased in size. Because cell shrinkage and chromatin deterioration are hallmarks of apoptosis, we used TUNEL staining to detect apoptotic cells in mutant embryos. Indeed, the detaching muscles in (z) embryos are positive for the apoptotic marker (), whereas heterozygous control embryos only show apoptotic signals in the CNS and other nonmuscle tissues (). Of note, the detached muscles in () mutant embryos, which lack functional β-integrin at their attachment sites, do not show any significant apoptotic signals and do not shrink, indicating that apoptosis is not an automatic consequence of muscle detachment (). Hence, we propose that the muscle detachment in mutants is a consequence of apoptotic events in these muscles.
To test this proposal further, we blocked apoptosis through forced expression of the caspase inhibitor p35 in muscle founders and their derived muscles. Blocking apoptosis in muscles of (z) mutant embryos leads to a significant reduction of muscle detachment and deterioration at early stage 16 (compare with ; no effects are seen with analogous expression of p35 in a wild-type background []). At late stage 16, some muscle degeneration still occurs in the p35-overexpressing mutant embryos, as evidenced by the slightly larger number of rounded muscles with decreased sizes and missing muscle fibers, although it is much less severe than in (z) mutants without blocked apoptosis (). A large number of muscle fibers are still present at stage 17 in these apoptosis-blocked (z) mutant embryos (unpublished data). Notably, the muscle degeneration phenotype is rescued in (z) mutant embryos that are also homozygous for , which deletes the apoptosis inducers , , and , (). In this background the majority of the muscles appear normal until at least late stage 16, although we do not know whether they change their morphology after cuticle formation. Together, these observations suggest that muscle detachment and degeneration in mutants are largely a consequence of triggered apoptosis.
Because RING fingers are implicated in protein ubiquitination, we sought to test whether the RING fingers of Mib2 are required for its activity and whether Mib1 and Neuralized (Neur), E3 ubiquitin ligases that have been shown to be modulators of the Notch pathway by ubiquitinating Delta (; ; ; ; ; ), could substitute for Mib2. Overexpression of full-length Mib2 in muscle founders and the derived muscles of wild-type embryos does not affect the pattern and stability of the muscles, although there is an increased number of unfused myoblasts at late stages (). In the mutant background, forced expression of full-length Mib2 leads to essentially complete rescue of the muscle detachment and deterioration phenotype (compare with ), although an excessive number of unfused myoblasts is also evident. Notably, forced expression of a Mib2 version lacking both RING fingers (Mib2; see ) in the mutant background also allowed a significant, albeit incomplete, rescue of the muscle defects. In these embryos there is only occasional detachment of muscles and very few signs of apoptotic decay, although the muscles sometimes appear shorter and thicker as compared with normal muscles (compare with ). Analogous overexpression of this mutated Mib2 version in a wild-type background does not have any effects on muscles (unpublished data). Based upon the significant degree of rescue with Mib, we conclude that the RING fingers have a less prominent role in promoting muscle integrity as compared with the other domains, and that ubiquitination may not be the main activity of Mib2 that is required for muscle development.
In sharp contrast to full-length Mib2 and Mib2, Mib1 and Neur are not able to confer any rescuing activity under similar experimental conditions (compare with ), suggesting that Mib2 possesses important targets that are different from those of Mib1 and Neur, and that blocked Notch signaling is not the cause of the observed muscle defects in mutant embryos. This latter point is underscored by our results from experiments with the allele, which never yield any embryos with muscle phenotypes that are similar to those of mutants (unpublished data; see also ). However, we do not exclude the possibility that Mib2 can act in the Notch pathway in other contexts, such as in post-embryonic tissues, which we have not yet examined. It has been shown in cell culture that vertebrate Mib2 is capable of ubiquitinating Delta and Jagged, and Mib2 was also identified as a binding partner of Delta in yeast two-hybrid screenings (; ; ; ).
In summary, Mib2 appears to have a unique and Notch- independent role in “protecting” differentiated body wall muscles from entering apoptosis, undergoing detachment, and being subject to degradation. We speculate that Mib2 is required in a yet undefined pathway for the establishment of specific functional features of the sarcomeres or other structures of the myofibers. In this context, it is interesting to note that mouse Mib2 (also known as skeletrophin) was identified as a binding partner of α-actin and is expressed in skeletal muscles (). The disruption of these unknown structural and functional features in the absence of activity could become detrimental upon stimulation of contractility and trigger entry into apoptosis. Apoptotic degradation of multiple components of the muscle fibers could first result in detachment because the contractile force renders muscle attachment more sensitive to disruptions, and leads to the degradation of the entire syncytia. The identity of the functional target(s) of Mib2 in muscles is currently unknown, as the expression of all markers examined to date, including founder cell markers (Fig. S2, available at ), muscle attachment proteins (), and differentiation markers (; ; and unpublished data), is unaffected in mutant embryos. Future studies, including the identification of interaction partners or mutations in other genes with similar phenotypes, will help to elucidate the pathway in which Mib2 acts to protect muscle integrity. Whether this pathway is involved in preventing skeletal muscle atrophies in which caspase-3 activation contributes to the breakdown of actomyosin complexes of myofibrils () could also be explored.
(), (made by Bruce Hay, CalTech, Pasadena, CA; ), (), P (),
), - (), (), (), (), and the alleles and (), which were induced by EMS and EMS plus γ-rays, respectively, were obtained from the Bloomington stock center. Other fly stocks used include: (), (), (), (), and ().
To generate and Δ transgenic lines, we subcloned the regions coding for amino acids 1–1,050 and amino acids 1–907, respectively, from the EST clone LP14687 (obtained from the Berkeley Genome Project/BDGP) into the pUAST vector and injected the resulting constructs into embryos by using standard protocols. Multiple independent insertions were obtained and analyzed for each construct.
The gene maps genetically at 37B10, a genomic region that was characterized genetically and molecularly in the context of the gene (). By comparing the data of with those from the BDGP, we determined that is uncovered by the overlapping deficiencies , , and (). Based upon additional molecular and genetic mapping data, we identified a complementation group, , of originally five embryonic lethal alleles as the most likely candidate for the gene (; ). The supporting evidence includes: genomic rescue experiments done by with a construct, which according to our analysis only contains and a neighboring gene called , rescued the lethality of and mutations; and one allele, , which no longer exists, was shown to be associated with an ∼800-bp deletion of sequences that we now have identified as being part of the gene.
EST clone LP14687 was fully sequenced and the derived ORF was identical to that of For allele sequencing, the alleles were balanced with a “blue balancer”. Fixed embryos were stained with an antibody against β-galactosidase, and homozygous or mutant embryos were identified by the absence of LacZ expression from the “blue balancer”. Hand-picked embryos of the appropriate genotype were incubated in a solution of 10 mM Tris-HCl, 1 mM EDTA, 25 mM NaCl, and 200 μg/ml proteinase K. Amplified products were purified and subjected directly to automated sequencing. Specific primers were used for sequencing all exons and exon–intron boundaries. For confirmation, the fragment that contained a sequence aberration was reamplified from genomic DNA and resequenced. The mouse Mib2 sequence data are based on our sequencing of the cDNA clone IMAGE:6516763.
For rescue and overexpression experiments the following stocks were generated and used: ΔΔ--Δ----
pET30-Mib2(COOH) was generated by cloning the PCR fragment that code for amino acids 650–1,038 of Mib2 in frame with the 6xHis tag of the pET-30a vector (Novagen). The fusion protein was expressed in and purified with Ni-affinity chromatography under denaturing conditions (QIAGEN). Antiserum production in guinea pigs was done by Covance Research Products and affinity purification was performed against bacterially expressed S-tag (Novagen)-Mib2 fusion protein.
The embryonic mRNA expression was first described in the BDGP in situ hybridization database (). We confirmed and extended these expression data with the use of a digoxygenin-labeled RNA probe that was generated by using the LP14687 cDNA clone and published protocols (). Embryos were photographed with Nomarski DIC optics on a microscope (AX70; Olympus) with a 20× Uplan Fl/0.5 NA objective and a color camera (5.0 RTV; QImaging). Images were acquired with QImaging software and processed with Adobe Photoshop.
Immunocytochemistry was performed essentially as described () and the TSA amplification system was used as needed. Cy3 and FITC were used as fluorochromes. Embryo stainings were analyzed on a confocal microscope (TCS-SP 4D; Leica) with a HC Plan Apo20×/0.7 NA and a HCX Plan Apo40×/0.75–1.25 NA oil objective at 20°C. Generally, Z-scans were taken at 1- to 1.5-μm steps and four to eight Z-scans were merged via maximum projection using the Leica TCS software package or Adobe Photoshop CS. Except for the final adjustment of contrast and brightness with Adobe Photoshop CS, no other processing of the imaging was performed.
Antibodies were used as follows: mouse anti-βgalactosidase (1:100; developed by J. Sanes, Washington University, St. Louis. MO, and obtained through DSHB, Iowa University, Iowa City, IA), rabbit anti-β-galactosidase (1:1500; ICN), rat anti-Tropomyosin (1:50; Babraham Tech), mouse anti-Myosin (1:200; ); rabbit anti-Mef2 (1:700; ); rat anti-αPS2 (1:10, TSA; ); rabbit and mouse anti-Talin (1:750, TSA, and 1:20, TSA, respectively; ); rabbit anti-Pinch (1:15,000, TSA; ); rabbit anti-FAK[pY397] (1:300; Biosource International); and rabbit anti-GFP (1:10,000, TSA; Molecular Probes). Biotinylated (1:200, Vector Laboratories) and fluorescent (1:100, Jackson ImmunoResearch Laboratories) secondary antibodies were also used. The Apoptag kit (Intergen) was used for detecting apoptotic cells as described in .
Figure S1 shows protein sequence alignment of Mib2 with Mib1 and mouse Mib2. Figure S2 shows absence of any effects of mutation on muscle founder marker expression. Figure S3 shows gut phenotype and phospho-FAK staining in mutant embryos. Online supplemental material is available at . |
During cell migration, the protrusive leading edge plays a key role in directional movement (). The leading edge of the migrating cells consists of the two types of actin cytoskeletal architectures, lamellipodia and filopodia. Filopodia is the structure protruding from the edge of the cells that plays an essential role in the wide range of cell motile activities, including cancer cell migration (; ) and neuronal path finding (; ). Although many studies have been conducted on the role of actin-binding proteins in the actin dynamics at membrane protrusion (; ; ), little is known about the role of the actin-based motor protein myosin in filopodia formation.
Recent studies have revealed that myosin-X (myoX) has an important role in the elongation of filopodia (; ; ; ; ; ; ). The N-terminal domain of myoX functions as a motor domain, which is followed by a neck region. The predicted coiled-coil segment is present at the C-terminal side of the neck region (). However, a recent study suggested that this domain does not form a stable coiled coil but instead forms a stable α helix (SAH; ). The C-terminal end of the molecule is the tail domain that was reported as a binding portion to the specific cargo molecules (; ; ; ). Because myoX moves toward the tip of filopodia and transports the cargo molecules, the function of myoX was thought to simply be that of a cargo carrier.
In this study, we report that the motor activity of myoX is itself critical for the initiation of filopodia formation. Using the inducible dimer-forming technique, we found that dimer formation of myoX without the cargo-binding domain can trigger the initiation of microspikes/filopodia in lamellipodia in living cells. Furthermore, the elimination of myoX abolished the actin bundles and microspikes in lamellipodia, and the dimerized myoX can move laterally at the leading edge of lamellipodia. These findings suggest that the motor activity of myoX plays a role in the convergence of actin fibers in lamellipodia, thus forming the base for the initiation of filopodia.
To eliminate the effect of the tail-binding molecules on filopodia formation, we constructed a GFP-tagged tailless myoX (). We produced constructs having the stable two-headed structure because that structure is necessary for continuous movement of the processive myosins (; ; ). We added the coiled-coil domain of myosin-Va (myoV) at the C-terminal end of the SAH domain of myoX (). GFP signals of this construct showed a distinct localization at the tip of filopodia in COS7 cells (). Interestingly, the construct lacking the SAH domain () failed to show the tip localization, although it formed the two-headed structure. On the other hand, GFP-M10MoIQ3SAH and GFP-M10MoIQ3, having no stable coiled-coil domain, failed to localize at the tip (unpublished data). These results suggest that the formation of the two-headed structure and the presence of the SAH domain are required for the movement of myoX toward the tip of filopodia.
To further elucidate the relationship between the dimer formation of myoX and the initiation of filopodia in the living cells, we used the regulated homodimerization system (described in Materials and methods). In the NIH3T3 cells expressing GFP-M10MoIQ3SAH–FK506-binding protein (FKBP), GFP signals were distributed throughout the cell body (). After the addition of the homodimerizer AP20187, GFP signals became distinctly concentrated at the tip of filopodia (). In contrast, GFP-M10MoIQ3SAH, having no FKBP domain, did not show the AP20187-induced tip localization (unpublished data). When we used the heterodimerizer AP21967 as a control, GFP-M10MoIQ3SAH-FKBP did not localize at the tip (unpublished data). These results further support the notion that dimer formation is critical for the movement of myoX to the tip of filopodia.
To further investigate the potential role of the monomer to dimer transition of myoX in filopodia initiation, we performed quantification of microspikes/filopodia in COS7 cells (). The production of microspikes/filopodia of GFP-M10MoIQ3SAH-M5CC was four- to fivefold greater than that of GFP-M10MoIQ3-M5CC (). Furthermore, addition of the dimerizer considerably induced the microspike formation in GFP-M10MoIQ3SAH-FKBP–expressed cells, and the number of microspikes was four- to fivefold greater than without dimerizer (). These results suggest that the initiation of microspikes/filopodia takes place with the dimer formation of myoX.
The aforementioned results indicate that three domains (motor, IQ, and SAH) of myoX are essential for the induction of filopodia. It was hypothesized that the motor activity of myoX plays a critical role in the induction of filopodia. To address this idea, we produced the two constructs having motor-dead mutation in switch I (R220A) or switch II (G437A) of GFP-M10MoIQ3SAH-FKBP. These two mutations were found to inhibit both the localization of myoX at the tip of filopodia and the induction of filopodia even after the addition of dimerizer (). These results support a critical role for the motor activity of myoX in filopodia formation.
It has been proposed that the SAH domain may function as part of the myoX lever (). To investigate a correlation between the neck length and filopodia formation, we constructed the myoX vectors that have different neck lengths by introducing a different number of IQs from myoV (). The introduction of one IQ domain from myoV to the SAH-deleted construct () considerably (four- to fivefold) increased the initiation of microspikes, and the expressed molecules were concentrated at the tip of filopodia (unpublished data). There was no difference in the number of microspike/filopodia between the constructs having one or two additional IQs (). The addition of even more IQs (M5, M5, and M5) did not further increase the number of microspikes (unpublished data).
It was reported that myosin-VIIa (myoVII) also has a SAH domain in the predicted coiled-coil region (). To investigate whether the function of the SAH domain of myoX (M10SAH) is caused by the specificity of the M10SAH structure, we swapped M10SAH for M7SAH of the original construct (). The swapped construct induced the microspike/filopodia after the addition of dimerizer, similar to the original construct (). It was calculated that the length of the SAH domain used in this study was ∼5.4 nm, and one IQ domain was 3.5 nm (). Based on this calculation, the lengths of three IQs and a SAH (), four IQs without the SAH (), and five IQs without the SAH () were 15.9 nm, 14.0 nm, and 17.5 nm, respectively. These results suggest that deletion of the SAH hampered the proper movement of myoX because of the lack of sufficient neck length. The results also indicate that it is important for myoX movement to have the certain minimum neck length. Therefore, it is likely that the SAH domain provides enough span and flexibility for myoX heads to search for the proper binding sites on actin filaments.
To examine whether AP20187 actually induces dimer formation in the cells, we expressed myc-M10MoIQ3SAH-FKBP along with GFP-M10MoIQ3SAH-FKBP-HA in COS7 cells (). The cells cotransfected with the aforementioned two constructs were incubated with or without AP20187. The myoX construct immunoprecipitated with anti-HA antibodies was recognized by both anti-myc and anti-HA antibodies only when the cells were incubated with AP20187 (, lane 4). The result indicates that AP20187 induces the dimer formation of M10MoIQ3SAH-FKBP in cells. It should be noted that the dimer formation of the same construct was not detected with AP21967 (unpublished data). To directly visualize the structure of the molecules of the myoX constructs, the isolated myoX molecules were subjected to electron microscopic observation. Approximately 70% of the molecules of M10MoIQ3SAH-FKBP were two headed in the presence of AP20187, whereas all of the molecules were monomeric in the absence of AP20187 (). These results clearly show that AP20187 induces the two-headed structure of M10MoIQ3SAH-FKBP.
Using the induced dimerization technique described in , we examined whether the formation of the two-headed structure of myoX directly related to the initiation of filopodia in the living cells. The representative images of spreading cells plated on the fibronectin (FN)-coated coverslip are shown in and Video 1 (available at ). Before the addition of the dimerizer, GFP-M10MoIQ3SAH-FKBP showed diffuse localization throughout the cytosol, and fewer than three of the fluorescent puncta (the tip of filopodia) per cell appeared from the cell periphery (, arrowhead). After the addition of the dimerizer, >20 additional fluorescent puncta appeared within 7 min (, bottom). Note that the induced new filopodia were produced only from the active ruffling area and quickly retracted to the edge (Video 1).
and Video 2 (available at ) show representative images of the migrating cells on the FN-coated coverslip. The cell shown in and Video 2 was observed migrating from the bottom to the top of the frame (, arrow). Before the addition of AP20187 (, top), substantial accumulation of GFP-M10MoIQ3SAH-FKBP at the leading edge was observed, and the predominant structure of the leading edge was lamellipodia. Approximately 3 min after the addition of AP20187, several short filopodia appeared from the leading edge and elongated in the direction of the migration (, bottom). However, the short filopodia retracted quickly to the edge of lamellipodia. The protrusion and retraction of the filopodia continued while the cell was moving forward. It should be noted that newly produced filopodia only protruded from the leading edge but never appeared from the lateral or the rear side of the migrating cells (Video 2). In contrast, AP21967 had no inducible effect on the production of filopodia (unpublished data).
It was also discovered that after the addition of AP20187, GFP-M10MoIQ3SAH-FKBP moved laterally along the leading edge and fused with another tip of filopodia ( and Video 3, available at ). It should be emphasized that the lateral movement was not observed before addition of the dimerizer. Lateral movement of the full-length myoX has been reported using another cell line (). The present results suggest that the tail domain is not necessary for the lateral movement of myoX along the leading edge of lamellipodia and that formation of the two-headed structure is critical not only for intrafilopodial movement but also for lateral movement at the leading edge.
To further clarify the function of myoX in filopodia formation, experiments were conducted to eliminate the expression of endogenous myoX, and the effect of the deletion of myoX in actin dynamics was examined. The myoX-specific siRNA markedly reduced the expression of myoX as revealed by both Western blotting (Fig. S1 A) and immunocytochemistry (Fig. S1 B, available at ). In control double-stranded RNA–treated cells, the endogenous myoX localized at the tips of detectable actin filaments (bundles) that are aligned radially at the leading edge in lamellipodium. The elimination of myoX abolished not only the myoX localization at the tip of lamellipodia but also the radial arrangement of actin bundles at the cell periphery (Fig. S1 B). We also used pSIREN-DNR-DsRed plasmid that coexpresses small hairpin RNA (shRNA) and DsRed simultaneously to monitor the transfected cells. Two target sequences were used as a control: mouse specific (siM10m) and human specific (siM10h; Fig. S1 C). 3 d after the transfection, immunostaining showed that the expression of siM10m but not siM10h shRNAs markedly decreased levels of myoX in mouse NIH3T3 cells (Fig. S1 D). It should be noted that siM10m showed an effect on the actin structure similar to myoX-specific siRNA ().
To evaluate the specificity of the myoX knockdown phenotype, we performed rescue experiments. GFP-tagged bovine myoX (GFP-M10MoIQ3SAH-FKBP) that is refractory to siM10m shRNA restored microspike formation in cells expressing siM10m shRNA after the addition of AP20187 (; rescue). On the other hand, GFP-M10F also restored the formation of protrusion from lamellipodium. However, actin bundles were elongated to the long filopodia (unpublished data), which is consistent with a previous study showing that the full-length myoX induces long, stable filopodia (). These results indicate that the siRNA effect is specific to the loss of myoX but not as a result of off-target silencing. These results support the idea that myoX is important for promoting filopodia initiation in lamellipodia.
Based on these results, we propose the following model (). MyoX is present as a dimer and monomer in the cells. The monomeric (single headed) myoX does not localize at the edge of lamellipodia. Once the dimer is produced, myoX moves to the tip of the actin filaments, presumably as a result of its ability to walk on the actin filaments toward the barbed end. The tips move laterally along the leading edge with actin filaments, and the mechanical activity of myoX plays a role in this process. The lateral movement of myoX convergences on the barbed end of the actin filaments, thus producing the base of filopodia where the actin polymerization system might gather to induce parallel actin bundles.
According to the convergent elongation model of filopodia formation, the initiation step of filopodia consists of actin filament convergence and the barbed-end interaction in lamellipodia (). We think that the lateral movement of dimerized myoX powers these movements and the structural changes of actin cytoskeleton in the lamellipodia. The function of the SAH domain can be related to the step size of myoX or the flexibility of the neck domain to search for an appropriate binding site on an actin protofilament. It has been reported that the shortening of the neck length of myoV markedly diminishes the run length (). Because the neck length (the number of IQ motifs) of myoX is one half of myoV, it is thought that the SAH domain functions to help myoX to find the proper actin monomer in the filament, thus facilitating the continuous movement.
Although the tailless myoX can initiate filopodia formation, the filopodia produced are short and unstable. These results suggest that the tail portion of myoX is important for the elongation and stabilization of filopodia and that these processes are likely to be controlled by the cargo molecules binding to the tail of myoX. Recently, it was reported that the unconventional myosin myosin-VI is dimerized after binding to Disabled-2 or the lipids phosphatidylinositol 4,5-disphosphate (). The pleckstrin homology domain of myoX also binds to the lipid phosphatidylinositol 3,4,5-trisphosphate in vitro (). Thus, it is possible that the dimer formation of myoX is induced by lipid binding. The tail domains of myoX have binding partners such as microtubules (), integrins (), and VASP (vasodilator-stimulated phosphoprotein; ). It is plausible that these binding proteins also control the dimer formation of myoX. Understanding the spatio-temporal regulation of the monomer-dimer transition of myoX in cells is a critical problem requiring further study.
The construction of expression vector GFP-M10F was described previously (). The cDNA encoding the first 811 amino acids, including the motor domain and the three IQ motifs, was amplified by PCR and subcloned in frame to pEGFP-C1 (pEGFP-M10MoIQ3). Inclusion of the SAH domain (the first 861 amino acids) was also constructed using the same method (pEGFP-M10MoIQ3SAH). The sequence of the coiled-coil region of mouse myoV (M5CC; 907–1,090 amino acids) was amplified by PCR and fused to the aforementioned expression vectors (pEGFP-M10MoIQ3-M5CC and pEGFP-M10MoIQ3SAH-M5CC). Using the same methods, IQ6 to the coiled-coil region (885–1,090 amino acids) or IQ5 to the coiled-coil region (863–1,090 amino acids) was fused to pEGFP-M10MoIQ3 to create pEGFP-M10MoIQ3-M5-M5CC and pEGFP-M10MoIQ3-M5-M5CC, respectively.
Based on human FKBP and its small molecular ligands (), an FKBP was fused to the C-terminal end of the SAH domain of myoX (GFP-M10MoIQ3SAH-FKBP). The membrane-permeable drug created by chemical cross-linking of the two monomeric ligands with short linker (AP20187) can specifically bind to FKBP. If two FKBPs are present, AP20187 binds to both FKBPs, thus creating a dimer of the targeting molecule. As a control, we used AP21967, a chemically modified derivative of rapamycin that can induce the heterodimerization of FKBP and FRB-containing fusion proteins. To create pEGFP-M10MoIQ3-FKBP and pEGFP-M10MoIQ3SAH-FKBP, a fragment encoding FKBP was isolated from pC4-Fv1E (provided by ARIAD Pharmaceuticals) by restriction digestion and subcloned into pEGFP-M10MoIQ3 and pEGFP-M10MoIQ3SAH. pEGFP-M10MoIQ3-M7SAH-FKBP was created by swapping M10SAH for the rat myoVII SAH domain (M7SAH; 869–926 amino acids).
The sequence of the switch I loop (NNNSSRFG; residues 215–220) of myoX with the exception of the second N is conserved in all nearby myosins sequenced so far (). The R to A mutation in the switch I loop results in loss of the ATP binding ability of skeletal myosin-II () and the actin filament sliding activity of myosin-II in the in vitro motility assay (). The motor domain of myoX also has a conserved sequence of the switch II loop (DIFGFE; residues 434–439) that has a charge interaction with switch I loop. This interaction is critical for ATP hydrolysis by myosin (). Although the G to A mutation in switch II does not abolish ATP binding to the active site, it causes the loss of ATP hydrolysis () and actin sliding activity (). According to the aforementioned results, we made two types of motor-dead constructs having a mutation in switch I (R220A) or in switch II (G437A) by site-directed mutagenesis (, ) of pEGFP-M10MoIQ3SAH-FKBP ().
African green monkey kidney COS7 cells and NIH3T3 fibroblasts (American Type Culture Collection) were cultured in DME supplemented with 10% FCS. Transient transfections were performed with Fugene-6 (Roche Biochemicals) or LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions. At 16 h after transfection, cells were trypsinized and replated on FN for 3 h, fixed, and stained with rhodamine-phalloidin.
Immunofluorescence microscopy was performed as described previously (). In brief, cells were cultured on 10 μg/ml FN-coated coverslips fixed with 4% formaldehyde, 2 mM MgCl, and 1 mM EGTA in PBS for 10 min at RT, treated with 0.2% Triton X-100 in PBS for 5 min, and washed with PBS. Blocking was performed by incubating the fixed cells with 5% BSA in PBS for 60 min at RT. After the antibodies had been diluted with the blocking solution, the cells were incubated at 4°C overnight with the primary antibody and subsequently for 30 min with the secondary antibody. For actin staining, AlexaFluor phalloidin (Invitrogen) was added to the secondary antibody. Specimens were observed at RT using a laser-scanning confocal microscope (DM IRB; Leica) controlled by a confocal microscope system (TCS SP II; Leica) equipped with a Plan-Apochromat 60× 1.40 NA oil immersion objective (Leica) with appropriate binning of pixels and exposure time. The images were processed using Photoshop 7.0 software (Adobe). Immunofluorescence video microscopy was imaged using the same system of immunofluorescence to control illumination shutters and camera exposure (Leica). Time-lapse images were obtained by sequential epifluorescent and phase illumination. The intervals were 10 s, and exposure times were 100–300 ms depending on the time-lapse interval and level of fluorescence. Cells were imaged over periods of 1–30 min at room temperature (25–30°C). Video files were created using QuickTime (Apple).
Spread cells containing prominent lamella with leading edges were chosen from phalloidin-stained samples. The number of actin bundles touching (microspike) or crossing (filopodium) the edge was counted. Only bundles that have fluorescence intensity of at least 1.2 times above the background were considered. Data were analyzed using a test. 50 cells ( and ) and 20 cells () were quantified for each sample. For the rescue experiments, analysis of the number of microspikes/filopodia used cells expressing GFP-M10MoIQ3SAH-FKBP.
Cells were lysed in lysis buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 2 mM MgCl, 0.2 mM EGTA, 1 mM ATP, 0.5% NP-40, 1 mM PMSF, 10 μg/ml leupeptin, 2 μg/ml pepstatin A, and 1 μg/ml trypsin inhibitor). The samples were centrifuged at 10,000 for 20 min at 4°C, and the supernatants were incubated with antibodies conjugated to Affi-Prep Protein A (Bio-Rad Laboratories) for 2 h at 4°C. The precipitates were washed five times in ice-cold lysis buffer. The precipitates were dissolved in 5% SDS and 0.5 M NaHCO buffer. These samples were then subjected to Western blotting as described previously ().
M10-MoIQ3SAH-FKBP-GFP was subcloned into pFastBac-HT baculovirus transfer vector (Invitrogen) containing a hexa-His-tag. The recombinant M10-MoIQ3SAH-FKBP-GFP protein was copurified with calmodulin as reported previously () with an additional incubation with or without 100 nM AP20187 (ARIAD Pharmaceuticals) for 30 min at RT before the cell lysis step. Rotary metal-shadowing electron microscopy of M10-MoIQ3SAH-FKBP-GFP was performed as described previously (). In brief, myoX proteins diluted to ∼4 nM were absorbed onto a freshly cleaved mica surface for 30 s. Unbound proteins were rinsed away, and the specimen was stabilized by brief exposure to uranyl acetate as described previously (). The specimen was visualized by the rotary shadowing technique according to a previously described method () with an electron microscope (model 300; Philips) at 60 kV.
The mouse myoX Stealth siRNA (target sequence 5′-GGAUGUCGGGCUGAUUGAUUCUGUA-3′ corresponding to nt 4,011–4,035 relative to the start codon) was generated by Invitrogen. Control siRNA was purchased from Dharmacon. pSIREN-DNR-DsRed-siM10h and -siM10m were constructed according to the manufacturer's instructions (BD Biosciences). The selected target sequences for siM10h was nt 135–153 of human myoX (GenBank/EMBL/DDBJ accession no. ) and for siM10m was nt 135–153 of mouse myoX (GenBank/EMBL/DDBJ accession no. ). siM10h and siM10m had three base mismatches; thus, siM10h served as a negative control of siM10m. The rescue construct (GFP-M10MoIQ3-SAH-FKBP) made from bovine myoX had four base mismatches to siM10m and were refractory to siM10m siRNA. The siRNA and plasmid transfection were performed using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions. The cells were analyzed 3 d after transfection.
Fig. S1 shows the depletion of myoX in NIH3T3 cells. Video 1 shows the movement of tailless myoX (GFP-M10-MoIQ3SAH-FKBP) in a spreading COS7 cell. Video 2 shows the movement of tailless myoX in a migrating COS7 cell. Video 3 shows lateral movement of the tailless myoX along the leading edge of a migrating cell upon addition of the dimerizer. Online supplemental material is available at . |
Actin cytoskeletal dynamics require the cycling of actin between monomeric and polymeric pools, a transition that is mediated in large part by the actin-binding and filament-severing protein cofilin (; ). Actin assembly is regulated through the generation of free high-affinity actin filament ends, also known as free barbed ends (FBEs). Three mechanisms regulate the availability of FBEs in vivo (). In the first, de novo nucleation mediated by the Arp2/3 complex results in new actin FBEs (ARP2/3-dependent FBEs [ARP]). Severing of noncovalent bonds of existing F-actin () is the second mechanism described for generating FBEs. This mechanism is known to be mediated by members of the actin depolymerizing factor/cofilin family, and so results in cofilin-mediated FBEs (COF; ). The third mechanism for actin assembly involves uncapping of existing actin filament barbed ends. The uncapping mechanism, which results in the removal of actin filament capping proteins such as CapZ and gelsolin, leads to the exposure of FBEs, called uncapping-mediated FBEs (UNCAP) that drive the polymerization of existing filaments.
Members of the Rho family of small GTPases have been demonstrated to play key roles in the regulation of the actin cytoskeleton (; ; ). Rac and Cdc42 have been shown to regulate Arp2/3 de novo nucleation through Wiskott-Aldrich syndrome protein and the Wiskott-Aldrich syndrome protein family verprolin-homologous protein (; ; ; ; ). Using a permeabilized neutrophil model we have previously shown that although Cdc42 regulates the Arp2/3 complex, it is unclear how Rac, and more specifically Rac1 and Rac2, regulate actin assembly (). The development of Rac1 and Rac2 neutrophil-knockout models has allowed us to dissect the specific regulatory roles of these proteins in neutrophil functions (). Previous studies have shown that the Rac1 and Rac2 isoforms carry out distinct roles in the regulation of neutrophil functions, including chemotaxis compass regulation (Rac1 specific) and actin polymerization (predominantly Rac2; ; ; ).
In the present investigation, we used a previously described cell permeabilization technique (; ) in Rac1 and Rac2 knockout neutrophils to further dissect the specific roles of these small GTPases in generating FBEs. We describe here that Rac1 and Rac2 differentially regulate the generation of actin FBE through separate pathways: Rac1 mediates FBE generation during neutrophil chemotaxis through uncapping of existing filaments; Rac2 is responsible for the majority of FBE formation through its mediation of cofilin activation and the ARP2/3 complex.
To assess the relative roles of Rac1 and Rac2 in FBE generation downstream of the methionyl-leucyl-phenylalanine (fMLP) receptor, wild-type (WT), Rac1 null (Rac1N), and Rac2N neutrophils were subjected to a previously described pyrene-actin nucleation assay that measures FBE generation after neutrophil activation (). Analysis of the pyrene-actin polymerization curves, where the slope is proportional to the FBE numbers, demonstrate that both Rac1N and Rac2N neutrophils display a defect in fMLP-mediated FBE generation compared with WT neutrophils (). Although Rac1N cells displayed a modest 30% defect in FBE generation, Rac2N cells exhibited a 70% defect in fMLP-mediated FBE generation, which was consistent with our previous work demonstrating that Rac2N cells are unable to migrate because of an inability to assemble actin during chemotaxis ().
To investigate the mechanisms through which Rac1 and Rac2 regulate actin assembly we used a partial permeabilization assay (). This assay allows us to assess actin filament barbed end uncapping by measuring the release of actin filament capping proteins from existing barbed ends after neutrophil activation (; ). As shown in (A and B), Rac1N neutrophils demonstrated no release of CapZα after fMLP stimulation, whereas WT and Rac2N neutrophils showed a clear release of this protein after fMLP stimulation. Similarly, Rac1 defective mutants failed to release the barbed end capping proteins gelsolin () and adducin (). The release of these two additional uncapping proteins was similar to the CapZα release kinetics (not depicted). Interestingly, Rac2N neutrophils showed a small reduction in gelsolin release compared with the WT cells, suggesting that Rac2 may have a minor role in the regulation of gelsolin uncapping. These results demonstrate that Rac1 is the Rac small GTPase responsible for efficient actin filament uncapping after neutrophil activation (Rac1→UNCAP).
It is clear from several studies that the actin binding protein cofilin has an essential role in the actin-remodeling process and is an essential element in cells undergoing rapid actin cytoskeletal turnover (; ; ). It is clear that cofilin generates FBEs while generating free actin monomers that add on to FBEs at the leading edge of migrating cells (; ; ). To determine if the Rac small GTPases regulate cofilin activity, cofilin phosphorylation at serine 3 was assessed before and after fMLP stimulation in WT, Rac1N, and Rac2N neutrophils. We observed that although WT and Rac1N neutrophils were able to efficiently activate cofilin downstream of fMLP, Rac2N neutrophils failed to dephosphorylate cofilin (activate) after fMLP stimulation (). These data demonstrate that Rac2 is the small GTPase responsible for cofilin dephosphorylation/activation in neutrophils (Rac2→COF).
To verify that cofilin dephosphorylation was required for generating FBE as demonstrated previously (; ), we used a novel approach, which we have recently described (), to transfect primary murine neutrophils with chronophin (CIN). CIN is a haloacid dehydrogenase–type phosphatase that directly dephosphorylates cofilin (). Considering that Rac2N neutrophils had an impaired dephosphorylation of cofilin, we asked whether an overexpression of CIN would rescue the Rac2N phenotype by dephosphorylating cofilin and generating FBEs. Cells were transfected with the constitutively active CIN-WT (GFP-CIN-WT) construct and analyzed for the formation of FBE (; ). The overexpression of GFP-CIN-WT led to increased FBE formation both in fMLP-stimulated cells and in resting cells (). Importantly, Rac2N cells transfected with GFP-CIN-WT significantly increased FBE formation (P < 0.03), displaying levels comparable to the fMLP-stimulated WT-nontransfected group. To verify that the transfection protocol did not influence FBE generation, neutrophils were transfected with a GFP vector control and displayed no significant differences when compared with their nontransfected counterparts (P > 0.05). The overexpression of active CIN confirms that cofilin dephosphorylation () leads to increased FBE generation in neutrophils ().
Previous studies have demonstrated the importance of cofilin in FBE formation at the leading edge and in generation of protrusive force in migrating cells (; ). To further confirm that Rac2 regulates cofilin activation and subsequent FBE generation at the leading edge of migrating cells, we investigated the subcellular localization of total cofilin and phosphorylated cofilin (P-cofilin) in WT, Rac1N, and Rac2N neutrophils. Epifluorescence microscopy images show that WT and Rac1N neutrophils display abundant cofilin and complete absence of P-cofilin in the F-actin rich leading edge during fMLP-mediated chemotaxis (). However, Rac2N cells show no loss of P-cofilin from the same leading edge of fMLP-activated neutrophils (). These results confirm that Rac2 is required for mediating cofilin dephosphorylation and subsequent FBE generation at the leading edge of chemotaxing neutrophils. As we have described previously in and reflected in , Rac knockouts demonstrate distinct morphological changes when compared with WT cells. Although Rac1N neutrophils display an abnormally elongated morphology caused by impaired Rho activation (), Rac2N neutrophils polarize and orient toward the source of chemoattractant but fail to migrate efficiently, resulting in a poorly defined uropod ().
Although we recognize that there is likely to be interplay and interdependence between the three FBE generation mechanisms, we have used our findings and observations to model and approximate the relative roles of uncapping, cofilin severing, and Arp2/3 nucleation in FBE generation downstream of the fMLP receptor. At present, three primary mechanisms have been identified as being responsible for generating FBE in rapidly migrating cells such as neutrophils and cancer cells (; ). In WT neutrophils, FBEs are generated through either Arp2/3 (ARP), uncapping (UNCAP), and/or cofilin severing (COF): WT = ARP + UNCAP + COF. To determine the proportion (%) that each of these mechanisms contributes to total fMLP-mediated FBEs we used the five following observations to solve the above equation (, top left).
First, neutrophils without both Rac1 and Rac2 (double null) show an uncapping and a cofilin dephosphorylation–severing defect. Double null neutrophils treated with the CA domain peptide from WASP, which inhibits Arp2/3-mediated nucleation and was previously described in and , do not show any detectable increase in FBEs downstream of the fMLP receptor (unpublished data), confirming the validity of the equation WT = ARP + UNCAP + COF.
Second, because we know that the CA peptide inhibits de novo nucleation through the Arp2/3 complex, we could determine the relative role played by Arp2/3 in the FBE formation downstream of fMLP stimulation by comparing the CA-pretreated and nontreated neutrophils after fMLP stimulation (ARP = FBE − FBE).
Third, because we know that Rac1 is responsible for uncapping (UNCAP; ), then Rac1
= ARP + COF. We are now able to eliminate ARP from the equation by using the CA peptide. As demonstrated in , this allows us to conclude that Rac1
+ CA = COF = 10% of total FBEs.
Fourth, based on our findings that Rac2 is responsible for COF ( and ), FBE formation in Rac2N cells is attributed to filament uncapping and Arp2/3 (Rac2
= ARP + UNCAP). We are able to eliminate ARP from the equation by using the CA domain peptide. As demonstrated in , this allows us to conclude that Rac2
+ CA = UNCAP = 15% of total FBEs. In addition to this observation, we also show that Rac2 is the primary regulator of Arp2/3, responsible for ∼60% of total ARP.
In the present investigation, we analyzed the specific roles of Rac1 and Rac2 on FBE formation downstream of the fMLP receptor in neutrophils. We describe for the first time that Rac1 and Rac2 differentially regulate FBE formation in neutrophils through uncapping and Arp2/3/cofilin, respectively (). Here we show that Rac1 is responsible for UNCAP formation (∼15%). We also show that Rac2 is the key regulator of FBE formation in murine neutrophils (∼70%) by regulating both actin depolymerizing factor/cofilin (COF, ∼10%) and the Arp2/3 complex (ARP, ∼60%). Previous studies have demonstrated that cofilin is a major generator of FBEs in migrating cancer cells (; ). Recently, also showed that cofilin directly initiates FBE formation and works synergistically with Arp2/3 to create a burst of actin nucleation. Using a different approach, demonstrated that cofilin is also required to generate the necessary supply of actin monomers, which add on to FBEs during actin assembly at the leading edge of migrating cells. Thus, cofilin appears to have dual roles in the leading edge of migrating cells through its generation of FBEs and its critical role in supplying free monomers required for actin assembly. Our experimental approach measuring FBE generation, using incorporation of pyrene actin in Rac knockouts treated with CA peptide, has allowed us to confirm and quantify the relative role of cofilin in FBE generation. From our data we are able to demonstrate that cofilin does make a small but considerable contribution to FBEs (COF, ∼10%) in neutrophils downstream of fMLP activation, independent of its critical role in regulating the availability of free actin monomers. Importantly, our observation that Rac2 is the primary regulator of the Arp2/3 complex is consistent with previous work showing that Cdc42 activation in these murine neutrophils is downstream of Rac2 (). Future studies will focus on the mechanisms through which Rac2 regulates cofilin activity.
All procedures described were performed in accordance with the Guide for the Humane Use and Care of Laboratory Animals and were approved by the University of Toronto Animal Care Committee. Rac1-conditional null and Rac2 mice were generated according to a protocol described in . In brief, Rac1 was selectively disrupted in granulocytes/neutrophils by using a conditional Rac1LysM in which the Cre recombinase is expressed under the control of the murine lysozyme M gene regulatory region. This approach generated Rac1 deletion in neutrophils at birth (). Rac1 conditionals were bred with Rac2N mice () and the resulting offspring enabled the generation of mice with neutrophils deficient in either Rac1, Rac2, or both. Genotyping for Rac1, Rac2, and LysM alleles was performed as described in . Neutrophil isolation was performed as described in . More than 85% of cells isolated were neutrophils as assessed by Wright-Giemsa staining. Viability as determined by trypan blue exclusion was >90%.
Isolated bone marrow neutrophils were exposed to 1 μM fMLP at 37°C for 60 s and immediately subjected to 15% SDS-PAGE. Nitrocellulose membranes were incubated overnight with 1:1,000 phosphocofilin (Ser3) and cofilin antibodies (Cell Signaling Technology) in TBS-Tween 20 solution and 5% fat-free milk. Membranes were incubated with goat anti–rabbit IgG peroxidase conjugates (GE Healthcare) followed by chemiluminescence visualization (ECL; PerkinElmer). Immunoblots were scanned (300 dpi; Perfection 1250; Epson) and analyzed by densitometry (ImageJ 1.35s; National Institutes of Health). For cofilin and P-cofilin immunostaining, 10 neutrophils were cultured on BSA-coated slides for 10 min at 37°C and stimulated with fMLP for 1 min followed by fixation with 4% PFA. Fixed cells were washed in PBS, permeabilized with 0.1% Triton X-100 in PBS for 4 min, and blocked with 1% BSA for 30 min. Cells were incubated with anti-phosphocofilin or anti-cofilin (Cell Signaling Technology), diluted 1:50, and detected with Alexa Flour 488 goat anti–rabbit IgG (Invitrogen). Cells were also stained with 1:400 rhodamine phalloidin at room temperature for 30 min followed by epifluorescence microscopy analysis. Images were visualized using an Eclipse E100 (Nikon), a 40/0.95 Plan Apo objective (Nikon), and a digital camera (C4742-80; Hamamatsu). Images were acquired using Simple PCI software version 5.3 (Compix). All figures were created using CorelDRAW 12.0.
To analyze actin nucleation activity, we determined the ability of permeabilized neutrophils to accelerate spontaneous actin assembly measured as enhancement of pyrene-actin fluorescence with polymerization (). We permeabilized neutrophils (5 × 10/ml) for 10 s using 0.2% OG (PHEM buffer containing 10 μM phallacidin, 42 nM leupeptin, 10 mM benzamidine, and 0.123 mM aprotinin). We stopped the permeabilization process by diluting the detergent with 3 vol of buffer B (1 mM Tris, 1 mM EGTA, 2 mM MgCl, 10 mM KCl, 5 mM β-mercaptoethanol, and 5 mM ATP; pH 7.4) and then stimulated the cells with 1 μM fMLP for 60 s. We then assayed for FBEs by adding pyrene-labeled rabbit skeletal muscle actin () to a final concentration of 1 μM and followed the fluorescence increase using a microplate reader (FLUOstar optima; BMG Labtech) with fluorescence excitation and emission wavelengths of 355 and 405 nm, respectively. For some experiments cells were stimulated with fMLP for 60 s after cell permeabilization. There were no differences in FBE numbers whether the cells were stimulated before or after permeabilization, as this technique maintains receptor signaling to actin assembly after the brief OG permeabilization (). To confirm that we were measuring FBEs and not pointed ends, 2 μM cytochalasin D (CD) was added to block barbed ends. In all experiments >97% of all fluorescence increase was inhibited by the CD (). The CA peptide was used to block Arp2/3 complex–mediated nucleation (). As described in , 1 μM CA peptide was present in the media at the time of cell permeabilization to allow for peptide entry into the permeabilized cells, and this was then followed by fMLP stimulation and pyrene incorporation quantification as described in this paragraph.
Gelsolin and CapZ were measured in the supernatant of OG-permeabilized cells and on the respective cell lysates. One million murine neutrophils were permeabilized with 0.2% OG buffer and then stimulated with 1 μM fMLP for up to 60 s. The supernatants were collected at 10, 30, and 60 s and subjected to SDS-PAGE electrophoresis. The remaining cells were lysed and collected separately, followed by SDS-PAGE electrophoresis. Proteins were transferred to a membrane and blotted with 1:2,000 anti-gelsolin, 1:2,000 anti–adducin α (Santa Cruz Biotechnology, Inc.), or 1:5,000 anti-CapZ primary antibodies (BD Biosciences). The anti-gelsolin antibody was provided by C.A. McCulloch (University of Toronto, Toronto, Canada). The release of capping proteins was evaluated by the amount of capping proteins found in the supernatant of permeabilized cells. Results were analyzed with ImageJ 1.35s.
Primary neutrophils were transfected using a previously described protocol (). In brief, murine neutrophils were suspended in 100 μl of Nucleofector Solution V (Amaxa Biosystems) and supplemented with 6 μg of vector DNA of WT CIN tagged with GFP (GFP-CIN-WT) or GFP control vector (). The GFP-CIN-WT construct was supplied by G.M. Bokoch (The Scripps Research Institute, La Jolla, CA). The pmaxGFP DNA construct (Amaxa Biosystems) was used as the control. The cells were transfected using the program Y-001 (Amaxa Biosystems). The cells were carefully recovered with 1,000 μl of 37°C Iscove's Modified Dulbecco's medium and transferred to 1 ml of Iscove's Modified Dulbecco's medium with 10% FBS, followed by a 2-h recovery time in a humidified 37°C 5% CO incubator. Cells were allowed to recover on nontissue culture 12-well plates. In some experiments, the cells were subjected to SDS-PAGE electrophoresis and immunoblotted for P-cofilin to monitor cofilin activation (dephosphorylation). A >45% transfection efficiency was achieved using this protocol.
Statistical analysis was performed using the 14.0 software (SPSS Inc.). Multiple comparisons were performed by analysis of variance associated with the Bonferroni and Tukey's honestly significant difference tests for post hoc testing. All results represent at least three independent experiments. Statistical significance was defined as P < 0.05. Data are expressed as mean ± SEM. |
Cerebral cavernous malformations (CCMs) are vascular malformations mostly found within the central nervous system. CCMs can occur in sporadic or autosomal dominant inherited forms, the latter of which map to three loci, KRIT-1/CCM1, MGC4607/OSM/CCM2, and PDCD10/CCM3 (; ; ; ). KRIT-1 protein was detected in endothelial cells by Western blot, immunofluorescence, and immunohistochemistry (; ,), but this work was challenged because KRIT-1 mRNA was not detected in the endothelium (). However, CCM lesions are composed of endothelial cells (; ) and can occur outside the brain (). Furthermore, mice lacking KRIT-1 die because of defective vascular development but have apparently normal brain development (), all of which suggest that the primary defect in CCM lesions is in the endothelial compartment.
KRIT-1 possesses four ankyrin repeats, a band 4.1/ezrin/radixin/moesin (FERM) domain, and multiple NPXY sequences, one of which is essential for integrin cytoplasmic domain-associated protein-1α (ICAP1α) binding () and all of which mediate binding of CCM2 (; ). KRIT-1 was identified as a Rap1-binding protein in yeast two hybrid experiments (), and the FERM domain of KRIT-1 binds with 10-fold higher affinity to Rap1 than to H-Ras (). Although the interaction of Rap1 and full-length KRIT-1 has been disputed (), we have confirmed the association by coimmunoprecipitation (unpublished data). Rap1 regulates cell–cell junctions in both endothelial and epithelial cells (; ; ). The disruption of endothelial cell junctions in CCM suggests that KRIT-1 may have a role in the capacity of Rap1 to regulate endothelial cell–cell junctions. Here, we report that KRIT-1 is expressed in endothelial cells where it is present in cell–cell junctions and associated with junctional proteins. The junctional localization of KRIT-1 is mediated by its FERM domain and regulated by Rap1 activity. Furthermore, we find that KRIT-1 is required for the stabilizing effect of Rap1 on endothelial cell–cell junctions. Together, these data establish that KRIT-1 is a Rap1-binding protein that regulates endothelial junction integrity and may provide a molecular explanation for aspects of the CCM phenotype.
Immunoprecipitation with an mAb anti–KRIT-1 (15B2) followed by Western blotting with an affinity-purified pAb (Rb6832) was performed to increase the sensitivity of detection of endogenous KRIT-1. We detected a band of ∼80 kD in bovine aortic endothelial cells (BAECs; ) and human umbilical vein endothelial cells (HUVECs; ). The mobility of this band is consistent with the calculated molecular mass of KRIT-1 (81 kD). Moreover, CHO cells also exhibited this 80-kD band, and when transfected with authentic HA-tagged human KRIT-1 exhibited a second intense band of slightly lower mobility consistent with the presence of the HA tag (). An unrelated antibody (glyceraldehyde 3-phosphate dehydrogenase [GAPDH]) did not immunoprecipitate KRIT-1, nor did nonimmune mouse IgG. We used siRNA-mediated knockdown of endogenous KRIT-1 to confirm that the endogenous 80-kD polypeptide was authentic KRIT-1. Transfection with KRIT-1–specific siRNA (530; ) completely eliminated this band from BAECs and reduced it by 70% in HUVECs, but had no effect on the expression of a variety of cell–cell junction proteins (). An irrelevant (, GAPDH) and a negative control siRNA (NC) had no effect on the expression of this polypeptide.
Having shown the antibodies' specificity in immunoprecipitation blotting experiments, we then used the mAb anti–KRIT-1 to detect endothelial cell KRIT-1 by immunofluorescence. KRIT-1 staining was enriched at sites of cell–cell junctions and the nucleus in confluent bovine aortic endothelial monolayers (), but was absent from free membranes (not depicted), suggesting cell junction localization consistent with the reported localization of recombinant KRIT-1 in transfected HEK293 cells (). Staining specificity was confirmed by its marked reduction after transfection with the KRIT-1 siRNA 530 (). Similar staining was seen in HUVECs and was also observed with the affinity-purified pAb anti–KRIT-1. No staining was seen with nonimmune rabbit or mouse IgG (unpublished data). Thus, KRIT-1 protein is expressed in both arterial and venous endothelial cells in culture, which is consistent with its reported presence in endothelium in vivo ().
Confocal microscopy demonstrated that KRIT-1 staining was enriched in cell junctions and colocalized with β-catenin, a cell–cell junction protein (; ). In KRIT-1 siRNA 530–treated cells, β-catenin localization to the cell–cell junction was disrupted but global expression of junctional proteins was unchanged ( and ). β-catenin also physically associated with KRIT-1, as did p120-catenin and the junctional scaffold AF-6 (also called afadin; ), as judged by their presence in KRIT-1 immunoprecipitates (). The coimmunoprecipitation was done under conditions that garnered near complete recovery of KRIT-1 from the cell lysate. From this, we estimate that KRIT associates with ∼2% of total cellular β-catenin and p120-catenin and 6% of total AF-6. Nevertheless, the intensities of the KRIT-1 bands were similar to those observed for the junctional proteins, suggesting that KRIT-1 is not an abundant protein. This idea is supported by the requirement for prior immunoprecipitation to visualize KRIT-1 by immunoblotting (). The association of KRIT-1 with AF-6 is noteworthy as AF-6 can inhibit Rap1 activity (; ) through recruitment of Rap GTPase activating proteins (GAPs). Thus, the interaction of KRIT-1 with AF-6 could act as a negative feedback modulator of Rap activity. Also present in coprecipitates, but not shown here, were vascular endothelial cadherin and α-catenin. In contrast, ZO-1, a tight junction marker, and talin, a focal adhesion protein absent from adherens junctions (), were not detected in KRIT-1 immunoprecipitates ().
KRIT-1 associates with Rap1 small GTPase and exhibits a preference for Rap versus Ras in vitro (; ). We therefore asked whether Rap1 could regulate the localization of KRIT-1 to the junctions and its association with junctional proteins. Transfection of cells with an activated Rap1 (Rap1A-G12V, referred to as RapV12) increased the association of endogenous KRIT-1 with β-catenin and AF-6 (). Expression of the inhibitor of Rap activity (Rap1GAP) inhibited the association of KRIT-1 with β-catenin and AF-6 to a greater extent than the increase elicited by activated Rap1A. The strong effect of Rap1GAP suggests the presence of basal Rap activity in BAECs. Neither RapV12 nor Rap1GAP expression affected KRIT-1 expression (). Furthermore, thrombin treatment led to a dramatic loss of KRIT-1 from the junctions () but did not alter KRIT-1 expression (Fig. S1 A, available at ). Thrombin's effect was counteracted by the addition of an activator () of the RapGEF Epac-1, 8-pCPT-2′--methyladenosine-3′,5′-cAMP (8-pCPT-2′--Me-cAMP; ) . RapV12 also reversed the thrombin-stimulated loss of KRIT from the junctions, supporting the idea that effect of 8-pCPT-2′--Me-cAMP is caused by the activation of Rap1 and that KRIT-1 association with cell–cell junctions is increased by Rap1 activation.
To explore the mechanism of KRIT-1 targeting of endothelial cell–cell junctions, we examined the capacity of different domains of the protein to mediate junctional localization. The isolated KRIT-1 FERM domain (GFP-KRIT-F123; ) colocalized with β-catenin in cell–cell junctions (). Furthermore, the KRIT-1 FERM domain was physically associated with β-catenin (). In sharp contrast to the intact protein, the interaction of GFP-KRIT-F123 with β-catenin was insensitive to the activation state of Rap1 (), suggesting that active Rap1 may increase the association of KRIT-1 with junctional proteins by increasing the availability of this domain. In contrast to the FERM domain, the KRIT-1 N terminus () did not colocalize with either junctional β-catenin () or ZO-1. Furthermore, the N terminus did not associate with β-catenin in coimmunoprecipitation experiments (). Moreover, the KRIT-F23 fragment of the FERM domain failed to localize to junctions, suggesting that the F1 region is required for localization to the junction. Thus, KRIT-1 localization to endothelial junctions is mediated by the KRIT-1 FERM domain and regulated by Rap1.
CCM lesions display characteristics typical of a loss of endothelial cell–cell junctions (). As noted previously, KRIT-1 is a relatively specific Rap1 effector (; ), and Rap1 regulates the integrity of cell–cell junctions in endothelial and epithelial cells (; ; ), leading us to examine the role of KRIT-1 in Rap1-induced stabilization of endothelial junctions. We noted that there was extensive redistribution of β-catenin out of cell–cell junctions in KRIT-1–depleted BAECs (), suggesting that KRIT-1 might be required for the stability of cell–cell junctions in these cells. As a test for the function of these junctions, we examined the permeability of endothelial cell monolayers to HRP (). KRIT-1 siRNA transfection caused an approximately twofold increase in permeability () that was reversed by reconstitution with recombinant KRIT-1. Furthermore, treatment of KRIT-1–depleted cells with 8-pCPT-2′--Me-cAMP had little effect on permeability (). In sharp contrast, 8-pCPT-2′--Me-cAMP reversed the increase in permeability caused by thrombin as expected (; ). Thrombin treatment caused a loss of KRIT-1 from the junctions () and a concomitant redistribution of β-catenin away from the junction (Fig. S1 B); however, thrombin treatment did not disrupt the interaction of KRIT-1 and β-catenin, suggesting that they remain associated as they redistribute away from the junction (Fig. S1 A). Thrombin increased permeability further in KRIT-1–depleted cells. Furthermore, overexpression of KRIT-1 partially reversed the increased permeability induced by expression of Rap1GAP (). Thus, KRIT-1 regulates endothelial permeability and is required for Rap1-mediated stabilization of endothelial junctions.
We have shown here that Rap1 activation leads to association of KRIT-1 with endothelial cell–cell junction proteins. The KRIT-1 N terminus has binding sites for ICAP1α and CCM2 (, ; ). Consequently, recruitment of KRIT-1 through its FERM domain has the potential to bring these proteins to the junction. Both ICAP1α and CCM2 are involved in the regulation of Rho family GTPases (; ), which are known to control the integrity of endothelial junctions via the actin cytoskeleton (). Furthermore, activation of Rap1 by 8-pCPT-2′--Me-cAMP inhibits thrombin-induced RhoA activation in endothelial cells (). Hence, we examined F-actin distribution after depletion of KRIT-1 with siRNA. In control cells, actin was distributed around the circumference of each cell, and few stress fibers were observed. This contrasts strongly with KRIT-1 siRNA–treated cells, which had abundant stress fibers. Overexpression of HA-KRIT in control cells increased cortical actin staining, and HA-KRIT reexpression restored cortical actin morphology in KRIT-1 siRNA–treated cells (). As changes in actin morphology are linked to altered permeability (; ; ), this result suggests that KRIT-1 may act as a scaffold that regulates endothelial junctions by recruiting modulators of the Rho family GTPases that control the actin cytoskeleton.
CCM lesions are composed of a bed of leaky capillaries (; ). Thus, our finding that the loss of KRIT-1 increases endothelial permeability provides a direct link between the pathogenesis of CCM and KRIT-1 function. KRIT-1 is a relatively specific effector of Rap1, which is a regulator of endothelial permeability and the stability of cell– cell junctions (; ). We now find that the capacity of Rap1 to stabilize endothelial cell junctions depends on KRIT-1 and that Rap1 regulates the junctional localization of the KRIT-1 protein and increases the association of KRIT-1 with junctional proteins. The KRIT-1 FERM domain mediates its association with junctional proteins and the isolated FERM domain associates with junctional proteins in a Rap1-independent manner. This suggests that the availability of the KRIT-1 FERM domain may be regulated by the binding of active Rap1. This would allow KRIT-1 to associate via the FERM domain with the junction complex, thus stabilizing cell–cell junctions, perhaps through modifying the activity of Rho family GTPases. In summary, Rap1 stimulates the junctional localization of KRIT-1 via KRIT-1's FERM domain, and KRIT-1 is required for Rap1 stabilization of endothelial junctions. These data provide a molecular linkage between KRIT-1 protein function and the CCM phenotype and identify a Rap effector that regulates cell–cell junctions.
BAECs were a gift of S. Shattil (University of California, San Diego, La Jolla, CA). BAECs were cultured in DME with 10% calf serum (CS) and 1% penicillin/streptomycin (Invitrogen). HUVECs were obtained from Cambrex and cultured in endothelial growth medium 2 (Cambrex). CHO cells were obtained from American Type Culture Collection. CHO cells were cultured in DME with 10% fetal bovine serum plus 1% penicillin/streptomycin, 1% -glutamine, and 1% nonessential amino acids (Invitrogen).
For permeability, immunofluorescence, and cotransfection studies, BAECs were transfected using a nucleofection device (Amaxa). In brief, 0.5 × 10 cells per transfection were suspended in Basic endothelial cell solution (Amaxa) together with 25 nM siRNA with or without 1 μg DNA. The cells were then nucleoporated using program M-003 (Amaxa). After recovery at 37°C for 10 min, the cells were plated as required. This method garnered transfection efficiencies from 70 to 90%.
Polyclonal anti–KRIT-1 6832 was developed against the recombinant KRIT-1 FERM (F123) domain coupled to keyhole limpet hemocyanin. Monoclonal anti–KRIT-1 antibodies were also developed using the recombinant KRIT-1 FERM (F123) domain as the antigen. Mice were immunized with the KRIT-1 FERM domain in incomplete Freund's adjuvant. Mouse sera were titered by ELISA against GST-F123 before fusion of splenic cells with myeloma cells. After fusion, single hybridoma cells were plated by limiting dilution, and antibody production was assayed in the hybridoma supernatant by ELISA. Titers were done in parallel on GST-coated plates to assess background binding. Hybridomas with high titer were selected for subcloning, and the process was repeated twice. Selected hybridoma supernatants were further purified by affinity chromatography on a protein G–Sepharose column.
For endogenous coimmunoprecipitation experiments, cells expressing KRIT-1 were scraped into 500 μl of lysis buffer containing 50 mM Tris, pH 7.4, 150 mM NaCl, 0.5% NP-40, and 5 mM MgCl plus a protease inhibitor cocktail (Roche). After resting 5 min on ice, the lysate was incubated for 15 min at 4°C with rocking. The lysate was spun down at 14,000 rpm for 10 min and the protein concentration of the supernatant was determined using a bicinchoninic acid assay (Pierce Chemical Co.). 500 μg of precleared total cell protein was added to 2 μg of immunoprecipitating antibody and incubated at 4°C with continuous rocking for 2 h. 10 μl of a 50% slurry of protein G–Sepharose beads (GE Healthcare) was then added and the rocking of the samples was continued overnight. The immunoprecipitated samples were washed three times with lysis buffer and solubilized with 10 μl SDS-PAGE sample buffer. Samples were resolved on 4–20% polyacrylamide gels (Invitrogen) in SDS-PAGE buffer and transferred to nitrocellulose membranes.
Concomitant to the leak assay, a portion of the transfected cells was plated on FN-coated glass coverslips and grown to confluence (∼24 h) in low-glucose DME/10% CS. Cells were serum starved (0.5% CS) for 4 h then treated for 30 min with 100 μM 8-pCPT-2′--Me-cAMP or the vehicle alone (control), and indicated cells were treated with 2 U/ml thrombin for 60 min. Slips were fixed with 3.5% formaldehyde for 2 h, permeabilized for 10 min with 0.15% Triton X-100 and washed with TBST. Slips were blocked with 10% NGS for 60 min at RT and washed again. 200 μl/slip of primary antibody was incubated overnight at 4°C in a humidified chamber. The control stain was mouse IgG at 1:1,000. After washing, donkey anti–mouse IgG Alexa 488 (Invitrogen) was added at a 1:1,000 dilution overnight at 4°C. Coverslips were washed six times in alternating PBS/TBST rinses and mounted on 10 μl Prolong Gold mounting medium (Invitrogen) and photographed using a confocal microscopy system. Images were processed using Photoshop. For actin staining (), cells were incubated with a 1:200 dilution of Alexa 488 or Alexa 568–phalloidin (6 μM in methanol; Invitrogen) overnight at 4°C. Coverslips were washed and mounted as above and imaged using a microscope (DMLS; Lecia NPlan 40× 0.65 objective) with a camera (SPOT RT Color-2000; Diagnostic Instruments).
The permeability of the endothelial monolayer was evaluated using a leak assay originally described in . In brief, BAECs were grown to semiconfluence. Cells were transfected with 25 nM of negative control siRNA or KRIT-1 siRNA 530 with or without 1 μg pcDNA3.1 HA–KRIT-1. Transfected cells in phenol-free DME/10% CS were plated into 3-μm-pore polyester FN-coated transwell filters (Corning). Filter-plated cells were incubated for 48 h at 37°C to full confluence. Cells were then incubated in serum-free, phenol-free DME for 2 h. As indicated in the figure legends, in some conditions cells were treated with 100 μM 8-pCPT-2′--Me-cAMP for 30 min, and then half were treated with 2 U/ml thrombin (GE Healthcare) for 30 min. 50 μl of phenol-free DME containing 1.5 μg/ml HRP (Sigma-Aldrich) was added to upper wells for an additional 30 min. Filters were removed from outer wells and fixed in 3.5% formaldehyde and later stained with 0.25% Coomassie blue and examined by phase-contrast microscopy to reconfirm the integrity of cell monolayers.
The HRP content of the lower chamber medium was measured using a microplate peroxidase colorimetric assay. 100-μl-per-well guaiacol/sodium phosphate assay buffer (1:1) was added to 25 μl of each sample in triplicate. 25 μl of freshly made 0.6-mM HO in ddHO was added to each well for ∼15 min or until color developed. Reaction was stopped with 10-μl-per-well 2N HSO. A was acquired and raw absorbance values were normalized as a percentage of control untransfected (vehicle) cell sample absorbance. Data was analyzed for statistical significance using analysis of variance and SigmaStat software (Jandel).
Fig. S1 shows the effect of thrombin treatment on KRIT-1 expression, association with β-catenin, and β-catenin localization. Online supplemental material is available at . |
Accurate segregation of replicated chromosomes during mitosis is essential for the maintenance of genomic integrity. To ensure faithful chromosome segregation, eukaryotic cells have developed a surveillance network called the mitotic checkpoint that delays anaphase onset until sister kinetochores of duplicated chromosomes are properly attached to microtubules emanating from opposite spindle poles (for reviews see ; ). Early in mitosis, various mitotic checkpoint proteins, including Bub1, Bub3, BubR1, Mad1, Mad2, and Mps1, are recruited to unattached kinetochores. These kinetochore-associated checkpoint proteins promote the formation of diffusible Mad2, BubR1, Bub3, and Cdc20 protein complexes that inhibit the anaphase-promoting complex/cyclosome (APC/C), an E3 ubiquitin ligase that drives cells into anaphase by targeting securin and cyclin B for destruction by the 26S proteasome (for reviews see ; ; ). After all chromosome pairs are properly attached to the spindle and aligned in the metaphase plate, mitotic checkpoint proteins dissociate from the APC/C, thus triggering the ubiquitin-mediated destruction of securin and cyclin B. Separase, a protease that is inhibited by securin binding and cyclin B/Cdk1-mediated phosphorylation, then cleaves the kleisin subunit Scc1 of cohesin, thereby allowing sister chromatid disjunction and anaphase onset (for reviews see ; ).
The discovery of the mitotic checkpoint led to speculation that mutations in mitotic checkpoint genes might play a role in the development of aneuploidy in human cancers (; ). Over recent years, mutant mitotic checkpoint genes have indeed been identified in various human cancers, although at relatively low frequency (; for reviews see ; ). The Bub1 kinase is mutated in several cancer types, including colorectal, lung and thyroid cancer, and T cell leukemia (; ; ; ). In addition, Bub1 expression is frequently reduced in several human cancers, including colorectal, gastric, and esophageal cancers (; ; ).
Bub1 is a serine/threonine protein kinase that targets to unattached kinetochores at the onset of mitosis (; ; ). There it is thought to phosphorylate Cdc20, thereby preventing Cdc20 from activating the APC/C (; ; ). Additionally, Bub1 is required for Mad1–Mad2 localization to unattached kinetochores. These complexes function to prevent premature APC/C activation by changing the conformation of monomeric Mad2 such that it can efficiently bind to and inhibit the APC/C coactivator Cdc20 (, ; ). Besides Mad1 and Mad2, Bub1 also recruits BubR1, Bub3, centromere protein E (CENP-E), and CENP-F to unattached kinetochores (; ). Several of these proteins are important for microtubule-kinetochore attachment, which may explain why Bub1-depleted cells have chromosome congression defects (). In addition, Bub1 contributes to the stability and inner centromere localization of Shugoshin (Sgo1), a protein that functions as an adaptor for phosphatase PP2A (, ; ). At the inner centromere, PP2A counteracts the Plk1-mediated release of cohesin until anaphase onset, thus preventing the premature separation of sister centromeres (). Bub1 also controls the stability and correct positioning of the chromosomal passenger complex to the inner centromeric region of sister chromosomes, a function that appears to be critical for the recruitment of Sgo1 to centromeres ().
Although the molecular mechanisms of Bub1 action are beginning to emerge, the physiological role of Bub1 in higher eukaryotes is still unknown. The most definitive way to address this role would be to generate Bub1 knockout mice by homologous recombination in embryonic stem (ES) cells. However, previous gene knockout studies for Mad1, Mad2, BubR1, and Bub3 revealed that these mitotic checkpoint proteins are essential for cell proliferation, causing mice to die during the early stages of embryonic development (; ; ; ; ; ). Anticipating that Bub1-null mice would be embryonically lethal as well, we generated a series of mice in which the expression of Bub1 protein is reduced in a graded fashion from normal to zero. We find that Bub1-null mice are indeed embryonically lethal but that mice with very low levels of Bub1 protein are viable. Here, we show that Bub1 deficiency is associated with aneuploidy and spontaneous tumorigenesis in a dose-dependent fashion. Furthermore, we provide evidence for a novel role of Bub1 in eliminating cells that have undergone chromosome missegregation.
By homologous recombination, we inserted a neomycin-resistance (Neo) gene flanked by a site into intron 8 and a site into intron 9 of the mouse Bub1 gene (). This created a hypomorphic allele (called Bub1) because the Neo gene harbors a cryptic exon that is known to reduce the level of normally spliced messenger RNA (; ; ). Correctly targeted ES clones were injected into blastocysts, and Bub1 offspring were obtained from the resulting chimeras (). Bub1 mice were established by crossing Bub1 males with transgenic females that express Cre recombinase in the germline (). Both Bub1 and Bub1 mice were healthy and indistinguishable from wild-type littermates. Subsequent intercrosses of Bub1 mice produced no Bub1 newborn mice. Further analysis revealed that Bub1 embryos died between days 4.5 and 6.5 of development (unpublished data), which is in agreement with other mitotic checkpoint gene knockout mice. Also, no Bub1 pups were born from intercrosses of Bub1 mice, implying that the level of wild-type Bub1 protein produced by these hypomorphic alleles was not sufficient for successful embryonic development.
To bypass this problem, we created a Bub1 hypomorphic allele by the use of an alternative method. This method takes advantage of a hygromycin B phosphotransferase expression (Hyg) cassette that causes a high incidence of premature transcriptional termination when inserted into intronic sequences (). We constructed a targeting vector to introduce this Hyg cassette into intron 9 of the endogenous Bub1 gene (). Properly targeted ES clones were used to produce Bub1 mice. Intercrosses of Bub1 mice yielded viable Bub1 offspring at the expected Mendelian frequency. Furthermore, interbreeding of Bub1 and Bub1 mice yielded viable Bub1 offspring with normal Mendelian frequency (). Like Bub1 mice, Bub1 and Bub1 mice exhibited no changes in development or appearance when compared with wild-type mice. We performed Western blotting to measure the level of wild-type Bub1 protein in mouse embryonic fibroblasts (MEFs) derived from Bub1, Bub1, Bub1, Bub1, and Bub1 mice (). We assessed that Bub1 signals from Bub1, Bub1, Bub1, and Bub1 MEFs were ∼75%, 50%, 30%, and 20% of those from Bub1 MEFs, respectively. Truncated forms of Bub1 encoded by the – and H alleles were undetectable even after overexposure of the Western blots (Fig. S1, available at ). Together, these results demonstrated that we had produced a series of mice with decreasing Bub1 protein dosage.
To determine whether the reduced expression of Bub1 protein affects the accuracy of chromosome segregation, we collected splenocytes from Bub1, Bub1, Bub1, Bub1, and Bub1 mice at 5 mo of age and prepared metaphase spreads for karyotype analyses. Chromosome counts showed that <1% of wild-type splenocytes were aneuploid (). In contrast, splenocytes from Bub1, Bub1, Bub1, and Bub1 mice had a 6%, 16%, 35%, and 39% incidence of aneuploidy, respectively, revealing an inverse correlation between the level of Bub1 protein and the percentage of aneuploidy in this cell type. Moreover, the range of abnormal chromosome numbers broadened with the decreasing expression of Bub1 protein (). We observed the premature separation of sister chromatids (PMSCS) in 14 and 15% of the mitotic figures from Bub1 and Bub1 splenocytes but only in 4% of the mitotic figures from Bub1 splenocytes (). Thus, there seems to be no clear link between PMSCS and Bub1 dosage in splenocytes.
We further investigated the effect of Bub1 insufficiency on chromosome number stability by performing chromosome counts on metaphase spreads from Bub1, Bub1, Bub1, Bub1, and Bub1 MEFs at passage 5. We found that the percentage of aneuploid metaphases was much higher in Bub1 and Bub1 MEFs than in Bub1 and Bub1 MEFs, which, in turn, had a higher percentage than Bub1 MEFs (). PMSCS was not increased in Bub1, Bub1, Bub1, and Bub1 MEFs compared with Bub1 MEFs (). These data confirm that a high percentage of cells with low levels of Bub1 become aneuploid without the apparent requirement of PMSCS.
Many of Bub1's critical functions during mitosis occur at the kinetochore. Therefore, we tested whether the graded reduction of Bub1 expression corresponds to a graded reduction in Bub1 levels at kinetochores. Immunostaining of Bub1, Bub1, and Bub1 MEFs with affinity-purified Bub1-specific antibody showed that fluorescence signals at kinetochores progressively declined with decreasing cellular levels of Bub1 expression (). To examine how this graded reduction in kinetochore-bound Bub1 affected the localization of mitotic checkpoint proteins whose targeting to kinetochores is Bub1 dependent, we immunostained Bub1, Bub1, and Bub1 MEFs with antibodies against the mitotic checkpoint proteins Mad1, Mad2, BubR1, and CENP-E. In Bub1 prometaphase cells, Mad1 staining was concentrated on kinetochores visualized by antikinetochore antibody (). However, kinetochore-associated Mad1 signals were much less abundant in the corresponding Bub1 and Bub1 cells. As Mad1 is required for the kinetochore localization of Mad2 (, ), we anticipated that Mad2 staining patterns would also be reduced in Bub1 and Bub1 prometaphase cells. We tested this prediction, but despite numerous attempts, we were unsuccessful in obtaining kinetochore-associated Mad2 signals in Bub1 MEFs with antibodies that are known to detect Mad2 at kinetochores of human prometaphase cells (see Materials and methods for details). Unlike Mad1, kinetochore-associated BubR1 and CENP-E signals were unaffected in Bub1 cells during prometaphase (). However, kinetochore signals for both of these proteins were dramatically reduced in prometaphase Bub1 cells. Western blot analysis showed that Mad1, BubR1, and CENP-E protein levels were similar in Bub1 and Bub1 cells (Fig. S2 A, available at ), excluding the possibility that the reduction in kinetochore localization of these proteins in Bub1 cells is caused by reduced protein stability.
Next, we tested how the graded reduction in kinetochore-bound Bub1 levels affected the subcellular localization of Sgo1 and Aurora B, both of which have been reported to require Bub1 for their proper localization to the centromeres (, ; ; ). Fewer Sgo1-positive centromeres were observed in Bub1 prometaphases than in Bub1 prometaphases (). In contrast, no such decrease was observed in Bub1 prometaphase cells (unpublished data). Immunostainings for Aurora B revealed that the localization of this protein was normal in both Bub1 and Bub1 prometaphase cells (Fig. S2 B). Thus, whereas most proteins that require Bub1 for proper localization to kinetochores/centromeres are mislocalized in Bub1 MEFs, only Mad1 is mislocalized in Bub1 cells.
To analyze the activity of the mitotic checkpoint in MEFs with a graded reduction in Bub1 expression, we performed a nocodazole challenge assay (; ). In this assay, MEFs were first transduced with a retrovirus encoding a YFP-tagged H2B fusion protein to allow the visualization of chromosomes by fluorescence microscopy (). MEFs were then challenged with nocodazole, and 20–30 cells undergoing nuclear envelope breakdown (NEBD) were marked and monitored at 15-min intervals to determine when their chromatin decondenses. The duration of arrest in mitosis, which is defined as the interval between NEBD (onset of mitosis) and chromatin decondensation (exit from mitosis without cytokinesis), was calculated and plotted. The time at which 50% of the cells have exited mitosis was used for comparison. Nocodazole- challenged Bub1 MEFs typically remained arrested in prometaphase for 7.2 h (). Bub1 and Bub1 MEFs were impaired in their ability to maintain this arrest, with 50% of the cells exiting around 5.4 h. However, Bub1 and Bub1 MEFs exhibited a more profound defect, with 50% of the cells exiting mitosis at 3.8 h and 3.5 h, respectively. Thus, the mitotic checkpoint appears to be considerably weaker in Bub1 and Bub1 MEFs than in Bub1 and Bub1 MEFs.
Next, MEFs with graded reductions in Bub1 expression were screened for chromosome segregation defects. In essence, we followed YFP-H2B–positive MEFs through an unchallenged mitosis by live cell imaging and determined the fraction of mitotic cells with chromosome segregation abnormalities. Two known defects underlying chromosome missegregation, congression failure and chromosome lagging (), were observed at higher rates in Bub1, Bub1, Bub1, and Bub1 MEFs than in Bub1 MEFs (). The combined incidence of the aforementioned defects was remarkably similar in Bub1, Bub1, Bub1, and Bub1 MEFs (). Furthermore, anaphases with centrophilic chromosomes that segregate faster than the other chromosomes () were observed at an approximately two- to fivefold higher frequency in Bub1, Bub1, Bub1, and Bub1 MEFs than in Bub1 MEFs (). Whether this type of abnormality leads to chromosome missegregation is unclear, but even with the inclusion of this defect, the overall incidence of chromosome segregation abnormalities remains very similar in Bub1, Bub1, Bub1, and Bub1 MEFs (). Irrespective of Bub1 genotype, most cells with abnormal chromosome segregation events involved a single chromosome (or a duplicated chromosome). Occasionally, two or three chromosomes were implicated ( and Table S1, available at ). Thus, the aforementioned analyses suggest that the accuracy of chromosome segregation is highly dependent on a full complement of Bub1 protein and that both small and large reductions in Bub1 cause chromosome missegregation at comparable rates.
Initially, we were surprised that chromosome missegregation rates were similar in Bub1, Bub1, Bub1, and Bub1 MEFs because the percentage of aneuploid cells was much higher in Bub1 and Bub1 cultures than in Bub1 and Bub1 cultures (). One explanation could be that cell survival after chromosome missegregation increases with decreasing Bub1 levels. To explore this possibility, we infected Bub1, Bub1, Bub1, Bub1, and Bub1 MEFs with the YFP-H2B virus and monitored the fate of cells undergoing chromosome missegregation for up to 12 h by live cell imaging. Typically, 95% of Bub1 MEFs died within several hours after chromosome missegregation ( and Videos 5 and 6, available at ). This percentage declined progressively and sharply as Bub1 expression decreased, with only 32% of Bub1 MEFs dying after a missegregation event (, , and Videos 1 and 2). Cells dying after chromosome missegregation consistently showed nuclear fragmentation and/or cytoplasmic blebbing (; and Videos 3, 4, 7, and 8). Cells with accurate segregation rarely died during the 12-h observation period, irrespective of Bub1 genotype ( and Videos 9 and 10). From this, we conclude that although the rates of chromosome missegregation are comparable at various levels of Bub1 reduction, aneuploid cells accumulate to higher steady-state levels in cultures with low amounts of the protein because cells in these cultures are more likely to survive after chromosome missegregation. Consistent with this interpretation, we found that micronuclei, which we observed by live cell imaging to result from misaligned, centrophilic, or lagging chromosomes, accumulated steadily with decreasing Bub1 levels (Fig. S3).
To explore whether Bub1 plays a more general role in cell death signaling, we measured cell survival to various kinds of DNA-damaging agents. MEFs with graded reduction in Bub1 protein levels were exposed to increasing concentrations of doxorubicin, mitomycin C, or paraquat for 48 h. Cell survival was then determined by using the MTS assay. Cell survival in these agents was similar for Bub1, Bub1, Bub1, Bub1, and Bub1 MEFs (Fig. S4, A–C; available at ). In addition, decreased Bub1 expression also did not increase survival to prolonged exposure to nocodazole, a spindle poison that induces tetraploidization by driving prometaphase cells into G1 without chromosome segregation (Fig. S4 D). These experiments suggest a rather specific role for Bub1 in mediating cell death after the missegregation of one or a few chromosomes.
To determine the long-term consequences of Bub1 down- regulation, we created and monitored cohorts of Bub1 ( = 160), Bub1 ( = 142), Bub1 ( = 137), and Bub1 ( = 238) mice on a mixed 129 × C57BL/6 background. Earlier, we reported that BubR1 hypomorphic mice have a short lifespan, are infertile, and develop various early aging–associated phenotypes (). We note that no such phenotypes were observed in any of our Bub1 mutant mice (unpublished data). However, we found that Bub1 and Bub1 mice were significantly more prone to spontaneous tumors than Bub1 mice (). Bub1 mice had a significantly shorter median tumor-free survival (530 d) than Bub1 mice (676 d), which, in turn, had a significantly shorter median tumor-free survival than Bub1 mice (772 d; ). Moreover, Bub1 and Bub1 mice developed a different spectrum of tumors than did Bub1 mice (). Bub1 mice developed significantly more sarcomas, lymphomas, and lung tumors. Bub1 mice were also prone to develop sarcomas but not lymphomas and lung tumors. Bub1 mice were highly susceptible to hepatocellular carcinomas, a tumor type that was not significantly increased in Bub1 mice. In contrast to Bub1 and Bub1 mice, Bub1 mice showed a trend toward decreased tumor formation, particularly in liver and lung tissue (). Collectively, these data establish a causal relationship between the down-regulation of Bub1 expression and cancer development and suggest that there is a threshold level of Bub1 below which the incidence of neoplastic transformation progressively increases. Our data further imply that Bub1 reductions above the threshold may slightly inhibit tumor formation in particular tissues.
Based on the aforementioned data, we conclude that Bub1 mice have enough Bub1 protein to protect themselves against spontaneous tumorigenesis. To determine whether this level is sufficient to guard against carcinogen-induced tumors, we administered a single dose of 0.5% DMBA (9,10-dimethylbenz-A-athracene) in acetone to the dorsal skin of 3–5-d-old pups generated from Bub1 × Bub1 intercrosses. 5 mo after treatment, we killed the mice and screened for tumors. Irrespective of the genotype, tumors were exclusively detectable in the lungs. Bub1 mice exhibited a two- to threefold higher incidence of lung tumors than in Bub1 mice (). Moreover, the tumor burden of Bub1 mice was increased approximately threefold (). From this experiment, we conclude that Bub1 heterozygous knockout mice are prone to carcinogen-induced tumorigenesis.
In this study, we produced a series of mutant mice in which the expression of Bub1 is reduced in a graded fashion from normal to zero by the use of wild-type, hypomorphic, and knockout alleles to determine the physiological role of Bub1. As anticipated, we find that the complete loss of Bub1 leads to embryonic lethality. Strongly reduced Bub1 expression (up to approximately fivefold reduction) does not interfere with embryogenesis and allows for the development of adult mice that are overtly indistinguishable from their wild-type littermates. However, the reduction of Bub1 levels does have adverse consequences on genomic stability in these cells. Karyotyping of splenocytes and MEFs from our series of mutant mice established an inverse correlation between Bub1 expression level and aneuploidy. Failure of chromosome congression during metaphase is the main chromosome segregation defect resulting from Bub1 insufficiency. Although small and large reductions in Bub1 levels cause similar rates of chromosome missegregation, rates of cell survival after aberrant chromosome segregation increase considerably with declining Bub1 levels, providing a plausible explanation for why large reductions cause more aneuploidy than small ones. Furthermore, the reduction of Bub1 protein affected the strength of the mitotic checkpoint and loading of certain proteins onto centromeres or kinetochores. Bub1 haploinsufficiency in mice does not cause spontaneous tumors, but, as Bub1 levels drop further, animals become highly susceptible to a variety of spontaneous tumors, with the highest rate of tumorigenesis seen at the lowest level of Bub1 expression.
Bub1 is known to be required for the binding of several other mitotic checkpoint proteins to kinetochores (; ; ). Our analysis of MEFs with graded reduction in Bub1 expression now reveals that these proteins require different levels of Bub1 protein for their normal recruitment to kinetochores. In particular, the recruitment of Mad1 to kinetochores is dramatically reduced when Bub1 is down-regulated. We speculate that Mad2, which forms a complex with Mad1 at kinetochores (), is similarly sensitive to Bub1 down-regulation, although we were unable to confirm this because of the lack of an antibody that detects mouse Mad2 at kinetochores. Besides Mad1, the recruitment of BubR1 and CENP-E to kinetochores is also sensitive to Bub1 down-regulation, but not as sensitive as Mad1, as their localization is normal in Bub1 heterozygous MEFs. Recent studies have presented evidence that Bub1 functions to recruit Sgo1 to centromeres to prevent the precocious separation of sister kinetochores (; ). Consistent with these studies, we find that centromeric Sgo1 levels are reduced in Bub1 mutant MEFs, but only when Bub1 expression is strongly down-regulated. However, this did not result in premature sister kinetochore separation, implying that an even further drop in centromeric Sgo1 is required to trigger the cleavage of cohesin molecules that link sister centromeres. A recent study has implicated Bub1 in the targeting of chromosomal passenger complexes to centromeres in early mitosis (). Our finding that very low amounts of Bub1 are sufficient for directing these complexes to centromeres suggests that near complete Bub1 depletion is required to dislocate the passenger complex from mitotic centromeres.
One of our more surprising findings is the observation that a relatively small reduction in Bub1 expression has a major impact on the accuracy of chromosome congression. What could be the explanation for this observation? Although the precise role of Bub1 in chromosome congression is currently not known, it is believed that this role involves kinetochore assembly (). Of the mitotic checkpoint proteins whose recruitment is Bub1 dependent, only CENP-E has so far been implicated in chromosome congression. Thus, one explanation for the congression failure in Bub1 mutant MEFs might be a CENP-E recruitment defect. Consistent with this, we find that the targeting of CENP-E to kinetochores is perturbed in Bub1 hypomorphic MEFs. On the other hand, Bub1 haploinsufficient MEFs, which display similar rates of congression failure as Bub1 hypomorphic MEFs, exhibit normal CENP-E recruitment to kinetochores, implying that the mechanism of congression failure is CENP-E independent. This conclusion is further supported by data of demonstrating that the depletion of Bub1 in HeLa cells by RNA interference causes chromosome congression defects in the absence of CENP-E mislocalization. Therefore, it remains unclear how Bub1 promotes proper chromosome congression. Nonetheless, we suspect that it involves a known or novel kinetochore- associated protein that functions in microtubule capture and whose recruitment to kinetochores is highly dependent on a full complement of Bub1.
Our analysis of MEFs with graded reduction in Bub1 expression indicates that relatively small shortages in Bub1, such as those seen in Bub1 and Bub1 MEFs, weaken the mitotic checkpoint considerably. It is plausible that the impaired recruitment of Mad1 (and presumably Mad2) to kinetochores undermines the mitotic checkpoint in these cells, as kinetochore-associated Mad1–Mad2 complexes generate soluble Mad2–Cdc20 complexes that bind to and inactivate APC/C. Larger reductions in Bub1, as present in Bub1 and Bub1 MEFs, had an even more profound impact on mitotic checkpoint activity. We propose that this is caused, at least in part, by the added loss of CENP-E and BubR1 from kinetochores, as kinetochore-bound CENP-E and BubR1 molecules have been implicated in the assembly of various inhibitory protein complexes that target APC/C (, ). The Bub1 kinase also can inhibit APC/C directly through the phosphorylation of Cdc20 (; ; ). We have not addressed whether the phosphorylation of Cdc20 is affected in our mutant series of MEFs as a result of the current lack of antibodies that recognize phosphorylated mouse Cdc20.
Although it has been well established that gross abnormalities of chromosome segregation (frequently referred to as mitotic catastrophe) often cause cell death (), the fate of cells undergoing the random missegregation of only one or a few chromosomes has been unknown. Here, we show by the use of live cell imaging that wild-type primary MEFs die at very high rates after minor abnormalities in chromosome segregation. The implication of this finding is that aneuploidy rates in cultured wild-type MEFs are substantially higher than metaphase spread karyotypes reveal. Our discovery that cell death rates after chromosome missegregation dramatically decline with decreasing levels of Bub1 creates a molecular entry point for studying the underlying cell death mechanism. Whether Bub1 plays a unique role in this mechanism or whether there is a broader connection between mitotic checkpoint damage and decreased cell death after chromosome missegregation is an important question for future analysis. Bub1's dual function as a guardian of high fidelity chromosome segregation and as a mediator of cell death after aberrant segregation is reminiscent of proteins such as ATM (ataxia telangiectasia mutated) and p53 that function in both DNA repair and apoptosis in response to DNA damage (). A recent study showed that Bub1-depleted cancer cell lines display increased mitotic cell death when they are exposed to agents that perturb kinetochore-microtubule attachment, such as nocodazole (). We observed no such effect in nocodazole-treated MEFs with graded reduction in Bub1 expression (Fig. S4 D), suggesting that the impact of the Bub1 level of expression on mitotic cell death induced by spindle poisons is cell type and/or transformation status dependent.
Bub1 expression is reduced in several human cancers, including colorectal, gastric, and esophageal tumors (; ; ); however, it was unknown whether the reduced expression of this mitotic checkpoint protein is causally implicated in tumorigenesis. Analysis of our series of Bub1 mutant mice firmly establishes that the reduced expression of Bub1 leads to the development of spontaneous tumors in mice, but only when Bub1 levels fall below a threshold level. Those with the most drastic reductions of Bub1 expression have the shortest tumor latency and the highest incidence of tumors. The level of Bub1 required to prevent spontaneous tumorigenesis appears to vary per tissue, as illustrated by the fact that only mice with the most profound reduction in Bub1 are predisposed to lymphomas and lung tumors. In the liver, the optimal level of Bub1 down-regulation is not the lowest level, as Bub1 mice but not Bub1 mice are prone to hepatocellular carcinomas. Adding even more complexity is the discovery that Bub1 haploinsufficiency exerts a slight tumor-suppressive effect in both liver and lung tissue. This finding is consistent with the recent discovery that CENP-E haploinsufficiency inhibits tumorigenesis in certain mouse tissues (). However, unlike CENP-E insufficiency, Bub1 haploinsufficiency does not inhibit DMBA-induced tumorigenesis. In fact, Bub1 haploinsufficient mice are highly susceptible to lung tumors when challenged with this carcinogen. This observation implies that the loss of one Bub1 gene copy acts to accelerate the development of tumors initiated by particular cancer gene mutations.
Because Bub1 hypomorphic mice have a high percentage of aneuploid cells and are predisposed to spontaneous tumors, whereas Bub1 haploinsufficient mice have a relatively low percentage of aneuploid cells and are not tumor prone, it is tempting to speculate that it is the increase in aneuploidy that drives tumorigenesis in Bub1 hypomorphic mice. However, the fact that both Rae1/Bub3 and Rae1/Nup98 double-haploinsufficient mice develop aneuploidy at rates very similar to that of Bub1 hypomorphic mice but are not prone to spontaneous tumors argues against this idea (; ; , ). One possible explanation for this discrepancy could be that as a result of the decreased cell death in response to chromosome missegregation, Bub1 hypomorphic mice may develop a wider variety of abnormal karyotypes than Rae1/Bub3 and Rae1/Nup98 double-haploinsufficient mice, thereby perhaps increasing the incidence of karyotypes that have the ability to drive tumorigenesis. However, the role of aneuploidy in tumorigenesis is clearly highly complex, and it will be necessary to carefully examine each individual regulator of chromosome segregation for its involvement in tumorigenesis through the use of animal models. We expect these efforts to allow the identification of a subset of mitotic regulators that are particularly important for tumor prevention. Among them may be mitotic regulators that serve as molecular hubs within the mitotic checkpoint or other networks that regulate proper chromosome segregation or mitotic regulators with connectivity to other pathways that guard against neoplastic transformation.
In this study, we have used a series of mutant mice to demonstrate that only after reducing Bub1 levels beyond a threshold level do mice start to develop spontaneous tumors. Had we used only Bub1 haploinsufficient mice rather than a series of mice with graded reduction in Bub1 expression, our conclusions would have been dramatically different in that we would conclude that Bub1 does not act as a tumor suppressor itself. Heterozygous knockout models for several other mitotic checkpoint genes are also not predisposed to spontaneous tumorigenesis. For a more definitive understanding of the roles these genes have in tumor prevention, it will be useful to use hypomorphic alleles to further reduce their level of expression in mice.
An 8.5-kb Bub1 129Sv/J genomic DNA fragment was used to generate both targeting vectors used. Gene-targeting procedures were performed as previously described (). We identified targeted ES cell clones by Southern blot analysis using a 3′ probe on BamHI-cut genomic DNA (). Mutant mice were derived from targeted ES cell clones through standard procedures. These mice were maintained on a mixed 129Sv/E × C57BL/6 genetic background. Mice in tumor susceptibility experiments were observed daily for the development of overt tumors or signs of ill health. Moribund mice were killed, and all major organs were screened for overt tumors using a dissection microscope (SZX12; Olympus). Tumors that were collected were processed by standard procedures for histopathology. Prism software (GraphPad Software, Inc.) was used for the generation of tumor-free survival curves and for statistical analyses. DMBA treatment was performed as previously described (; ). All major organs were screened for overt tumors using a dissection microscope (SZX12; Olympus). Harvested tumors were routinely processed for histopathological confirmation. We note that all mice were housed in a pathogen-free barrier environment.
Western blot analyses and indirect immunofluorescence were performed as previously described (). A laser-scanning microscope (LSM 510 v3.2SP2; Carl Zeiss MicroImaging, Inc.) as well as a microscope (Axiovert 100M; Carl Zeiss MicroImaging, Inc.) with a c-Apochromat 100× oil immersion objective (Carl Zeiss MicroImaging, Inc.) was used to analyze immunostained cells and to capture representative images. Primary antibodies were visualized with appropriate secondary antibodies conjugated to AlexaFluor594, -488. or -647 (Invitrogen). Primary antibodies used for Western blotting and indirect immunofluorescence were as follows: rabbit anti–human Bub1(25–165), rabbit anti–human BubR1(382–420) (), rabbit anti–human Mad1 (provided by T. Yen, Fox Chase Cancer Center, Philadelphia, PA), rabbit anti–human SgoI([1–262][177–351]) (provided by H. Yu, University of Texas Southwestern, Houston, TX; ), mouse anti–Aurora B (BD Biosciences), human anticentromeric antibody (Antibodies, Inc.), and rabbit anti–CENP-E (provided by D. Cleveland, Ludwig Institute for Cancer Research, La Jolla, CA). Mad2 antibodies tested were as follows: rabbit anti–human Mad2(FL-205) (Santa Cruz Biotechnology, Inc.), mouse anti–human Mad2 (BD Biosciences), and rabbit anti–human Mad2 (Covance). None of these Mad2 antibodies detects Mad2 at kinetochores of mitotic MEFs.
Chromosome counts on metaphase spreads were performed as previously described (). We note that cells were scored as diploid ( = 40 chromosomes), tetraploid ( = 80 chromosomes), or aneuploid ().
To allow the visualization of chromosomes by fluorescent microscopy on living cells, we used a retrovirus expressing YFP-tagged H2B (). Passage 2 MEFs were seeded in T25 flasks at 75% confluence and cultured in DME/10% FBS at 3% oxygen. 12 h after seeding and again every 12 h for at least three times, the medium was replaced with medium harvested from EcoPACK pMSCV-puro-H2B-YFP viral producer cell lines. Cells were then seeded onto 35-mm glass-bottomed culture dishes (MatTek Corp.) and cultured in DME/10% FBS. Approximately 24 h later, experiments were performed using a microscope system (Axio Observer; Carl Zeiss MicroImaging, Inc.) with CO Module S, TempModule S, Heating Unit XL S, a plan Apo 63× NA 1.4 oil differential interference contrast III objective (Carl Zeiss MicroImaging, Inc.), camera (AxioCam MRm; Carl Zeiss MicroImaging, Inc.), and AxioVision 4.6 software (Carl Zeiss MicroImaging, Inc.). The imaging medium was DME/10% FBS. The temperature of the imaging medium was kept at 37°C. The exposure times in nocodazole challenge experiments were 100 ms at 2 × 2 binning. Time of arrest in mitosis was defined as the interval between NEBD (onset of mitosis) and chromatin decondensation (exit from mitosis without cytokinesis). Interframe intervals were 15 min for nocodazole challenge. Analysis of mitotic defects was performed as previously described (). For analysis of the incidence of cell death after chromosome missegregation, MEFs undergoing abnormal chromosome segregation were marked and followed with an interframe interval of 30 min for up to 12 h. Cell death was preceded by severe nuclear blebbing and cytoplasmic fragmentation. For each of the aforementioned experiments, we examined at least three independent clones per genotype unless otherwise noted. Prism software (GraphPad Software, Inc.) was used for statistical analyses. To evaluate the incidence of micronuclei formation, at least 600 YFP-H2B–expressing interphase MEFs were screened for the presence of micronuclei by live cell microscopy.
Analyses of cell survival in response to doxorubicin, mitomycin C, and paraquat were performed as described previously () with the exception that passage 3 MEFs were used instead of passage 2 MEFs. For analysis of cell death in response to nocodazole treatment, 10 passage 3 MEFs were seeded in duplicate for three independent cell lines of each genotype. After ∼12 h, normal medium was replaced with medium containing 100 ng/ml nocodazole, and cells were cultured for an additional 72 h. All cells were collected after this time, and tryptan blue exclusion was used to count living cells.
Fig. S1 shows that truncated proteins encoded by the Bub1 knockout and hypomorphic alleles are undetectable by immunoblotting. Fig. S2 shows that Mad1, BubR1, and CENP-E protein levels were similar in Bub1 and Bub1 cells and that Aurora B is not mislocalized in Bub1 cells. Fig. S3 shows that the incidence of micronuclei increases with declining levels of Bub1. Fig. S4 shows that the Bub1 level of expression has no impact on MEF cell survival to DNA-damaging agents and prolonged nocodazole exposure. Videos 1 and 2 show videos of the Bub1 MEF presented in A. Videos 3 and 4 show videos of the Bub1 MEF presented in B. Videos 5 and 6 show a Bub1 MEF undergoing chromosome missegregation in mitosis. Both daughter cells die after exit from mitosis. Videos 7 and 8 show a Bub1 MEF undergoing chromosome missegregation. One of the two daughter cells undergoes cell death. Videos 9 and 10 show a Bub1 MEF undergoing normal chromosome segregation. Both daughter cells survive. Table S1 presents data about the number of chromosomes that are abnormally segregated in cells with mutliple segregation defects. Online supplemental material is available at . |
Gene expression in diploid cells is generally biallelic: RNA is transcribed from both alleles of a gene in each cell. However, it is becoming clear that a substantial subset of genes is expressed monoallelically, despite having identical DNA sequences. Monoallelically expressed genes fall into two major classes: imprinted genes and random monoallelic genes.
Imprinted genes are expressed exclusively from either the maternally or paternally inherited chromosome. For example, only the maternal allele of the mouse locus () and the paternal allele of the gene () are expressed. For imprinted genes, the identity of the expressed allele is predetermined, often by differential DNA methylation established in the male and female gametes ().
Random monoallelic genes, on the other hand, can be expressed from either the maternal or paternal chromosome. A dramatic example of random monoallelic expression is mammalian X inactivation, in which one of the two X chromosomes in a female cell is transcriptionally silenced (). The choice of which X chromosome to silence is made early in embryonic development. Subsequently, the inactive X is clonally inherited, resulting in adult females with mosaic expression of X-linked genes from the maternally and paternally inherited X chromosomes. In addition to X-linked genes, an increasing number of random monoallelic genes are being identified on autosomes. These genes include olfactory receptors (ORs) (), several immune-system genes including natural killer cell receptors and interleukins (; ; ; ), and the cell adhesion molecule p120 catenin (). OR gene choice has an added layer of complexity in that only one allele of one of the ∼1,000 olfactory loci located in tandem arrays across the genome is expressed in each olfactory neuron ().
The mechanism by which one and only one allele of a gene is chosen at random to be expressed remains mysterious. We have recently shown that the homologous X chromosomes in female mouse embryonic stem (ES) cells adopt different, mutually exclusive states even before X inactivation is initiated (). Furthermore, these two states correlate with the fate of the chromosome upon differentiation of ES cells carrying mutations that predetermine the fates of the active and inactive X chromosomes. In wild-type ES cells, the maternal and paternal X chromosomes can switch back and forth between these states, but upon X inactivation, the states appear to be fixed as the active and inactive X chromosomes. The two states are detected as a tendency for replicated loci on one chromosome to appear as single pinpoints by FISH in paraformaldehyde (PFA)-fixed cells, while loci on the homologous chromosome tend to appear as doublet signals. We refer to this phenomenon as singlet/doublet signals independent of asynchronous DNA replication, or SIAR.
In this paper, we find that SIAR is a general characteristic of random monoallelic genes in ES cells. Establishment of SIAR is dependent on the Polycomb Group protein Eed, an essential component of the histone H3 lysine 27 methyltransferase complex. Furthermore, Eed is also required for asynchronous replication of random monoallelic genes in differentiated cells. Together, these results suggest that a common mechanism, involving chromatin modifications, underlies both X inactivation and autosomal random monoallelic expression.
We wished to examine whether SIAR was peculiar to the X chromosome, or whether it was a more general characteristic of monoallelically expressed genes. To determine whether future monoallelic genes on autosomes behave similarly to X-linked genes, FISH was performed on PFA-fixed mouse ES cells using probes to random monoallelic genes, imprinted genes, and biallelically expressed controls (). Genes destined to be randomly monoallelically expressed in differentiated cells displayed a singlet FISH signal on one allele and a doublet signal on the other allele in a high percentage of S-phase ES cells (). Imprinted and biallelically expressed genes displayed a lower frequency of singlet/doublet (SD) cells than the future random monoallelic genes ().
The high frequency of SD FISH signals for X-linked genes in female ES cells () does not reflect asynchronous replication of these loci. Imprinted genes, in contrast, have been shown to replicate asynchronously in ES cells (). We asked whether the elevated frequency of SD FISH signals for autosomal random monoallelic loci could be attributed to asynchronous replication. First, we assayed the replication timing of , an OR in the OR11-5 array, and the imprinted gene ES cells were released from a late G1 block, DNA was isolated at 12 one-hour intervals, newly replicated BrdU-labeled DNA was immunoprecipitated, and or sequences were detected by qPCR. Replication of both and occurred around the 3–5-h window after release ().
We then assayed the cell cycle window in which cells with SD FISH signals occurred to determine whether the cells with SD signals peak at a single point. Such a result would be consistent both with our replication timing data and with the slight replication asynchrony that has been reported for these genes (Singh et al., 2003; ). ES cells were FACS sorted by DNA content and fixed with PFA, and the proportion of cells with SD FISH signals was determined in each of the six fractions (). For the percentage of cells with SD signals peaked in the third fraction and then decreased, consistent with the documented asynchronous replication of this locus (). For the OR array OR11-5, the percentage of cells with SD signals increased between the first two fractions and remained at a relatively constant level throughout the remaining fractions.
In combination, the analyses of replication timing and SD FISH signals across the cell cycle indicate that one of the two OR11-5 alleles exhibits a singlet FISH signal in a significant fraction of ES cells, even when both loci have replicated. Thus, a substantial fraction of the observed SD FISH signals for OR11-5 cannot be attributed to asynchronous replication.
In differentiated cells, random monoallelic genes do display asynchronous replication (; ). When multiple random monoallelic genes occur on the same chromosome, they replicate early on one homologue and late on the other, despite the fact that intervening biallelically expressed genes do not exhibit asynchronous replication (Singh et al., 2003). We therefore tested whether loci on the same chromosome are coordinated in their behavior even before differentiation. To do this, we performed FISH in ES cells with pairs of probes to random monoallelic genes on the same chromosome (). Two pairs of OR arrays on chromosome 2, a pair of OR arrays on chromosome 11, and an OR array and interleukin-4 on chromosome 11 were tested pairwise (). All four pairs displayed singlet signals on one chromosome and doublet signals on the other in ∼65% of cells in which each locus exhibited one singlet and one doublet allele (). This is significantly different from the 50% of cells that are predicted to exhibit this pattern if behavior of loci on the same chromosome is not coordinated.
The identity of the homologue that contains the early replicating random monoallelic genes is fixed in clonally derived mouse embryo fibroblasts (MEFs) (Singh et al., 2003), prompting us to examine whether the same chromosome always shows an elevated frequency of singlet FISH signals in clonally derived ES cells. FISH was performed in an allele-specific manner, using an ES cell line containing a transgene on one copy of chromosome 11. In ES cells derived from a single progenitor, the singlet signal for either of the two OR probes on chromosome 11 appeared on the transgene-containing chromosome in approximately half the cells (). Thus, the identity of the chromosome exhibiting the singlet FISH signal for autosomal random monoallelic loci switches in ES cells, in contrast to the fixed behavior of these loci in differentiated cells.
The appearance of an already-replicated allele as a singlet FISH signal suggests that the sister chromatids remain closely apposed such that the individual chromatids cannot be distinguished. To test whether intact chromatin structure is necessary to observe the high percentage of SD cells for autosomal monoallelic genes, we compared two fixation methods. PFA fixation, as used above, preserves nuclear organization, while fixation with methanol/acetic acid (MeOH) removes proteins from DNA and disrupts chromatin organization (). FISH for autosomal random monoallelic genes in MeOH-fixed samples revealed a lower percentage of SD cells than in PFA-fixed samples; imprinted genes, on the other hand, displayed similar frequencies of SD cells in MeOH- and PFA-fixed samples (). These results confirm that SIAR requires relatively intact chromatin structure ().
SIAR at X-linked loci is restricted to undifferentiated ES cells (). To determine whether SIAR at autosomal loci is also limited to undifferentiated cells, before the establishment of monoallelic expression, we performed FISH for two OR arrays, OR2-1 and OR11-1, in MEFs. Both probes displayed a lower frequency of SD cells in MEFs compared with ES cells, and there was no significant difference in the percentage of SD cells in PFA- and MeOH-fixed MEFs (). The SD cells observed in MEFs presumably reflect the asynchronous replication of OR arrays in differentiated cells ().
The chromatin difference that underlies the difference in appearance of future random monoallelic loci by FISH remains unknown. The loss of SIAR in MeOH-fixed cells implies that intact chromatin structure is required; for this reason, possible candidates for the cause of SIAR include proteins that play a role in chromatin modification. We performed FISH on ES cells mutant for either (), which is required for histone H3 methylation on lysine 27, or the maintenance DNA methyltransferase () to see if they are required for SIAR. ES cells mutant for did not display a significant difference in the frequency of SD cells for OR2-1 or OR11-1 probes compared with wild-type cells (Fig. S4, available at ). Mutation of , in contrast, did result in a significant reduction in the frequency of SD signals for OR arrays (). In addition, the frequency of SD signals was the same for PFA-fixed and MeOH-fixed mutant ES cells (), further suggesting that these cells no longer display SIAR, because disruption of nuclear organization no longer affects SD signal frequency. Together these data indicate that Eed is necessary for SIAR at OR2-1 and OR11-1.
We examined whether Eed was also necessary to establish asynchronous replication timing of OR genes in differentiated cells.
ES cells were differentiated as embryoid bodies for 17 d. FISH for OR2-1, OR11-1, , and was performed on MeOH-fixed differentiated cells. Differentiated wild-type cells showed a high frequency of SD FISH signals at monoallelically expressed loci (), consistent with previously published results in MEFs ().
cells, the frequency of SD signals for the OR arrays was reduced to the level seen for the biallelically expressed .
cells (). In combination, our results show that Eed plays dual roles in regulation of random monoallelic autosomal genes: it is necessary for SIAR before differentiation and asynchronous replication timing of these loci after differentiation.
The similar behavior of autosomal and X-linked genes before random monoallelic expression lends support to the idea that the mechanism of random choice is at least partially conserved between autosomes and X chromosomes (). On both the X chromosome () and autosomes, alleles of future random monoallelic genes differ from each other in a switchable fashion that is dependent on intact nuclear structure and coordinated among loci on the same chromosome. Our current model () is that the observed differences in frequency of singlet and doublet FISH signals reflect an underlying difference in chromatin structure between the chromosomes that affects the likelihood that the replicated loci on each chromosome will separate enough to appear as a doublet signal by FISH.
As there are no known cell lines in which OR choice is predetermined, it is not possible to correlate autosomal SIAR with future expression states, as was done with X-linked loci (). However, autosomal random monoallelic loci display asynchronous replication in differentiated cells, even in cell types in which they are not expressed, and this asynchronous replication has been proposed to underlie the choice of which allele will be expressed in specific differentiated cell types (). mutant ES cells that lack SIAR also lose asynchronous replication after differentiation (), suggesting that the chromatin difference underlying SIAR of autosomal random monoallelic loci may be required for the later asynchronous replication of the loci.
Although it remains possible that Eed affects SIAR in ES cells and asynchronous replication in differentiated cells through two independent pathways, our results are consistent with the hypothesis that SIAR is a precursor to both random monoallelic expression and asynchronous replication.
embryos (). However, maternal stores of Eed present earlier in embryogenesis may have already established the underlying chromatin differences that are visualized as SIAR. Alternatively, it is conceivable that SIAR of X-linked and autosomal loci is mediated by different genes, or that SIAR is important for asynchronous replication, but not monoallelic expression.
ES cell lines currently exist, it remains to be seen whether SIAR of X-linked loci is affected by loss of Eed.
ES and differentiated cells suggests that Eed-mediated histone methylation is involved in asynchronous replication in differentiated cells as well as in the differences between homologous alleles of future random monoallelic genes in ES cells. Although it is well-established that histone methylation, expression status, and replication timing are closely correlated, it has been less clear whether particular histone modifications are a cause or effect of replication timing (). The histone methyltransferase Suv39h1, which methylates histone H3 lysine 9, has recently been shown to affect the timing of replication of pericentric heterochromatin in the mouse (). Together with our results, this suggests that a combination of histone modifications may play a causal role in establishing replication timing of particular loci.
Mouse cell lines used in this study included: ES2-1, wild-type female ES cells (); E14, wild-type male ES cells (); RRR379, male ES cells carrying an insertion of pGT0Lxf at the locus on chromosome 11 (BayGenomics);
male ES cells (),
male ES cells (), and wild-type female MEFs.
MEFs and EBs were cultured in Diff medium: Knockout DME (Invitrogen) with 10% fetal bovine serum (FBS), 1× nonessential amino acids (UCSF Cell Culture facility), 1× -glutamine, 1× penicillin/streptomycin (UCSF Cell Culture facility), and b-mercaptoethanol. ES cells were cultured in ES medium: same as Diff medium, plus 1,000 U/ml leukemia inhibitory factor (LIF), following standard protocols. Differentiation of EBs was tested using the ELF Phosphatase Detection kit (American Type Culture Collection).
Wild-type female ES cells were labeled with BrdU (GE Healthcare) and stained with 40 ug/ml Hoechst 33342 (Molecular Probes) for 45 min before harvesting for flow cytometry. The cells were resuspended in ES medium containing 40 ug/ml Hoechst 33342, 7% Cell Dissociation Buffer (Invitrogen), and 10 mM EDTA, and sorted using a FACSDiVa Cell Sorter (Becton Dickinson). DNA content was measured based on the intensity of Hoechst emission using a HQ445/50 bandpass filter. Cells were sorted into six fractions containing similar numbers of cells onto multiwell slides pretreated with 1 mg/ml poly--lysine and allowed to settle and adhere.
All cells were labeled with BrdU for 30 min before fixation. Cells to be PFA fixed were cytospun onto slides, washed 30 s with ice-cold CSK buffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl, and 10 mM Pipes, pH 6.8), 30 s with cold CSK + 0.5% Triton X-100, and 30 s with cold CSK, then fixed for 10 min at room temperature in 4% PFA, 1× PBS (). For MeOH fixation, trypinized cells were treated with 0.075 M KCl for 10 min on ice, washed four times with a 3:1 MeOH/acetic acid solution, and then dropped onto slides ().
FISH for genomic DNA was performed essentially as previously described (). PFA-fixed samples were pretreated with 0.01% pepsin in 0.01 M HCl for 4 min at 37°C, fixed for 5 min in 4% PFA/1× PBS at room temperature, and treated for 30 min in 0.1 mg/ml RNaseA at 37°C. After dehydration through an ethanol series, samples were denatured for 3–8 min on an 80°C heat block. For combined RNA/DNA-FISH, the RNaseA treatment was omitted. MeOH-fixed slides were not pretreated, and were denatured for 30 s on the heat block. BACs () were directly labeled with Cy3-dCTP by random priming for use as probes. The frequency of nuclei displaying SS, SD, and DD signals for single probes were scored in S-phase (BrdU positive) cells. To ensure that singlet and doublet signals were being scored consistently, three of the authors independently scored several blinded slides for SIAR, obtaining comparable results. Linked sequences can be reliably scored as being on the same chromosome over distances of up to 50 Mb (); all pairwise DNA-FISH experiments in this study were performed within this distance range.
All microscopy was performed at room temperature, on slides mounted with Vectashield (Vector Laboratories). FISH results were analyzed using a fluorescent microscope (BX60; Olympus) with a 100× oil immersion objective, NA 1.30. Images were captured with a digital camera (ORCA-ER; Hamamatsu) and Openlab 4.0.1 software. Grayscale images were combined into an RGB image using Photoshop, with Cy3, FITC, and DAPI images pasted into the red, green, and blue channels, respectively. The Photoshop Levels tool was used to adjust the upper and lower input levels for each channel to match the upper and lower boundaries of the image histogram. No gamma adjustments were made.
Immunoprecipitated DNA from each time point and a dilution series of genomic DNA were used as templates for qPCR amplification of sequences from OR11-5, , and the human DNA control. PCR primer sequences are given in . The results were quantitated using the relative standard curve method as described in ABI User Bulletin #2, normalizing to the human DNA control.
An unpaired test was used to examine which genes displayed a significantly greater percentage of SD signals than the biallelic control , with P ≤ 0.001. A test was also used to compare the frequency of SD FISH signals in PFA- and MeOH-fixed cells (Fig. S2 B). All P values to determine statistical significance of coordination or switching were calculated using a χ–square test, with a null hypothesis of a 50:50 (random) distribution.
Figure S1 shows full scoring of FISH signal appearance in PFA-fixed ES cells and ES cell FACs profile. Figure S2 shows FISH signal appearance in MeOH-fixed ES cells. Figure S3 shows full scoring of FISH signal appearance in MEFs.
and
mutant cells. Online supplemental material is available at . |
Telomeres, which are composed of tandem repeats of the TTAGGG sequence and associated proteins, are nucleoprotein structures that cap the ends of chromosomes (for reviews see ; ; ). Telomere length is maintained by telomerase, a reverse transcriptase that counteracts telomere shortening associated with cell division by de novo addition of telomere repeats onto chromosome ends (for review see ). Telomerase is expressed in the stem cell compartment of several adult tissues, where it is thought to compensate for telomere shortening associated with cell proliferation and tissue regeneration (for reviews see ; ; , ; ). Interestingly, telomerase activity levels are not sufficient to maintain telomere length during human aging, and telomeres progressively shorten with increasing age in the context of the organism (; ; for review see ) and are also associated with different disease states (; ; ; ; ). Indeed, telomerase levels in humans and mice are thought to be rate limiting for organismal life span. In particular, reduced telomerase activity caused by mutations in telomerase components in the human diseases dyskeratosis congenita (; , ; ), aplastic anemia (; ), and idiopathic pulmonary fibrosis (; ) leads to accelerated telomere shortening, premature loss of tissue regeneration, and premature death. Some of these phenotypes are shared by telomerase-deficient mice (
mice; ), which show a reduction in both the median and the maximum life span already within the first mouse generation ().
mouse generations with progressively shorter telomeres concomitant with premature loss of tissue regeneration and organismal survival (; ; ; ; ). Similarly, disease anticipation has also been reported with increasing generations of human dyskeratosis congenita patients ().
The telomerase-deficient mouse model has been instrumental in understanding the effects of telomere shortening on stem cell biology (for review see ). In particular, we recently showed that telomere shortening results in an impaired capacity of hair follicle (HF) stem cells to regenerate the hair and the skin because of a defective mobilization of the HF stem cells out of their niche (). This defective stem cell behavior anticipates the fact that telomerase-deficient mice show premature skin-aging phenotypes such as decreased wound healing, hair loss, and hair graying (; ; ), as well as decreased skin cancer, as indicated by the fact that they are resistant to skin carcinogenesis protocols (). These findings suggested that the progressive telomere shortening that occurs in human tissues with increasing age might directly impact the ability of different adult stem cell populations to maintain tissue homeostasis. Furthermore, these results opened the possibility that restoration of telomerase activity may be sufficient to correct stem cell defects associated with short telomeres and to extend the organismal life span.
Here, we demonstrate that restoration of a copy of the Terc gene into late generation G3/G4 telomerase-deficient mice is sufficient to elongate critically short telomeres in skin keratinocytes from these mice, prevent end-to-end chromosome fusions, and rescue both HF stem cell defects in vivo and the impaired proliferative capacity of epidermal stem cells ex vivo. Finally, telomerase reintroduction was able to extend the normal life span of G4 telomerase-deficient mice by preventing degenerative pathologies in the absence of increased cancer. These findings support the notion that telomerase activators would be sufficient to correct stem cell defects in tissues with critically short telomeres in the absence of undesired effects.
× G2/G3
intercrosses (Materials and methods). The progeny of these crosses was divided into two mouse cohorts according to their telomerase status: G3/G4
telomerase-deficient mice and G3/G4
telomerase-reconstituted mice (referred to here as G3/G4
mice). Importantly, both mouse cohorts inherited the same telomere length from the parents; however, G3/G4
mice lack telomerase activity and G3/G4
mice are telomerase proficient (). To address whether telomerase activity in G3
mice was able to rescue short telomeres compared with G3
cohorts, we measured telomere length using quantitative FISH (Q-FISH) on primary keratinocytes obtained from newborn mice (; ; Materials and methods). G3
skin keratinocytes showed, on average, longer telomeres than those from the corresponding G3
littermates (P < 0.001; ). Furthermore, this telomere elongation coincided with a significant reduction of the percentage of short telomeres (<100 arbitrary units of fluorescence [a.u.f.]) in G3
mice compared with the corresponding G3
littermates (P < 0.001; ), which is in agreement with previous findings showing that telomerase preferentially elongates short telomeres both in yeast and mammals (; ; ). The percentage of longest telomeres (>1,000 a.u.f.) was also significantly increased in telomerase-reconstituted G3
mice compared with the corresponding G3
littermates (P = 0.015; ), in agreement with the increase in mean telomere length.
Finally, we determined whether the elongated telomeres in G3
keratinocytes correlated with a significant rescue of chromosomal aberrations associated with critically short telomeres. For this, we performed a full karyotypic analysis using telomere Q-FISH on metaphases (Materials and methods). As shown in , telomerase-reconstituted G3
keratinocytes showed a significant rescue of chromosomal abnormalities associated with critically short telomeres, such as signal-free ends and end-to-end fusions, compared with the corresponding G3
controls (P < 0.001 for all comparisons), suggesting that these types of aberrations are the direct consequence of critical telomere shortening and that when short telomeres are reelongated by telomerase they are completely prevented. In contrast, breaks and fragments are not rescued by telomerase reintroduction, indicating that they are not the direct consequence of critical telomere shortening.
To investigate whether elongation of short telomeres by telomerase in skin keratinocytes was sufficient to rescue epidermal stem cell defects in late generation telomerase-deficient mice (), we compared the number and mobilization ability of epidermal stem cells in the HF stem cell niche before and after mitogenic activation in G3
mice with that of the corresponding G3
littermates (). To visualize HF stem cells, we used a labeling technique previously shown to mark self-renewing and multipotent epidermal cells, the so-called label retaining cells (LRCs; for review see ; ; ). Of notice, these experiments were performed in young mice (0–2 mo old) from both genotypes before any skin phenotypes associated with short telomeres were detectable. Confocal microscopy revealed that LRCs are enriched at the bulge area of the HF in the two genotypes, which is in agreement with the known location of the HF stem cell niche (; ; ; ; ).
In control resting skin conditions, we did not detect significant differences in the numbers of LRCs at the hair bulge of G3
mice compared with the corresponding G3
littermates (P = 0.090; ). To test whether the hair bulge stem cells were able to mobilize (exit their quiescence state and migrate) out of the niche, we studied the response of G3
and G3
LRCs to 12--tetradecanoylphorbol-13-acetate (TPA) treatment, a potent tumor promoter that activates LRCs to give numerous progeny (). TPA treatment results in rapid disappearance of LRCs (), skin hyperplasia (), and entry of HFs into their anagen (growing) phase (). After TPA treatment, only 34% of the LRCs mobilized out of the HF niche in G3
mice () compared with the previously reported 70% mobilization in wild-type mice of different genetic backgrounds (), thus confirming the defective stem cell mobilization associated with short telomeres. In contrast, 65% of LRCs mobilized in TPA-treated G3
mice (P = 0.006; ). A similar rescue in HF stem cell mobilization defects was obtained when comparing G4
mice to the corresponding G4
littermates (Fig. S1, A–C, available at ). Collectively, these results demonstrate that telomerase reintroduction in late generation G3 and G4
mice significantly rescues epidermal HF stem cell mobilization defects compared with the G3 and G4
littermates.
In agreement with the defects in HF stem cell mobilization associated with short telomeres, HF length was not significantly increased in G3
mice after TPA treatment (P = 0.80; ), reflecting a defective HF anagen response in these mice after TPA treatment (). Again, telomerase reintroduction rescued this defect in G3
littermates, where HF length was significantly increased in response to TPA treatment compared with resting nontreated skin (P < 0.001; ), suggesting that telomerase is sufficient to restitute skin homeostasis in these mice. Similar results were obtained for increased interfollicular epidermis (IFE) thickness in response to TPA treatment. Again, telomerase reintroduction in G3
littermates resulted in increased IFE hyperplasia compared with littermate G3
mice in response to TPA treatment (P < 0.05; ). To study whether the increased IFE hyperplasia in G3
mice compared with the G3
controls was associated with significant differences in cell proliferation or apoptosis, we performed immunohistochemistry of skin sections with antibodies against Ki67 and caspase 3 to detect proliferating and apoptotic cells, respectively (Materials and methods). As shown in Fig. S2 (available at ), we did not detect considerable differences in the percentage of Ki67-positive cells between G3
and G3
either in resting skin conditions or upon TPA treatment. Similarly, we were unable to detect caspase 3–positive cells in the skin of G3
and G3
mice, suggesting that apoptosis is not a major cellular response to critically short telomeres in the skin (Fig. S3). Finally, we studied whether there were differences in skin differentiation markers between G3
and G3
mice by performing immunohistochemistry with antibodies against K14 and p63, two skin basal-layer markers whose expression is normally reduced at the suprabasal skin layers (Materials and methods). We could not detect significant differences in the percentage of cells or in the number of keratinocyte layers positive for these markers (Figs. S4 and S5).
Next, we used hair-plucking experiments as an independent way to induce entry of HFs into their anagen phase (; Materials and methods). In control resting skin conditions, we did not detect differences in back skin HF length and dermis thickness between G3
and G3
littermates (P = 0.71 and P = 0.49, respectively; ). Upon hair plucking, however, G3
mice showed a significantly increased back skin HF length and dermis thickness compared with the corresponding G3
littermates (P = 0.04 and P = 0.09, respectively; ), again demonstrating that telomerase reintroduction in mice with short telomeres is able to improve the ability of epidermal HF stem cells to mobilize and regenerate the hair and skin.
It has been previously described that the epidermal stem cell defects observed in late generation telomerase-deficient mice are cell autonomous, as indicated by a defective clonogenic potential of these cells ex vivo (). Individual colonies in clonogenic assays have been proposed to derive from single stem cells (; Materials and methods). Here, we performed clonogenic assays to compare the proliferation potential of telomerase-deficient G3
and telomerase-reconstituted G3
epidermal stem cells (Materials and methods). In agreement with the in vivo results shown in , primary keratinocytes from newborn G3
mice formed fewer and smaller colonies than those from wild-type controls (P < 0.001; ), reflecting the previously described defective capacity of late generation -null epidermal stem cells to proliferate ex vivo (). Interestingly, the defective clonogenic potential of these cells was significantly corrected in telomerase-reconstituted G3
keratinocytes compared with the corresponding G3
controls (P = 0.006; ), demonstrating that telomerase reintroduction ameliorates the ex vivo proliferative capacity of epidermal stem cells from late generation telomerase-deficient mice.
We have previously described that late generation telomerase-deficient mice show a small body-size phenotype at the time of birth, which is associated with shorter telomeres in these mice as well as with a decreased clonogenic potential of epidermal stem cells in vitro (). These observations opened the intriguing possibility that body size and stem cell proliferative potential may be mechanistically related. Here, we first confirmed that newborn G3
mice showed a significantly lower body weight at the time of birth compared with the wild-type controls (P < 0.001; , left), which again was concomitant with a lower clonogenic potential of keratinocytes derived from G3
mice compared with the wild types (). Interestingly, the small body-size phenotype of newborn G3
mice was significantly corrected by telomerase reintroduction into G3
newborns, paralleling the rescue of stem cell phenotypes in these mice (
). This significant rescue of the small body-size phenotype observed in newborns was maintained when comparing age-matched adult (2 mo old) G3
mice to the corresponding G3
littermates (P < 0.001; , right). These results support that the small body size of telomerase-deficient mice may be linked to decreased stem cell functionality.
The results presented here for epidermal HF stem cells open the possibility that telomerase reintroduction into mice with critically short telomeres may be sufficient to restore stem cell functionality in different tissues, thus rescuing life span and long-term survival.
mice ().
controls, which went from ∼130 and 140 wk, respectively, to <50 wk in the case of G4
mice (P < 0.001; ).
controls (P < 0.001; ).
controls (, NS; P = . 0.56), indicating that telomerase reintroduction into mice with critically short telomeres is sufficient to restore a normal long- term survival in these mice.
heterozygous mice.
heterozygous mice (), which is in agreement with a potent tumor suppressor role for short telomeres in the context of telomerase deficiency (; ; ).
heterozygous controls, respectively (), suggesting that telomerase reintroduction is sufficient to sustain normal tumorigenesis in G4
mice.
heterozygous controls (P < 0.001; ) may be related to the fact that not all chromosomal defects associated with short telomeres are rescued by telomerase reintroduction (i.e., fragments and breaks; ).
Finally, telomerase-reconstituted G4
mice also showed a significant rescue of degenerative pathologies compared with G3 and G4
mice. In particular, atrophies of the small intestine appeared in only 6% of G4
mice at time of death compared with 100% of the G3 and G4
mice ().
controls with a similar dose of the Terc allele, illustrating a complete rescue of degenerative pathologies associated with late generation
mice. Collectively, these results suggest that telomerase reconstitution into mice with critically short telomeres is sufficient to confer a normal life span and normal aging in the absence of abnormally increased tumorigenesis.
controls in both age-matched adult skin keratinocytes (10 mo old; ) and primary splenocytes (12–24 mo old; ), suggesting that telomerase activity with a normal telomere length is able to provide homeostasis during the life span of these mice.
The mechanisms by which short telomeres negatively impact on organismal aging and life span are still far from being understood. One of the proposed mechanisms, which has gained increasing experimental support, is the progressive loss of stem cell functionality associated with critically short telomeres. Evidence for this comes from the study of the telomerase-deficient mouse model, which shows impaired stem cell functionality in several tissues including the bone marrow, the brain, and the skin (; ; ; ). In particular, we have recently shown that late generation telomerase-deficient mice show an impaired ability of epidermal stem cells to mobilize out of their niches and to regenerate the skin and the hair. This defective stem cell behavior anticipates the fact that telomerase-deficient mice show premature aging of the hair and the skin as well as an increased resistance to developing skin cancer (; ; ; ), supporting the notion that short telomeres provoke aging by impairing the functionality of stem cells.
Here, we provide further support for a stem cell theory of telomere-mediated aging by showing that telomerase reintroduction in late generation telomerase-deficient mice is sufficient to restore a normal behavior of epidermal HF stem cells and a normal skin functionality in these mice, therefore supporting the notion that stem cells are important players in the known role of telomeres and telomerase in aging.
and
mice. Indeed, the defective mobilization ability of epidermal HF stem cells anticipates the premature skin aging phenotypes of these Terc-deficient mice. Furthermore, we demonstrate that telomerase reconstitution in the context of very short telomeres not only corrects epidermal HF stem cell defects in newborn mice but is also sufficient to sustain a long-term normal organismal life span in these mice by preventing organismal aging in the absence of increased cancer. It is important to highlight that telomerase-reconstituted mice show a telomere length that is indistinguishable from that of normal, nonreconstituted, Terc heterozygous mice, indicating that telomerase activity not only elongates short telomeres but is able to restitute a normal telomere-length homeostasis during the life span of these mice.
Finally, these observations support the idea that therapies based on telomerase activation may be effective in correcting the proaging effects of short telomeres in the absence of increased risk of carcinogenesis. This is of particular relevance in the case of premature aging diseases characterized by decreased levels of telomerase activity and shorter telomeres, such as some cases of dyskeratosis congenita and aplastic anemia, which result in premature death associated with a defective tissue renewal capacity (bone marrow and skin) and increased cancer ().
females (). Genotyping was performed as described in . Note that littermate G3/G4
and G3/G4
mice are of an exact genetic background (C57BL6) as if they were derived from the same parents.
Mouse colonies were generated in a pure C57BL6 background and maintained at the Spanish National Cancer Center under specific pathogen-free conditions in accordance with the recommendations of the Federation of European Laboratory Animal Science Associations.
Primary mouse embryonic fibroblasts (MEFs) were trypsinized and washed in PBS, and S-100 extracts were prepared as described in . Three protein concentrations were used for each sample (5, 2, and 1 μg). Extension and amplification reactions and electrophoresis were performed as described in . A negative control was included by preincubating each MEF extract with RNase for 10 min at 30°C before the extension reaction. An internal control for PCR efficiency was included (TRAPeze kit; Oncor).
To induce LRC mobilization, IFE hyperplasia, and anagen entry, tail skin from 71-d-old mice in the telogen (resting) phase of the hair cycle was topically treated every 48 h with TPA (20 nM in acetone) for a total of four doses. The control mice were treated with acetone only. 24 h after the last TPA treatment, mice were killed and the tail skin was analyzed. To induce anagen by physical stimulation, dorsal HFs in the telogen phase of the hair cycle were plucked from the back skin of 60-d-old G3
mice and the corresponding G3
controls. 10 d after plucking, dorsal skins were harvested and prepared for histology.
LRCs were obtained as described in , , and , with some modifications. In brief, litters of neonatal mice were injected with 50 mg/kg of bodyweight BrdU (Sigma-Aldrich) diluted in PBS. Each animal received a daily injection beginning at day 4 of life for a total of 5 d. After the labeling period, mice were allowed to grow for 60 d before the initiation of any treatment. Cells retaining the label at the end of the treatment were identified as LRCs.
Whole mounts of mouse tail epidermis were prepared as previously described in . In brief, after mice were killed with CO and their tails were amputated, skin was peeled from the tails and incubated in 5 mM EDTA in PBS at 37°C for 4 h. Using forceps, intact sheets of epidermis were separated from the dermis and fixed in neutral-buffered formalin for 2 h at room temperature. Fixed epidermal sheets were maintained in PBS containing 0.2% sodium azide at 4°C before labeling.
To detect LRCs in whole mounts of the tail skin, fixed epidermal sheets were blocked and permeabilized by incubation for 30 min in a modified phosphate buffer () containing 0.5% BSA and 0.5% Triton X-100 in TBS. Subsequently, epidermal sheets were immersed for 30 min in 2 M HCl at 37°C, incubated overnight with a mouse anti-BrdU antibody conjugated with fluorescein (Roche) at 1:50 in modified PB buffer, washed four times in PBS containing 0.2% Tween 20, and mounted in Vectashield (Vector Laboratories).
A laser scanning confocal microscope (TCS-SP2-AOBS; Leica) was used to obtain fluorescence images. Image stacks of 60–80 μm were obtained through the z dimension at steps 1.0 μm apart, using a PL APO 20×/0.70 PH2 (Leica) as lens. Maximum intensity projections of the image stacks were then generated using LCS Software (Leica).
Mice were killed when they showed signs of poor health, such as reduced activity or weight loss, and subjected to exhaustive histopathological analysis. The organs we analyzed for age-related degenerative pathologies were the intestine (atrophy of the small and large intestine), kidney (glomerulonephritis and tubular degeneration), spleen (atrophy, hemosiderosis, and myeloid and lymphoid hyperplasia), liver (congestion, vacuolar degeneration, microgranuloma, and steatosis), testis (atrophy and ectasis of seminal vesicles), ovary (atrophy), uterus (cystic endometrial hyperplasia), skin (benign hyperplasia), lung (congestion), heart (congestion and cardiomyopathy), and brain (calcification).
2-d-old mice were killed and soaked in betadine (5 min), in a PBS antibiotic solution (5 min), in 70% ethanol (5 min), and again in a PBS antibiotic solution (5 min). Limbs and tail were amputated and the skin was peeled off using forceps. Skins were then soaked in PBS (2 min), PBS antibiotic solution (2 min), 70% ethanol (1 min), and again in PBS antibiotic solution (2 min). Using forceps, each skin was floated on the surface of 1× trypsin solution (4 ml on a 60-mm cell culture plate; Sigma-Aldrich) for 16 h at 4°C. Skins were transferred to a sterile surface. The epidermis was separated from the dermis using forceps, minced, and stirred at 37°C for 30 min in serum-free Cnt-02 medium (CELLnTEC Advanced Cell Systems AG). The cell suspension was filtered through a sterile Teflon mesh (Cell Strainer 0.7 m; BD Biosciences) to remove cornified sheets. Keratinocytes were then collected by centrifugation (160 ) for 10 min and seeded on collagen I–precoated cell culture plates (BD Biosciences).
1,000 mouse keratinocytes per genotype were seeded onto 10 μg/ml mitomycin C (2 h), treated with J2-3T3 fibroblasts (10 per well, 6-well dishes), and grown at 37°C/5% CO in Cnt-02 medium. After 10 d of cultivation, dishes were rinsed twice with PBS, fixed in 10% formaldehyde, and then stained with 1% Rhodamine B to visualize colony formation. Colony size and number were measured using three dishes per experiment.
Freshly isolated splenocytes were obtained by squeezing the spleen through a cell strainer (70 μm; Nylon; BD Biosciences). Red cells were lysed by osmotic shock, and the splenocytes were resuspended in RPMI 1640 containing 10% FBS and 0.55 μM β-mercaptoethanol. Concanavalin A (Sigma-Aldrich) was added to a concentration of 5 μg/ml and splenocytes were grown for 48 h. The cells were incubated with 0.1 μg/ml colcemide (Invitrogen) for 2 h and fixed in methanol/acetic acid (3:1). Q-FISH was performed as described in and . To correct for lamp intensity and alignment, images from FluoroSpheres (fluorescent beads; Invitrogen) were analyzed using the TFL-Telo software (provided by P. Lansdorp, Terry Fox Laboratory, Vancouver, Canada). Telomere fluorescence values were extrapolated from the telomere fluorescence of lymphoma cell lines LY-R (R cells) and LY-S (S cells) with known telomere lengths of 80 and 10 kb, respectively. There was a linear correlation (r = 0.999) between the fluorescence intensity of the R and S telomeres. We recorded the images using a camera (CCK; COHU) on a fluorescence microscope (DMRb; Leica). A mercury vapor lamp (CS 100 W-2; Philips) was used as a source. We captured the images using the Q-FISH software (Leica) in a linear acquisition mode to prevent oversaturation of fluorescence intensity. We used the TFL-Telo software () to quantify the fluorescence intensity of telomeres from at least 10 metaphases for each data point.
Exponentially growing primary keratinocytes were fixed in methanol/acetic acid, and Q-FISH of interphase nucleus was performed. For Q- FISH in tail skin, paraffin-embedded tail sections were deparaffinated. Both keratinocytes and deparaffinated sections of tail skin were hybridized with a PNA-telomeric probe and telomere fluorescence was determined as described in and . More than 60 nuclei from each mouse and condition were captured at 100 magnification using a microscope (CTR MIC; Leica) and a camera (High Performance CCD; COHU). Telomere fluorescence was integrated using spot IOD analysis in the TFL-TELO program ().
Metaphases from keratinocytes of the indicated genotypes were obtained by adding 1 μg/ml colcemide (Invitrogen) to primary keratinocytes during 5 h and then fixing in methanol/acetic acid (3:1). Q-FISH was performed as described in and . For analysis of chromosomal aberrations, 50 metaphases per genotype were analyzed by superimposing the telomere image on the DAPI image using the TFL-telo software.
Unless otherwise indicated, data are given as mean values ± SEM of and have been analyzed for statistically significant differences using test.
Fig. S1 shows rescue of HF stem cell mobilization defects in late generation telomerase-reconstituted G4
mice. Fig. S2 shows similar proliferation rates in G3
and G3
tail skin. Fig. S3 shows no detectable apoptosis in the skin of G3
mice and G3
siblings. Fig.
and G3
tail skin. Fig.
and G3
tail skin. Online supplemental material is available at . |
The human immunodeficiency virus (HIV) type 1 genome consists of a single transcription unit that is integrated into cellular chromatin. HIV-1 is highly dependent on transcriptional regulation: acutely infected cells synthesize high levels of virus, whereas latently infected cells transcribe little or no viral RNAs. This tight regulation is a critical feature of viral pathogenicity because it allows the virus to remain silent in the organism and prevents clearance by the current antiretroviral regimens (for review see ; ).
The HIV-1 promoter is located in the U3 region of the 5′ long terminal repeat (LTR). Its transcription is performed by the cellular machinery, but it is boosted by the viral protein Tat (for reviews see ; ). In latently infected cells that do not produce Tat, the polymerases initiating at the HIV-1 promoter are unprocessive and unable to transcribe the entire viral genome. Lymphocyte-activating stimuli induce the HIV-1 promoter to produce small amounts of Tat, which starts a positive feedback loop by stimulating viral transcription (). Tat recruits to the HIV-1 promoter the active form of the positive transcription elongation factor P-TEFb (), which consists of a complex between cyclin T1 and Cdk9 (for review see ). Tat binds to both cyclin T1 and the trans-activation–responsive region, an RNA element present at the 5′ end of all viral transcripts. This induces the formation of a tertiary complex on nascent RNAs, which brings Cdk9 into position to phosphorylate several components of the transcription machinery, including the C-terminal domain (CTD) of the large subunit of RNA polymerase II (RNAPII), and elongation factors DSIF and negative elongation factor (NELF; for review see ). This converts RNAPII into a highly processive enzyme, which can transcribe the entire viral genome.
Production of HIV-1 mRNAs is not only regulated at the level of transcription but also at the level of splicing and polyadenylation (; ). HIV-1 possesses two polyadenylation sites, one in each LTR. Polyadenylation at the first site is suppressed by binding of the small nuclear RNP U1 to the major splice donor (SD1; ), whereas polyadenylation at the second site is activated by upstream sequences present in the U3 region of the 3′ LTR (; ).
Many of the regulatory events that determine the fate of HIV-1 RNAs occur at the transcription site. This includes the decision to initiate transcription, to elongate, to splice, and to process the RNA at the 3′ end. Most importantly, the relative kinetic rate of each process appears to be critical for their final outcome. For instance, the rate of cleavage at the first polyA site and the rate of U1 binding at SD1 determine the amount of read-through at this polyA site (). Similarly, the rate of splicing versus 3′-end formation at the second polyA site determines the amount of unspliced RNA available for Rev-mediated export (). In these two cases, the kinetic competition is influenced by the rate of transcription elongation. For instance, cleavage at the first polyA site cannot be suppressed until the polymerase reaches SD1, and splicing cannot be competed out by 3′-end formation until the polymerase arrives at the second polyA site. Importantly, the rate of transcription elongation can be regulated, and this can control gene expression (; ).
Like HIV-1, many cellular genes are regulated at the level of transcription, and transcriptional regulation has been the subject of a large number of studies. However, although transcription by RNAPII is a fundamental process, we still lack a precise view of how transcription occurs in vivo. In particular, we lack detailed kinetic models describing mRNA synthesis and the cycle of RNAPII. In this study, we used FRAP to study these processes directly in living cells.
Previous studies have shown that tagging RNA with 24 binding sites for the coat protein of phage MS2 allows its detection in living cells with excellent sensitivity (; ). To analyze the biogenesis of HIV-1 mRNA, we inserted 24 MS2-binding sites in the 3′ untranslated region of an HIV vector that carried the elements required for RNA production (): the 5′ LTR, the major splice donor (SD1), the packaging signal Ψ, the Rev-responsive element, the splice acceptor A7 flanked by its regulatory sequences (exonic splicing enhancer and ESS3), and the 3′ LTR that drives 3′-end formation. Stable arrays of this reporter construct (pExo-MS2×24) were integrated into U2OS cells. Two clones (Exo1 and Exo2) showed robust trans-activation by Tat and other stimuli known to induce transcription of integrated HIV-1 promoters (Fig. S1, available at ). When expressed, the RNAs were distributed homogenously in the cytoplasm and concentrated in a bright spot in the nucleoplasm. This spot corresponded to the transcription site as it colocalized with RNAPII and was labeled with probes directed against the nontranscribed strand of the vector (). Several active genes localize near speckles (; ). However, the HIV-1 transcription site rarely colocalizes with speckles labeled by the marker protein SC35 ().
To visualize nascent HIV-1 RNAs, a nuclear MS2-GFP fusion was expressed in Exo1 or Exo2 cells. In live cells, MS2-GFP was diffused in the nucleoplasm and concentrated in a spot at the transcription site (). FRAP is a powerful technique to study the dynamic properties of a fluorescent molecule, and we used it to study mRNA synthesis by photobleaching the nascent RNAs labeled with MS2-GFP. Indeed, when a transcription site is bleached, incoming polymerases will synthesize new MS2 sites, and this will result in recovery of the fluorescent signal. In this system, RNAs are visualized indirectly through MS2-GFP, and this may complicate the FRAP analysis (). First, a slow diffusion rate of MS2-GFP may mask the neosynthesis of nascent RNAs. Second, a rapid dissociation of the RNA-bound MS2-GFP may lead to rapid recovery rates unrelated to the synthesis of new RNAs.
To obtain an estimate for the rate of exchange of the MS2-GFP protein on its target site, we used an abundant noncoding RNA, U3, which was modified to incorporate a single MS2-binding site. U3 is synthesized in the nucleoplasm and accumulates in nucleoli, where it plays essential functions during ribosomal RNA biogenesis (). Previous work has shown that a fraction of the proteins associated with U3 exchange slowly between nucleoli and the nucleoplasm (), and it was therefore suspected that a fraction of U3 would also be stably associated with nucleoli. When nucleoli of cells expressing only MS2-GFP were bleached, fluorescence recovery was nearly complete within seconds (Fig. S2, available at ). In striking contrast, when nucleoli of cells expressing both U3-MS2 and MS2-GFP were bleached, only a fraction of the fluorescence was recovered, even after 10 min. This indicated that a part of U3-MS2 was immobile and that bleached MS2-GFP molecules stayed stably bound to the RNA during the course of the experiment. Furthermore, diffusion of MS2-GFP was rapid (15 μm/s) and much faster than recovery at transcription sites (). Thus, the diffusion or dissociation of MS2-GFP was neglected in the analysis of FRAP experiments.
Recovery of Exo1 transcription sites showed that the synthesis of new RNAs occurred in <3 min and that virtually no RNAs were retained at the transcription site for a long time (immobile fraction of 3%; ). A second clone expressing the same MS2-tagged RNA yielded similar kinetics (Exo2; ), indicating that this was principally a property of the reporter.
RNA biogenesis occurs in a series of steps: transcription, 3′-end formation (cleavage and polyadenylation), and release from the transcription site (). To test whether elongation was rate limiting, we used a slow version of RNAPII, the hC4 mutant (). This α-amanitin–resistant mutant was transfected in Exo1 cells, endogenous enzymes were inactivated with α-amanitin, and remaining transcription sites were analyzed by FRAP. As a control, we used another α-amanitin–resistant version of RNAPII that has no elongation defect (wild type [WT]; ). FRAP curves obtained with WT or the endogenous polymerase were nearly identical (). In contrast, recoveries were markedly slower with the hC4 mutant: 129 s for half-recovery versus 75 s for the WT enzyme. To confirm that transcriptional elongation took a substantial part of the time that nascent RNAs spent at their transcription sites, we used camptothecin. This drug targets topoisomerase I, and it transiently cross-links it to DNA. A previous study has shown that this creates steric road blocks for polymerases, which result in increased pausing and a net slowdown of transcriptional elongation (). Indeed, when cells were treated with this drug, the recovery of nascent RNAs was slower: 170 s for half-recovery in Exo1 cells compared with 65 s in the case of untreated cells ().
To further establish that elongation was rate limiting, we developed a cell line (ExoLong) that integrated a second reporter, pExo-MS2 × 24-Long, which differed by 2.4 kb in the length of the sequence separating the MS2 repeat and the end of the gene. If elongation is rate limiting, the recovery should be slower for the long reporter, and the difference should represent the time taken to transcribe the additional sequences. As expected, transcription sites of ExoLong recovered more slowly: 126 s for half-recovery versus 65 s for the short reporter (). Altogether, these observations demonstrated that elongation took a substantial fraction of the time dedicated to mRNA production.
In kinetic terms, transcription elongation is the repetition of an elementary step: the addition of one nucleotide. The stochastic basis of the process implies that the time taken to add a single nucleotide is variable and follows an exponential described by the rate constant k. In contrast, when polymerases transcribe a large number of nucleotides, the repetition of the elementary step creates a statistical averaging such that the time taken to synthesize nucleotides will be virtually constant for all polymerases and equal to /k. This feature of transcription elongation is distinctive, and it predicts a linear increase in signal during FRAP recovery, whereas single-step processes should result in exponentials (Fig. S3 B, available at ).
Elongation is the first step of mRNA biogenesis that can be observed in our FRAP experiments. Indeed, once the polymerase reaches the polyA site, the pre-mRNA is cleaved, polyadenylated, and released. These processes occur as a series of single steps and, thus, may be modeled as exponentials. We attempted to fit the FRAP curve with two components: a straight line for elongation followed by a single exponential for 3′-end formation and release. This two-step model fitted experimental data substantially better than a single exponential (, G and H; and 3). The short reporter gave 65 to 73 s for elongation and 44 to 31 s for the half-time of the exponential in Exo1 and Exo2 cells, respectively (). The long reporter yielded 148 s for elongation and 65 s for the exponential. This translated into similar elongation rates of 1.79–2.03 kb/min. Cells treated with camptothecin or transfected with the slow mutant of RNAPII could also be fitted to this two-step model ( and ). For the hC4 mutant, it yielded an elongation rate more than twofold slower than with WT RNAPII (0.8 kb/min), which is in agreement with in vitro data ().
The proportion of time dedicated to elongation versus 3′-end processing can also be estimated by direct comparison of the short and long reporters. Indeed, if initiation rates are equal, the intensity of the MS2 signal at various transcription sites should be proportional to the time that the RNA spends there. When the curves of the short and long reporters were normalized to their initiation rate, that is, to their initial slopes, the long reporter accumulated 1.5 times more RNAs at the transcription site (). If one assumes that the time required for elongation is proportional to the RNA length, whereas other steps are identical for the two reporters, this value corresponds to half of the time spent on elongation for the short reporter and 71% for the long one (see Materials and methods). This was in agreement with the values obtained by fitting the FRAP curves with the two-step model ().
Next, we performed quantitative in situ hybridization with oligonucleotide probes that hybridized along the gene. Incompletely transcribed RNAs should yield more signals with probes hybridizing at their 5′ end, whereas full-length RNAs should yield equal signals with 5′ and 3′ probes (). Because incompletely transcribed RNAs correspond to elongating molecules, whereas full-length RNAs are at the stage of 3′-end processing or transcript release, the ratio of 5′ to 3′ probes can be used to estimate the relative time taken by elongation versus 3′-end processing and release. Four sets of Cy3-labeled probes were used (): the first hybridized in exon 1 (E1), the second in the intron (I), the third at the splice acceptor site (I-E2), and the last in exon 2 (E2), immediately before the active polyA site. When signals were normalized with a Cy5-labeled probe against the MS2 repeat, 5′ probes gave 2.2-fold more signal than 3′ probes (). To confirm this, we hybridized E1-Cy5 and E2-Cy3 probes simultaneously, and signals at the transcription sites were normalized to signals in the cytoplasm. Again, we found twofold more E1 probe at transcription sites (Fig. S4, available at ). From these values, we could estimate that half of the polymerases that have reached the MS2 repeat are elongating toward the polyA site, whereas the other half are at the stage of 3′-end formation (see Materials and methods), which is in agreement with FRAP experiments.
To further characterize RNA species present at transcription sites, we used a probe specific for polyadenylated HIV-1 mRNAs. This probe yielded robust signals in the cytoplasm but only faint signals at transcription sites (11% of E2 signal; ). This was unlikely the result of a failure of the probe to hybridize to its target because the nuclear polyA-binding protein PABN1 was also not detected there (). This indicated that RNAs were released rapidly once polyadenylation started. Indeed, with a total 3′-end processing time of 63.5 s (), we calculated that cleavage/polyadenylation took 54.7 s and mRNA release took 8.8 s (see Materials and methods).
The CTD of the large subunit of RNAPII has been proposed to connect the polymerase with the 3′-end processing machinery (for review see ). In addition, RNAPII lacking the CTD have defects in transcription initiation and in elongating through nucleosomal templates (). Remarkably, transcription sites generated by this RNAPII mutant recovered very slowly after photobleaching (half-recovery of 550 s vs. 75 s for WT; ). A defect in initiation should result in lower amounts of fluorescent signal at the transcription site, but it should not affect the time that polymerases spend on the gene. Thus, the defects we observed arose from a reduced ability to synthesize or process nascent RNAs, which is in agreement with biochemical data.
To confirm that alteration of 3′-end processing could slow down the release of HIV-1 RNA, we made a reporter containing mutations near the polyA site. Besides the canonical AAUAAA and G/U-rich element, HIV-1 contains upstream elements that are required for efficient 3′-end formation. These elements are present in the U3 region of the LTR, partly explaining why the upstream polyA site in the 5′ LTR is not used (; ). The removal of these activating sequences does not prevent 3′-end formation but renders the process inefficient (). For instance, polyA sites inserted downstream of the HIV-1 3′ LTR are normally not used, but they become active when the U3 region of the 3′ LTR is removed (). Thus, we created cell lines expressing HIV-1 reporters lacking the U3 region in the 3′ LTR (clone pTRIP_1_13). Remarkably, when MS2-GFP was bleached at these transcription sites, the signal recovered much more slowly than for WT HIV-1 reporters: half-recovery took 400 s instead of 65 s (). Thus, slowing down the rate of 3′-end formation correspondingly increased the time that nascent RNAs remained at their transcription site.
To better understand HIV-1 RNA biogenesis, we simulated the behavior of individual polymerases in silico. For each time interval, each polymerase could initiate, elongate by one nucleotide, cleave and polyadenylate, and eventually release their associated RNA. The probabilities to perform these steps were the means of the measured rate constants (see Materials and methods). Predicted FRAP curves approximated those obtained experimentally with both the short () and long reporters (). Simulated SDs were also in range with experimental ones. Close examination of simulations indicated that a large part of the SD was caused by stochastic variation in the total number of polymerases on the gene, which affected the prebleach values and the extent of recovery.
Interestingly, the parameters measured with the short reporter were slightly too rapid to accurately predict the recovery curve of the long one ( and ). An explanation for this observation might reside in the fact that RNAPII can pause because pausing is expected to be more frequent with longer sequences. Another interesting possibility would be that the site of integration of the reporter in the genome generates some variation in the kinetics of mRNA synthesis and processing.
Although the MS2-GFP FRAP assay provides a direct measurement of RNAPII activity, it provides little information on the events that occur either before the polymerase reaches the MS2 repeat or after it releases its mRNA. To gain a complete view of the transcription cycle, we repeated the FRAP assay with fluorescent subunits of RNAPII. HIV-1 transcription sites were identified with a red variant of MS2 (MS2-mCherry), and GFP-tagged subunit C of RNAPII was bleached in the nucleoplasm or at HIV-1 transcription sites. Recoveries at the HIV-1 transcription site were much slower than in the nucleoplasm, suggesting that most polymerases present at the HIV-1 gene array were engaged in transcription (). Interestingly, recovery of the polymerase was substantially slower than recovery of nascent RNAs: in Exo1 cells, half-recovery took 200 s for the polymerase but only 66 s for nascent RNAs.
To extract more information from the recovery curves, they were fitted with the diffusion/reaction model developed by . This model assumes that the bleached spot contains identical binding sites that are equally distributed in space, and it allows one to derive diffusion coefficients, binding time (t), and the delay between two binding events (t). The recovery curves obtained with the subunit C of RNAPII were fitted to this model (see Materials and methods). In Exo1 cells, we found that the polymerase resided for 333 s at the HIV-1 transcription site and diffused for 10 min before engaging a second transcription cycle.
When the residency time of the polymerase was compared with that of nascent RNAs, it was obvious that the polymerase remained on the gene longer than expected. Indeed, elongating through the reporter should take 114 s, and 3′-end formation should take 63 s. Thus, 156 s were missing to match the 333 s of the residency time of the polymerase. One possibility could be that RNAPII proceeds after the 3′ LTR before terminating transcription. To test this possibility, we performed chromatin immunoprecipitation experiments. As expected, RNAPII was enriched within the HIV gene after Tat induction (primer sets A and B; ). Surprisingly however, a PCR fragment located 220–420 bases downstream of the polyA site did not show a comparable enrichment (primer set C), indicating that RNAPII stops rapidly after the end of the gene and is released without proceeding much through neighboring sequences.
FRAP is a powerful tool to analyze the dynamics of tagged proteins. In this study, we photobleached MS2-GFP to analyze the turnover of nascent mRNAs tagged with the MS2 repeat. This supposes that there is little dissociation of the MS2-GFP–RNA complex during the course of the experiment. The MS2 protein variant we use has a mutation that increases its affinity by 7.5-fold (Table S1, available at ; ). In addition, the RNA-binding site also contains a mutation that decreases the off rate of the protein by 100-fold such that the complex has a half-life of 7 h at 4°C (Table S1; ). To directly analyze the stability of the complex in vivo, we tagged an RNA stably associated with nucleoli (U3) and found by FRAP analysis that little dissociation occurred within the first 10 min of the experiment. Although it could be argued that the complex might be less stable in the nucleoplasm than in the dense nucleoli, analysis of mutant mRNA (clone pTRIP_1_13) or polymerase (ΔCTD) set a minimal value of 10 min for the half-life of the complex in the nucleoplasm. MS2-GFP has also been previously used to analyze the diffusion of MS2-tagged mRNA in the nucleoplasm (). Interestingly, similar values (although not identical) were obtained when diffusion was measured by single-particle tracking or photoactivation, which is also in agreement with a slow dissociation rate of MS2-GFP in vivo. Thus, the contribution of the dissociation of MS2-GFP in the FRAP curves modeled in this study is expected to be small and was neglected in the analyses.
To analyze the biogenesis of mRNA in live cells and real time, we engineered cell lines that contained many copies of the reporter integrated in a single place within the chromatin (75 copies for Exo1 cells). This amplifies the signal such that many events become visible by microscopy analysis. For instance, this system can be used to analyze the dynamic of the Tat–pTEFb complex on nascent HIV-1 RNAs (). However, the repeated structure of the transgene might affect mRNA biogenesis in unknown ways. Clearly, the next frontier will be to perform similar analysis on single-copy genes.
In this study, we were able to visualize mRNA transcription and processing in real time and single cells by fluorescence tagging of HIV-1 RNAs. By performing FRAP analysis on RNAPII and nascent RNAs through MS2-GFP, we could obtain a view of the entire HIV-1 transcription cycle. For this, the polymerase recovery curves were fitted with a diffusion- binding model and compared with that of nascent RNAs. This indicated that during the 333 s that the polymerase resided at the HIV-1 transcription site, 114 s could be attributed to elongation, and 63 s could be attributed to 3′-end processing and transcript release. The remaining 156 s could be the result of initiation or termination. In an attempt to discriminate these possibilities, we investigated the localization of Xrn2, a 5′ → 3′ exonuclease involved in transcription termination (). Xrn2 is loaded cotranscriptionally at the end of genes (), and it degrades the 3′ cleavage product generated by the 3′-end maturation of mRNAs (). Interestingly, Xrn2 was present only in a minute amount at the HIV-1 transcription site (unpublished data), suggesting that termination and polymerase release may be rapid. Thus, a large fraction of the missing 156 s could be the result of initiation. In this view, as many as 47% of the polymerases could be initiating transcription at the promoter, 34% would be undergoing processive elongation, and 19% would be processing the RNA at the 3′-end (). However, it is equally possible that the steps between pre-mRNA cleavage and the entry of Xrn2 account for some of the missing 156 s. Likewise, an alternative hypothesis to explain the long residency time of the polymerase might be that a fraction of the polymerase would reengage transcription on the same gene by a looping mechanism, as described in yeast (; ). Finally, we also cannot exclude that some polymerases present at the HIV-1 transcription sites are stalled or are involved in as yet uncharacterized but slow processes.
We found that although the residency time of the polymerase at the HIV-1 gene was 333 s, the diffusion time between two binding events was 10 min. Previous FRAP experiments of the large subunit of RNAPII at random nucleoplasmic sites have indicated that for cellular genes, polymerases should spend only a third of their time engaged in transcription (). Because the half-life of transcription was estimated as 20 min, it was deduced that polymerases spend as much as 90 min diffusing between two transcriptional cycles (). Thus, initiation at the HIV-1 promoter is nearly 10 times more frequent than at cellular genes. This could be caused by a high efficiency of transcription initiation at the HIV-1 site or, alternatively, by a much higher density of active genes at the HIV-1 transcription site. Interestingly, we did not detect a rapid component in the recovery curves of RNAPII (), which corresponds to polymerases undergoing rapid binding and dissociation from the promoter (). In contrast, the vast majority of polymerases appeared transiently immobilized on the HIV-1 gene and engaged in productive transcription. This suggested that the initiation of HIV-1 transcription was indeed very efficient. A high efficiency of initiation would be consistent with biochemical studies that have shown that the HIV-1 promoter is constantly occupied by the polymerase (; ). In addition, it is well established that a major mode of trans-activating the HIV-1 promoter is at the level of elongation through the binding of Tat to nascent RNAs, and this requires that the promoter can efficiently initiate transcription (for reviews see ; ).
RNAPII is a fundamental enzyme in the cell. However, few studies were able to directly measure its activity in vivo. In vitro kinetic analyses of human RNAPII have yielded elongation rates between 0.9 and 1.8 kb/min and have suggested that the chemical step is rate limiting (). Our in vivo data are in the upper range of these values, indicating that RNAPII may work close to its maximal speed. The giant gene of human dystrophin and a long yeast gene have also been used to evaluate elongation rates (; ). These studies yielded values of 2.4 and 2 kb/min, which is in agreement with our measures. The range of values obtained in different biological systems have observed elongation rates ranging from 1.5 kb/min for living bacteria to 5.7 kb/min for eukaryotic ribosomal RNA genes (; ). Our model system provides a tool of choice for the quantitative analysis of transcription in real time and living cells.
In vitro, single-molecule analyses of bacterial RNA polymerases have shown that the time taken to transcribe a DNA segment can vary substantially between individual molecules, and at least part of this variability can be attributed to polymerase pausing (). In higher eukaryotes, pausing is also a well-known phenomenon, and there is some evidence that the elongation rate can regulate gene expression (). We describe elongation as the repetition of an elementary step and model it as a straight line in the recovery curves. This does not take into account the polymerase heterogeneity that can be caused by pausing and may directly contribute to the exponential component of the curve. Although determining polymerase pausing time may not be easy in many cases, it seems that pausing occurs when cells were treated with camptothecin. Indeed, in this case, the MS2-GFP FRAP assay indicates a large increase in the exponential component of the recovery curve. Because camptothecin is believed to physically arrest the polymerase, it is tempting to speculate that this increase represents pausing.
3′-end formation is a critical step during mRNA biogenesis, yet the rates of this reaction are not known. Our study indicates a rapid release of mRNA once polyadenylation is initiated (9 s). In contrast, the preceding steps were relatively slow (55 s). These steps probably correspond to the cleavage of the pre-mRNA, as indicated by the much longer residency time of a mutant RNA known to have a slow rate of 3′-end cleavage (pTRIP_1_13). However, we cannot rule out other possibilities. For instance, cleavage could be rapid, but a quality control step could occur before polyadenylation and could prevent mRNA release. Indeed, yeast data indicate that a quality control step occurs at the level of transcript release (), and detailed localization studies in mammalian cells have shown that some transcript can accumulate at the transcription site after detaching from the gene (; ). It is also possible that the cleaved RNA could be released before being polyadenylated.
However, with the most likely possibility being that cleavage is slow and rapidly followed by polyadenylation, we estimated that cleavage and release occurred in 55 s and 9 s, respectively. Remarkably, when the U3 sequences of the 3′ LTR were removed, the rate of 3′-end formation was dramatically reduced. The half-life of these RNAs at the transcription site was 400 s, against 65 s for WT RNAs. This indicates that the cleavage reaction may occur as much as 10 times more slowly for the mutant RNAs. This was consistent with previous biochemical analyses that have shown that U3 contains sequences that can markedly stimulate 3′-end formation by enhancing the recruitment of cleavage and polyadenylation specificity factor (). Thus, we expect that the rates of 3′-end formation and polymerase read-through may differ markedly from gene to gene depending on the strength of the polyA site. In particular, situations in which alternative splicing regulates polyA site usage implies that 3′-end formation at the first site is slow enough to let the polymerase reach the splice junctions.
Remarkably, ChIP analysis of the distribution of RNAPII along the HIV-1 gene indicated that the density of polymerases dropped sharply a few hundred bases after the polyA site, similar to what has been observed recently with an HSP70 gene of (). Because cleavage is estimated at 55 s and elongation is estimated at 2 kb/min, this suggests that polymerases pause and/or loose their processivity after passing the polyA site. This would be consistent with several previous studies. First, run-on assays and in vitro transcription reactions have suggested that the polymerase looses its processivity as it passes the polyA site (). Second, pause sites have been found adjacent to polyA sites (). Third, it has recently been shown that the 3′-end processing factor Pcf11 and the exonuclease Xrn2 have a role in promoting transcription termination at the end of cellular genes (; ) by dismantling paused elongation complexes (; ). Altogether, our data are in agreement with previously proposed models in which the 3′-end processing machinery would assemble on nascent RNAs, whereas the polymerase would pause after passing the polyA site. Completion of assembly and pre-mRNA cleavage would occur in about one minute, after which the mRNA would be rapidly released and the polymerase would be dismantled.
U2OS cells were cultivated at 37°C in DME containing 10% FCS. For live cell imaging, cells were maintained in the same medium, except it did not contain riboflavin and phenol red (). Stable transformants were obtained with the calcium-phosphate procedure by cotransfecting a 20-fold excess of the vector of interest with Ptk-Hygro and selecting cells with 132 μg/ml hygromycin. Individual clones were expanded, and their gene copy number was measured by quantitative PCR using DNA from U1 cells as reference (two copies per genome). Exo1 and ExoLong contained 75 and 70 copies, respectively.
For live cell experiments, cells were plated on glass, transiently transfected with LipofectAMINE with vectors expressing Tat and MS2-GFP, and analyzed 24 h later at 37°C in a nonfluorescent media (). For polymerase replacement, cells were treated for 2.5 h with 100 μg/ml α-amanitin before FRAP. Control experiments showed that this was sufficient to induce the disappearance of transcription sites in cells that did not express α-amanitin–resistant forms of the polymerase. Treatment of cells for 2 h with 10 μg/ml actinomycin D resulted in the disappearance of the spots in all cases.
Plasmids expressing MS2-GFP and the hC4 and WT mutant of RNAPII have been described previously (; ). Plasmid expressing PABPN1-GFP was a gift from M. Carmo-Fonseca (Institute of Molecular Medicine, University of Lisbon, Portugal). MS2-Cherry and GFP-PolII-C were created with the Gateway system (Invitrogen). U3-MS2 was created by inserting a single MS2 site in the apical loop of the rat U3B.7 gene. pExo-MS2×24 plasmid was derived from the plasmid pEV731 () by cloning 24×MS2 repeats into the ClaI–XhoI sites. pExo-MS2×24-Long was constructed by inserting a cassette coding for CFP with the peroxisome localization signal SKL, the internal ribosome entry site from encephalomyocarditis virus, and the thymidine kinase from HSV-1 into the unique XhoI site (). The clone pTRIP that lacked U3 sequences in the 3′ LTR originated from a similar vector.
Chromatin immunoprecipitation analysis was performed on Exo1 cells treated with GST-Tat essentially as previously described (; ). Primer set A corresponds to primer sets for Nuc1 in and map to the promoter-proximal region. Primer set B corresponds to a region of the vector proximal to the 3′ LTR (primer Bfw [5′-CATGGAGCAATCACAAGTAGC-3′] and primer Brv [5′-ATCTTGTCTTCGTTGGGAGTG-3′]). Primer set C maps 3′ of the 3′ LTR within the backbone of the vector (primer Cfw [5′-AGCATCTGGCTTACTGAAGCAG-3′] and primer Crv [5′-ATCGGTGATGTCGGCGATATAG-3′]). Primer set B13 corresponds to an unrelated genomic region, as described in . Quantification of immunoprecipitated material was performed by semiquantitative PCR and normalized for input DNA and for B13. The antibody against RNAPII was purchased from Santa Cruz Biotechnology, Inc. (N-20).
In situ hybridization was performed as previously described (). The formamide concentration was 50% in the hybridization and washing mixture except for the pA+ probe (hybridized and washed at 30% formamide) and the MS2 probe (10% formamide). The sequences of the probes were as follows (X stands for amino-allyl-T): HIV_I_E2 (5′-A X GGGTTGGGAGGTGGGTC X GAAACGATAATGGTGAAT X A); HIV_E1a (5′-A X GAGAGCTCCTCTGG X TTCCCTTTCGCTT X CAAGTCCCTGTTC X A); HIV_E1b (5′-A X TCTTGCCGTGCGCGCT X CAGCAAGCCGAGTCCTGCGT X A); HIV_intron1 (5′-A X TCTCGCACCCATC X CTCTCCTTCTAGCC X CCGCTAGTCAAAAT X A); HIV_intron2 (5′-A X AACTGCGAATCGT X CTAGCTCCCTGCT X GCCCATACTATATGTT X A); HIV_Exon2a (5′-A X GTGGCTAAGATC X ACAGCTGCCTTGTAAG X CATTGGTCTTAAA X A); HIV_Exon2b; (5′-A X ATCTTGTCTTCG X TGGGAGTGAATTAGCCC X TCCAGTCCCCC X A); HIV_pA+ (5′-A X TTTTTTTTTTTTTTTTTTTTTTTTTT X TGAAGCACTCAAGGCAAGC X A); MS2 (transcribed strand; 5′-A X GTCGACCTGCAGACA X GGGTGATCCTCA X GTTTTCTAGGCAAT X A); and MS2 (nontranscribed strand; 5′-A X AGTATTCCCGGG X TCATTAGATCC X AAGGTACCTAATTGC X A). The modified oligonucleotide probes for RNA FISH were synthesized by J-M. Escudier (Plateforme de synthèse d'Oligonucléotides modifiés de l'Interface Chimie Biologie de l'ITAV).
For quantitative measurements, 3D image stacks were collected and deconvolved with Hyugens (Bitplane AG). Background was removed, and the total light intensity at the transcription site was calculated and divided by the number of planes. The number of molecules was then computed from a calibration curve of the probes (), or, alternatively, transcription site signals were normalized to the ones of the cytoplasm. For each probe, 15–40 transcription sites were analyzed.
For Cy3/Cy5 quantification relative to the MS2-Cy5 probe (), cells were hybridized sequentially with the Cy3 and MS2-Cy5 probes (the MS2 probes hybridized at the lower stringency than the other probes). Images were taken in both channels, and the amount of light at the transcription site was calculated using the same mask for both colors. The Cy3/Cy5 ratios were then corrected for the specific activity of each probe by measuring the signals of an equimolar solution of the Cy3 and Cy5 probes under the microscope. The SDs were E1 (0.55), I (1.55), I-E2 (0.35), and E2 (0.16).
Immunofluorescence was performed as previously described (). Anti–Pol II (all isoforms) was used at the following dilutions: 8WG16 at 1:100 and anti-SC35 (Sigma-Aldrich) at 1:100. Fluorescent images of fixed cells were captured on a 100× NA 1.4 wide-field microscope (DMRA; Leica) equipped with a camera (CoolSNAP HQ; Roper Scientific) and was controlled by MetaMorph software (Universal Imaging Corp.). Stacks of wide-field images were deconvolved with Huygens (Bitplane AG) and mounted with Photoshop (Adobe).
For live cell imaging, cells were maintained at 37°C in appropriate medium (). Two microscopic setting were used to perform FRAP. For analysis of rapid recoveries (), we used a confocal microscope (Meta LSM510; Carl Zeiss MicroImaging, Inc.) with a 100× NA 1.4 objective. MS2-GFP at transcription sites or in the nucleoplasm was bleached at 488 nm in a circle of 1.5-μm diameter at full laser power and for one passage (bleaching time of 100 ms). Recoveries were measured at a high frame rate (one image for ≤160 ms) and for a short time (up to 10 s) using ≤1% of the power of the 488-nm laser line. Images were analyzed as previously described () by recording the fluorescence of the bleached region. Background was removed, intensities at each time point were corrected for bleaching by dividing them by the total cell fluorescence, and these values were finally normalized by dividing them with the fluorescent intensity before the bleach. In (E–I), 3, and 6 B (right), postbleach values were additionally set to zero.
When recovery of transcription sites had to be recorded for periods exceeding 10 s, we used an adapted microscopic setting. A microscope (TE200; Nikon) equipped for both confocal and wide-field imaging was used with a 100× NA 1.45 objective. Transcription sites were bleached with the confocal port using a circular region of 2.5-μm diameter (bleaching time of 1 s). Recoveries were then recorded in the wide-field port using MetaMorph (Universal Imaging Corp.) with an excitatory light of low intensity. Images were recorded with an EM-CCD camera (Cascade 512K; Roper Scientific). Stacks of nine images 0.5 μm apart were collected every 3 s (one stack took 0.5–1 s). For image analysis, fluorescence intensities were measured in a small parallelepiped (1 × 1 × 1.5 μm) placed at the most intense area of the transcription site. This operation was performed automatically by a macro that was created in ImageJ software (National Institutes of Health). This automatic tracking of transcription sites in 3D allowed us to correct cell movements and to minimize signal from diffusing MS2-GFP. When a nucleoplasmic region devoid of transcription site was bleached, the cube was placed at the center of the bleached region. The values obtained were then treated and normalized as in the previous paragraph except that the postbleach value was taken at the 5-s time point. Indeed, at this time, the diffusing pool of MS2-GFP had come back very close to its equilibrium, allowing us to neglect the diffusion of MS2-GFP in the analysis (). In all figures except , the postbleach value was set to zero to facilitate comparison of the curves.
Diffusion coefficients were measured by FRAP on transfected HeLa cells. For 21 μm/s of free GFP, we exploited the solution for free diffusion described by . For 15 μm/s MS2-GFPnls, we exploited the solution proposed for the reaction-diffusion model by .
For the MS2-GFP FRAP experiment, the curve of 10–20 cells were averaged and fitted with a straight line followed by an exponential: f(t) = A + α × t, t ≤ t; f(t) = A + α × t + (B − A − α × t) × (1 − exp[−α × (t − t)/(B −A − α × t)]), t > t. t is the time point at which the line converts into an exponential and corresponds to the length of the linear phase. A is the intensity at the zero time point, α is the slope of the initial linear part, and B is the immobile fraction. The minimization routine of the C++ GNU Scientific Library () was used for finding the minimum of the chi square. Chi square values were 0.06, 0.06, 0.04, 0.01, and 0.04 for Exo1, Exo2, ExoLong, Wt, and hC4, respectively.
Recovery curves of RNA polymerase at the transcription site were fitted with the binding-only simplification of the diffusion/reaction model developed by : f(t) = C × exp[−K × t], where K is the binding off rate and C is the ratio of on and off rates. This model was fitted with the R () nonlinear least squares function. This yielded a residency time of 333 s and a delay between two binding events of 660 s.
The software was written in Metal, a BASIC emulator for Mac computers (). It simulated a small population of RNA polymerase molecules using a stochastic model: during an elementary time period, polymerases in a given state had a certain probability to perform the next reaction in the pathway. We used the following circular scheme: inactive → first nucleotide transcribed → second nucleotide transcribed…last nucleotide transcribed → pre-mRNA is cleaved → mRNA is released and polymerase is inactive.
The kinetic rates of each step were determined from the experimental data: elongation by one nucleotide, 31.5 s (1.89 kb/min); cleavage, 0.018 s (54.7 s); and mRNA release, 0.11 s (8.8 s). To estimate initiation rates, we measured the number of RNA molecules by quantitative in situ hybridization using the MS2 probe. We found that a mean of 105 RNA molecules was present at the transcription site of Exo1 cells (±50). Because FRAP experiments estimated that the MS2-tagged RNAs stayed a mean of 128 s at this site, the initiation rate was deduced to 0.99 s for the entire array. The total number of polymerases simulated was calculated such that 105 molecules of MS2-tagged RNA were present at the transcription site at equilibrium. At the start of the simulation, the population of polymerase was distributed according to the equilibrium. Then, at each time point of the simulation and for each polymerase, a random draw determined whether the next step of the pathway occurred or not. The probabilities of success were calculated to match the rate constants, and the time resolution was small enough to ensure that a single event per polymerase could occur. A variable fluorescent value was attributed to each polymerase, which corresponded to the number of nucleotides transcribed in the MS2 repeat. For the simulation of FRAP, all fluorescence was set to zero, and recovery was plotted as a function of time. The values obtained were treated as experimental data (i.e., normalized between the pre- and postbleach values).
Curves normalized for their initiation rate (the initial slope of the curve) are shown in and indicate that 1.5 times more RNAs accumulated at the transcription site of the long reporter than for the short one. If initiation rates are constant, the amount (i) of fluorescent RNA at the transcription site is proportional to the time interval (t) between the moment polymerase reaches the MS2 site and the moment RNA leaves the transcription site. Thus, t = k × i, where k is a constant. However, t can be decomposed in two components: a variable one as a result of elongation (t1), which is proportional the length (l) of RNA that remains to be transcribed, and a constant time (t2), representing the time required for 3′-end processing and release. Thus, k × i = a × l + t2, where a is a constant. Because l is 2,180 for the short reporter and 4,580 for the long one, it follows that t1 is 1.2 times t2 for the short reporter and 2.5 times t2 for the long one.
The ratio between 5′ and 3′ probes depends on the relative amount of time that the polymerase spends on the transcription site once it reaches these hybridization sites. The same hypothesis was made as above by assuming that the time that a nascent RNA remains at the transcription site is decomposed into a variable part related to elongation and a constant part related to all other processes. In this case, the intensity (i) at a given hybridization site can be written as k × i = a × l + t2, where l is the length separating the hybridization site from the polyA site and k and a are constants. Assuming an elongation rate of 2.03 kb/min and a 3′-end processing time of 63.5 s (in Exo1 cells), one can calculate the values given in .
The FRAP curves yield a total time for 3′-end formation and release of 67.3 s on average. 3′-end formation can be decomposed in cleavage, polyadenylation, and transcript release. This latter rate can be estimated from the amount of polyadenylated mRNA present at the transcription site of Exo1 cells. The polyadenylated species represents 11% of the exon 2 signal, meaning that when a polymerase reaches exon 2, the nascent RNA then spends 89% of its time completing transcription and 3′-end processing and 11% as a polyadenylated species. Because the exon 2 probe is 560 nucleotides away from the polyA site, it should take polymerases 16.6 s to go from there to the polyA site (at 2.03 kb/min; value from Exo1 cells) and then a further 63.5 s to process the RNA (value from Exo1 cells). Thus, polyadenylated RNAs should remain at transcription sites 11% of 80.1 s (i.e., 8.8 s, yielding a rate of 0.11 s). Because a total time of 63.5 s is required for 3′ processing, the time required for cleavage/polyadenylation was then estimated at 54.7 s (0.018 s).
Fig. S1 shows that transcription of HIV-1 mRNA is induced by Tat and PMA/ionomycin in Exo1 cells. Fig. S2 shows that MS2-GFP is stably bound to its target RNA in vivo. Fig. S3 shows that elongation can be modeled with a straight line. Fig. S4 shows quantification of exon 1 versus exon 2 at the HIV-1 transcription site. Table S1 provides RNA binding properties of the coat protein of phage MS2. Online supplemental material is available at . |
Skeletal muscle tissue is characterized by a very slow turnover. Turnover increases, however, upon certain physiological stimuli or in pathological conditions, such as primary myopathies, leading to an extensive repair process aimed at preventing the loss of muscle mass.
The initial phase of muscle repair is characterized by necrosis of the damaged tissue and activation of an inflammatory response (). Local cues, produced by growth factors and inflammatory cytokines released by infiltrating cells, lead to the activation of quiescent myogenic cells, the satellite cells located beneath the basal lamina of muscle fibers that start to proliferate, differentiate, and fuse, leading to new myofiber formation and reconstitution of a functional contractile apparatus.
Muscle satellite cell activation resembles embryonic myogenesis in several ways, including the de novo induction of the myogenic regulatory factors (MRF). Genetic studies have defined a temporal and functional hierarchy among the myogenic basic helix-loop-helix transcription factors during somitic myogenesis, the first of which are Myf5 and MyoD (; ). Quiescent satellite cells express another important transcription factor, Pax7, which is involved in myogenic specification (; ). Upon injury, satellite cells are activated and start proliferating and expressing Myf5 and MyoD, whereas Pax7 expression is progressively reduced. At this stage they are often referred to as myogenic precursor cells. Subsequently, expression of myogenin and MRF4 (MRF members) is up-regulated in cells beginning their terminal differentiation program, which is followed by permanent exit from the cell cycle, activation of muscle-specific proteins such as sarcomeric myosin, and fusion with damaged muscle fibers or with themselves, to produce new fibers that replace the dead ones. A fraction of activated cells down-regulates expression of Myf5/MyoD and returns to a quiescent state to maintain a more or less constant pool of satellite cells. A delicate balance between cell proliferation and exit from cell cycle, differentiation, and fusion is required for the correct muscle regeneration to occur, and many proteins have been found to play a crucial role in these processes, including insulin-like growth factor, myostatin and follistatin (; ; ; ), intracellular mediators like p66 (), and transcription factors other than MRF, such as E2F1 and myocyte nuclear factor (; ; ).
Necdin is a member of the melanoma antigen-encoding gene family (), a large family of proteins initially isolated from melanomas. The predominant feature of these proteins is a large central region termed the melanoma antigen-encoding homology domain. Deficiency of necdin in transgenic mice causes neonatal respiratory distress that is usually fatal, and surviving mice exhibit behavioral and cognitive disruptions that bear some resemblance to the phenotype observed in humans affected by the Prader Willi syndrome (; ). Studies in neuroblastoma cell lines and dorsal root ganglia cultures have suggested that necdin may act as a growth suppressor, facilitating cell cycle exit and differentiation while inhibiting apoptosis in neurons (; ; ; ). The precise cellular mechanism leading to cell cycle arrest is still unclear, although evidence points to interaction with proteins involved in cell cycle progression such as p53 and E2F (; ; ; ). We have also shown that necdin cooperates with the homeobox transcription factor Msx2 during smooth muscle differentiation of mesoangioblast cells (). Whether necdin acts as a transcriptional cofactor or as a direct transcriptional repressor or activator in these different scenarios is still unknown.
Necdin is also expressed in developing skeletal muscle, which suggests that it plays a relevant role in this tissue (; ; ). Its function in skeletal muscle and myogenesis, however, has not been investigated in detail.
Here, we provide the first evidence that necdin plays an important role in skeletal muscle growth and repair in vivo, as shown by using loss- and gain-of-function experiments in necdin knockout mice and mice overexpressing necdin specifically in skeletal muscle. We also demonstrate that necdin activity is mediated through stimulation of myoblast differentiation and protection from apoptotic cell death.
To study the expression of necdin in postnatal skeletal muscle we performed immunofluorescence on cryostat sections of tibialis anterior (TA) muscle from postnatal day (P) 5 and P10 pups () and 2-mo-old mice (not depicted). In P5 () and P10 () muscle sections, we observed colocalization of necdin with Myf5-positive cells, indicating that these cells are myogenic precursor cells. No other cells in the muscle appeared to express necdin. We did not detect staining for necdin in 2-mo-old muscle (not depicted). Using Western blot, we also analyzed the profile of necdin expression in TA muscle from mice at different ages. As indicated in , necdin expression was high in muscle from 5–10-d-old pups, it decreased in pups from 15 d to 1 mo old, and it was undetectable at 2–3 mo, in agreement with the immunofluorescence data. This suggests that necdin is restricted to active myogenic progenitor cells.
We isolated myoblasts from newborn mice and studied the temporal expression profile during differentiation. Proliferating myoblasts expressed high levels of necdin, which decreased as differentiation proceeded, along with increased expression of sarcomeric myosin and decreased expression of the proliferation marker proliferating cell nuclear antigen (PCNA; ).
We also isolated single muscle fibers from 2-mo-old mice and analyzed the expression of necdin, together with that of Pax7, a marker of quiescent satellite cells and early myogenic progenitors, and of MyoD, a marker of activated satellite cells and myogenic precursor cells (; ; ; ). Expression of necdin and these two markers was analyzed from cells on a single fiber or from cells that migrated away from the fiber, which were subsequently maintained in proliferating or differentiating medium. We found that necdin was coexpressed with Pax7 in very few cells (; and Fig. S1, A–D; available at ). In most cases it was coexpressed with MyoD (). Necdin localization was in the cytoplasm but staining was also observed in some nuclei () and in both cells on the fibers and cells leaving the fibers. In differentiated myotubes, the level of necdin expression was lower and predominantly in the cytoplasm ().
We first generated a transgenic mouse overexpressing necdin in skeletal muscle. We cloned the ORF of , downstream of the myosin light chain (MyLC) 1F promoter and upstream of the MyLC enhancer (; ). The linearized vector was used to generate transgenic lines by pronuclear injection. Two founders, MlcNec2 and MlcNec9, screened by PCR and selected on the basis of transgene transmission to the progeny, were maintained as two independent lines on a CBA/C57BL6 background (). We tested the expression of the transgene on tissues and myoblasts by RT-PCR using specific primers () and the overexpression of the protein by immunofluorescence and Western blot (). The transgenic animals showed increased expression of necdin in isolated adult myogenic progenitors, and the protein was also found in different adult muscles (diaphragm, soleus, and TA) where levels of expression in the wild type (wt) were undetectable (). Of notice, the MyLC1F promoter, derived from the locus of a fast myosin-type gene, was also able to drive necdin expression in the soleus, a muscle with an abundance of slow fibers. The transgene was expressed in both proliferating myoblasts and differentiating myotubes, indicating that the promoter also allows expression of this transgene in cells not terminally differentiated (). In addition, it was possible to observe expression of necdin in nuclei and also faintly in the fiber cytoplasm on sections of adult muscle (; and Fig. S2 E, available at ). In some cases, transgenic necdin colocalized with rare Myf5-positive activated satellite cells (). As in the case of wt, expression of necdin in myoblasts was both in the nucleus and cytoplasm (), and we could detect overexpression of necdin in the cytoplasm and nucleus of cells on single muscle fiber coexpressing MyoD (). The tissue specificity of the promoter and the transgene expression was verified by testing the expression of necdin in other tissues and organs where necdin is expressed at very low, barely detectable levels (lung, kidney, and heart; ). The transgenic mice from both lines showed no overt differences in terms of weight, fertility, behavior, and life span with respect to the littermate controls (not depicted).
We decided to compare the results in these mice to the results of necdin loss-of-function transgenic mice, i.e., the Ndn mice () obtained from Ndn intercrosses. Necdin has been described as an imprinted gene, paternally expressed only, but the lack of expression from the maternal chromosome in myogenic cells has never been investigated. We performed analysis of necdin RNA and protein expression in brain and myoblasts from Ndn, Ndn, Ndn, and wt. We observed that heterozygous mice carrying the mutated gene on the paternal chromosome showed expression of necdin neither in brain nor in myoblasts, indicating that necdin also undergoes complete maternal silencing in the myogenic lineage ().
To determine whether necdin plays a functional role in skeletal muscle in vivo, we examined myofiber size in wt, Ndn, and MlcNec2 muscles from P10 pups () and 2-mo-old mice (). The TA muscles were collected from adult wt, Ndn, and MlcNec2 mice. In the adult we also collected the soleus. Sections were stained with hematoxylin and eosin (H&E; ). As confirmed by cross-sectional area (XSA) analyses (), Ndn myofibers were significantly smaller than wt myofibers in both P10 and adult mice, whereas MlcNec2 myofibers were consistently larger, particularly in the adult (). We also determined the mean number of myofibers in TA and soleus muscle in the adult. We observed that MlcNec2 muscle contained a similar number of myofibers, whereas Ndn showed a higher number of fibers (). Finally, we measured the maximum XSA of the TA and soleus muscle; the results indicate that MlcNec2 mice have an increased XSA (). The effects of loss and gain of function of necdin on morphological parameters of muscle was greater in the TA with respect to the soleus, reflecting the relative amount of necdin expression in the two muscles. All these data suggest that necdin is involved in myofiber growth and maintenance.
In wt mice, new myotube formation in TA muscle was observed after 7 d (), even if inflammatory infiltrates were still present. In contrast, MlcNec2 mice displayed a more advanced rate of regeneration and formation of newly regenerated fibers by 3 d after the CTX injection (). The accelerated regeneration in MlcNec2 mice was accompanied by high expression of neonatal myosin, a marker of muscle regeneration in the adult (). Conversely, Ndn mice showed a marked delay in the regeneration process with increased inflammatory infiltrates, decreased numbers of new regenerating, centronucleated fibers (), and no expression of neonatal myosin () with respect to the wt muscle (). 14 d after the crush, the overall architecture was completely restored in the wt, MlcNec2, and Ndn muscle, and they were undistinguishable from uninjured muscle except for centrally located nuclei ().
To further investigate whether delayed regeneration in Ndn mice may indicate a major defect in this process, we caused more severe damage by injecting 100 μM CTX, a higher concentration than before. We analyzed the mice after 14 d and observed that in wt animals, regeneration was occurring normally, even if inflammatory infiltrates were still present (). In contrast, MlcNec2 mice showed a complete regeneration of the muscle (). A dramatic effect was seen on the Ndn animal. At dissection, the injured TA already appeared atrophic (not depicted) and no regenerating fibers were seen (). The muscle was mostly composed of necrotic fibers in the presence of massive inflammatory infiltrates. As expected, no expression of neonatal myosin could be observed even by RT-PCR (). We also extended the analysis at 30 d after severe damage and observed similar results (Fig. S2 and not depicted).
Myoblasts from newborn transgenic MlcNec2 mice, Ndn, and wt littermates were isolated as described in and seeded at clonal density. Cells were either maintained in proliferation medium or switched to differentiation medium. At different time points we either obtained cell lysates for Western analysis or we fixed the cells and performed immunofluorescence using the MF20 antibody that recognizes the sarcomeric myosin heavy chain (MyHC). Myoblasts cultured from MlcNec2 mice initiated differentiation more rapidly and formed myotubes with an increased fusion index with respect to wt controls (). Myoblasts cultured from Ndn mice consistently showed a reduced differentiation and fusion index ().
We also investigated the effect of necdin overexpression in myoblasts by infecting wt newborn myoblasts with a lentiviral vector expressing necdin under the cytomegalovirus constitutive promoter (pLentiNecdinIRESGFP) or the empty vector (pLentiIRESGFP). Cells were infected 24 h after plating at clonal density, and then maintained in proliferation medium for 72 h. We did not observe any effect on myoblast proliferation or number and size of the clones generated (not depicted), but necdin-overexpressing myoblasts underwent spontaneous differentiation, forming myotubes with an increased fusion index with respect to controls (Fig. S3, A–F, available at ). A similar effect on differentiation was also seen in C2C12 cells after transfection with a necdin-expressing vector (Fig. S3, G–L).
Increased or reduced differentiation of myoblasts may depend on changed expression of MRF and/or decreased cell death. We decided to investigate which of these parameters was altered by first analyzing the protein levels of MyoD, myogenin, and sarcomeric MyHC (). The increased and decreased differentiation levels we observed in the myoblast cultures from MlcNec2 and Ndn mice, respectively, were accompanied by increased and decreased expression of MyHC, respectively. Interestingly, MyoD was more highly expressed in Ndn satellite cells, possibly reflecting an accumulation of myoblasts, whereas myogenin expression at the beginning of differentiation in the MlcNec2 cultures was significantly increased in comparison with wt and even more so in comparison with Ndn.
Necdin has been described as acting both as a coactivator and a corepressor of transcription mediated by other transcription factors (; ; ). To determine whether the modified expression of myogenic markers was caused by a direct transcriptional activation of these genes by necdin, we tested the ability of this protein to transactivate the myogenin promoter–driving expression of a chloramphenicol acetyltransferase (CAT) reporter gene by cotransfecting a necdin expression construct (pIRES-GFP-necdin; ) together with the reporter construct pMyo-1565CAT () in 10T1/2 or C2C12 cells. We observed that in C2C12, where the myogenin promoter already had a high basal activity caused by the presence of endogenous myogenic factors such as MyoD, necdin alone was able to activate four- to fivefold the expression of the CAT reporter (). In contrast, necdin was unable to activate the reporter gene when transfected alone in 10T1/2, but when necdin was cotransfected with a MyoD-expressing construct (), it increased CAT activity fivefold in comparison with MyoD alone. These results indicate that necdin is involved in the transcriptional regulation of myogenic genes by cooperating with other MRFs.
We decided to investigate the mechanism of action of necdin by analyzing its ability to bind and/or interact with the promoter. C2C12 cells were transfected with the necdin overexpressing construct pCMV3B-Myc-necdin, and chromatin immunoprecipitation was performed using anti-necdin, anti-Myc, and anti-MyoD–specific antibodies. As expected, we found that MyoD was able to bind to the promoter (). Interestingly, necdin was also equally able to bind this DNA sequence, which was precipitated both by anti-necdin or anti-Myc antibodies ().
We then wanted to investigate whether necdin was able to directly interact with other proteins known to bind the promoter, e.g., MyoD, myogenin, and Mef2A. We performed coimmunoprecipitation experiments on C2C12 cells transfected with the same necdin-Myc–overexpressing construct. As shown in. Fig. S4 A (available at ), we did not detect any direct interaction between necdin and MyoD, myogenin, or Mef2A in these conditions.
In neural cells, necdin protects against apoptosis, even if the mechanism is not completely clear (; ; ; ). Based on this observation in neural cells, we studied whether necdin protected myoblasts from apoptosis, and whether such protection contributed to accelerated regeneration in vivo in the MlcNec2 mice and delayed regeneration in the Ndn animals.
We used TUNEL assay to measure the level of cell death in CTX-injured muscles (). We observed an increased TUNEL staining in Ndn TA muscle with respect to wt (), whereas MlcNec2 showed a decreased number of TUNEL-positive cells ().
background (Fig. S4, B–E; and not depicted).
We then investigated whether overexpression of necdin had prosurvival effects on myoblasts in vitro. To this end, wt, Ndn, and MlcNec2 myoblasts were exposed for a further 24 h to 20 μM of the cytotoxic stimuli staurosporine or 100 μM of the reactive oxygen species generating agent HO, or exposed for 1 min to UV radiation. Cell death was determined 24 h later by measuring both Annexin V staining of phosphatidylserine exposed on the outer leaflet of the plasma membrane and propidium iodide incorporation. Results obtained are summarized in , and representative dot plot analyses showing the results using HO as the death-inducing stimulus are shown in . Cell death induction by all stimuli was decreased in MlcNec2 cells with respect to wt. On the contrary, cell death was increased in all conditions in the Ndn cells, including basal cell death. To investigate which cell death pathway necdin was involved in, we measured the levels of expression of the activated effector caspase 3 and the initiating activated caspase 9, mainly involved in mitochondrial-mediated cell death (). We observed that Ndn myoblasts challenged with either staurosporine or HO display a higher level of activated caspase 9 and 3, with respect to wt cells, whereas activated caspase 3 and 9 were almost undetectable in MlcNec2 myoblasts (). We observed a similar effect when C2C12 myoblasts transfected with necdin where treated with the same apoptotic stimuli (Fig. S4 F).
Stem cells are capable of both self-renewal and generating progenitors that finally undergo terminal differentiation to ensure the regeneration of tissues whose differentiated cells cannot divide. In tissues with a high cell turnover such as blood and skin, a constant flow of precursors for terminal differentiation is needed. In contrast, adult skeletal muscle needs only new myofibers for repair. In accordance, satellite cells are normally quiescent in adult muscle and activated only in response to specific signals provoked by physiological or pathological stimuli. The activation of satellite cells and their subsequent progression through cell cycle, differentiation, and fusion need to be finely regulated and tuned to ensure correct growth and maintenance of the fibers; therefore, understanding the molecular pathways involved in myoblast differentiation and survival is crucial to the development of treatments for impaired muscle growth associated with age, disease, and atrophy.
Here, we demonstrate that necdin is an important player in skeletal muscle differentiation and maintenance. We found that necdin acts on two different pathways: it acts on myoblast differentiation through direct transcriptional regulation of myogenin, in cooperation with MyoD; and it protects myoblasts from cell death.
The following reagents were purchased as indicated: anti-necdin pAb from Upstate Biotechnology; anti–glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mAb from Biogenesis; anti-MyoD mAb from DakoCytomation; anti-PCNA mAb, anti-MyoD, anti-Mef2A, and anti-Myf5 pAbs from Santa Cruz Biotechnology, Inc.; antisarcomeric myosin MF20, antimyogenin, and anti-Pax7 mAbs from Developmental Studies Hybridoma Bank; antidevelopmental MyHC (NCL-MyHCd) from Novocastra; antilaminin and anti-Myc from Sigma-Aldrich; anti-GFP from Invitrogen; and antiactivated caspase 3 pAb and anti–caspase 9 mAb (recognizing activated and nonactivated forms) from Cell Signaling. In immunofluorescence analysis, primary antibodies were detected by appropriate Alexa-conjugated (Alexa 489 or Alexa 594) secondary antibodies (Invitrogen). In immunoblot analysis, primary antibodies were detected by chemiluminescence with appropriate horseradish peroxidase–conjugated secondary antibodies (Bio-Rad Laboratories). Cell culture media and sera were purchased from Cambrex. Bicinchoninic acid–based assay was purchased from Perbio. The Hoechst dye, CTX, and other chemicals were purchased from Sigma-Aldrich.
Transgenic mice were generated by injecting the purified linearized construct represented in (as in ) into the pronuclei of fertilized oocytes (CBA/C57BL6 F1).
Ndn mice (provided by F. Muscatelli) were genotyped by PCR analysis of tail DNA using the following primers: Ndn-KO-S: 5′-TCTCATGCTTGAACTGCA-3′; Ndn-KO-AS: 5′-CAGGTAATTCTGCTGGAC-3′; NECF1: 5′-GTCCTGCTCTGATCCGAAGG-3′; and TgNec reverse: 5′-GGT- CAACATCTTCTATCCGTTC-3′.
Mice were also crossed with CBA mice and maintained on either a pure C57BL6 or a CBA/C57BL6 background.
mice (provided by S. Tajbakhsh,Pasteur Institute, Paris, France).
Isolation of primary myoblasts. Primary myoblasts from newborn mice of the different strains were isolated as described in and plated at clonal density. Cells were grown either in proliferation medium (DME supplemented with 20% FBS, 3% chick embryo, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μg/ml gentamycin) or in differentiation medium (DME supplemented with 2% horse serum, 100 U/ml penicillin, and 100 μg/ml streptomycin). We observed that already after 24 h in proliferation medium some cells started to express MyLC and MyHC.
TA, diaphragm, or soleus muscles were dissected from adult mice (2–3 mo) and frozen in liquid N–cooled isopentane. 8-μm serial muscle sections were either stained with H&E or Azan Mallory or immunostained as described in .
To evaluate the ability of the satellite cell to participate in the regeneration process in vivo, injury was performed on the TA of 3-mo-old MlCNec2, Ndn, and wt mice by injecting 25 μl of 10 μM or 100 μM CTX (four animals per group). Mice were killed at 1, 3, 7, 14, or 30 d after the CTX injection, and the muscle was collected and either sectioned for histological analysis or subject to protein or RNA extraction.
Morphometric analyses were performed on sections collected from similar regions of each TA muscle and from the belly of each soleus muscle. Two images were captured from each section and Image 1.63 (Scion Corporation) was used to determine the XSA of 700–1,000 myofibers per section.
C2C12 and 10T1/2 cell lines were obtained from American Type Culture Collection. C2C12 and 10T1/2 cells were cultured in DME supplemented with 15% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin (proliferation medium). C2C12 cells were differentiated in DME supplemented with 2% horse serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (differentiation medium) as described in .
10T1/2 and C2C12 cells were transiently transfected with Lipofectamine Plus reagent (Invitrogen). In brief, 10 cells were seeded in triplicate on 35-mm dishes, and the next day transfected with 0.5 μg of the reporter construct pMyo-1565CAT () and 0.5 μg pCMV-βGal (CLONTECH Laboratories, Inc.), alone or together with various combinations of plasmids mixed as follows: 1 μg pIRESNecdin; 1 μg pEMSV-Myf3 (Myod1); and 1 μg pIRESNecdin + 1 μg pEMSV-Myf3 (Myod1). Cells were grown for 2 d in DME, 10% FCS (or in the case of C2C12 cells, in differentiation medium, DME, and 2% horse serum); proteins were then extracted by repeated freezing–thawing. The levels of CAT protein were determined using an enzymatic immunoassay (CAT-ELISA; Roche Diagnostics). Cell extracts were normalized for protein concentration and CAT expression was further normalized to β-galactosidase activity.
Immunofluorescence on the cell cultures and cryosections was performed according to , using antibodies specific for necdin, sarcomeric myosin MF20, MyoD, MyHCd, laminin, Pax7, Myf5, and GFP. For fluorescent detection, we used appropriate secondary antibodies conjugated with either Alexa 488 (green) or Alexa 594 (red; Invitrogen).
Satellite cells from the different genotypes were incubated with or without 100 μM HO or 5 μM staurosporine for 24 h, or exposed to UV light for 1 min. Cells were detached and stained with fluorescein isothiocyanate–Annexin V and propidium iodide, according to the manufacturer's instructions (Bender MedSystems), and analyzed by flow cytometry as described in . Cells that showed single staining for Annexin V or double staining for Annexin V and propidium iodide were considered dead cells. Alternatively, cells were lysed 3 h after treatments and analyzed for expression of activated caspases by Western blot.
Cells were homogenized in 50 mM Tris/HCl, pH 7.4, 1 mM EGTA, 1 mM EDTA, 1% Triton X-100, and protease inhibitor cocktail (Complete; Roche Diagnostics) and centrifuged at 1,000 for 20 min at 20°C to discard cellular debris.
Muscle tissues were dissected and homogenized in 100 mM NaHCO, 1 mM EDTA, 2% sodium dodecyl sulfate, and protease inhibitor cocktail and centrifuged at 1,000 for 10 min at 4°C to discard cellular debris. Sample preparation and Western blot analyses were performed as described in .
6 × 10 of differentiated C2C12 cells, previously transfected with Myc-tagged necdin, were treated with 1% formaldehyde and sonicated in 600 μl of protease inhibitor–containing buffer. The chromatin was then immunoprecipitated with, respectively, mouse anti-MyoD (Santa Cruz Biotechnology, Inc.), mouse anti-Myc (Sigma-Aldrich), and rabbit anti-necdin (Upstate Biotechnology), as in . The input DNA was an aliquot of the supernatant from centrifuged sonicate (DNA size range: ∼200–1,000 bp), and the preimmune chromatin was immunoprecipitated with normal rabbit and mouse IgG (Santa Cruz Biotechnology, Inc.). After an overnight incubation with the antibodies, 10 μl of 20 mg/ml tRNA, 20 μl of 10 mg/ml salmon sperm DNA, and 20 μl of protein G–agarose beads were added to the 1-ml samples and incubated for 3 h at 4°C with constant agitation. The purified immunoprecipitated DNA was dissolved in 60 μl of 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA. PCRs were performed with GoTaq (Promega) for 40 cycles. The primers used were myogenin forward (5′-GAATCACATGTAATCCACTGGA-3′), myogenin reverse (5′-ACGCCAACTGCTGGGTGCCA-3′), β-actin forward (5′-GCTTCTTTGCAGCTCCTTCGTTG-3′), and β-actin reverse (5′-TTTGCACATGCCGGAGCCGTTGT-3′).
For detection of necdin interactors, C2C12 cells were transfected with pCMV3B-Myc-necdin. Cells were harvested after 48 h in differentiating medium and lysed in RIPA buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 0.5% deoxycholate sodium, 1% NP-40, and 1× protease inhibitor mixture (Roche). The 2 mg of lysates were incubated overnight at 4°C with 5 μg of polyclonal antibody against necdin (Upstate Biotechnology) or rabbit anti-IgG. The complexes were pelleted with protein G–Sepharose (Invitrogen), separated by 10% SDS-PAGE, and transferred to nitrocellulose membrane by electroblotting. The membranes were incubated with antibodies against MyoD, Mef2A, myogenin, or Myc, and washed and incubated with horseradish peroxidase–conjugated goat anti–rabbit and anti–mouse IgGs (Bio-Rad Laboratories). Proteins were visualized by the chemiluminescence method (ECL Western blot detection reagent; GE Healthcare).
1 μg RNA, collected from cells or tissues using RNeasy mini (or micro) kit (QIAGEN), was converted into double stranded cDNA using the cDNA synthesis kit ThermoScript RT-PCR system (Invitrogen), according to the manufacturer's instructions. The primers used were the following: GAPDH forward: TGAAGGTCGGTGTGAACGGATTTG; GAPDH reverse: CATGTGGGCCATGAGGTCCACCAC; dMyHC forward: ACTGAGGAAGACCGCAAGAATG; dMyHC reverse: AAGTAAACCCAGAGAGGCAAGTGACC; MyLC1 forward: GATCACCTTAAGTCAGGT; MyLC1 reverse: gcaacgcttctacctctt; necdin NEC F2: 5′-CACTGATAGTTTCTGACCCATAC-3′; NEC R1: 5′-GCCAGTTGAAGTCATATGGAG-3′; necdin transgene TgNec reverse: 5′-GGTCAACATCTTCTATCCGTTC-3′; and TgMex2 forward: 5′-GTGTCAAGGTTCTATTAGGCACTA-3′.
Fluorescent and phase-contrast images were taken on microscopes (Eclipse E600 [Nikon]; Plan Fuor lenses: 10×/0.33, 20×/0.50, and 40×/0.75; [Leica AF6000]; lenses: HCX PL FLUOTAR L40×/0.60). Images were acquired using digital cameras (DXM1200 [Nikon]; DFC 350 FX [Leica]) and acquisition software (ACT-1 [Nikon]; AF6000 [Leica]). The imaging medium was PBS buffer; images were taken at room temperature. Images were assembled in panels using Photoshop 7.0 (Adobe). Images showing double or triple fluorescence were separately acquired using the appropriate filters, and the different layers were merged with Photoshop 7.0.
The results are expressed as means ± SEM; represents the number of individual experiments. Statistical analysis was performed using the analysis of variance test. Asterisks in the figures refer to statistical probabilities versus wt controls, respectively. Statistical probability values of <0.05 were considered significant.
Fig. S1 shows coexpression of necdin with Pax7 on single fiber, expression of necdin on a TA section of adult MlcNec2 mice, and expression of necdin and Myf5 on a TA section of Ndn P5 mice. Fig. S2 shows histology of TA of wt, MlcNec2, and Ndn, 30 d after injection of 100 μM CTX. Fig. S3 shows the differentiation of primary myoblasts or C2C12 cells after overexpression of necdin. Fig.
, Ndn mice after CTX injection; and expression of activated caspase 3 and 9 in C2C12 cells overexpressing necdin. Online supplemental material is available at . |
The primary cilium (PC) is a microtubule-based structure that protrudes from the surface of most vertebrate cells. It generally comprises a membrane-bound 9 + 0 ciliary axoneme, which consists of nine outer doublet microtubules but lacks both the central microtubule pair and dynein arms. Thus, with few exceptions, PC are nonmotile and instead function as sensory organelles (; ; ). They play important roles during development, particularly with regard to the establishment of left–right asymmetry, as well as later in life when they are required for the processing of mechanical or chemical signals in many organs (; ). For instance, in kidney epithelial cells, PC sense fluids flow within the lumen of the nephron, which is critical for normal epithelial development and function. Proteins localizing to the ciliary membrane, known as polycystins, play an important role in mediating this mechanosensory function, and mutations in the corresponding genes cause polycystic kidney disease (). Similarly, retinal degeneration can be caused by dysfunction of the connecting cilium, a highly specialized PC connecting the inner and outer segments in vertebrate photoreceptors (; ). Moreover, recent studies implicate PC in various signal transduction pathways, including sonic hedgehog, platelet-derived growth factor receptor α, and Wnt signaling (; ). Ciliary defects have also been causally linked to several pleiotropic disorders, including Bardet-Biedl syndrome (BBS), Alstrom syndrome (ALMS), oral-facial-digital syndrome type I, and nephronophthisis (; ; ).
The assembly of the PC requires a basal body, which in turn is derived from one of the two centrioles that constitute the centrosome. During ciliogenesis, this basal body is positioned close to the plasma membrane and ciliary microtubules elongate from its distal end. Ciliogenesis requires the assembly of multiple soluble and membranous protein complexes. In particular, the so-called intraflagellar transport (IFT) system is then responsible for moving cargo (IFT particles) to and from the tip of the growing axoneme. IFT, first described in the algae (), is now known to be mediated by the association of IFT particles with kinesin II and dynein microtubule–based motors for antero- and retrograde movement, respectively (; ). The signaling networks that control PC function during cell cycle progression remain to be elucidated, but several studies concur to identify a key role for the von Hippel-Lindau tumor suppressor in PC formation (; ; ). Furthermore, Aurora A kinase has recently been implicated in PC resorption ().
In this study, we have sought to identify centrosomal proteins (Ceps) that are required for ciliogenesis. Taking advantage of the fact that PC formation can be induced in cultured cells by serum starvation (; ), we depleted individual centrosomal proteins by siRNA and examined the consequences on subsequent PC formation. This siRNA screen identified several proteins that affected PC formation, albeit to different degrees. A very strong effect was observed upon depletion of Cep164, a protein that had not previously been studied. Our characterization of Cep164 leads to conclude that this protein is not only required for PC formation but also constitutes an excellent marker for distal appendages on mature centrioles or basal bodies.
To search for proteins involved in PC formation, an siRNA screen focusing on centrosomal proteins () was performed. After the depletion of individual proteins from retinal pigment epithelial (hTERT-RPE1) cells, PC formation was induced by serum starvation () and monitored by staining with antibodies against acetylated tubulin (). Depletion efficiency was assessed by quantitative real-time PCR (qRT-PCR) and, whenever possible, immunofluorescence (IF) microscopy and/or Western blot analysis (Table S1, available at ). Because depletion of outer dense fiber (Odf) 2 and pericentrin had previously been found to impair PC formation (; ), these proteins were reexamined to provide points of reference for classifying phenotypes. Of 41 proteins analyzed, depletion of 25 proteins produced clear phenotypes, suggesting that these proteins might be involved in PC formation and/or maintenance (). Depletion of 23 proteins (including Odf2 and pericentrin) considerably reduced the proportion of cells that assembled PCs in response to serum starvation, from 90–95% in controls to 25–60% in depleted cells (; and not depicted). Furthermore, the depletion of Cep57 or ALMS1 led to the formation of morphologically abnormal, stunted cilia without considerably reducing the efficiency of ciliogenesis (), whereas the depletion of Cep131 and Cep152 resulted in both a clear reduction in ciliogenesis and morphological aberrations in those cilia that did form (). In contrast, no qualitative or quantitative effects on PC formation were observed upon siRNA-mediated depletion of 16 other proteins analyzed, including fibroblast growth factor receptor 1 oncogene partner, rootletin, and centrosomal Nek2-associated protein 1 (C-Nap1; , gray; and not depicted).
To exclude the possibility that depletion of centrosomal proteins might have influenced cell cycle profiles, FACS analyses were performed 48 or 72 h after siRNA treatments (unpublished data). These analyses provided no evidence for major deviations from controls, except for an increase in the proportion of G2/M cells upon depletion of Plk4 and other proteins implicated in centriole duplication, as expected (). Although the depletion of several centrosomal proteins has recently been reported to induce a p53-dependent G1 arrest (), we found no evidence to indicate that depletion of any of the centrosomal proteins analyzed here produced a G1 arrest. Interestingly, in the course of the siRNA screen we occasionally observed splitting of centrosomes, concomitant with the impairment of PC formation (, right [inset]; note that centrosome splitting is defined here as the separation of parental centrioles by >2 μm). However, this splitting was only seen in serum-deprived cells and not in cycling cells. Furthermore, there was no strict correlation between centrosome splitting and PC formation. The depletion of BBS4 did not induce splitting although this protein was required for ciliogenesis, whereas the depletion of proteins implicated in centrosome cohesion (rootletin and C-Nap1) induced splitting without detectably impairing PC formation. Thus, the functional relationship, if any, between cell cycle–dependent centrosome splitting and the inhibition of PC formation is not presently clear.
This screen points to the involvement of many proteins in the biogenesis and/or maintenance of the PC, which is consistent with the known complexity of this structure (; ; ). With regard to several proteins previously implicated in centriole duplication, we note that depletion of hSas-6 considerably reduced PC formation, in agreement with a recent independent paper (). In contrast, the depletion of other proteins required for centriole biogenesis, notably Plk4, Cep135, or CPAP (), produced little adverse effects on PC formation, presumably because the limited duration of the present siRNA experiments prevented a marked reduction in centriole numbers. Remarkably, depletion of only three proteins almost completely prevented PC formation in our screen. The three proteins identified here as essential for ciliogenesis were pericentrin, Cep290, and Cep164. A key role of pericentrin in PC formation has previously been reported () and a requirement for Cep290 in ciliogenesis falls in line with the implication of this protein in cilia-related diseases, including Joubert syndrome, Senior-Loken syndrome, and Meckel syndrome (; ; ). In contrast, Cep164, the focus of this paper, has not previously been analyzed.
The availability of anti-Cep164 antibodies () made it possible to monitor the depletion of Cep164 in siRNA experiments. Focusing on cells that were effectively depleted of Cep164, we found that only 3.6% of cells formed a PC compared with 95% of GL2-treated control cells, clearly demonstrating that Cep164 is indispensable for PC formation (). Cep164 was originally identified in a proteomic inventory of human centrosomes (). The corresponding gene maps to human chromosome 11q23.3 (GenBank/EMBL/DDBJ accession nos. and ) and database analyses suggest the existence of potential isoforms. The Cep164 protein studied here is made up of 1,460 residues, resulting in a predicted molecular mass of 164 kD. As predicted by the SMART protein domain database (), the protein comprises a putative N-terminal WW domain (57–89) as well as three coiled-coil regions (589–810, 836–1047, and 1054–1200; ). Orthologues of Cep164 clearly exist in other vertebrates (e.g., mouse [XP_929307] and zebrafish [XP_697015]) and potentially also in (NP_611787; ). No obvious structural homologues could be identified in , , or , but this does not exclude the existence of functional homologues. Considering the importance of flagellated and ciliated single-cell eukaryotes for the study of basal body function (; ; ), the identification of Cep164-related proteins in a genetically tractable organism would be of obvious interest.
Antibodies raised against an N-terminal fragment (1–298) of recombinant Cep164 recognized a band of ∼200 kD on Western blots performed on either centrosomes purified from KE37 lymphoblastoid cells or lysates from HeLaS3, hTERT-RPE1, and 293T cells, whereas preimmune sera showed no specific reactivity (). The reason for the unexpectedly slow migration of Cep164 is not entirely clear but may relate to the relatively acidic isoelectric point of the protein (5.32). In any case, we emphasize that a myc-tagged Cep164 protein expressed in human 293T cells showed a similarly retarded migration (). Moreover, siRNA-mediated depletion of Cep164 (using two different siRNA oligonucleotide duplexes) abolished immunoreactivity (), making us confident that the 200-kD protein detected by the anti-Cep164 antibody represents endogenous Cep164. IF microscopy demonstrated that Cep164 localizes to the centrosome (), confirming earlier results based on the localization of GFP- and myc-tagged Cep164 constructs (). Interestingly, when compared with the staining produced by antibodies against γ-tubulin (), the anti-Cep164 antibody stained only one of two centrioles (, middle, inset). Preimmune serum did not produce any centrosomal staining (, top) and Cep164 staining was abolished by siRNA-mediated depletion of Cep164 (, bottom), attesting to the antibody's specificity.
To determine whether the localization of Cep164 to only one centriole relates to the known difference in maturity between the two centrioles, we performed costaining with Cep170, an appendage-associated protein and established marker for the mature parental centriole (). Although Cep164 and Cep170 did not colocalize exactly (a point to which we will return later), the two proteins clearly localized to the same single centrioles (Fig. S1, A and B [GL2 controls], available at ). This indicates that Cep164, like Cep170, associates specifically with mature centrioles. To corroborate this conclusion, we also examined the localization of Cep164 in hTERT-RPE1 cells that had been induced to form a PC by serum starvation (). Costaining of the PC with antibodies against acetylated tubulin revealed that Cep164 was always associated with the one centriole that was located at the base of the PC, in agreement with the fact that only the mature centriole is able to initiate PC formation (). Interestingly, anti-Cep164 antibodies stained bar- or ringlike Cep164-positive structures at the base of the PC, depending on the angle of viewing (), whereas the axoneme itself was unstained.
Next, we examined Cep164 localization during cell cycle progression. The protein was detectable at centrioles at all stages, including mitosis, although staining was more intense during interphase (Fig. S2 A, available at ). When compared with centrin, which is a marker for individual centrioles (), Cep164 was restricted to only one centriole within both G1 and duplicated G2 centrosomes as expected (). Increased staining of a second parental centriole could then be detected at the onset of centrosome separation during prophase (Fig. S2 A, prophase), supporting the conclusion that Cep164 associates with structures that form concomitant with centriole maturation. During subsequent stages of mitosis, Cep164 was associated at similar levels with one centriole at each spindle pole ( and S2 A). This is in stark contrast to Cep170 and ninein, both of which were displaced from centrioles during mitosis (Fig. S2, B and C).
We also examined Cep164 expression during the cell cycle using biochemical approaches. As determined by qRT-PCR, Cep164 mRNA levels showed little change throughout G1 and S phase, but levels increased beginning in G2 and peaked in mitotic cells, reaching a twofold increase compared with interphase (). A similar fluctuation was also seen at the protein level. Whereas Cep164 protein was present at a low level throughout interphase, it became slightly more abundant in mitosis (). Furthermore, in mitotic samples Cep164 displayed a retarded electrophoretic mobility, suggestive of a mitotic modification (most likely phosphorylation).
Most proteins that specifically localize to mature centrioles, including Cep170 (), ninein (), centriolin/CEP110 (; ), ɛ-tubulin (), and Odf2 (), are known to associate with appendage structures. To examine whether Cep164 might similarly localize to appendages, we first asked to what extent Cep164 colocalizes with some of these proteins. Although Cep164 and ninein stained the same centriole, they did not colocalize exactly (). As already noted in the previous section, the same was true for Cep164 and Cep170 (Fig. S1). In contrast, ninein and Cep170 showed extensive colocalization as expected (). Next, we asked whether these proteins depend on each other for correct localization. siRNA-mediated depletion of Cep164 did not detectably affect the localization of ninein, Cep170, or ninein-like protein ( and S1 A; and not depicted). Conversely, depletion of ninein did not affect Cep164, although it caused the loss of Cep170 from the mature centriole (), and depletion of Cep170 did not detectably influence Cep164 localization (Fig. S1 B). These data suggest that Cep164 does not directly interact with any of the appendage proteins investigated here, raising the possibility that it localizes to a distinct appendage structure altogether.
To explore this possibility, immunogold electron microscopy (immuno-EM) was performed. This study revealed that Cep164 localizes specifically to the appendages of mature parental centrioles, whereas immature parental centrioles as well as progeny centrioles were unstained (), which is consistent with the IF results. In the past, two distinct sets of appendages, distal and subdistal, have been described (). To the best of our knowledge, most appendage proteins characterized so far localize primarily to subdistal appendages, although depletion of Odf2 was found to abolish distal as well as subdistal appendages (). In contrast, our data suggest that Cep164 is a genuine component of distal rather than subdistal appendages. In support of this interpretation, immunogold labeling of Cep164 decorated structures at the very tip of centrioles (, arrowheads), whereas electron-dense material possibly representing subdistal appendages could occasionally be seen at the proximal side of Cep164-positive structures (, arrows). Furthermore, whereas siRNA-mediated depletion of Cep164 abolished Cep164 staining, attesting to its specificity (), the immunolocalization of ninein was not detectably affected in Cep164-depleted cells (; compare with and ). Collectively, these data suggest that Cep164 is indispensable for PC formation and represents a genuine marker for distal appendages.
The present study identified several proteins whose depletion interfered with PC formation, confirming the view that ciliogenesis is a complex process (; ; ). In addition to proteins identified in independent studies, notably pericentrin, Odf2, PCM-1, and BBS4 (; ; ), we observed a role in PC assembly for ninein, a protein involved in microtubule nucleation and anchoring (), chTOG, a protein required for stabilizing microtubules and spindle pole organization (), and Cep290 (also called NPHP6), mutations in which have been linked to several human disease syndromes (; ; ; ). The exact roles of these proteins in ciliogenesis remain to be elucidated, but they are likely to relate to intracellular transport processes that depend on functional centrosome–microtubule interactions. In addition, we identified several proteins, notably Cep57, Cep131, Cep152, and ALMS1, whose depletion resulted in the formation of short or stunted cilia (). Little is presently known about the functions of Cep57, Cep131, and Cep152, but ALMS1 is attracting considerable interest because of its implication in ALMS (; ). What precise molecular defects lead to stunted cilia remains to be elucidated, but it is plausible that the functions of Cep57, Cep131, Cep152, and ALMS1 relate to IFT. In support of this proposal, we emphasize that truncated cilia have frequently been seen in and mutants carrying defects in the IFT machinery (; ).
The main focus of the present study was the characterization of Cep164, a novel protein of previously unknown function whose depletion severely impaired PC formation. By both IF microscopy and immuno-EM we were able to show that Cep164 localizes specifically to very distally located appendage structures on the mature centriole. On the basis of morphological analyses, centriolar appendages have in the past been classified as distal or subdistal (). From this perspective, it is remarkable that Cep164 did not colocalize exactly with proteins previously shown to associate with subdistal appendages, notably ninein and Cep170. Furthermore, no reciprocal dependencies for centriolar localization could be observed between Cep164 and ninein or Cep170 and, unlike ninein and Cep170, Cep164 persisted at the mature centriole throughout mitosis. Collectively, these data strongly indicate that Cep164 constitutes a genuine component of distal appendages.
A priori, Cep164 could be required for PC assembly or maintenance. Considering that Cep164 does not localize to the axoneme, it is unlikely to play a direct role in IFT, although it is possible that it contributes to intracellular transport between the interior of the cell and the cilium. Perhaps more likely, Cep164 may form part of a terminal plate (), which mediates the interaction between the basal body and the plasma membrane, or provide a docking site for the formation of transition fibers. Further insight into the molecular function of Cep164 will require identification of its interaction partners. In particular, it will be interesting to search for proteins binding to the WW domain that is present within the N-terminal end domain of the protein. One possibility is that Cep164 might contribute to the mediation of interactions with the cytoskeleton. Such interactions may in turn be critical for basal body positioning during PC formation. Alternatively, Cep164 may be indispensable for the formation of distal appendages. Careful ultrastructural analyses of Cep164-depleted centrioles will be required to explore this intriguing possibility.
In conclusion, our study identifies a novel protein, Cep164, as being critically required for PC formation. Furthermore, our results indicate that Cep164 provides an excellent marker for distal (rather than subdistal) appendages, strengthening the conclusion that centriolar appendage proteins are crucial for ciliogenesis (). Finally, considering the rapid emergence of data emphasizing the importance of PC formation and function for human health, it will be interesting to explore a possible relationship between Cep164 and ciliary disease syndromes. In this context, it is intriguing that the Cep164 gene locus (11q23.3) maps close to a region implicated in Jacobsen syndrome (). Thus, it may be rewarding to explore a possible connection between impaired functionality of Cep164 and diseases possibly related to defective ciliogenesis.
PCR was used to amplify a full-length human Cep164 cDNA from clone KIAA1052 (Kazusa DNA Research Institute). The cDNA was subcloned into a mammalian expression vector providing a C-terminal myc tag and the construct was verified by sequencing. For expression of a recombinant Cep164 protein fragment, bp 1–894 of the coding sequence were amplified by PCR, inserted into the expression vector pET28b+ (Novagen), and verified by sequencing. His-tagged N-terminal Cep164 was expressed in strain BL21 (DE3) and purified under denaturing conditions according to standard protocols (QIAexpressionistsystem; QIAGEN).
Rabbit anti-Cep164 antisera (R171 and R172) were raised against an N-terminal fragment spanning aa 1–298 (Charles River Laboratories). Similar results were obtained with both sera but R171 was used for most experiments. Antibody R171 was affinity purified using Affigel according to standard protocols (Affigel-10; Bio-Rad Laboratories). The anti-Cep170 mAb and anti-ninein mAb (79–160) were produced by immunization of BALB/c mice with recombinant fragments of human Cep170 (aa 15–754) and ninein (aa 1,110–2,662) purified from . Spleen cells were fused with PAIB3Ag81 mouse myeloma cells, and positive hybridoma clones were subcloned by limiting dilution. The anti-Cep170 mAb is an IgG1 and the anti-ninein mAb is an IgG2a.
Cells were grown at 37°C under 5% CO. U2OS, HeLaS3, and 293T cells were grown in DME and supplemented with 100 IU/ml of 10% FCS and 100 μg/ml penicillin-streptomycin. hTERT-RPE1 cells were grown in DME nutrient mixture, Ham's F12 (Sigma-Aldrich) supplemented with 10% FCS, penicillin-streptomycin, 2 mM glutamine, and 0.348% sodium bicarbonate. 293T cells were transfected using the calcium phosphate precipitation method ().
For preembedding immuno-EM, U2OS cells were grown on coverslips, fixed with 4% formaldehyde for 10 min, and permeabilized with PBS + 0.5% Triton X-100 (Roth) for 2 min. Blocking and primary antibody incubations were then performed as described for IF microscopy followed by incubation with goat anti–rabbit IgG-Nanogold (1:50; Nanoprobes). Nanogold was silver enhanced with HQ Silver (Nanoprobes) and cells were further processed as described previously (; ).
Cells were prepared for IF microscopy as described previously (). The primary antibodies used were 1 μg/ml affinity-purified rabbit anti-Cep164 IgG (R171) and anti–C-Nap1 IgG (), mouse anti-Cep170 mAb, anti–γ-tubulin mAb (1:1,000, GTU-88; Sigma-Aldrich), anti-centrin mAb (1:3,000, 20H5; provided by J.L. Salisbury, Mayo Clinic, Rochester, MN; ), and anti-acetylated tubulin mAb (1:2,000; 6-11B-1; Sigma-Aldrich). Secondary antibodies were Alexa Fluor 488/555–conjugated (1:1,000; Invitrogen) and cy2/cy3-conjugated donkey IgGs (1:1,000; Dianova). DNA was stained with 0.2 μg/ml DAPI.
Cells were analyzed using a microscope (Axioskop-2; Carl Zeiss MicroImaging, Inc.) equipped with a 63× NA 1.4 plan apochromat oil immersion objective and standard filter sets (Carl Zeiss MicroImaging, Inc.), a 1,300 × 1,030 pixel cooled charge-coupled device camera (CCD-1300-Y; Princeton Instruments), and Metavue software (Visitron Systems). Alternatively, for the data shown in and , a microscope (Deltavision) on a base (Olympus IX71; Applied Precision) equipped with an apo 100× 1.35 oil immersion objective, a camera (CoolSnap HQ; Photometrics), and a 37°C chamber were used for collecting 0.18–0.2-μm-distanced optical sections in the z axis.
Immunoblotting was performed as described previously (; ; ). Primary antibodies were used at the following concentrations: 1 μg/ml rabbit anti–Cep164 affinity-purified IgG (R171) or corresponding preimmune serum (1:1,000), mouse anti–α-tubulin mAb (1:5,000; DM1A; Sigma-Aldrich), anti–polo-like kinase 1 mAb (1:5; ), anti-cyclin E mAb (1:5; HE-12; provided by J. Bartek, Danish Cancer Society, Copenhagen, Denmark). Secondary antibodies were HRP-conjugated goat anti–rabbit (1:7,000; Bio-Rad Laboratories) or anti–mouse (1:7,000; Bio-Rad Laboratories) IgGs.
Proteins to be tested in the siRNA screen were depleted using siRNA duplex oligonucleotides (Qiagen and Dharmacon) targeting the sequences described in Table S1. Cep164 was efficiently depleted using two different siRNA duplex oligonucleotides (278 and 279). A duplex targeting luciferase (GL2; ) was used for control. RNA oligonucleotides were used at 20 μM, and cells were analyzed 48 or 72 h later.
RNA for qRT-PCR analysis was prepared as described in the text. cDNA was synthesized from RNA samples using random hexamers and Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. PCR reactions contained cDNA, Power SYBR Green Master Mix (Applied Biosystems) and 300 nM of forward and reverse primers. Primers were designed with Primer Express software (Applied Biosystems) and the amplified fragment corresponded to an exon–exon junction. qRT-PCR was performed in optical 384-well plates and fluorescence was quantified with a sequence detection system (Prism 7900 HT; Applied Biosystems). Samples were analyzed in triplicate and the raw data consisted of PCR cycle numbers required to reach a fluorescence threshold cycle. Raw fluorescence threshold cycle values were obtained using SDS 2.0 (Applied Biosystems). The relative expression level of target genes was normalized with geNorm software (Primer Design Ltd.; ) using eukaryotic translation elongation factor a-1 and β-glucuronidase genes as references to determine the normalization factor. The thermal profile recommended by Applied Biosystems was used for amplification (50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 1 min). To verify the specificity of amplification, a melting curve analysis was included according to the thermal profile suggested by the manufacturer (95°C for 15 s, 60°C for 15 s, and 95°C for 15 s). The generated data were analyzed with SDS 2.2 software.
Human centrosomes were isolated from KE37 cells as described previously ().
Table S1 indicates the siRNA oligonucleotide sequences used for protein depletion. Figs. S1 and S2 illustrate the localizations and dependencies of appendage proteins. Online supplemental material is available at . |
Appropriate functioning of the mature nervous system relies on the correct development of neuronal circuitry. In the spinal cord, second order neurons integrate sensory input from a large number of primary afferents. A prerequisite to form such a high degree of connectivity is the multiple ramification of primary axon projections to allow the innervation of several distinct targets. Dorsal root ganglion (DRG) axons enter the spinal cord at the dorsal root entry zone (DREZ), where they bifurcate into a rostral and a caudal arm. These arms extend longitudinally over several segments but remain confined to the oval bundle of His. Collaterals are then generated from these stem axons to penetrate the gray matter (; ). Cutaneous sensory collaterals are confined to the dorsal horn, whereas collaterals of muscle spindle Ia afferents grow to the ventral cord (). Thus, from a structural point of view, sensory axons display at least two types of ramifications within the cord: bifurcation at the DREZ and interstitial branching from stem axons to generate collaterals.
So far, the signaling cascades that underlie axonal branching in vivo have remained poorly understood, although neurotrophins, semaphorin 3A, and Slit proteins were implicated in the branching of axons or dendrites in vitro (; ; ; ; ). Our earlier studies suggested that the bifurcation of sensory axons at the DREZ depends on cyclic guanosine monophosphate (cGMP) signaling via the serine/threonine kinase cGMP-dependent protein kinase I (cGKI, also termed PKGI). cGKI is strongly expressed in embryonic sensory axons at the DREZ, and its absence was shown to cause axonal misprojections at the DREZ, reduced axon numbers in the developing dorsal funiculus, premature growth of some sensory axons toward the central canal, and consequently a reduction of ventral root potentials (). cGMP, a common second messenger that is produced by soluble or particulate guanylyl cyclases (sGCs or pGCs, respectively), controls a broad spectrum of physiological responses such as smooth muscle relaxation, phototransduction, olfactory transduction, bone growth, sperm motility, platelet spreading, electrolyte and water balance, and axonal pathfinding (; ; ; ).
Here, we identify the cGMP-producing receptor GC Npr2 as a molecule essential for sensory axon bifurcation. 1,1′- dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) labeling analysis of a loss-of-function mutant of Npr2, as well as a constitutive knockout and crossbreeding experiments of these mutant mice with a mouse line expressing EGFP in sensory neurons under control of the promotor, demonstrate at the single axon level that the interruption of cGMP signaling in sensory neurons results in a selective bifurcation error of DRG neurons. As a likely consequence a reduction in the degree of coupling with second order neurons in the dorsal horn was observed. Thus, these findings demonstrate that cGMP signaling involving Npr2 and cGKI is crucially important for proper sensory axon bifurcation at the DREZ during nervous system development.
Because our previous studies () gave evidence for an erroneous projection of DRG axons at the DREZ in the absence of cGKI, we analyzed the trajectories of single axons in cGKI knockout mice at different spinal levels. The lipophilic tracer DiI was applied to embryonic day (E) 12–14 DRG in a manner that allowed us to follow the trajectories of single sensory axons in whole mount dissections of the spinal cord. Interestingly, visualization of single growth cones at the DREZ in wild types indicated that the formation of the rostral and caudal arms occurs directly by splitting of the tip of the axon, the growth cone (), whereas interstitial branches, the major branch type within the brain, sprout from the axon shaft (). All sensory axons in wild-type and heterozygous mice bifurcated in a T- or Y-shaped manner in caudal and rostral directions; in contrast, sensory axons of cGKI-deficient mice at all spinal levels almost always lacked a bifurcation. The ingrowing axon ran either caudally, or, with slight preference, rostrally (). These results indicate that sensory axon bifurcation at the DREZ depends on cGKI.
To identify components responsible for cGMP synthesis in embryonic sensory neurons we surveyed the expression of GCs in embryonic DRG neurons. cGMP is generated from Mg-GTP by s- or pGCs. sGCs are heterodimeric enzymes composed of α and β subunits that are activated by nitric oxide (NO) and found to be colocalized with NO synthases (NOS) in several cases (, ; ). In contrast, the pGCs are transmembrane proteins that contain extracellular ligand-binding domains and exist as homodimers on the plasma membrane. In mammals, seven distinct pGCs are known that share a common overall domain structure (; ).
In a PCR screen using cDNA from E12 mouse DRG as an amplification template, we could detect transcripts for the receptor-type GCs (also called or ) and () but not for the other five receptor GCs (). The expression of could be confirmed by in situ hybridization where mRNA was found to be expressed in embryonic DRG in a pattern overlapping with that of (), whereas mRNA was found to be beyond the detection limits using two distinct probes for hybridization (unpublished data). transcripts were also not detectable using cDNA from DRG or spinal cord tissue as an amplification template. Spinal cord neurons were negative or faintly positive for , whereas cGKIα was found in DRG cells, in preganglionic spinal neurons, and faintly in motoneurons at the mRNA as well as at the protein level (). It has been reported that NOS1 (nNOS) colocalized with cGKI in rat DRG, suggesting that sGCs might be relevant to the generation of cGMP in embryonic DRG (). Our PCR screen, as well as in situ analysis of the different and s, revealed no or only weak expression of the and different forms of s in embryonic DRG, making it unlikely that, at this age, the NOS/sGC system is functional in DRG (Fig. S1, available at ).
Based on these results on the expression pattern and other published findings (), we focused on Npr2 as a candidate protein for the generation of cGMP and thereby activation of cGKIα in embryonic DRG neurons. To test whether Npr2 might be involved in axonal bifurcation at the DREZ, we made use of mice. These express an inactive form of Npr2 because of an amino acid substitution in the GC domain and lack an increase of the intracellular cGMP level upon stimulation (). DiI tracing experiments revealed a complete phenocopy of the lack of bifurcation of sensory axon as observed in the constitutive cGKI knockout mice. Again, sensory axons turned without bifurcation in the rostral or caudal direction with a slight preference to the rostral direction (). The majority of the axons stayed within the oval bundle of His.
A lack of bifurcation at the DREZ would result in a significant reduction of axon numbers in the developing dorsal funiculus. As a measure of axon number, anti-trkA immunoreactivity was quantified in transverse sections of E13 spinal cords (). The trkA-labeled area in the dorsal funiculus of E13 embryos amounted to 64.1 and 59% of the wild-type controls, determined at thoracic and lumbar trunk levels, respectively. Although this method only gives an indirect estimate of axon number, the results strongly suggest that mice lacking a bifurcation in the DREZ contain substantially fewer axons within the dorsal funiculus. In addition to these branching errors we also observed a small group of trkA-positive axons penetrating prematurely into the dorsal horn growing further in the direction of the central canal (). This erroneous trajectory of trkA-positive axons resembles the previously described misguidance in the absence of cGKI (). However, in both cases the majority of axons turns and grows in the developing funiculus. Thus, the identical phenotypes in axon bifurcation at the DREZ of and the constitutive cGKI knockout mice suggest that in embryonic DRG neurons activation of cGKIα by cGMP generated by Npr2 is crucial for triggering the bifurcation process at the DREZ. Analysis of bifurcation of double heterozygotes did not reveal significant bifurcation errors (93% T-shaped branches in double heterozygotes and 97% in wild type), indicating that there was no detectable concentration dependency in this system.
It is important to note, however, that the inactivation of cGMP signaling in Npr2 loss-of-function mutants disrupts neither the overall organization of the spinal cord nor the pathfinding of TAG-1–positive commissural axons to the ventral midline or the formation of L1-positive lateral and ventral axon tracts within the developing spinal cord ().
After a waiting period, collaterals extending into the gray matter of the spinal cord are generated by interstitial branching along the longitudinal stem axons (). Although the collaterals of nociceptive afferents are confined to the dorsal cord, collaterals of muscle afferents grow into the ventral horn where further branching occurs. Staining of transverse sections by antibodies to trkA, parvalbumin, or peripherin indicated that nociceptive as well as proprioceptive collaterals extend into the dorsal and ventral horn of mutant mice, respectively. The pattern of target innervation of both collateral systems was indistinguishable from that of wild-type mice (). Furthermore, DiI tracing experiments revealed that neither blocked cGMP formation in mice with inactive Npr2, nor did the constitutive cGKI knockout prevent the formation of collaterals, although the distances between the points of origin from the stem axon were slightly reduced in mutant mice (). Thus, cGMP signaling induced by Npr2 and mediated by cGKIα is not required for collateral formation, and its loss does not influence the overall trajectories of sensory collaterals (). It should be noted that the overall growth pattern of sensory axons within the periphery was unchanged as well (Fig. S2, available at ).
To seek further confirmation of the bifurcation error observed at the DREZ and clarify whether compensatory mechanisms exist to correct the bifurcation error at later developmental stages, we examined the bifurcation behavior of DRG neurons by a or allele crossed into the mutants or the constitutive cGKI knockout mice. Only small fractions of all sensory modalities of DRG neurons are labeled in the mouse line, designated M, which allows one to follow the trajectories of single sensory axons at the DREZ (). Because of the late activation of the promoter, dissected spinal cords were analyzed at postnatal day (P) 21. Consistent with the results of the DiI tracing, we found in whole mounts from or cGKI-deficient mice a selective bifurcation error at the DREZ. In contrast to the T- or Y-shaped axons observed in wild-type mice, the main axons of mutant mice turned either caudally or rostrally, with some preference for the rostral direction (). We conclude that the bifurcation error caused by the lack of cGMP signaling at a stage when axons enter the cord persists to mature stages, and no compensation seems to exist at the DREZ, at least until P56 ( and not depicted), when the connectivity of the spinal cord is quasimature in mice. However, in addition to the bifurcation error, we sometimes observed aberrant sensory axon projections at the DREZ as illustrated in . Axons initially turned in one direction but then looped backward to extend in the opposite direction () or formed short processes of 20–50 μm in length at the DREZ (). Furthermore, these data were confirmed by using the mouse line crossed into mutant mice ( and not depicted).
Primary nociceptive afferents establish synaptic connections with neurons in the superficial laminae of the dorsal horn. The absence of bifurcation in mice prompted us to examine the functional consequences of this alteration. The overall layering within the dorsal spinal cord of these mice appears not to be severed by the inactive Npr2, as indicated by staining using the isolectin B4 antibodies to the calcitonin gene-related peptide (CGRP) or the vesicular transport protein of glutamatergic synapses VGlut1 (). As an indicator for neuronal connectivity we recorded glutamatergic miniature excitatory postsynaptic currents (mEPSCs) from neurons in the superficial laminae of the dorsal horn in slice preparations from P10–14. To measure activity originating from nociceptive afferents we applied capsaicin, a compound preferentially selective for polymodal nociceptor cells in the superficial dorsal horn that activates presynaptic TRPV1 receptors on primary afferents (). It was found that the fraction of neurons responding to capsaicin was significantly higher in wild-type slices (wild type, 57.14% [16/28]; , 22.22% [6/27]; P < 0.01, χ test), which is consistent with the observation that the number of DRG axons was lower in the oval bundle of His in mice (). Furthermore, dorsal spinal cord neurons of mice displayed a significantly weaker response to capsaicin than wild-type neurons (; P = 0.032, Mann-Whitney U test). Even under resting conditions, i.e., in the absence of capsaicin, mESPC frequency tended to be lower in neurons (P = 0.055, Mann-Whitney U test). Interestingly, capsaicin-insensitive neurons did not differ with respect to their frequency. The basic postsynaptic parameters of the glutamatergic mEPSCs in the dorsal spinal cord, i.e., amplitude and time constant of decay, were not affected in mice (; amplitude: wild type, 21.36 ± 1.39 pA, = 24; , 24.19 ± 2.17 pA, = 24; decay time constant: wild-type, 2.09 ± 0.15 ms; = 24; , 1.90 ± 0.12 ms; = 24).
Branch formation is one principle process that defines the pattern of axonal trajectories. Despite intensive research efforts, the molecular signaling pathways underlying neuronal branching have remained poorly understood. The data presented in this paper illuminate the role of cGMP signaling comprising the receptor GC Npr2 and the serine/threonine protein kinase cGKI for bifurcation of primary afferents at the DREZ. In this paper, we observed that Npr2 is expressed in embryonic DRG, largely overlapping the expression of cGKIα. We showed that interruption of cGMP signaling caused by a loss of function of Npr2 or the absence of cGKI prevents sensory axon bifurcation at the DREZ. Interestingly, interstitial branching, i.e., the sprouting of collaterals, is not affected by the interruption of cGMP-mediated signal transduction, suggesting that different sets of molecules are responsible for sensory axon bifurcation and interstitial branching. Consistently, in retinotectal axons, interstitial branching appears to depend on ephrin-EphA signaling (; ). The absence of bifurcation at the DREZ in turn results in a substantial reduction in the number of axons running in the dorsal funiculus and the synaptic input received by second order neurons within the superficial dorsal horn, which is the first relay station of nociceptive sensory axons ().
Several genetic instructions are likely to act in concert to induce bifurcation and longitudinal growth: (a) a mechanism causing the sensory axons to avoid the gray matter when reaching and entering the cord; (b) a signal at the point of bifurcation that allows the two main branches to become distinct and enables further segregation; (c) the growth cones of the two main branches should be able to navigate independently to detect specific signals for rostral or caudal growth and therefore may express distinct sets of guidance receptors; (d) main branches growing in the same longitudinal direction must recognize each other to fasciculate in the oval bundle of His; and (e) it might be essential that after branching at the DREZ, further bifurcation is suppressed. Additional guidance cues are then required to regulate the mediolateral position of the main axon branches in the dorsal funiculus and to induce interstitial branching to generate collaterals.
Most of these processes are not fully understood, although recent investigations on the trajectories of proprioceptive and cutaneous axons and their collaterals suggest that repulsive factors, such as semaphorins and netrins, or branching/repulsive factors, such as Slit2, may shape axon growth. For instance, in the absence of plexinA1, main axons of proprioceptive neurons invade the superficial dorsal horn, whereas the expression of netrin-1 within the dorsal spinal cord prevents ingrowth and intraspinal projections of both proprioceptive and cutaneous afferents (; ). Slit proteins that bind to robo receptors are reported to influence branching of sensory axons in a collagen culture (). Notably, in the absence of Slit1 and 2, mice display a partial overshooting in axonal growth toward the midline of the spinal cord, but all of the sensory axons in double mutants still bifurcated at the DREZ, although at a slightly different angle (). Furthermore, the transcription factors Runx3 and Er81 may regulate the expression of some of the surface receptors that direct proprioceptive sensory collaterals to the intermediate and ventral spinal cord (; ; ). Finally, antibody perturbation experiments in the chick indicated that members of the immunoglobulin superfamily influence the projection patterns of collaterals ().
The deficits observed in this paper suggest a signaling mechanism where, after the activation of the receptor GC Npr2, cGMP is generated from GTP, which triggers cGKIα to phosphorylate yet-unresolved targets converging in cytoskeletal rearrangements () in the growth cones of sensory afferents (). From other cellular systems it is established that the binding of C-type natriuretic peptide (CNP), the extracellular ligand of Npr2, leads to the generation of the second messenger cGMP that, in turn, activates cGMP-dependent kinases as one of its major cellular targets (). Because transcripts of CNP are abundant in the dorsal spinal cord of E12.5 mice (), CNP might be an attractive candidate serving as a “bifurcation signal.” However, using a variety of in vitro conditions we did not see any action of CNP (either provided as a point source or homogeneously) on the growth cone behavior of DRG axons (Fig. S3, available at ). There are several explanations for this finding, including the following: (a) in DRG neurons CNP is inappropriate for activation of Npr2; (b) CNP only acts in concert with other factors, e.g., repulsive components such as the Slit or netrin proteins to avoid ingrowth into the gray matter (; ); and (c) the in vitro systems do not reflect the in vivo situation, which makes it impossible to reproduce the conditions required for sensory axonal bifurcation. In this context it is noteworthy that activation of receptor GCs can also occur via the small GTPase Rac and the p21-activated kinase pathway (), which opens the possibility that cGMP formation by Npr2 can be induced by extracellular factors other than CNP. cGMP formation by transmembrane GCs can also be regulated via protein kinase C through a so-called heterologous desensitization mechanism leading to dephosphorylation of conserved stretches of intracellular segments of pGCs (). The situation becomes more complex in that on the one hand, cGMP signaling can regulate its own degradation by binding to GAF domains present in several cyclic nucleotide phosphodiesterases (PDEs), where they cause allosteric activation of their catalytic domain (). Alternatively, this association can also affect the level of intracellular cAMP depending on the type of PDE present in sensory growth cones. In vitro studies showed that the relation between cAMP and cGMP modulates the responses of growth cones to external signals ().
Outside the nervous system, a well-characterized physiological action of CNP/Npr2 via cGKII but not GKI is the induction of long bone growth, where CNP acting on chondrocytes induces endochondral ossification. Inactivating mutations of the genes encoding for CNP () or Npr2 (Npr-B) in mice (; ) or humans () causes dwarfism, whereas overexpression of CNP as a transgene (; ) or reduced clearance of CNP (; ) was shown to cause skeletal overgrowth. Interestingly, -deficient mice displayed self-clasping, priapism, and seizures, suggesting neuronal disorders (). Abnormalities of bone growth caused by enhanced or decreased CNP action only evolve after birth because of changes in the proliferative zones in the growth plates. Accordingly, cGKII deficiency or spontaneous inactivating mutations of this gene in rats produce dwarfism (; ; ). Although it is unclear whether dwarfism as seen in mutants also influences neuronal connectivity, it is unlikely to explain the present bifurcation deficits in DRG sensory axons by a reduced bone growth because cGKI-deficient mice lack the respective symptoms (). cGMP signaling has also been implicated in cell proliferation, but cell counts in the DRG have revealed no change in the constitutive cGKI knockout mice. The reduction of axon numbers as reflected by the stained area in the dorsal funiculus is therefore not a result of neuronal cell death ().
A deeper understanding of the cGMP signaling mechanisms at the DREZ requires further knowledge on the activation and desensitization processes of the receptor GC Npr2, the regulation of cGMP concentration by PDEs, and the identification of phosphorylation targets of cGKIα in growth cones. The latter might include components regulating the actin cytoskeleton such as proteins of the Ena/VASP family () or the myosin phosphatase (). It will be interesting to explore whether cGMP-mediated bifurcation is unique to sensory afferents of the spinal cord and how the complete absence of one longitudinal main branch affects the sensory-motor coordination of spinal reflex activity.
mice were obtained from Jackson ImmunoResearch Laboratories and their genotyping as well as that of cGKI-deficient mice has been described previously (; ). These mouse lines were crossed with transgenic or mice under the control of the promoter () to get GFP-expressing and cGKI-deficient mice, respectively.
For fluorescence analysis of sensory axon morphology in the offspring of crossbreedings between and cGKI-deficient mice or mice, spinal cords were removed from P21 and fixed in PBS containing 4% paraformaldehyde. After mounting, spinal cords were examined using an inverted flourescence microscope (for details see Immunohistochemistry).The evaluation was performed blind with regard to the genotype. Only labeled axons that were unambigously identified as single axons were counted. Because of the limitations of the DiI tracing only distances between developing collaterals were counted from E14 embryos. Axons that had no or only one collateral were ignored for the quantification.
mRNA isolated from mouse E12 spinal cord or DRG, respectively, using Dynabeads Oligo(dT) (Invitrogen) according to the manufacturer's instructions was subsequently reverse transcribed using SuperScript II Reverse Transcriptase (Invitrogen). DNA fragments of elements of the cGMP signaling pathway were amplified using oligonucleotides 5′ (5′-AAAAATGAGCGAACTGGAG-3′), 3′ (5′-GACCTCTCGGATTTAGTGAAC-3′), 5′ (5′-GTTCGGAAGAGTGGAGCTTG-3′), 3′ (5′-CTTTGTTAAGAATGACCTCGGG-3′), 5′ (5′-CGTCAATGATCGGCCCCTGGTA-3′), 3′ (5′-CTTTGGCTGGTCCCCCTCTGTTG-3′), 5′ (5′-GGGACCTGGCCACCTTGTTCA-3′), 3′ (5′-TGCGGCTGGACTTTTCACTCTGC-3′), 5′ (5′-CAAAGGGGGCAGGCATCACCAG-3′), 3′ (5′-CACCGCTCGAGCAAAGGCACAGA-3′), 5′ (5′-AAGGGTCAACCTGGACTCAC-3′), 3′ (5′-ATAGATTCTCTTTCCTGCAGCC-3′), 5′ (5′-AGAAAGACAAGCCGCAACAGAGT-3′), 3′ (5′-GCAGCCGCTTTAATGATACCAG-3′), 5′ (5′-ACTGGACCAGTCTTAGCAGG-3′), 3′ (5′-TCTCATCTGGTGATGACTGG-3′), 5′ (5′-GAGAAGGGGCCATGAAGATTGTC-3′), 3′ (5′-CCTCCGTGCCTGTATTTTTCCTG-3′), 5′ (5′-CTTCCACCTGGATGTGTTTG-3′), 3′ (5′-CACAGCCATCAGCTCCTG-3′), 5′ (5′-TGGCGCCTTCCCCTCCTGACT-3′), 3′ (5′-CTGGGCCTGTTCCTGGGTTCG-3′), 5′ (5′-TGCTGATTGCCCTCCTTGTGCT-3′), 3′ (5′-ATTTGAGACGCCCGTGGACTTC-3′), 5′ (5′-CTGTGGTATTTTGGAGTGTCC-3′), 3′ (5′ TGTCCTGGTGAGAGGACAC-3′), 5′ (5′-GATCACTGCACCCCAAGAC-3′), and 3′ (5′-TGCTTCAGGTTGAGGTCAAG-3′). The second set of primers for Gucy2e were 5′ (5′-GGACTGGATGTTCAAGTCTTCC-3′), 3′ (5′-CAGGTCTACCACCTCAATAGGC-3′), 5′ (5′-GGCCTCAGGATTTGTTGG-3′), 3′ (5′-TACATCATAGGGGACAAAGACG-3′), 5′ (5′-TCGTTTTCATCCTCTTGCAG-3′), and 3′ (5′-ATGAGGTGTTCGAGTCATTCC-3′). For control purposes all primer pairs were tested with a cDNA template from a whole mouse E17 embryo. The amplification products were subsequently cloned into a pBluescriptKS vector (Stratagene) and their identity was verified by sequencing.
In situ hybridization studies on 25-μm transversal sections of mouse E12 spinal cord using DIG-labeled riboprobes to and were performed as described previously ().
For immunohistochemical detection, cryostat sections of formaldehyde-fixed embryos were stained by indirect immunofluorescence using rabbit antibodies to trkA (Upstate Biotechnology), parvalbumin (Swant), CGRP (Peninsula Laboratories, Inc.), L1 (), peripherin (Chemicon), guinea pig anti-VGlut1 (Chemicon), mouse monoclonal anti–TAG-1 (4D7; Developmental Studies Hybridoma Bank; ), or isolectin GS-IB Alexa Fluor conjugate (Invitrogen). The mAb 4D7 was purified from hybridoma supernatant by affinity chromatography and applied at a concentration of 4 μg/ml. Secondary Cy3-conjugated antibodies were obtained from Dianova. All images were obtained at room temperature using a microscope (Axiovert 135 or 200) equipped with Neofluar/Acroplan objectives (5, 10, 20, or 40× magnification with numerical apertures 0.15, 0.25, 0.5, or 0.75, respectively), a charged-coupled device camera (Axiocam HRC), and acquisition software (Axiovision 3.1; all from Carl Zeiss MicroImaging, Inc.). The area labeled by anti-trkA was measured using Image 1.37v software (National Institutes of Health). Contrast and brightness were adjusted in some images using Photoshop (Adobe) but no further processing was performed. Figures were assembled using CorelDraw (Corel).
The spinal column was removed from P10–14 mice and placed in an ice-cold dissection solution with a reduced calcium concentration consisting of 125 mM NaCl, 4 mM KCl, 10 mM glucose, 1.25 mM NaHPO, 25 mM NaHCO, 0.1 mM CaCl, and 3.0 mM MgCl (; ). After laminectomy, the spinal cords were removed and embedded in 2.5% agarose (SeaPlaque; Cambrex Bio Science), and 180-μm transversal slices were prepared by vibratome cutting. Slices were maintained at room temperature for at least 1 h before recording. Whole-cell patch-clamp recordings of mEPSCs of neurons located in the superficial dorsal horn were performed as described previously (). AMPA receptor–mediated mEPSCs were isolated pharmacologically by blocking glycinergic, GABAergic input, and NMDA receptor–mediated currents (1.0 μM strychnine; 100 μM picrotoxin; 100 μM APV). Action potential–dependent neurotransmitter release was blocked by 1 μM tetrodotoxin. All experiments were performed at room temperature. During recordings, slices were perfused at a flow rate of 2 ml/min with a bath solution of 125 mM NaCl, 4 mM KCl, 10 mM glucose, 1.25 mM NaHPO, 25 mM NaHCO, 2.0 mM CaCl, and 1.0 mM MgCl. The patch pipette solution contained 120 mM CsCl, 4 mM NaCl, 5 mM glucose, 5 mM ethylene glycol-bis (β-aminoethyl ether) N,N,N′,N′-tretraacetic acid, 10 mM N-2-hydroxyethylpiperzine-N′-2-ethanesulfonic acid, 0.5 mM CaCl, and 4 mM MgCl, pH 7.3. Capsaicin (Sigma-Aldrich) was applied by a superfusion pipette for 100 s at a concentration of 10 μM. During recordings at a holding potential of −70 mV the effective access resistance was in the range of 10–40 MOhm and was controlled throughout the experiment by using a short depolarizing pulse. Recordings were only accepted if the access resistance was <40 MOhm and did not change more than 20% during the experiment. All electrophysiological experiments were performed blind with regard to the genotype of the animals. Recordings were made using a patch clamp amplifier (EPC-9; HEKA Electronik). Signals were sampled at a rate of 10 kHz and analyzed off-line using WinTida 5.0 (HEKA Electronik). Postsynaptic currents were filtered at 3 kHz and analyzed by MiniAnalysis (Synaptosoft, Inc.).
Fig. S1 shows that NOS1, NOS3, and NOS1/NOS3-deficient mice do not reveal a reduced trkA-positive dorsal funiculus that is indicative of a branching error at the DREZ (A–C). The localization of NOS1 at the mRNA and protein level is revealed in D–H. Fig. S2 shows whole mounts of embryonic mice stained by an anti-neurofilament antibody that indicate no pathfinding errors of sensory axons in the periphery in the absence cGKI. Fig. S3 analyzes the growth cone behavior of sensory axons in response to the presence of CNP in in vitro cultures. Online supplemental material is available at . |
The plasma membrane of the cell is organized into cell surface domains that include clathrin-coated pits, lipid rafts, and caveolae. Lipid rafts have been proposed to be transient and dynamic nanodomains of <10 nm in size (; for review see ). Caveolae are invaginated lipid raft macrodomains (50–150 nm) whose stability at the plasma membrane is attributable, in large part, to the formation of highly stable oligomers of their coat protein, caveolin-1 (; ; ). Clathrin-coated pits (100–150 nm) internalize rapidly upon formation at the same plasma membrane site, and their lateral cell surface mobility is enhanced by actin cytoskeleton depolymerization (). The membrane skeleton, which is associated with the cytoplasmic face of the plasma membrane, is composed of a meshwork of actin filaments and associated proteins that form fences and pickets that restrict molecular diffusion and partition the membrane into compartments that vary in size from 50 to several hundred nanometers in size. Large-scale movements require the traversing of these compartmental boundaries via a process called hop diffusion, providing an explanation for the reduced diffusion of macromolecular complexes in biological membranes (; ; ; ; ). However, the identification of physiological processes regulated by plasma membrane domain compartmentalization remains limited.
Molecular cross-linking of raft components has been proposed to generate stabilized domains that promote transmembrane signaling and interaction with the cytoskeleton (; ). Clustering of the glycosyl-phosphatidylinositol–anchored receptor CD59 was recently shown to result in the transient recruitment of Gαi2 and Lyn and immobilization through binding to F-actin, which was termed stimulation-induced temporary arrest of lateral diffusion (STALL); the recruitment of PLCγ to CD59 clusters undergoing STALL results in local IP3-Ca signaling events (,). Single-particle tracking analysis has shown that movement of cell surface–bound murine polyoma-like virus particles is actin restricted, cholesterol dependent, and not associated with caveolae or clathrin-coated pits (). However, the transient anchorage of cross-linked glycosyl- phosphatidylinositol–anchored proteins was found to be dependent on caveolin in addition to cholesterol and Src family kinases (). Ras clustering upon activation supports a role for macromolecular complex formation in signal transduction (; ).
Glycan-based domains generated by galectin binding to cell surface glycoproteins have been proposed (). The gene encodes β1,6-acetylglucosaminyltransferase V (GlcNAc-TV), a Golgi-processing enzyme that modifies N-glycans, generating high affinity ligands for galectins (). The galectins are a family of β-galactoside–binding proteins with affinities for N-glycans proportional to GlcNAc branching () that can cross-link glycoproteins to form molecular lattices (). Close interactions of galectin-3 on the cell surface have recently been shown by fluorescence resonance energy transfer, demonstrating that galectin-3 can oligomerize to form a lattice (). We have shown that binding of Mgat5-modified N-glycans on EGF and TGF-β receptors to galectin-3 opposes receptor loss to constitutive endocytosis and, thereby, sensitizes cells to cytokines. Blocking clathrin-coated pit endocytosis with potassium depletion and lipid raft endocytosis with nystatin rescued cytokine EGF and TGF-β sensitivity in Mgat5 tumor cells (). This suggested that galectin binding protects receptors from negative regulation through interaction with clathrin-coated pits and lipid rafts. Herein, we demonstrate that the reduction of EGF receptor (EGFR) lateral mobility by Mgat5-dependent galectin-mediated cross-linking limits interaction of the receptor with stable inhibitory domains of oligomerized caveolin. Our data indicate that recruitment to positive regulatory Mgat5/galectin-dependent macromolecular complexes limits the large-scale macrodiffusion of EGFR, effectively competing with receptor recruitment to other plasma membrane domains.
gene expression and its β1,6GlcNAc-branched N-glycan products increase with oncogenic transformation in human cancers of the breast and colon and contribute directly to tumor progression and metastasis in mice (). In transgenic mice expressing the polyoma middle T antigen (PyMT) transgene under the control of the mouse mammary tumor virus (MMTV) long terminal repeat, Mgat5 mice show delayed tumor development and considerably fewer lung metastases compared with their Mgat5 littermates (). In contrast to the Mgat5 background, MMTV-PyMT mammary tumorigenesis is accelerated in Cav1 mice (; ).
Cav1 has been shown to act as a negative regulator of growth signaling (; ) that, via its scaffolding domain (; ), recruits EGF, PDGF, and TGF-β receptors to caveolae and suppresses responsiveness to these cytokines (; ; ). The CAV1 gene maps to a tumor suppressor locus (D7S522; 7q31.1) that is frequently deleted in human carcinomas, including breast cancer (). Up to 16% of human breast cancers express a CAV1 P132L mutation that correlates with breast tumor progression and acts as a dominant negative for scaffold domain–dependent growth suppression (; ). However, contrasting with its apparent tumor suppressor function, Cav1 expression is associated with a poor prognosis in multiple tumor types, including breast tumors (; ; ).
The demonstration here that Mgat5 expression overrides the tumor suppressor function of Cav1 identifies the latter as a conditional tumor suppressor. The Mgat5-dependent galectin-glycoprotein lattice is a positive signaling environment that regulates EGFR mobility and acts dominantly to protect receptors from negative regulation and immobilization through interaction with oligomerized Cav1. Mgat5-dependent expression of the galectin lattice relieves Cav1 suppression in Mgat5 PyMT mammary tumor cells, and responsiveness to EGF is rescued in Mgat5 tumor cells by reducing Cav1 levels below a threshold. Our results demonstrate the competitive recruitment of EGFR to the extracellular galectin lattice and stable caveolin-1 microdomains and show that the integrity of these domains determines signaling potential and tumor progression.
The majority of PyMT Mgat5 tumors are small (<2 cm), but a minority (5–10%) of breast tumors in PyMT Mgat5 mice display an acceleration of growth, suggesting escape from the suppressive effects of Mgat5 deficiency (). Mgat5 and escaper Mgat5 tumor cell lines were established from small and large tumors, respectively, of MMTV-PyMT Mgat5 mice. Compared with Mgat5 mammary carcinoma cells, the Mgat5 cell line is markedly less sensitive to EGF and TGF-β and is deficient in epithelial-mesenchymal transition (EMT) and fibronectin matrix deposition (; ). In contrast to Mgat5 cells, Mgat5 cells display levels of responsiveness to EGF that are comparable with that of wild-type Mgat5 cells (). However, responsiveness to TGF-β in both Mgat5 and Mgat5 cells is impaired relative to Mgat5 cells (). Furthermore, both Mgat5 and Mgat5 cells are deficient in EMT and fibronectin matrix deposition compared with Mgat5 cells (). Therefore, the phenotypic rescue of Mgat5 cells, which is permissive for tumor growth in the absence of Mgat5, is associated with increased EGFR signaling but not TGF-β signaling, EMT, and fibronectin remodeling associated with Mgat5-dependent invasive tumor cell phenotypes.
Both Mgat5 cell lines show the reduced expression of Cav1 relative to Mgat5 cells by both quantitative immunofluorescence and Western blotting (). However, Cav1 and total Cav levels were reduced to a significantly greater extent (P < 0.05) in Mgat5 than in Mgat5 cells (). Stable infection of Mgat5 and Mgat5 cells with an Mgat5 expression vector generated rescued Mgat5 (Rescue) and Mgat5 (ESC-Rescue) cell lines that present restored β1,6GlcNAc-branched N-glycan expression, as verified by labeling with L-PHA–FITC that binds the β1-6 branch of N-glycans (). However, Cav1 expression was restored to wild-type levels only upon the rescue of Mgat5 but not Mgat5 cells (). In Mgat5 cells, Cav1/2 levels were reduced by swainsonine, an α-mannosidase II inhibitor that blocks N-glycan branching, and by β-lactose, a competitive inhibitor of galectin binding at the cell surface (). Expression of Mgat5 and the N-glycan processing pathway can therefore impact Cav1. This suggests that Mgat5 and the expression of β1,6GlcNAc-branched N-glycans results in positive feedback to increase Cav1 levels. The failure of Mgat5 to restore Cav1 expression in Mgat5 cells suggests that additional genetic modifications may occur in the larger escaper Mgat5 tumors that block Cav1 up-regulation and suppression of growth.
Electron microscopic analysis of the cells shows a dramatic reduction in the number of cell surface–associated caveolae in both Mgat5 and Mgat5 cells relative to Mgat5 or Rescue cells that express elevated Cav1 levels without affecting the number of clathrin-coated pits (). Cav1 is required for caveolae formation (), suggesting that the threshold level of Cav1 required for caveolae formation is not attained in either Mgat5 cell line. To probe for possible functional interactions between Cav1 and the Mgat5-dependent lattice in tumor cells, we compared Cav1 protein levels and mammary tumor volume in MMTV-PyMT Mgat5 and MMTV-PyMT Mgat5 mice at 12 wk of age. Cav1 levels in Mgat5 tumor lysates correlated inversely with increased tumor growth, whereas no correlation with tumor size was observed in Mgat5 tumors (). Therefore, reduced Cav1 is observed after either chemical disruption of the galectin lattice in Mgat5 mammary tumor cells in vitro and spontaneously with tumor progression in PyMT Mgat5 tumors in vivo.
To determine whether lower endogenous Cav1 levels in Mgat5 cells were permissive for cytokine responsiveness, Mgat5 cells were infected with an adenoviral vector for the expression of Cav1 (). Overexpression of Cav1 protein levels in Mgat5 cells suppressed EGF-dependent Erk1/2 phosphorylation and nuclear translocation to levels comparable with untreated Mgat5 cells (). Cav1 adenoviral infection of Mgat5 cells, Rescue, and ESC-Rescue cells did not inhibit EGF responsiveness (). This suggests that the expression of β1,6GlcNAc-branched N-glycans protects EGFR from negative regulation by Cav1. This was confirmed by reducing Cav1 with siRNA in Mgat5 cells that enhanced the EGF response, but this had no such effect in Mgat5, Mgat5, or Rescue cell lines (). Thus, Mgat5 expression in tumor cells blocks the ability of Cav1 to act as a suppressor of EGFR signaling.
To explore the effects of Cav1 on cell surface raft dynamics, we measured the lateral diffusion rate of GM1-bound cholera toxin b subunit (CT-B) at the cell surface using FRAP at room temperature to limit the internalization of GM1-bound CT-B ( and ; ). Mgat5 cells exhibited increased CT-B diffusion relative to the other cell lines. Both the rate of diffusion and fractional recovery of CT-B into the FRAP region was increased in Mgat5 cells, suggesting that Cav1 contributes to the reduced mobility of CT-B. Indeed, surface diffusion of CT-B in Mgat5 cells was reduced upon transfection with Cav1–monomeric RFP (mRFP). Conversely, CT-B diffusion was enhanced by Cav1 siRNA treatment of Mgat5 and Mgat5 cells and presented a profile equivalent to that of Mgat5 cells. Therefore, the increased Cav1 expression in Mgat5 cells relative to Mgat5 cells correlates inversely with raft dynamics.
As observed for CT-B, EGFR-YFP showed enhanced cell surface diffusion and reduced immobile fraction in Mgat5 cells relative to Mgat5 cells by FRAP analysis ( and ). Cav1 transfection of Mgat5 cells reduced the diffusion rate and increased the immobile fraction, whereas Cav1 siRNA transfection of Mgat5 cells had the opposing effect. In contrast, modulation of Cav1 levels in Mgat5 cells by either adenoviral Cav1 expression or Cav1 siRNA did not impact EGFR-YFP diffusion ( and ). Therefore, by limiting the exchange of EGFR-YFP between the bleached zone and the rest of the plasma membrane, Cav1 restricts EGFR mobility, but only in the absence of the Mgat5/galectin lattice.
In both Mgat5 and Mgat5 cells, Cav1 migrates at the bottom of a 5–40% sucrose gradient (; ), which is in contrast to monomeric RhoA that migrates in lower density fractions (). In blue native gels, Cav1 migrates at ∼500 kD in both cell lines (), corresponding to the high molecular mass oligomers of Cav1 reported previously (; ). Therefore, in spite of the reduced expression of Cav1 and caveolae, Cav1 still forms stable oligomers in Mgat5 cells. To compare the behavior of Cav1 at different expression levels, Mgat5 cells were transfected with Cav1-mRFP, and the diffusion of Cav1 was tested by FRAP (). The transfected cells were subsequently fixed and labeled for Cav1 to compare transfected Cav1-RFP expression levels with endogenous Cav1 levels in Mgat5 and Mgat5 cells. Even when expressed at low levels, below those of Mgat5 cells, Cav1 remained highly immobile.
of recovery increased significantly (P < 0.01) with Cav1 intensity. These results indicate that even when expressed at low levels, Cav1 still forms stable oligomers at the plasma membrane.
Mgat5 cells were transfected with mutants of Cav1, either a Y14F mutation of the tyrosine phosphorylation site or an F92A/V94A mutation of the scaffolding domain (; ). Cav1-dependent inhibition of cell surface diffusion of both CT-B and EGFR-YFP ( and ) and of EGFR signaling () in Mgat5 cells is independent of Y14F phosphorylation but requires an intact scaffolding domain. Colocalization of EGFR and Cav1 is similarly increased in Mgat5 cells transfected with wild-type Cav1 and Cav1-Y14F, but the Cav1 scaffolding domain mutant shows reduced colocalization with EGFR ().
The mobile EGFR-YFP fraction was significantly greater (P < 0.01) and the rate of recovery was significantly faster (P < 0.01) after photobleaching in Mgat5 cells treated with lactose than in untreated or sucrose-treated cells and was comparable with that observed in Mgat5 cells ( and ). Lactose treatment of Mgat5 also reduces Cav1 levels (). To determine whether reduced Cav1 levels are responsible for increased EGFR-YFP mobility, Cav1-transfected Mgat5 cells were treated with lactose. Cav1 overexpression increased the recruitment of EGFR to the immobile fraction but did not affect the first-order diffusion rate of EGFR compared with cells treated with lactose alone ( and ). Disrupting the lattice with lactose did not alter CT-B diffusion in Mgat5 cells, suggesting that GM1 is not restricted by galectins (). Mgat5 retroviral rescue of Mgat5 cells did not impact EGFR-YFP mobility, but the rescue of Mgat5 cells, which did not restore Cav1 levels, reduced the rate of recovery and the mobile fraction of EGFR-YFP ( and ). This confirms that galectin binding to N-glycans restricts the rate of diffusion of EGFR-YFP.
Colocalization of EGFR with Cav1 is increased in Mgat5 cells relative to Mgat5 cells, and disruption of galectin binding in Mgat5 cells with lactose but not sucrose increases EGFR colocalization with Cav1 to levels observed in Mgat5 cells (). To test Cav1 association with EGFR in live cells, Mgat5 and ESC-Rescue cells were cotransfected with Cav1-CFP and EGFR-YFP, and time-lapse videos were acquired every 10 s over 5 min. Cells were fixed and labeled for Cav1 to determine relative Cav1 levels, and the mean colocalization of Cav1-CFP and EGFR-YFP was determined over the course of the video ( and Videos 1–4; available at ). In ESC-Rescue cells, Cav1 association with EGFR was significantly lower (P < 0.01) than in Mgat5 cells, irrespective of Cav1 levels. These data are consistent with the competitive exchange of EGFR between two cell surface domains: a signaling-competent Mgat5-dependent lattice and a negative regulatory Cav1-enriched microdomain.
To assess the role of the actin-based membrane skeleton on EGFR-YFP diffusion, the actin cytoskeleton was disrupted by treatment with latrunculin A (LatA). Phalloidin labeling of LatA-treated cells shows a loss of actin stress fibers and a reduction of total F-actin in Mgat5 and Mgat5 cell lines (). Disruption of the actin cytoskeleton with LatA significantly increased (P < 0.05) the mobile fraction of EGFR-YFP in Mgat5 cells ( and ). The effect of LatA on EGFR stabilization in Mgat5 was also observed in Mgat5 cells overexpressing Cav1 or transfected with Cav1 siRNA ( and ). LatA treatment had no effect on either the rate of diffusion or the immobile fraction of EGFR in Mgat5 cells treated with lactose, or in either Mgat5 cell line (). Therefore, lattice-associated EGFR shows preferential interaction with the actin cytoskeleton relative to EGFR in the absence of the lattice. However, disruption of the actin cytoskeleton with LatA does not alter the first-order rate of EGFR diffusion. This suggests that the galectin lattice and F-actin act together to regulate the mobility of the non-Cav1–associated fraction of EGFR.
In this study, we show that negative regulation of EGFR through recruitment to Cav1 microdomains is opposed by the expression of β1,6GlcNAc-branched N-glycans. Therefore, Mgat5 and Cav1 interact to regulate growth signaling and tumor progression. The galectin lattice impedes the diffusion rate of EGFR, confirming our earlier report that the lattice represents a surface microdomain that limits EGFR down-regulation by endocytosis (). Lactose-mediated disruption of the galectin lattice in wild-type Mgat5 cells increases EGFR first-order diffusion, whereas restoration of the lattice by Mgat5 expression in Mgat5 cells restricts EGFR diffusion independently of Cav1 expression. Cav1 overexpression reduces EGFR in the mobile fraction but does not suppress the effect of lactose on the first-order rate of diffusion of EGFR dynamics. The Cav1 microdomain and galectin lattice are therefore distinct cell surface domains that differentially regulate the distribution and dynamics of EGFR.
Deletion of the EGFR cytoplasmic domain did not impact EGFR lateral mobility, leading to the suggestion that extracellular interactions constrain the lateral diffusion of EGFR (). Mgat5 deficiency and lactose competition have previously been shown to inhibit galectin binding to EGFR (), identifying a requirement for extracellular galectin binding to N-glycans in the regulation of EGFR diffusion at the cell surface. We envisage that recruitment to the galectin lattice reduces the propensity of EGFR to perform hop diffusion. Disruption of the actin cytoskeleton with LatA increased the mobile fraction but not the rate of diffusion of EGFR-YFP, distinguishing its action from lactose-mediated disruption of the lattice. Although we cannot exclude the possibility of incomplete disruption of the submembrane actin cytoskeleton by LatA (), our data are consistent with a role for membrane protein density as a key regulator of diffusion rates in biological membranes (; ). Importantly, the effect of actin cytoskeleton disruption was not observed in Mgat5 or Mgat5 cells treated with lactose where the galectin lattice is reduced. Therefore, galectin-bound EGFR is also stabilized by the actin cytoskeleton. It is likely that galectin cross-links EGFR to other actin-associated membrane glycoproteins, generating actin- stabilized signaling domains ().
It is important to note that the scale of measurement with respect to both time and domain size using FRAP is dramatically larger than that measured by single-particle tracking (). Thus, it is not clear whether the Cav1- dependent immobilization of EGFR that we have measured by FRAP is equivalent to the cholesterol-dependent transient anchorage observed by single-particle tracking (; ; ; ; ). The immobile fraction detected by FRAP in our study reflects those EGFR-YFP molecules whose interaction with other plasma membrane components, such as Cav1, the galectin lattice, or the actin cytoskeleton, constrains their ability to exchange freely with fluorescent EGFR-YFP outside the bleached region. This may not necessarily reflect stable or even direct molecular interactions but rather the preferential recruitment of EGFR to microdomains that restrict protein exchange across membrane barriers ().
Activation of EGFR has been shown to occur in noncaveolar raft domains that associate with nascent coated pits (). Similarly, in Mgat5 cells, blocking coated pit endocytosis by K depletion leads to the precocious activation of Erk and growth signaling that can be suppressed by the disruption of rafts with nystatin (). Negative regulation of EGFR diffusion and signaling by Cav1 oligomers is consistent with the previously reported stable interaction of EGFR with Cav1 and caveolae (; ; ). However, the ability of Cav1 to form immobile oligomers that associate with and regulate EGFR diffusion and signaling at levels below the threshold for caveolae formation argues that oligomerized Cav1 can functionally sequester EGFR independently of caveolae formation. Indeed, in endothelial cells, the overexpression of Cav1 inhibits endothelial nitric oxide synthase activity without increasing caveolae expression, suggesting that a pool of Cav1 outside of caveolae may be responsible (; ).
Freeze-etch experiments have identified a striated caveolin coat on flat membrane domains as well as caveolae (), and threshold levels of Cav1 in cell surface domains are required for caveolae formation (). Furthermore, in contrast to the caveolae-rich basolateral surface of MDCK cells, the apical surface expresses Cav1 but no caveolae (). Caveolin forms stable oligomers (; ), and the caveolin coat of vesicular transporters is highly stable (). Cav1 regulation of EGFR signaling and dynamics in Mgat5 cells that express few caveolae suggests that the regulatory function of Cav1 is dependent on Cav1 oligomerization but not necessarily on caveolae formation.
Cav1 oligomers show a reduced mobility relative to larger (more intense) Cav1 structures (), perhaps reflecting increased dynamics and exchange of Cav1 in caveolae. However, in the absence of Mgat5 expression, Cav1 at varying expression levels functions equivalently to regulate the diffusion of CT-B and EGFR as well as EGFR signaling. In blue native gel analysis, Cav1 in Mgat5 cells migrates as a sharp band, which is indicative of a highly stable oligomeric configuration. Similar SDS stable oligomers were predicted to contain 15 caveolin molecules (). This is considerably less than the predicted 145 Cav1 molecules per caveolae () and is consistent with the reduced intensity and size of the Cav1 spots detected in cells expressing reduced levels of Cav1, such as Mgat5 cells. The stable interaction of EGFR with Cav1 oligomers argues that these domains form a stable platform for the recruitment of receptors and other interacting proteins. Although the spatial relationship of Cav1 oligomers, the galectin lattice, and the membrane skeleton remains uncertain, we suggest that the reduced mobility of both Cav1 oligomers and the galectin lattice is caused by the reduced ability of proteins and lipids recruited to these macromolecular domains to undergo hop diffusion ().
Spontaneous down-regulation of Cav1 in Mgat5 tumor cells argues that these conditions select for relief from the Cav1- mediated negative regulation of signaling at the cell surface. In Mgat5 cells, inhibition of the lattice with swainsonine or lactose treatment reduced Cav1 expression, whereas the Mgat5 rescue of Mgat5 cells restored Cav1 levels. This suggests that β1,6GlcNAc branching is an upstream regulator of Cav1. The inability of Mgat5 rescue to restore Cav1 levels in Mgat5 cells is suggestive of Cav1 loss caused by a stable genetic change. Moreover, the inverse correlation between Cav1 levels and tumor size in Mgat5 tumors suggests that reducing Cav1 expression is one mechanism that can relieve growth restriction imposed by Mgat5 deficiency.
PyMT Mgat5 mice display a dramatic reduction in the incidence of tumor metastasis, even in those animals that develop escaper fast growth tumors (). In this regard, although responsiveness to EGF is largely restored by Cav1 suppression in Mgat5 cells, it does not rescue the lattice-dependent deficiency in TGF-β signaling, EMT, or fibronectin fibrillogenesis. This suggests that Mgat5 and β1,6GlcNAc-branched N-glycans play additional roles that are distinct from Cav1 regulation in tumor cell polarity, motility, and invasion ().
In contrast to Mgat5 tumor cells, EGFR diffusion and responsiveness to EGF are not altered by either the overexpression or partial knockdown of Cav1 expression in Mgat5 cells. This suggests that β1,6GlcNAc-branched N- glycans and lattice retention of EGFR override negative regulation by Cav1. Although Mgat5-deficient tumor cells are partially depleted of surface EGFR as a result of constitutive endocytosis (), reducing Cav1 levels in the cells appears to compensate by increasing the availability of EGFR to ligand-dependent activation. We conclude that unlike Mgat5 cells, in which Cav1 expression is significantly higher (P < 0.05), Cav1 levels in Mgat5 cells are below the threshold required for suppression, leaving an estimated 10–15,000 surface EGFRs available for optimal activation of the MAPK activation (; ). Surface residency of EGFR in the lattice is therefore permissive for ligand activation and limits both constitutive endocytosis () and sequestration by inhibitory, immobile Cav1 domains (). Thus, Cav1 depletion enhances the availability of surface EGFR on Mgat5 cells, thereby removing a negative regulator of growth (). We suggest that Cav1 loss compensates for an approximately fivefold decrease in surface EGFR numbers observed in Mgat5 cells () and is thus epistatic for EGF sensitivity.
(∼2 h) compared with >10 surface EGFRs with a
> 6 h. TβRII has few N-glycans ( = 2) compared with EGFR ( = 8) and is therefore relatively more dependent on the branching of its N-glycans for residency in the lattice at the cell surface (). In contrast, EGFR has both a greater affinity for the lattice as a result of the higher N-glycan number and a greater sensitivity to regulation by Cav1 microdomains (). Therefore, our data argue that affinity for the galectin lattice and Cav1-enriched microdomains partners with receptor endocytosis rates to determine receptor availability to ligand.
Transgenic mice deficient in Mgat5 expression were crossed onto PyMT transgenic mice on a 129sv × FVB background (). Mammary tumor samples were dissected from Mgat5 and Mgat5 mice and snap frozen on dry ice for subsequent protein extraction. Cell lines were established from solid mammary carcinoma samples dissected from either Mgat5 or Mgat5 genotypes. The cell lines used herein are designated Mgat5(2.6), Mgat5(2.8), Mgat5 (also called Mgat5(22.9)), and Mgat5 (also called Mgat5(22.10)). Mgat5 and Mgat5 cells genetically rescued by infection with a pMX-PIE retroviral vector for the expression of murine Mgat5 (designated Rescue and ESC-Rescue, respectively) were selected by growth in medium containing 1 μg/ml puromycin (). All cell lines were grown in complete medium containing DME supplemented with 10% FBS, nonessential amino acids, glutamine, vitamins, and penicillin/streptomycin in a 5% CO/air incubator at 37°C. For signal transduction experiments, cells were rinsed twice and incubated overnight in serum-free DME at 37°C before performing the experiment. Disruption of the actin cytoskeleton was performed by treating cells with 0.5 μM LatA in complete DME for 20 min at 37°C before experiments.
Adenovirus expressing myc-tagged Cav1 under control of the tetracycline-regulated promoter was used to infect cells for 48 h as previously described (; ). Infected cells were visualized using anti-myc (Santa Cruz Biotechnology, Inc.) antibody. To knock down Cav1 expression, cells were cultured in complete medium for 2 d before transfection with specific mouse Cav1 siRNA oligonucleotides or with control siRNA (Dharmacon, Inc.). In brief, cells were rinsed twice with serum-free DME without antibiotics, transfected with siRNA for 4 h using Dharmafect 3 transfection reagent (Dharmacon, Inc.), washed twice with complete DME, and incubated in complete media for 48 h.
Blue native gels were performed as described previously (). In brief, cells were lysed at 4°C in lysis buffer (500 mM 6-amino caproic acid, 2 mM EDTA, and 25 mM Bistris, pH 7.0) containing 120 mM -octyl-glucoside for 30 min. Lysates were clarified by centrifugation at 13,200 rpm for 10 min. Supernatant were mixed with 1/10 vol of sample buffer containing 5% R-250 Coomassie blue and 1/10 vol glycerol. Proteins were separated on linear 4–15% acrylamide gels run at 100 V at 4°C until the dye reached the middle of the gel. Blue cathode buffer (50 mM Tricine, 15 mM Bistris, and 0.02% R-250 Coomassie blue) was then replaced with clear cathode buffer (with no Coomassie blue), and gels were run at 200 V until the dye reached the bottom of the gels. Proteins were then transferred to polyvinylidene difluoride membrane and processed for immunoblotting with Cav1 polyclonal antibody (Santa Cruz Biotechnology, Inc.).
Cells were rinsed with 0.1 mM sodium cacodylate, pH 7.3, fixed for 1 h with 2% glutaraldehyde at 4°C, rinsed with cacodylate buffer, scraped from the Petri dish, pelleted, and postfixed with 2% osmium tetroxide at 4°C. The cells were dehydrated and embedded in LR-White resin. Ultrathin sections were prepared, contrasted with uranyl acetate and lead citrate, and visualized with a CM902 or a H7600 transmission electron microscope (Carl Zeiss MicroImaging, Inc. or Hitachi, respectively). Smooth caveolar invaginations and clathrin-coated vesicles within 100 nm of the plasma membrane were counted per cell profile as previously described ().
Cells were plated in 96-well plates at 5,000 cells/well or on coverslips, serum starved for 24 h, and stimulated with EGF or TGF-β1 in DME plus 0.2% FBS. After various times with cytokine, cells were fixed for 10 min with 3.7% formaldehyde at 20°C, washed with PBS plus 1% FBS, and permeabilized using 100% MeOH for 2 min. The cells were washed three times and blocked in PBS plus 10% FBS for 1 h at 37°C. Mouse anti–phospho-Erk1/2 (Thr202/Tyr204; Sigma-Aldrich) or mouse anti-Smad2/3 (Transduction Laboratories) was added at 1/1,000 in PBS plus 10% FBS and incubated overnight at 4°C. The cells were washed three times with PBS plus 1% FBS and AlexaFluor488-labeled anti–mouse Ig (Invitrogen) added at 1/1,000 with Hoechst (1/2,000) for 1 h at 20°C. After washing three times, the plates were scanned using an ArrayScan automated fluorescence microscope (Cellomics Inc.). The difference in nuclear and cytoplasmic staining intensity was determined individually for 100 cells per well, and subtraction of total nuclear intensity values from cytoplasmic intensity values was used to represent the change in activation after the addition of cytokine. The SEM ( = 100) was generally <4% at each assay point.
Alternatively, cells were plated on coverslips for 24 h and transfected with either myc-tagged Cav1, Cav1Y14F, or Cav1F92AV94A or with Cav1 siRNA or control siRNA for 2 d. Cells were serum starved for 24 h before stimulation with 100 ng/ml EGF for 5 min and were fixed and labeled with mouse anti–phospho-Erk1/2 and either rabbit anti-myc (Santa Cruz Biotechnology, Inc.) or rabbit anti-Cav1 (Santa Cruz Biotechnology, Inc.) followed by Hoescht staining. Confocal images of cells mounted in Celvol 205 (Celanese Ltd.) were acquired on a confocal microscope (Fluoview 1000; Olympus) with a Uplan Apochromat 1.35 NA 60× objective (Olympus) with equivalent acquisition settings. The mean intensity of nuclear phospho-Erk was quantified by creating a mask based on Hoescht staining using ImagePro Plus software (Media Cybernetics, Inc.). Data from three independent experiments (>36 cells/condition) were compiled and normalized to Mgat5 cells stimulated for 5 min with EGF.
FRAP analysis of cells incubated with CT-B for 3 min or transfected with EGFR-YFP was performed in regular culture media without phenol red at room temperature. Images were acquired on a confocal microscope (FV1000; Olympus) with a Uplan Apochromat 1.35 NA 60× objective (Olympus) and fully opened pinhole. Photobleaching of CT-B–FITC was performed using 10 scans with the 488-nm laser at full power within a square area 20 pixels wide. EGFR-YFP photobleaching experiments were performed using 20 scans of a 405-nm scanner laser (SIM; Olympus) at full power within a circular region of interest of 27-pixel diameter. To study the effect of Cav1 on CT-B diffusion, cells were transfected with Cav1-mRFP, Cav1Y14F-mRFP, Cav1F92AV94A-mRFP, or Cav1 siRNA 2 d before experiments. Disruption of the galectin lattice was performed by treating the cells with 20 mM β-lactose or sucrose for 2 d. To study EGFR diffusion, cells were plated for 6 h and transfected with EGFR-YFP. The next day, cells were transfected with Cav1 or control siRNA or infected with Cav1 adenovirus. After 4 h, the media was changed to complete DME or complete DME containing 20 mM β-lactose or sucrose for 2 d. Recovery data (six to eight cells from each of three independent experiments) were analyzed with Prism software (GraphPad) using nonlinear regression with a bottom to bottom + span algorithm.
of recovery and mobile fraction were calculated as previously described (). For experiments using myc-tagged Cav1 constructs, cells were cotransfected with either EGFR-YFP or pOCT-dsRED to visualize the transfected cells. Similarly, recovery data for transfected Cav1-mRFP in Mgat5 cells were obtained by FRAP analysis at room temperature. Images were acquired with equivalent acquisition settings, and, to compare Cav1-mRFP intensity with endogenous Cav1 levels, Mgat5-, Mgat5-, and Cav1-mRFP–transfected Mgat5 cells were fixed and labeled in parallel with Cav1 polyclonal antibody (Santa Cruz Biotechnology, Inc.). A graph of Cav1 intensity versus Cav1-mRFP intensity was generated for fixed Mgat5 cells, and a linear regression was performed to determine the intensity of Cav1-mRFP relative to endogenous Cav1 levels in Mgat5 cells in the bleach zone of live cells that underwent FRAP analysis. FRAP data are presented in function of normalized Cav1-mRFP intensity for both the mobile fraction and half-time of recovery. r values were calculated from a linear regression performed with Prism software (GraphPad).
L-PHA and Cav1 expression levels were quantified from fluorescent images of cells mounted in Celvol 205 (Celanese Ltd.) acquired with the 60× 1.35 NA Uplan Apochromat objective (Olympus) of a confocal microscope (Fluoview 1000; Olympus). Quantification of L-PHA–FITC levels (mean density of fluorescence) was performed using ImagePro Plus software (Media Cybernetics, Inc.) from confocal images acquired with equivalent acquisition settings. Values from three independent experiments were normalized to the intensity of Mgat5 cells, and significance was determined by a test. To measure EGFR colocalization with Cav1, cells were plated on glass coverslips and preincubated for 48 h with 20 mM β-lactose or sucrose in complete DME, fixed, and labeled with rabbit anti-EGFR, mouse anti-Cav1 (Invitrogen), and Hoechst. Images were acquired with a 100× NA 1.4 Plan Apochromat objective (Olympus) of a microscope (DeltaVision Restoration; Olympus), and 64-layer stacks were acquired and deconvolved with softWoRx image analysis software (Applied Precision). Colocalization was quantified in Photoshop (Adobe) by defining a box of set dimensions and scoring the incidence of yellow stain within this box from six randomly selected regions within the cytoplasm of the cell. Alternatively, cells transfected with myc-tagged Cav1 constructs were labeled with anti-myc (Upstate Biotechnology) and anti-EGFR. From confocal images of cells mounted in Celvol 205 (Celanese Ltd.) acquired with the 60× 1.35 NA objective of a confocal microscope (FV1000; Olympus), the relative intensity of myc-Cav1 associated with EGFR labeling was determined using the colocalization coefficient of ImagePro Plus imaging software (Media Cybernetics, Inc.).
To quantify EGFR-Cav1 association in live cells, Mgat5 and ESC-Rescue cells were cotransfected with Cav1-CFP and EGFR-YFP. Time-lapse images of cells were acquired every 10 s for 5 min in regular culture media without phenol red at room temperature with the 60× 1.35 NA Uplan Apochromat objective of the FV1000 confocal microscope. Cav1 acquisition settings were kept constant, and high and low Cav1-CFP–expressing cells were determined relative to endogenous Cav1 levels in Mgat5 cells. The Pearson's coefficient for Cav1-CFP and EGFR-YFP was calculated from two random regions of the cell for each individual time frame, and the mean Pearson's coefficient over time was determined from three independent experiments ( >15) using ImagePro Plus software.
Cav1 oligomerization was determined using velocity sucrose gradient centrifugation as previously described (). In brief, cells were grown in 100-mm Petri dishes and lysed on ice in 500 μl of lysis buffer (25 mM MES, pH 6.5, 150 mM NaCl, 60 mM -octylglucoside, and protease inhibitor cocktail). This lysate was overlaid on top of 4.2 ml of a 5–30% discontinous sucrose gradient prepared in the same lysis buffer. The gradients were centrifuged in an SW55 rotor (Beckman Coulter) for 6 h at 53,000 rpm. 12 equal fractions of 392 μl were collected from the top of the gradient, and an equal volume of each fraction was analyzed by SDS-PAGE and transferred onto nitrocellulose membranes for immunoblotting with anti-Cav1 (Santa Cruz Biotechnology, Inc.) or anti-RhoA (Santa Cruz Biotechnology, Inc.) antibodies.
Videos 1–4 correspond to C. Mgat5 and ESC-Rescue cells were transfected with Cav1-CFP (red) and EGFR-YFP (green). Videos are of Mgat5 cells expressing high (Video 1) and low (Video 2) Cav1 levels and ESC-Rescue cells expressing high (Video 3) and low (Video 4) Cav1 levels. Online supplemental material is available at . |
As an undergrad, I read an interview in the Dutch equivalent of with one of the leading Dutch immunologists, Jon van Rood. I wrote him a letter asking whether I could work in his lab, but I never got an answer.
At the time, I was working for my undergraduate advisor, whose interests were in bacterial cell walls and membrane structure. He had some financial resources to allow his graduate students to go abroad to learn techniques and bring these back to his lab.
Since his two graduate students weren't particularly interested in this opportunity, he turned to me. He wrote letters explaining the purpose to a number of investigators, mostly in the U.S. One of the few who wrote back in the affirmative was Jack Strominger, who was working at Harvard on bacterial cell wall synthesis.
To my surprise—this was before Internet days—when I arrived, it turned out that half the Strominger lab was working on the biochemistry of transplantation antigens. Then and there, it became clear to me that that would be a great topic to return to. My six-month visit concluded with an agreement that I'd return as a graduate student and work on transplantation antigens.
For that, I got some financial support from the Dutch government. And who should be sitting on the selection committee but van Rood. As luck would have it, he even became my formal thesis advisor. I did the practical work for my graduate studies in Jack Strominger's lab, but defended my thesis in the Netherlands.
In 1980, I left the Strominger lab and took my first independent position.
Correct. My thesis project consisted of cloning cDNA for a major histocompatibility antigen. I think that work attracted the attention of Klaus Rajewsky, who was the head of immunology at the University of Cologne in Germany. He wondered whether I would be interested in taking a junior group leader position at that institute.
I think the first students you work with suffer the consequences. You have no experience leading a lab, you make all the mistakes that you can make. Objectively speaking, not much was produced, but I do think I learned how to run a lab.
No, my next slot was at the Netherlands Cancer Institute. I was invited to join in 1984 by Piet Borst. He had just become the institute's director and was interested in infusing some fresh blood. I led a group in the cell biology and biochemistry of antigen presentation.
In '92 I was approached by Susumu Tonegawa. He asked whether I might be interested in moving to MIT. That was an opportunity that I couldn't turn down.
I stayed there until '97, when Harvard Medical School wanted to know if I was interested in moving there. I had gotten to know several colleagues there when we served as editors for . I also had a very strong interest in the didactic missions of school, and they were looking for a person to take charge of the graduate program in immunology.
That lasted until 2005, and then I was recruited to Whitehead.
I think the scientific climate in the U.S. to this day is more vibrant and lively than most places in Europe. I'm not dinging Europe in the least, I enjoyed every moment I lived in Amsterdam, but I think the intensity with which people pursue their science is just on a different level. 15 years ago, that was certainly true, although the difference seems to be less pronounced today. The resources available, the intellectual firepower, the Boston academic climate with MIT, Harvard, and Harvard Medical School, it's just beyond compare.
Yes, I do. As it happens, I like to fish. I would say Boston is the saltwater fly-fishing capital of the world. I co-own a little fishing boat, , with my Harvard colleague, Fred Alt. We go out for striped bass, bluefish, bluefin tuna with varying degrees of success.
There are several issues associated with working at a medical school. Many have hospitals affiliated with them, and the economics and logistics of that imposes a certain organizational structure. The primary mission of a medical school is naturally to train doctors. To give you an example, I had to talk to the head of our public affairs department two years in a row before the annual Dean's report began highlighting the various graduate programs. It would showcase the medical students and the M.D./Ph.D. students, but the graduate programs seemed like an afterthought.
I felt that the Whitehead/MIT environment was more curiosity driven and less concerned with medically relevant research. There's nothing wrong with that in principle, but for me, it's like the difference between Italian and French cuisine. They're both outstanding if properly prepared, but some people prefer French, others, Italian. It has nothing to do with quality per se, just a difference in style.
My recent research interests are also better aligned with what MIT is really good at: chemistry, materials science, engineering, and so forth. Lately, I've become more and more interested in engineering-based approaches to biological questions.
If you enter the building where the mechanical engineers are, and you see the posters on the wall, you'll see so much ingenuity, such diversity of approaches. I'm always struck by the eagerness of faculty in other departments to get involved, get their hands dirty, think about possibilities for joint projects. There's a growing awareness that interdisciplinary interactions—where one can combine, say, physics with biology—can produce so much more. Compare that with your typical medical school, where everything of necessity is focused on either basic biology or translational stuff.
Yes, that is perhaps unusual. Most graduate students change tack when they do a postdoc and maybe again once they become independent, but I haven't really felt the need. The work I started as a graduate student still continues today, in a different incarnation, of course. Ever since I started working on these MHC products, I've been interested in how they're put together from their building blocks, how they travel from the site of synthesis to their destination.
True, I've never even applied for a job. I have been in a very privileged position, but I think a big part of it is just being enthusiastic about what you do.
What I've learned is that the single most important thing is passion. You also have to take pleasure in the details: the duplicates of a calibration curve falling right on top of one another or bands on a gel coming out razor sharp, in addition to real discoveries. To this very day, there's almost nothing I'd rather do than look at autorads. Get a life, right?
It's the most exciting profession I could imagine. It's like a stem cell, continuously self-renewing. The next day is almost always as exciting as the preceding one.
It's hard to imagine finding a working environment that is so dependent on the social aspects. Many students are under the mistaken impression that being a scientist means being somewhat of a recluse. Some future medical students think that because they see patients, they get a richer experience of interpersonal interactions. Sure, if you enjoy dealing with HMOs rather than working with your peers, debating puzzles, and thinking creatively… But I know which side my bread is buttered on. |
Cellular fate is predominantly determined by the processes of division, differentiation, and death. A cell is considered dead when the plasma membrane has lost its integrity or when it is fragmented into so-called apoptotic bodies (), but a plethora of definitions based on morphological parameters tries to capture the manifold types of mammalian cell death and the routes toward it. Originally, apoptosis was described as the type of cell death characterized by rounding and shrinking of the cell, chromatin condensation (pyknosis), nuclear fragmentation (karyorrhexis), and budding of discrete plasma membrane–lined portions of cytoplasm (blebbing; ). During apoptosis in mammalian tissues, the plasma membrane remains intact until late stages, thereby preventing an unwanted inflammatory response (). Autophagy is defined by a vacuolization of the cytoplasm. Autophagic vacuoles have two membranes and contain degenerating organelles and cytosolic content (). The third main type of cell death, necrosis, is characterized by cell swelling (oncosis), organelle dilation, and subsequent rupture of the plasma membrane ().
The acquisition of apoptotic morphology is, in most cases, associated with and depends on the activation of Cys-dependent Asp-specific peptidases (caspases; ; ). Caspases (clan CD, family C14) cleave their substrates after Asp, are synthesized as inactive zymogens, and can be divided into two types based on their overall structure and activation modes. Effectors or executioner caspases are activated by proteolytic separation of the large (p20) and small (p10) subunits, resulting in active (p20)(p10) heterotetramers. Initiator caspases have an N-terminal extension, the prodomain, that is needed to recruit them into protein complexes that function as activation platforms, called apoptosomes (). Their activation does not require proteolytic cleavage but relies on conformational changes after oligomerization (). Once triggered, initiator caspases can ignite a cascade by the proteolytic activation of effector caspase zymogens. The effector caspases ultimately cleave numerous substrates, thereby causing the typical morphological features of apoptosis (; ). Members of the CD clan of proteases are characterized by their specificity for the residue at the N-terminal side of the scissile bond, the P1 residue, in their substrates. For caspases, substrate recognition additionally requires three or more residues N terminal to P1-Asp. Based on the optimal substrate oligopeptide sequence, caspase activity can be specifically measured by synthetic peptides C-terminally coupled to a fluorogenic moiety, such as 7-amido-4-methylcoumarin (AMC). Upon cleavage by caspases, an increase of fluorescence is proportional to caspase activity. Despite their omnipresence during apoptosis, caspases are also involved in nonapoptotic events, including inflammation, cell proliferation, and cell differentiation. Therefore, the reciprocal conclusion that caspase activities are strictly correlated with apoptosis is invalid ().
As in animals, cell death is an essential part of the life cycle of plants. From seed germination until seed production, developmental cell death is manifested. A few well known examples are cell death during terminal differentiation of the vascular tracheary elements, leaf and flower senescence, elimination of reproductive organs in unisexual flowers, pollen rejection in the self-incompatibility response, fruit dehiscence, or pod shattering (). In addition, plants attempt to block the invasion of biotrophic pathogens via the hypersensitive response, leading to localized cell death at the site of infection (). Typical animal apoptotic features such as pyknosis, karyorrhexis, internucleosomal DNA cleavage, cell shrinkage, and the formation of apoptotic bodies have been observed in dying plant cells (). Importantly, during cell death, caspaselike activities are easily detected by synthetic fluorogenic oligopeptide substrates, and cell death can often be attenuated by synthetic caspase-specific inhibitors (). With the sequencing of the complete genome of the model plant (), these caspaselike activities have steered an intensive but frustrating search for caspase genes within plants.
italic
#text
In Eukaryota, paracaspases and caspases are restricted to animal genomes (kingdom Animalia), and metacaspases are present in the kingdoms Protozoa, Fungi, Plantae, and Chromista, whereas in Prokaryota, meta-/paracaspase-like proteins are found in both Archaea and Eubacteria (). Previous phylogenetic analysis of eukaryotic caspases, metacaspases, and paracaspases has suggested that these groups are about equally distant from each other. These findings have led to the hypothesis that eukaryotic metacaspases originate from a horizontal gene transfer (HGT) between the mitochondrial endosymbionts, α-proteobacteria, and the early eukaryotes (). Still, the origin of caspases and paracaspases remains elusive in such a scenario. Furthermore, meta-/paracaspase-like proteins can be found not only in α-proteobacteria but also in all groups of Bacteria, including cyanobacteria, the ancestors of chloroplasts in plants. Also, it is striking that only type I metacaspases can be found in Protozoa, Fungi, and Chromista, whereas both type I and II are present in Plantae, including green plants, glaucophyta, and rhodophyta. Therefore, an alternative hypothesis would be that caspases, paracaspases, and type I metacaspases have a common ancestor originating from HGT between mitochondrial endosymbionts and host eukaryotic cells. Type II metacaspases might possibly be derived from a second HGT event during the establishment of plastids from endosymbiotic cyanobacteria.
The alleged paracaspase of is a surprising element in this phylogenetic distribution because slime molds belong to the Protozoa kingdom. Phylogenetic analysis of the sequence of its putative catalytic p20 subunit reveals that it is almost equally related to that of caspases, metacaspases, paracaspases, and their bacterial homologues, making its classification as a separate paracaspase not well founded (our unpublished data). Also, its prodomain lacks a death domain and Ig domains, which is typical of animal paracaspases. Therefore, it is tempting to classify the protease as a metacaspase rather than a paracaspase.
In the genome, nine metacaspase genes are present: three of type I ( metacaspase 1 [] to ) and six of type II ( to ; ). Upon overproduction in , type II metacaspases autoprocess and display a Cys-dependent proteolytic activity against synthetic P1-Arg substrates, whereas AtMC9 also cleaves P1-Lys substrates, albeit with low efficiency (, ; ). Type I metacaspases from do not autoprocess upon recombinant overproduction and, like mammalian initiator caspases, possibly require induced oligomerization within an activation platform (). A positional scanning synthetic combinatorial library screening with purified recombinant AtMC9 confirmed the preference for P1-Arg. The optimized tetrapeptide substrate Ac-Val-Arg-Pro-Arg-AMC had a / of 4.6 × 10 M s and, thus, can be considered a very efficient substrate for AtMC9 ().
The P1 preference of clan CD proteases is dictated by conserved amino acids distributed throughout the mature protease that together form the S1 pocket (). In caspases, Arg, Gln, and Arg (according to caspase-1 residue numbering) form a basic S1 pocket for optimal binding of the acidic P1-Asp within their substrates (). When the available sequences of eukaryotic metacaspases, paracaspases, and bacterial meta-/paracaspase homologues are aligned to those of animal caspases, Gln is replaced by an Asp, and Arg is replaced by Asp or Glu in both para- and metacaspases. Arg of caspases aligns to Leu. Six residues more C terminal, a highly conserved Asp is present that aligns with Asp of bacterial gingipain R, which is known to coordinate binding of the P1-Arg of substrates of this peptidase (). Together, these residues are ideally positioned to create a highly acidic S1 pocket that is perfectly suited to accept the basic P1 residues Arg and Lys (). As the predicted S1 pocket–forming residues are strictly conserved in all known sequences of para- and metacaspases, the Arg/Lys specificity is very probably shared by all of them. The determined P1 specificity of metacaspases of other plants and of yeast and protozoa confirmed this hypothesis (; ; ).
The fact that no close bacterial caspase homologues have been identified yet would reflect an animal-specific evolutionary process of gene duplications and progression of the caspases from Arg/Lys toward Asp specificity. Until now, attempts to detect the protease activity of paracaspases have been unsuccessful (), but paracaspases may have retained their preference for basic P1 residues in their substrates. The reason for the shift in caspase P1 specificity remains unclear.
A biochemical particularity of AtMC9 is the presence of a second catalytic Cys, Cys. Mutation analysis revealed that the primary Cys is necessary for autocatalytic processing and concomitant activation of AtMC9. However, once activated either autocatalytically or by exogenous AtMC9, proteolytic activity almost completely depends on Cys because replacement of this residue by Ala reduces protease activity by 99%. Furthermore, Cys but not Cys can be inactivated by S-nitrosylation. Thus, in the presence of nitric oxide, AtMC9 remains inactive until S-nitrosylation is reversed or until upstream proteases convert pro-AtMC9 into its mature form ().
The identification of metacaspase genes prompted the assessment of their potential involvement in cell death events in fungi, protozoa, and plants. The first data came from studies in baker's yeast (). Overproduction of the single metacaspase YCA1 resulted in autocatalytic processing and rendered cells more sensitive to exogenous or aging-related oxidative stress, as determined by reduced clonogenicity (). However, it may not be surprising that overproduction of an active protease, causing endogenous stress, resulted in a higher sensitivity to exogenous stress. A yeast strain with a disrupted gene () was also shown to be threefold less sensitive to HO, and ∼5% of the cells escaped from aging-related cell death (). Whether this observation reflects a direct involvement of YCA1 in cell death or this desensitization is caused by indirect effects, such as an altered protein turnover disturbing the balance of pro– and anti–cell death mediators, remains unclear. Indeed, after treatment of cells with HO, levels of oxidized proteins were much higher than those of wild-type cells (). Concomitantly, the proteasome activity of cells increased and apoptosis decreased upon HO treatment, as measured by phosphatidylserine (PS) externalization and DNA fragmentation. The reduced capability of cells compared with wild-type cells to cope with damaged proteins might explain the considerable decrease in cell viability after extended culture (i.e., >30 d; ). Whereas PS exposure and DNA fragmentation are genuine apoptotic markers, clonogenicity assays might also reflect other cellular states such as cell cycle arrest or metabolic deficiencies. Therefore, the clonogenicity results themselves do not exclude functions of metacaspases other than cell death involvement. In animal cells, PS exposure by dying cells functions as an “eat-me” signal for phagocytotic cells. However, the physiological function of PS exposure by yeast and plant cells remains intriguing because they possess a rigid cell wall and, thus, are incapable of phagocytosis.
Extracts of HO-treated YCA1-overproducing yeast were highly active toward the synthetic caspase substrates Val-Glu-Ile-Asp-AMC and Ile-Glu-Thr-Asp-AMC, suggesting that the YCA1 metacaspase behaved as a bona fide caspase (). These results were later contradicted: lysates from bacteria and HO-stimulated yeast overproducing YCA1 were not active against synthetic caspase substrates but cleaved P1-Arg and, to a lesser extent, P1-Lys substrates similarly to plant metacaspases (). Thus, YCA1 involvement cannot be determined by using synthetic caspase substrates or inhibitors. Because YCA1-independent cell death (; for review see ) and YCA1-independent caspaselike activities (; ) have been reported, the involvement of metacaspase activity in yeast cell death remains debatable (for review see ). The identification of endogenous YCA1 substrates will be crucial in unraveling the signaling pathways regulated by this metacaspase.
The genome of the pathogenic filamentous fungus contains two type I metacaspases, CasA and CasB. With double knockout mutants, neither of these metacaspases was found to be necessary for virulence. In addition, stress-induced cell death did not depend on metacaspases despite the abrogation of apoptosis-related membrane PS exposure in CasA and CasB double knockout stationary-phase cultures. Interestingly, both CasA and CasB were required for growth in the presence of agents inducing endoplasmic reticulum stress, suggesting a prosurvival role for metacaspases rather than an involvement in cell death processes ().
Of the five type I metacaspases of , only TbMCA4 caused retardation in growth, loss of respiratory competence, and subsequent decrease in clonogenicity when overproduced in baker's yeast (). Surprisingly, TbMCA4, like TbMCA1, lacks a catalytic Cys at the canonical location, although an adjacent Cys is present. Nevertheless, using different synthetic tetrapeptides with Asn, Asp, Arg, or Lys at the P1 position, no proteolytic activity in lysates of or yeast overproducing TbMCA4 could be demonstrated. Triple-null trypanosomes for , , and had no altered cell death or enhanced susceptibility to stresses, but the rapid down-regulation of all three genes with induced RNAi resulted in an in vitro growth arrest (). In , two metacaspase genes, and , have been reported. In untreated epimastigotes, the encoded proteins were distributed in the whole cell, but, after exposure to fresh human serum, which induces rapid apoptosis-like cell death, relocalization to the nucleus was observed. Upon overproduction of TcMCA5, epimastigotes of were more sensitive to fresh human serum–induced cell death (). Overproduction in yeast of the single type I metacaspase from , LmjMCA, slightly enhanced sensitivity toward HO, as measured by PS exposure. Interestingly, extracts of LmjMCA-overproducing yeast cells were proteolytically active toward P1-Arg synthetic substrates, demonstrating the probably universal preference for basic P1 residues of metacaspases (). has three type I metacaspases, of which PbMC2 and PbMC3 lack one or both of the catalytic site residues, but knockout mutants of the gene did not display any obvious phenotype ().
As discussed in the section Phylogeny of the caspases, metacaspases, and paracaspases, we propose to classify the slime mold paracaspase as a metacaspase. Upon starvation, differentiates into multicellular fruiting bodies consisting of a spore mass supported by a stalk. During this process, stalk cells die in a caspase-independent autophagic cell death (; ). Both differentiation and cell death were demonstrated to be independent of meta-/paracaspase action ().
In Norway spruce (), metacaspases were studied in the context of developmental cell death during in vitro somatic embryogenesis. In this process, dying embryo suspensor cells contained elevated activity against the synthetic fluorogenic caspase-6 substrate Val-Glu-Ile-Asp-AMC. Accordingly, treatment with the synthetic inhibitor Val-Glu-Ile-Asp-fmk prevented differentiation of the suspensor and subsequent suspensor cell death (). Disruption of the type II metacaspase gene abrogated the terminal differentiation and death of the suspensor cells and drastically reduced caspaselike activity, suggesting that mcII-Pa had caspase activity and was involved in cell death (). Later in vitro experiments have shown that mcII-Pa had Arg but not Asp specificity (). Because knocking down mcII-Pa not only disrupted cell death but also blocked embryonic differentiation, we speculate that mcII-Pa might be primarily involved in suspensor differentiation rather than in suspensor cell death. Possibly, mcII-Pa regulates the actin reorganization observed during suspensor differentiation (), like mammalian caspases do in the cytoskeletal rearrangements during apoptosis ().
In , mere constitutive overexpression or disruption of metacaspase genes does not lead to an obvious phenotype (; ; our unpublished data), and, thus, a role for metacaspases in cell death or other processes has not been identified yet. Redundancy may exist between the various members of this family, or additional factors may be necessary to activate ectopically expressed metacaspases. A large amount of microarray data is available () describing the expression of >20,000 genes (). Analysis of these data for the nine metacaspase genes could at least give a hint to their functional roles. Several metacaspase genes are strongly induced in senescing flowers, in response to various pathogens and elicitors, and during various abiotic stresses (; our unpublished data). As a prominent role for cell death has been demonstrated in responses to biotic and abiotic stresses, it might be tempting to deduce from these expression profiles that metacaspases play a role in cell death signaling. Alternatively, plants might first try to cope with stresses by rapid adaptation before sacrifice. In vivo reporter systems will be necessary to specifically pinpoint those cells expressing particular metacaspases. Also, metacaspase activity can be regulated on multiple posttranslational levels. For example, the activity of AtMC9 is regulated by autoprocessing, pH, a protease inhibitor (AtSerpin1), and S-nitrosylation (, ; ). Likewise, AtMC4 and AtMC5 activity depends on calcium (; ). Therefore, specific in vivo activity assays would strongly contribute to our understanding of the role of individual metacaspases in plant development and stress response.
To date, metacaspases of plants, fungi, and protozoa have been shown to have Arg/Lys-specific activity (, ; ; ; ). Based on the available sequences, we speculate that all metacaspases and possibly also paracaspases share this specificity. As a consequence, the caspaselike activities reported to be involved in plant and fungal cell death most probably differ from the metacaspases. Other plant proteases exhibiting caspaselike activity and suggested to be involved in cell death include the legumains (also called vacuolar processing enzymes) and some subtilisins (; ; ).
Until now, the role of metacaspases in cell death still remained enigmatic, and both up- and down-regulation of metacaspases have yielded conflicting data. However, such approaches bear the risk that a constitutive perturbation of genes that are essential for normal cellular homeostasis leads to overinterpretation. Alternative routes toward unraveling the function of metacaspases could involve the identification of their substrates by using technologies that allow direct characterization of in vivo protein processing on a proteome-wide scale (). Knowing the degradome specificity of metacaspases could reveal their role in cellular and developmental processes, including cell death. Overproduction of the cleavage fragments and/or of uncleavable mutant proteins would help elucidate the functional consequences of substrate cleavage by metacaspases.
We conclude that although metacaspases, paracaspases, and caspases contain a caspase fold and probably originated from a common ancestor gene, metacaspases and paracaspases are clearly distinct from caspases for the following reasons. First, the fact that these different proteases contain a caspase fold might not be a valid argument to group them together in the caspase family (clan CD, family C14; ). Legumains (clan CD, family C13) and gingipains (clan CD, family C25) constitute separate families that also contain a caspase fold (; ). Second, the P1 preference of metacaspases is basic, whereas that of caspases is acidic. If metacaspases and caspases shared similar functions, we assume that both the proteases and their specific degradome would have coevolved. In view of the crucial functions of many of their substrates, this hypothesis is unlikely. Third, previous phylogenetic analyses of clan CD peptidases have shown that caspases constitute a separate group distinct from other clan CD peptidases, including metacaspases and paracaspases (). Therefore, we believe that it might be expedient to regroup metacaspases and paracaspases into a separate family in the CD clan of Cys peptidases. |
Bacteria, like other organisms with walled cells such as plants and fungi, must temporally and spatially control cell wall synthesis to regulate cell morphogenesis. An essential component of the bacterial cell wall is the peptidoglycan (PG), a meshwork of glycan strands cross-linked by peptide bridges that is synthesized and modified by enzymes collectively named penicillin-binding proteins (PBPs) because of their penicillin-binding property. Gram-negative bacteria mainly have one single layer of PG, whereas Gram-positive bacteria have multiple layers that are linked to each other via short peptides (). Either way, the mono- or multilayered PG forms one single, giant molecule that surrounds the cytoplasmic membrane and protects it from the turgor pressure exerted by the cytoplasm. The wall restrains the turgor pressure to avoid swelling and lysis, and the turgor pressure, in turn, is regarded as one of the primary forces that stretches the wall, favoring bond breaking and new PG insertion during cell growth (; ). Thus, bacteria, like other walled cell organisms, face two obstacles. First, the turgor pressure exerts equal force in all directions, which is problematic for achieving any nonspherical shape. Second, the integrity of the stress-bearing PG wall must be maintained at all times to avoid cell lysis, yet bonds must be broken to allow wall enlargement during growth and division. The latter is particularly challenging for Gram-negative bacteria with their single-layered PG.
A proposed solution is to restrict cell wall insertion to specific locales and to coordinate or couple PG synthetic activity with bond hydrolysis, perhaps in a holoenzyme complex comprising both synthetic and lytic enzymes (). Not only is this idea compatible with current models of glycan strand insertion, but affinity chromatography and plasmon resonance experiments have also demonstrated interactions in vitro among PG synthetic and lytic enzymes (; ; ). How do bacteria control cell wall synthesis and hydrolysis in time and space to produce specific shapes and sizes? Remarkably, despite the wide divergence in cell wall composition, structure, and metabolic activities between bacteria and walled eukaryotic cells, the spatio-temporal control of cell growth can be accomplished in both using a similar tool: the cytoskeleton. All three classes of cytoskeletal elements corresponding to eukaryotic tubulin, actin, and intermediate filament proteins are represented in bacteria (), with each playing an important role in cell morphogenesis.
In this paper, we will primarily discuss the most common and best-studied cell shapes—the sphere, the rod, and the curved rod—and describe the main control mechanisms involved in cell shape and size regulation.
xref
#text
xref
italic
#text
To achieve a rod shape, bacteria add an elongation phase between rounds of division. There are two main strategies that cells use to perform this elongation. One is to localize growth to the poles as in bacteria like (, middle; ; ). This strategy may depend on polarly localized proteins such as the coiled coil–rich protein DivIVA (). The second strategy, which is much more widespread in bacteria, is to halt most PG synthesis after pole formation at the end of division and switch to growth along the cylindrical sidewalls of the cell, keeping the poles relatively inert (, bottom; ; ). It should be noted that cell poles do not become inert immediately after division but appear to become progressively so (; ; ; ).
The MreB family of bacterial actin homologues has multiple functions (), including a critical role in controlling cell diameter during cell elongation. Depletion or drug-induced inactivation of MreB leads to a gradual cell widening during elongation, causing a growth-dependent rounding of the cell over time (; ; ; ). MreB is only found in nonspherical bacteria (), and some species have multiple MreB homologues ( has three: MreB, Mbl, and MreBH), but most have only one (MreB). How MreB homologues regulate cell width during elongation is not well understood. It was initially thought that MreB acted like FtsZ, controlling the localization of cell wall synthesis, because the helix-like pattern of nascent PG in () mimicked the helical localization of MreB homologues (; ; ; ). However, recent evidence suggests that the localization of nascent PG () as well as PBP localization patterns (; ) are largely preserved in cells lacking MreB homologues, although there is the suggestion that MreB is required for the establishment but not the maintenance of PBP2 localization in ().
One intriguing study demonstrated that MreBH of is required for the sidewall-specific localization of a PG hydrolase, LytE (), raising the possibility that the bacterial actin-like cytoskeleton may regulate PG insertion by controlling the availability of hydrolase-generated insertion sites in the existing wall. In , the PG precursor–synthesizing enzyme MurG, which accumulates at the midcell region during preseptal and septal growth (; ), is also found in a patchy pattern along the cell length, and the latter organization appears MreB dependent (). There is also evidence that MreB is in a complex with MurG in (). The actin-like cytoskeleton may be connected to cell wall enzymes through the membrane proteins MreC, MreD, and RodA (; ), although this coupling may be species specific (). In , MreC is required for the normal patchy pattern of a PG hydrolase, MltA (); collectively, these data hint at the existence of a morphogenetic network that coordinates PG precursor supply, synthesis, and hydrolysis.
Because rod-shaped bacteria depleted of tubulin-like FtsZ do not divide but continue to elongate into long filaments, FtsZ has been canonically viewed as being involved in division but not elongation. However, recent evidence shows that FtsZ can be responsible not only for division but also for a period of cell wall elongation from the central region of the cell (, bottom). In , the FtsZ ring localizes near the midcell well before the onset of division and, by recruiting MurG, spatially localizes the synthesis of PG precursors and directs a substantial portion of cell wall elongation before division (referred to as preseptal elongation; ). Observations in also support the existence of localized FtsZ ring–dependent wall synthesis before septation (). A recent study also suggests that FtsZ has an additional role in PG synthesis in the pole-proximal lateral sidewalls ().
There are many examples of bacteria that add another layer of morphological complexity to the common rod shape to produce a curved rod (vibrioid) or helical shape. There are a few known bacterial strategies for creating a curved shape. The spirochete appears to generate a helical shape with periplasmic flagella through an obscure mechanism (), and has no cell wall but produces helical morphology with contractile cytoplasmic fibrils (). In both cases, the fibers associated with helical shape are also involved with cellular motility. is the only walled bacterium with a curved rod-shaped morphology for which a cytoskeletal component of the curvature-generating mechanism has been identified thus far. This protein, crescentin, localizes specifically at the inner curvature of cells and shares striking similarities with eukaryotic intermediate filament proteins (, bottom; ). The consequence of crescentin loss is a straight rod cell shape ().
There are little data about how a cytoskeletal structure might produce a curved or helical shape, and, thus, current models are largely theoretical. One model states that a cytoplasmic fiber attached to the cell envelope will produce a curved or helical shape if the filament either actively shortens or grows more slowly than the cell as a whole (). This concept is borne out by physical modeling, and vibrioid cells, which elongate into helices after an extended time in stationary phase, obey one of the predictions of the model, namely that helical pitch will decrease as cells elongate (). An alternative hypothesis derives from the popular view that bacterial cytoskeletal elements have little or no mechanical role but are passive scaffolds, or platforms for the localization of proteins that regulate cell wall synthesis. In this model, a filamentous structure like that of crescentin might localize a negative regulator of PG synthesis along only one side of a rod-shaped cell, causing cell wall growth asymmetry between sidewalls and, thereby, curvature toward the shorter cell wall.
Surprisingly, little is known about the mechanical properties of whole cells, which is critical for a better understanding of cell morphogenesis. If mechanically deformed, would bacterial cells be elastic (returning to their original shape like a piece of rubber) or plastic (holding their new shape like a piece of clay)? Intriguingly, it has recently been demonstrated that simple mechanical force can change the growth patterns of bacteria so that they adopt specific morphologies. When cells are placed in agarose microchambers of defined shapes and cell division is inhibited, the elongating cells take the shape of the chamber in which they were confined (). Remarkably, when cells are released from the chambers, they retain the induced shape (), indicating that the cells act plastically; physical confinement alters the construction of the cell wall to produce a given shape. Continued growth after release leads to a gradual loss of the induced morphology (). This result raises the possibility that cytoskeletal elements may work in a similar manner, exerting local forces within the cell that lead to specific patterns of growth. Future experiments using micromanipulation techniques on single cells will surely be fruitful in exploring this possibility.
There are many complex bacterial shapes beyond what we have discussed above that remain exciting avenues of research. For instance, , which forms a funguslike mycelium of branched hyphae, grows only from the tips of branches (). Apical growth and morphogenesis seem to depend on localization of the DivIVA protein at hyphal tips, but its precise function is not yet clear (). Interestingly, MreB is not essential for vegetative growth in but is required for the development of aerial hyphae and spores (). Another instance of a complex shape is the prosthecate bacteria, which produce thin extensions of the cell envelope (referred to as stalks) to increase cell surface area and nutrient uptake. In , the stalk elongates specifically from the base of a cell pole (; ). How this occurs is unknown.
The Mollicutes are different from other bacteria in that they have defined shapes while having no cell wall; instead, they rely entirely on internal structures. For instance, the spiral shape of is produced by a helical cytoskeletal ribbon of protein fibrils that follows the shortest helical path within the cell (). Unequal local contraction of the constituent fibrils produces changes in helicity that propel the cell. Similarly, a cytoplasmic structure in produces the attachment organelle, a fingerlike projection that is important for cell motility (), but the exact structure and function of the cytoskeleton in these cells is unknown.
In addition to their geometry, bacteria need to carefully control their size. The placement and frequency of division can alternatively produce long filamentous cells or short rods, curved rods or spirals, and equally or unequally sized daughter cells. The position and timing of cell division must also be regulated to avoid chromosome guillotining or production of anucleate cells. In virtually all bacteria, division is initiated by the formation of a ring of the tubulin homologue FtsZ at the division site (), and, thus, regulation of the location and timing of FtsZ ring formation controls cell division.
There are multiple systems to control FtsZ assembly and ring formation that differ somewhat among species. The nucleoid occlusion system () prevents division at DNA-containing cellular locales. It uses the proteins SlmA in and Noc in , which bind to DNA and prevent FtsZ polymerization (; ). The Min system uses an FtsZ assembly inhibitor, MinC, that, through binding to the MinD ATPase, either oscillates (in ) or is tethered to the polar regions (in ; ; ; ). Therefore, the concentration of MinC is lowest at the midcell, favoring FtsZ assembly there. lacks the Min system and instead uses an FtsZ assembly inhibitor called MipZ, which binds to the chromosome-partitioning protein ParB (). Thus, MipZ follows ParB as the chromosomes separate, clearing MipZ from the central cell region and allowing the FtsZ ring to form (). Another system was recently uncovered in that couples nutrient availability to cell division (). It uses UgtP, a sugar transferase that acts both in the teichoic acid biosynthesis pathway and as an FtsZ assembly inhibitor, to ensure that fast-growing cells gain sufficient mass before division occurs (). Common to all these systems is a release of FtsZ assembly inhibition at the future division site.
Once established, the FtsZ ring is highly dynamic, with fluorescence recovery half-times of 8–9 s (), and it recruits several cell division proteins in hierarchical fashion (). As the cell divides, the FtsZ ring constricts until cell division is complete. It is unknown whether FtsZ generates a constrictive force itself or is merely compressed as it directs the synthesis of new cell poles. However, FtsZ is also present in wall-less bacteria, suggesting that it can itself generate a force (). In vitro, FtsZ forms straight protofilaments when GTP is bound and curved filaments when GDP is bound, suggesting that GTP hydrolysis–induced conformational changes may drive cell constriction (), although this remains unproven.
When it comes to cell morphogenesis, bacteria must meet the same challenges as walled eukaryotic cells, controlling the timing and location of cell wall synthesis to achieve specific shapes and sizes. Remarkably, similar to eukaryotes, bacteria use a highly regulated internal cytoskeleton for this spatio-temporal control to generate cell shapes ranging from the simple to the complex.
Still, many questions remain to be answered about the particulars of bacterial growth control. The cytoskeletal MreB family proteins are clearly important for cell width control and rod morphology, but how they achieve this function remains poorly understood. The chemical composition and bonding patterns of PG are well known, but the structure of PG has so far been inaccessible to observation techniques. How is the cell wall constructed at the molecular level in cocci, rods, and vibrioid or helical cells? Although many studies have focused on synthesis of the cell wall, few have considered membrane synthesis, which is also required for cell expansion and might be spatially or temporally regulated. A study of suggests that new membranes are primarily synthesized at the division septum (), opening the door for further experiments. It is remarkable that cells can grow into various shapes in response to external forces (). How can force alter cell wall growth? Although there is evidence that cytoskeletal structures (e.g., the FtsZ ring) serve as scaffolds for guiding cell wall deposition, can they produce force (and enough of it) to locally alter growth and, thus, shape?
The amazing ability of bacteria to generate, maintain, and transmit different shapes and sizes is intimately connected with a strikingly high degree of cytoplasmic organization, including distinct cytoskeletal elements. Many exciting discoveries surely lie ahead as the workings of bacterial morphology control systems are brought to light. |
The biogenesis of proteins requires the folding of newly synthesized polypeptide chains. In vivo, this process is assisted by molecular chaperones that control the folding of their client proteins in an energy-dependent reaction (for reviews see ; ). In the cytosol of bacteria and eukaryotes most of these chaperones rely on the hydrolysis of ATP as an energy source. In the periplasmic space of bacteria and the ER of eukaryotes, chaperone systems exist that stabilize the three-dimensional structure of their client proteins by oxidation, i.e., by the controlled formation of disulfide bridges between cysteine residues of the proteins (; ; ; ; ). In the bacterial periplasm, two components are critical for protein oxidation, DsbA and DsbB. DsbA is a soluble protein that directly interacts with the client proteins and transfers the electrons to the membrane protein DsbB. DsbB transfers the electrons further via the ubiquinone pool and the electron transport chain of the inner membrane to molecular oxygen from which water is then finally produced (; ).
Recently, the intermembrane space (IMS) of mitochondria was found to harbor a large number of proteins containing disulfide bonds (for reviews see ; ; ). Like in the periplasm of bacteria, proteins in the IMS of mitochondria are efficiently oxidized by a specific redox mechanism that is structurally not related to the bacterial system. The IMS contains two components, Mia40 and Erv1, that are essential for protein oxidation and the viability of eukaryotes (; ; ; ; ; ). Mia40 functions as an import receptor in the IMS. It transiently interacts with newly imported polypeptides, thereby converting them from a reduced and import-competent state to an oxidized, stably folded one (; ; ; ). During this interaction, Mia40 is reduced but subsequently reoxidized by the sulfhydryl oxidase Erv1 (also named augmenter of liver regeneration [ALR] in humans). Erv1 is a flavin adenine dinucleotide–binding protein that in vitro can directly pass its electrons on to molecular oxygen, giving rise to the production of hydrogen peroxide (). Interestingly, showed that, at least in vitro, the human homologue of Erv1, ALR, is able to reduce oxidized cytochrome . This observation led to the interesting hypothesis that the mitochondrial disulfide relay system might, like that of bacteria, interact with the electron transport chain of the inner membrane ().
In this paper, we present evidence that reoxidation of Mia40 in mitochondria indeed depends on the presence of oxidized cytochrome . As a consequence, the import of proteins into the IMS via the disulfide relay system depends on the energetic state of the respiratory chain. We show that this connection to cytochrome not only facilitates efficient reoxidation of Mia40 but also prevents the formation of potentially harmful hydrogen peroxide in the IMS of mitochondria.
Reduced and oxidized forms of Mia40 can be easily separated on nonreducing SDS gels (). This can be used to monitor directly the functionality of the disulfide relay system in isolated yeast mitochondria. In wild-type mitochondria ∼80% of the endogenous Mia40 is present in the oxidized, active form (). When the levels of Erv1 in the mitochondria are down-regulated, Mia40 is shifted to its reduced form, which shows a lower mobility on SDS-PAGE. Conversely, when the protein levels of Erv1 are up-regulated, basically all Mia40 is oxidized. Thus, the activity of Erv1 influences the redox state of Mia40.
Molecular oxygen was shown to serve as the final electron acceptor of Ero1 (). To test whether the redox relay in the mitochondrial IMS is influenced by oxygen, we assessed the redox state of Mia40 under oxygen-depleted and -saturated conditions at different glutathione concentrations. The precise glutathione concentration in the IMS is not known but the large diffusion limit of the porin channels in the outer membrane (∼5,000 D) should lead to similar levels of glutathione as in the cytosol, which in yeast is ∼13 mM (). In the experiment shown in , we assessed the redox state of Mia40 at glutathione concentrations from 0 to 40 mM. Under oxygen-depleted conditions, Mia40 is significantly more susceptible to reduction by glutathione than under oxygen-saturated conditions. It should be mentioned that the oxygen-depleted conditions used were ∼5–10% of fully saturated oxygen levels (Fig. S1, available at ). This is equivalent to physiological oxygen concentrations in animal mitochondria and still allows respiration (Fig. S2). In summary, the oxygen concentration in mitochondria has a direct influence on the redox state of Mia40, suggesting that in the mitochondrial disulfide relay system molecular oxygen serves as the final electron acceptor.
To address the dependency of the disulfide relay system on enzymes of the respiratory chain, Mia40 redox states were examined in mitochondria of different yeast mutants. The respiratory chain of yeast mitochondria () contains two proton-pumping enzymes, cytochrome reductase (complex III) and cytochrome oxidase (complex IV). Mitochondria were isolated from yeast strains lacking activity of either cytochrome reductase (, Δ and Δ) or oxidase (Δ and Δ). In addition, mitochondria were prepared from a strain in which both mitochondrial cytochrome isoforms (, Δ/Δ) were simultaneously deleted as well as from mutants lacking Atp1 or Atp10, respectively, which are subunits of the FF ATPase. We observed that the different mutants affected the redox state of Mia40 in different directions; in mitochondria devoid of cytochrome or cytochrome oxidase activity, Mia40 was considerably less oxidized than in wild-type mitochondria (). This suggests that oxidized cytochrome stimulates oxidation of Mia40. In contrast, the loss of cytochrome reductase activity shifted Mia40 to its oxidized state. The respiratory activity per se was not critical as the ATPase mutants did not influence Mia40. This is consistent with a specific function of oxidized cytochrome for oxidation of Mia40.
The Mia40 dependence on respiratory chain complexes was further studied in detail. In mitochondria of two exemplary yeast strains lacking activity of cytochrome reductase (Δ) or oxidase (Δ) the influence of externally added glutathione on the redox state of Mia40 was tested (). In wild-type mitochondria Mia40 remained largely oxidized even in the presence of up to 30 mM of reduced glutathione. The loss of cytochrome reductase activity even slightly increased the levels of oxidized Mia40. Conversely, in Δ mitochondria, Mia40 was rapidly reduced when glutathione was added, and at glutathione concentrations >15 mM virtually no oxidized Mia40 protein was found. This again points to a direct influence of the respiratory chain on the redox state of Mia40.
To exclude side effects in the various mutants, we tested the effect of inhibitors of the respiratory chain on the redox state of Mia40. To this end, wild-type mitochondria were incubated in the presence of 25 mM glutathione and subjected to antimycin A or potassium cyanide, which completely inhibit cytochrome reductase and oxidase activity, respectively. The addition of antimycin A increased the fraction of oxidized Mia40, whereas inhibition of cytochrome oxidase by cyanide led to a decrease of the oxidized form (). These shifts were not found in mutants lacking cytochrome (). This again indicates that cytochrome oxidase activity influences the state of Mia40. However, it should be noted that even in the absence of cytochrome or cytochrome oxidase activity, Mia40 is not completely reduced, and thus cytochrome and cytochrome oxidase are not essential for the oxidation of Mia40 under the conditions tested.
It has been shown that in vitro cytochrome can function as an electron acceptor for ALR (). In general cytochrome accepts electrons in vitro from a variety of sulfhydryl oxidases, including those that are not present in mitochondria. However, Erv1 and cytochrome are both located in the mitochondrial IMS, making a direct interaction both feasible and physiogically reasonable. Because the direct transfer of electrons by sulfhydryl oxidases to molecular oxygen yields hydrogen peroxide, a physiological interaction with cytochrome might protect the cell against oxidative damage. First, we tested whether yeast Erv1 is able to interact with cytochrome in vitro like its human homologue. To this end, we incubated Erv1 and oxidized cytochrome with DTT, which serves as a substrate for Erv1 (; ), and measured the reduction of cytochrome in a spectrophotometer at 550 nm. In the presence of Erv1, cytochrome was efficiently reduced (, squares). This was not caused by a direct interaction of DTT with cytochrome , as cytochrome remained almost entirely oxidized when Erv1 was omitted (, circles). Thus, Erv1 can efficiently shuttle electrons from DTT to cytochrome .
In contrast to Erv1 and Mia40, cytochrome is not essential for the viability of yeast cells, indicating that Erv1 can be oxidized in vivo even in the absence of cytochrome , at least to a certain degree. To test the relevance of cytochrome more explicitly, we assessed the redox states of Mia40 in wild-type or / double deletion mutants under oxygen-saturated and -depleted conditions. Upon saturation with atmospheric oxygen, Mia40 was largely oxidized even in cytochrome –deficient mitochondria (, left). The addition of low amounts of glutathione (7.5 mM) did not considerably influence the state of Mia40, indicating a stable state of the redox relay system as long as the oxygen concentration is high. Mia40 also remains largely oxidized in both strains under low oxygen conditions. In the wild type, the addition of 7.5 mM glutathione affected the redox state of Mia40 only slightly (, right). In contrast, in the / mitochondria, Mia40 was almost completely shifted to its reduced form. This suggests that cytochrome is especially important under oxygen-limiting conditions. Consistently it was reported that cytochrome is a 100-fold better electron acceptor relative to oxygen in the reoxidation of ALR ().
Next we examined whether the variation of the redox states of Mia40 observed in respiration-deficient mitochondria affects protein import into the IMS. It has been shown that the depletion of Erv1 renders protein import highly sensitive to DTT (). Import experiments were performed under oxygen-depleted conditions with mitochondria isolated from wild-type cells and respiratory chain mutants. Low amounts of DTT (2 mM) reduced protein import of Cox19 into isolated wild-type mitochondria to ∼50% and almost completely blocked import into mitochondria lacking cytochrome oxidase (, ). In contrast, import into mitochondria of a cytochrome reductase–deficient strain (Δ) was less affected than in wild-type mitochondria (). Similar results were obtained with other IMS proteins such as Tim10 (unpublished data).
Import into / mitochondria was hypersensitive to DTT similarly to import into mitochondria lacking cytochrome oxidase activity (). Moreover, the import of Tim10 was tested in the presence of antimycin A or potassium cyanide. Blockage of cytochrome oxidase by potassium cyanide rendered the import of Tim10 more sensitive to the addition of DTT, whereas the inhibition of cytochrome reductase with antimycin A had the opposite effect (). In addition, the DTT sensitivity of the import correlated with the DTT sensitivity of the binding of Mia40 to the newly imported Tim10 (, Mia40 • Tim10). This suggests that the activity of the respiratory chain complexes influences the activity of the Mia40 receptor and, as a consequence, the efficiency of protein import. In summary, the observed effects on the protein import into the IMS correlate with the redox levels of Mia40 found in the mutants: increased oxidation of cytochrome renders protein import more resistant toward DTT, whereas lower levels of oxidized cytochrome impair the import process.
Does the interaction of Erv1 with cytochrome really prevent the formation of hydrogen peroxide? To address this question, we developed an assay to monitor the Erv1-dependent production of hydrogen peroxide. To this end, we incubated purified recombinant Erv1 with its substrate DTT in the presence of Amplex red. This compound reacts in a 1:1 ratio with hydrogen peroxide, thereby forming resorufin, which can be easily detected by fluorescence. As shown in , mixing DTT and Erv1 led to rapid generation of hydrogen peroxide (top). Interestingly, the addition of oxidized cytochrome considerably delayed the production of hydrogen peroxide in a dose-dependent manner (, middle and bottom). This was not caused by a potential quenching effect of cytochrome because the fluorescence generated by hydrogen peroxide in the presence or absence of cytochrome was identical (). We thus conclude that electron transfer from Erv1 to cytochrome and cytochrome oxidase leads to the generation of water instead of harmful hydrogen peroxide.
In this paper, we describe a pivotal role of the respiratory chain for the activity of the disulfide relay system in the IMS of mitochondria. Based on our observations we propose a physical interaction of Erv1 with cytochrome (). This interaction directly connects the redox relay system to the respiratory chain that prevents the generation of hydrogen peroxide. The following observations support this model: (a) Erv1 efficiently reduces cytochrome in vitro; (b) in the absence of oxidized cytochrome , Mia40 is shifted to its reduced state; (c) the increase of oxidized cytochrome increases the oxidized form of Mia40; (d) cytochrome prevents the Erv1-dependent formation of hydrogen peroxide; and (e) protein import in the absence of cytochrome or cytochrome oxidase is hypersensitive to DTT.
Although oxidized cytochrome clearly facilitated oxidation of Mia40, it was found to be nonessential for this process at least under the conditions tested. However, under oxygen-limiting conditions, cytochrome efficiently prevented the reduction of Mia40 by glutathione. This suggests that the interaction of the disulfide relay to the respiratory chain might be particularly important under low-oxygen conditions. reported that the single deletion of the cytochrome isoform Cyc1 impairs growth under oxygen-depleted conditions. Low oxygen concentrations are common in many tissues of animals and humans (for reviews see ; ). It is very conceivable that the interaction of the disulfide relay system with the respiratory chain might be particularly important for multicellular organisms. The prevention of hydrogen peroxide production is presumably more critical for higher eukaryotes than for yeast cells. Therefore, it will be interesting to address the physiological relevance of the Erv1–cytochrome interaction in mammalian tissues in the future.
yeast strains Δ, Δ, Δ, Δ, and Δ/Δ were isogenic to the wild-type strain W303 (Mat α, , , ,, , and ; ). The cytochrome double deletion strain was provided by A. Barrientos (University of Miami, Miami, FL). Other mutants used in this study (Δ, Δ, Δ, and Δ) and the corresponding wild-type BY4742 (Mat α, Δ, Δ, Δ, and Δ0) were obtained from the yeast deletion collection (). All strains were grown in YP medium consisting of 10 g/liter of yeast extract and 20 g/liter of peptone adjusted to pH 5.5, to which 2% galactose was added. The mutant () was grown in liquid lactate medium () in the presence of 0.1% glucose or galactose to repress or induce the promoter, respectively. Mitochondria were isolated as described previously ().
Import of radiolabeled proteins into mitochondria was performed as described previously ().
Recombinant Erv1 (expression plasmid provided by T. Lisowsky, University of Düsseldorf, Düsseldorf, Germany) was purified as described previously (). Reduction of horse heart cytochrome was measured at 550 nm in a UV/visible light spectrophotometer (Ultrospec 2100 pro; GE Healthcare). The reaction was started by addition of 2 mM DTT (GERBU Biochemicals) to 8 μM of purified Erv1 in 0.5 mM EDTA (Merck) and 50 mM potassium phosphate buffer, pH 7.4. Software (Swift II; GE Healthcare) was used for data collection and quantification.
The production of hydrogen peroxide by Erv1 was measured using the fluorescence dye Amplex red (10-acetyl-3,7-dihydroxyphenoxazine) according to the manufacturer's instructions (Invitrogen). 2 μM of purified Erv1 was incubated in 100 mM potassium phosphate buffer, pH 7.4, with 50 μM Amplex red and 1 U/ml horseradish peroxidase (Sigma-Aldrich). Erv1 was activated with the artificial substrate DTT (). Cytochrome from horse heart (Sigma-Aldrich) was added in the concentrations described. Fluorescence was recorded in a spectrofluorometer (FluoroMax-2; HORIBA Jobin Yvon) with excitation at 550 nm and emission at 610 nm using a 1-nm slit. The integration time was 600 ms and the data was collected every 600 ms.
Fig. S1 shows that the oxygen concentration in the oxygen-depleted conditions used in this study is equivalent to ∼5–10% of full saturation. Fig. S2 shows that these oxygen concentrations still allow respiration and growth of cells on nonfermentable carbon sources. Online supplemental material is available at . |
In embryos, the oocyte has no developmentally important polarity (). After fertilization, the oocyte and sperm pronuclei are usually at opposite ends of the oblong embryo, and the oocyte pronucleus undergoes two meiotic divisions, extruding two polar bodies before polarity induction and mitosis (). Posterior polarity is induced via an unknown centrosome- dependent signal brought in by the sperm (; ; ; ; , ). This is thought to locally down-regulate actin contractility, leading to anterior movement of the anterior PAR (partitioning defective) polarity proteins PAR-3, PAR-6, and PKC-3 through the anterior contraction/movement of actin and nonmuscle myosin (; ). Anterior PAR protein localization leads to the posterior cortical localization of PAR-2 (). Establishment of the anterior and posterior cortical PAR domains is critical for all downstream polarized events ().
Sperm in which centrosome maturation is defective, such as and mutants, often fail to induce posterior polarity (; ; ). The defect in centrosome maturation in and mutants causes a delay in the nucleation of microtubules, which led to the proposal that microtubules might be required for inducing posterior polarity (; ). Consistent with this idea, showed using a variety of mutant backgrounds that a persistent meiotic (acentrosomal) metaphase spindle can induce posterior polarity, with the posterior polarity protein PAR-2 being found near the arrested meiotic spindle at the presumptive anterior end rather than near the sperm pronucleus. This posterior polarity induction depends on microtubules, supporting the view that microtubules can provide polarity signaling and that centrosomes are not required ().
Recent studies have challenged a role for microtubules in polarity induction (; ). After the knockdown of tubulin by RNAi or inhibition with the drug nocodazole, embryos can still establish embryonic polarity, with normal anterior and posterior PAR domains (; ). These results, together with work showing that removal of the centrosome by laser ablation leads to a failure of polarity induction, led to the conclusion that centrosomes induce polarity independently of microtubules (; ). However, this model fails to explain how a meiotic spindle, which lacks centrosomes, can provide a polarity signal. In addition, although greatly reduced, tubulin is still detectable at the centrosome after RNAi of tubulin or nocodazole treatment (; ; ), leaving open the possibility that microtubules could be required for the posterior signal.
In this study, we explore the relationship between polarity signaling and PAR protein levels and carry out a detailed analysis of embryos in which tubulin levels are reduced. Our results strongly support a role for microtubules in polarity establishment.
To identify embryonic polarity regulators, we tested a set of embryonic lethal genes by RNAi for inducing defects in the localization of GFP–PAR-2 in the one-celled embryo (unpublished data). One gene that showed a strong defect was , as previously reported (). SPD-5 is a centrosome component required for centrosome maturation and mitotic spindle assembly ().
We observed that most embryos showed reversed embryonic polarity, with GFP–PAR-2 near the meiotic polar bodies and opposite the sperm pronucleus (). Both PAR-3 and the nonmuscle myosin NMY-2 also show a reversed distribution in embryos, suggesting that a normal process of polarity induction is occurring ( and Fig. S1, available at ). Reversed embryonic polarity has previously been associated with mutants or RNAi knockdowns inducing a persistent or abnormal meiotic spindle (; ; ). However, no meiotic defects have been reported for mutants (). We confirmed that meiotic timing ( = 5) and polar body and spindle formation during meiotic divisions ( = 6) are normal in embryos. To test whether microtubules are responsible for the reversed posterior polarity signal, we used a β-tubulin mutant () combined with RNAi to inhibit tubulin function in embryos (see Materials and methods). In such embryos, anteriorly localized PAR-2 was never detected (, seventh column). In contrast, the total percentage of embryos that showed posterior PAR-2 in embryos was similar irrespective of the RNAi knockdown of tubulin (, last column). These results show that a normal meiosis produces a microtubule-dependent signal that can induce posterior polarity. Consistent with these data, wild-type embryos sometimes show a transient anterior cap of PAR-2 protein (). However, the timing of meiotic polarity induction differs from that of centrosome-dependent induction: reversed polarity in embryos occurs 10 min later than the initiation of normal posterior polarity in wild-type embryos (Fig. S2).
We observed a higher frequency of reversed polarity after RNAi of than in mutants (, second and third rows; ). A possible explanation for this is that embryos might have more SPD-5 activity because the mutant may not be a null allele. To test this idea, we used RNAi of to try to further reduce SPD-5 activity in embryos. Such embryos display a range of PAR-2 localization defects similar to those of mutants (). This indicates that the difference in penetrance of the reversal of polarity is not caused by a difference in SPD-5 activity.
Our experiments assaying polarity defects of embryos were performed in a background harboring GFP–PAR-2 in addition to endogenous PAR-2 and, thus, have increased PAR-2 protein levels (; ). This raised the possibility that the difference in penetrance of reversed polarity could be caused by a difference in the relative levels of anterior and posterior PAR proteins. To investigate this hypothesis, we compared the localization of PAR-2 in embryos carrying GFP–PAR-2, carrying no PAR transgenes, and in which the level of the anterior PAR protein PAR-6 was increased through a GFP–PAR-6 transgene (). Strikingly, we found that the embryonic polarity phenotypes induced by in the three genotypes strongly differ: PAR-2 is exclusively at the anterior end of 80% of GFP–PAR-2 embryos, of 52% of embryos with wild-type PAR levels, and of 0% of GFP–PAR-6 embryos (). Posterior polarity completely failed to be induced in 97% of these latter embryos, as they showed uniform GFP–PAR-6 and no PAR-2 on the cortex ( and not depicted). Because the polarity signaling events in the three genotypes are expected to be equivalent, these results suggest that the ability of a signal to induce polarity depends on the relative levels of the PAR proteins.
The similarity in the reversed polarity induced by a meiotic microtubule-dependent (acentrosomal) signal and a sperm centrosome-dependent signal prompted us to reexamine a role for microtubules in posterior polarity induction. We used RNAi to simultaneously deplete α and β tubulins (hereafter referred to as ; see Materials and methods). Extended RNAi of hermaphrodites led to maternal sterility, indicating that embryos completely devoid of tubulin cannot be produced (see Materials and methods). To achieve as strong a depletion as possible, we analyzed the embryos produced in a time window when 50–90% of double-stranded RNA (dsRNA)–injected hermaphrodites were sterile. To assess the effects of tubulin knockdown at different early embryonic stages, we carefully classified embryo age using the state of DNA condensation: stage 1, meiosis; stage 2, no DNA condensation; stage 3, initiation of DNA condensation; stage 4, intermediate DNA condensation; and stage 5, full DNA condensation (see for staging criteria). We then compared the pattern of tubulin staining in wild-type and embryos.
Wild-type embryos display a strong cortical network of microtubules at all of these stages (). At stage 1 (meiosis), microtubules are additionally visible in the meiotic spindle (). In stage 2, only the cortical network is observed (). In stage 3, most (92%) embryos have apparent centrosomal asters, which are found in 100% of stage 4 and stage 5 embryos (). In embryos, tubulin immunoreactivity is strongly diminished at all stages (). During meiotic stages, embryos lack cortical microtubules but show a small concentration of tubulin around the maternal DNA (). At stage 2, a weak network of cortical tubulin fibers is visible (). In contrast to wild type, only 22% of stage 3 and 34% of stage 4 embryos show a concentration of tubulin staining at the centrosome, with the remainder having weak cortical microtubules (). At stage 5, 89% have detectable microtubules at the centrosome. This analysis shows that after strong tubulin knockdown, a small centrosomal microtubule aster eventually forms, but at a later time than in wild-type embryos.
To confirm that tubulin knockdown does not impair the accumulation of other centrosomal proteins, we examined two centrosome markers in embryos. We found that the centrosomal accumulation of SPD-5 () and TAC-1 (; ; ) are normal in embryos ( = 13 and = 8, respectively; and not depicted).
We then compared polarity induction in wild-type and embryos. In wild-type embryos, polarity is initiated at stage 3 (initiation of DNA condensation; ). At this stage, we observed that 69% of such embryos had posterior PAR-2, all of which had visible asters; a further 23% had lower but detectable levels of tubulin at the centrosome but no obvious cortical PAR-2 (). At stage 4 (intermediate DNA condensation) and stage 5 (full DNA condensation), all wild-type embryos were polarized and had robust asters (). These results show that similar to a previous study (), the initiation of polarity is strongly correlated with growth of the sperm aster.
We found that polarity induction in embryos was delayed relative to wild type and was only apparent in embryos with centrosomal asters. Whereas the majority of wild-type embryos are polarized at stage 3, only 11% of embryos are polarized at this stage (). Similarly, at stage 4, when all wild-type embryos are polarized, only 17% of embryos are polarized (). Importantly, although only a minority of stage 3 and stage 4 embryos have asters (22% and 34%, respectively), all of the polarized embryos had visible asters (). At stage 5, 11% of embryos still lacked asters, and these were not polarized (). Therefore, there is a tight correlation between the time of polarity induction and microtubule aster growth in both wild-type embryos and embryos. Because centrosomal microtubule growth is delayed in the context of a normal centrosome in embryos, these results strongly argue that microtubules are required for polarity induction.
Previous work demonstrated that the polarity signal is dependent on the centrosome and that centrosome maturation is critical for the signal (; ; ). Mutants that impair centrosome maturation fail to polarize (; ). Regulation of centrosome maturation timing by cyclin E (CYE-1) is also important, as RNAi of delays maturation and prevents polarity induction (). Our work supports the view that the delay in microtubule growth caused by centrosome maturation defects in these mutants is responsible for the impairment in polarity induction.
Microtubule involvement in both the centrosome-dependent signal and a meiosis-dependent signal suggests that both processes may use a similar signaling mechanism. However, the timing of these signals appears to be different. We found that the meiotic signal induces PAR polarity ∼13 min after the completion of meiotic divisions and 10 min later than centrosome- dependent posterior polarity (Fig. S2). Therefore, it seems unlikely that it is the meiotic spindle itself that is delivering a signal. The meiotic divisions produce two polar bodies that have associated midbodies, or spindle remnants where they are attached to the embryo. Because these spindle remnants contain microtubules, one possibility is that this is the source of the meiotic polarity signal. The time of meiotic signaling is coincident with rapid mitotic centrosomal microtubule growth, which might induce growth/activity of the spindle remnant ends.
What could be the nature of the microtubule-dependent signal? It is surprising that components have not yet been found given the extensive genetic and RNAi screening that has been conducted for cell polarity genes. A possible reason for this is that polarity induction might involve partially redundant signaling pathways, which are commonly seen in other processes. Good candidates for involvement are microtubule plus end–binding proteins, many of which either regulate microtubule dynamics or influence cortical processes (). Identifying the signaling and receiving factors is the most crucial future task for understanding the mechanism of polarity induction.
The following strains were used, culturing by standard methods (): wild-type Bristol N2, JJ1473: [–NMY-2–GFP; +] (), zuEx69 [–GFP–PAR-6] (), JH1512: [–GFP–PAR-6] (), (), KK866 [pie-1–PAR-2–GFP] (), EU856: (), JA1390 [–PAR-2–GFP, +], and [–β-tubulin–GFP), +] (; ).
Synthesis of dsRNA and RNAi was performed by injection as described previously () using RNAi feeding clones from and as templates for RNAi synthesis (sjj_F56A3.4 for , sjj_C47B2.3 for , and sjj_C36E8.5 for ). RNAi of α- and β-tubulin genes was performed by combining dsRNA with both and . Because of the high level of sequence identity among tubulin genes, dsRNA will additionally target , and dsRNA will target . In all experiments, and dsRNAs were coinjected, and experiments were conducted at 25°C. 50–90% of hermaphrodites are sterile at 22–30 h after injection; at 40 h, all hermaphrodites are sterile even though sperm are still present. in was performed at 25°C, and embryos were dissected 30 h after injection. experiments were performed for 22–30 h at 25°C.
Antibody staining was performed as described previously (). Antibodies used were rabbit anti–SPD-5 (), rat anti–PAR-3 (), rabbit anti–PAR-2 (), rabbit anti–TAC-1 (), mouse antitubulin (clone DM1 A1; Sigma-Aldrich), and chicken anti-GFP (Chemicon). Secondary antibodies were purchased from Jackson ImmunoResearch Laboratories. Confocal images were taken on either an LSM 510 Meta microscope (Carl Zeiss MicroImaging, Inc.) or a Radiance instrument (Bio-Rad Laboratories). Wild-type and RNAi experiments were conducted in pairs, and images were taken using the same settings.
For live recordings in Fig. S2, [–PAR-2–GFP] and [pie-1– β-tubulin–GFP] were mounted in egg buffer (118 mM NaCl, 40 mM KCl, 3 mM CaCl, 3 mM MgCl, and 5 mM Hepes, pH 7.2) on 18 × 18-mm coverslips coated with 0.3% poly-lysine (Sigma-Aldrich), and the coverslips were inverted onto 3% agar pads and sealed with petroleum jelly. Paired images were taken every 10 s using a 63× lens on a fluorescence microscope (Axioplan 2; Carl Zeiss MicroImaging, Inc.) and 3DM software (Improvision).
Fig. S1 shows that embryos show a reversal of anterior and posterior PAR domains. Fig. S2 shows that reversed PAR-2 polarity in embryos is delayed relative to wild-type PAR-2 polarity. Online supplemental material is available at . |
xref
italic
sup
#text
To determine whether H efflux by NHE1 is necessary for activation of Cdc42, we used NHE1-deficient fibroblasts expressing wild-type (WT) NHE1 or a mutant NHE1 that contains an E266I substitution and lacks H efflux (). To biochemically determine Cdc42 activity in migrating cells, multiple wounds were created in a confluent monolayer with a multichannel pipette. The abundance of active Cdc42-GTP at the indicated times before and after wounding was determined by precipitation with GST–p21-activated kinase (PAK)–Cdc42/Rac-interactive binding domain (CRIB) and immunoblotting for Cdc42 (). After wounding, the abundance of Cdc42-GTP in WT cells increased twofold at 5 min and remained elevated at 5 h. In E266I cells, the abundance of Cdc42-GTP before wounding was 60% of that in WT cells, and after wounding there was no increase in Cdc42-GTP. The abundance of total Cdc42 was similar in WT and E266I cells before and after wounding.
Monolayer wounding triggers multiple stimuli, including activation of integrins and release of growth factors. Integrin engagement with extracellular matrix proteins activates Cdc42 () and stimulates NHE1 activity (; ). We found that H efflux by NHE1 is necessary for haptokinetic migration toward fibronectin (Fig. S1, available at ) and for integrin-induced activation of Cdc42 (). Plating on fibronectin for 1 h increased Cdc42-GTP fourfold in WT cells compared with cells in suspension. In E226I cells, the abundance of Cdc42-GTP in cell suspension was less than in WT cells and there was no increase after plating on fibronectin. Integrin affinity for fibronectin, determined by binding FITC-labeled fibronectin, and expression of β1 integrin, determined by immunoprecipitating lysates of biotinylated cells with β1 antibodies, were similar in WT and E266I cells (Fig. S1).
PDGF also increases Cdc42-GTP () and NHE1 activity (), and we found that H efflux by NHE1 is necessary for activation of Cdc42 by PDGF (). 50 ng/ml PDGF increased the abundance of Cdc42-GTP in WT cells, with a maximum of 2.2-fold increase at 2 min. In contrast, PDGF did not activate Cdc42 in E266I cells, although the abundance of total Cdc42 was similar in subconfluent WT and E266I cells. PDGF-induced activation of Cdc42 also was inhibited in NHE1-deficient PS120 cells, which were used to make WT and E266I cells, but not in parental CCL39 cells that express NHE1 (). Hence, H efflux by NHE1 is necessary for maintaining Cdc42 activity in quiescent cells and for increased Cdc42 activity with monolayer wounding, integrin engagement, and PDGF.
In migrating cells, Cdc42-GTP is predominantly localized at cell protrusions (). NHE1 is localized at the leading-edge membrane of migrating cells (; ), which suggests that H efflux might be necessary for Cdc42 activity at cell protrusions. This was confirmed by using a merocyanine–Cdc42-binding domain (MeroCBD) biosensor for Cdc42-GTP (; ). MeroCBD, an environmentally sensitive fluorescent dye covalently coupled to the Cdc42/Rac binding domain of the neural Wiskott-Aldrich syndrome protein, increases fluorescence intensity upon binding activated Cdc42, which enables detection of spatially localized endogenous Cdc42-GTP in living cells. EGFP attached to the Cdc42/Rac binding domain allows ratiometric image analysis, thereby normalizing for cell thickness and concentration artifacts. The MeroCBD probe, which is insensitive to pH 4.5–8.0, was injected into cells at the edge of a wounded monolayer 15 h after wounding, and images were acquired after 30 min. In WT cells, Cdc42-GTP was elevated in cell protrusions ( = 22 cells), however, in E266I cells, Cdc42-GTP was more uniform and notably reduced even where cells protruded ( = 52 cells; ). Acquired images were also used to quantify active Cdc42 in microinjected cells, and, like biochemical assays with GST-PAK-CRIB, they indicated attenuated Cdc42-GTP in E266I cells (). The ratio intensity was 344.4 ± 28.1 U in WT cells and 170.9 ± 7.0 U in E266I cells. The Mero/EGFP fluorescence ratio of cells injected with an insensitive control probe (MeroCBD mutated to greatly reduce Cdc42 binding; ) was 196.9 ± 24.8 U ( = 5 cells; unpublished data), indicating that activation in E2661 cells was near the minimum level detectable by the biosensor. Hence, biochemical and imaging analyses indicate that H efflux by NHE1 is necessary to maintain the abundance of Cdc42-GTP in quiescent and stimulated cells and to maintain the localization of Cdc42-GTP in migrating cells.
We previously reported that H efflux by NHE1 increases in fibroblasts expressing an active GTPase-deficient Cdc42-V12 and that H efflux by NHE1 stimulated by a constitutively active Gα13-QL is suppressed by coexpression of mutationally inactive Cdc42-N17 (). In wound-edge WT cells, expression of Cdc42-N17 inhibited H efflux by NHE1, resulting in decreased intracellular pH (pH) of 7.10 ± 0.09 ( = 43 cells), compared with pH of 7.30 ± 0.10 ( = 30 cells in cells not expressing Cdc42-N17. In subconfluent WT cells, PDGF stimulated NHE1 activity and increased pH from 7.15 ± 0.03 to 7.47 ± 0.05, which was attenuated in cells expressing Cdc42-N17 to 6.99 ± 0.02 and 7.10 ± 0.03 ( = 3 cell preparations).
Activation of Cdc42 requires release of inactive Cdc42-GDP from Rho GDP dissociation inhibitor (RhoGDI) in the cytosol, recruitment to the plasma membrane, and activation at the plasma membrane by a guanine nucleotide exchange factor (GEF) that catalyzes the exchange of GDP for GTP. Immunoprecipitating RhoGDI and immunoblotting for total Cdc42 indicated that PDGF induced a decrease in the abundance of coprecipitating Cdc42 that was similar in WT and E266I cells (), which suggests that NHE1 activity is not necessary for the regulated dissociation of Cdc42 from RhoGDI. Immunoblotting particulate fractions (P100) for Cdc42 showed that abundance in E266I cells was comparable to that in WT cells after PDGF stimulation (), which indicates that H efflux by NHE1 is not necessary for membrane recruitment of Cdc42.
These findings suggest that H efflux by NHE1 might be necessary for GEF-induced guanine nucleotide exchange, which we confirmed by determining GEF activity in cell lysates (). Exchange of [P]GTP by Cdc42 for cold GTP was used as an index of GEF activity. The amount of Cdc42-[P]GTP decreased to 49.2% in lysates from WT cells treated with PDGF (). Lysates from PDGF-stimulated E266I cells did not show a marked exchange of [P]GTP for cold GTP compared with unstimulated cells (). These data suggest that GEF activity is impaired in E266I cells.
We asked whether GEF activity might be pH sensitive because in WT cells the quiescent pH of 7.15 increases to 7.45 with growth factors, but in E226I cells, the quiescent pH of 7.00 does not change (; ). At least 20 members of the Dbl family of GEFs stimulate guanine nucleotide exchange by Cdc42, and we tested whether the activity of two GEFs for Cdc42, Dbs (Dbl's big sister) and intersectin, is pH dependent. Because Dbs contains a His residue (H814) in the α6 helix critical for interacting with switch two of Ccd42 and because it contacts a His residue in Cdc42 (H103 in the α3b region; ), we reasoned that pH-dependent titration of these histidines might regulate GEF activity or contact with Cdc42. Using the minimal Dbl homology (DH)–pleckstrin homology (PH) segment of Dbs necessary for activity and a GST-fusion of Cdc42 loaded with the fluorescent analogue methylanthraniloyl (mant)-GDP, which has reduced fluorescence when not bound to Cdc42 (), we found no change in activity from pH 6.5 to 8.0 for Dbs () or intersectin (Fig. S2, available at ). Additionally, in the absence of Dbs, the release of mant-GDP from Cdc42 was similar from pH 6.5 to 8.0 (), indicating that guanine nucleotide exchange by Cdc42 was pH insensitive.
Although in most Rho family GEFs the DH domain is sufficient to catalyze nucleotide exchange, a tandem PH domain that binds phosphoinositides is invariant. In cells, phosphoinositide binding by the PH domain can regulate activity of the DH domain of some GEFs (). We asked whether phosphoinositide binding to Dbs or intersectin might be pH sensitive because phosphoinositides bind to positively charged residues in PH domains, which might titrate with changes in pH, and phosphates on phosphoinositides have pKs near neutral (). Additionally, FYVE domains, which share structural similarity with PH domains for binding phosphoinositides at loops between β strands, have pH-dependent affinity for phosphoinositides (). By using liposome sedimentation, we found that the DH-PH domain of Dbs bound phosphotidylinositol 4,5–bisphosphate (PIP2), as previously reported (; ), and that binding was pH dependent (). Maximal specific binding seen at pH 6.5 (52 ± 13%) was significantly reduced at pH 7.5 and 8.0 (P < 0.05; = 4), suggesting a lower affinity at higher pH. Although PIP2 binding by the DH-PH domain of intersectin was maximal at pH 6.5 (42 ± 8%), binding was relatively insensitive to pH (). We speculate that pH-sensitive binding of PIP2 to Dbs is caused by the presence of a His (H843) in the same position as H355 in the Arf GEF Grp1 that is critical for binding phosphoinositides (; Barriero, G., and M. Jacobson, personal communication). Computational modeling (unpublished data) indicates that a spatially conserved His in close proximity to predicted PIP2–binding sites is present in other GEFs activating Cdc42, including Fgd1 (H985), αPix (H38), ASEF (H513; H505), and Dbl (H701; H756), but is absent in intersectin, Fgd3, Tiam1, and PDZRhoGEF (unpublished data; Barriero, G., and M. Jacobson, personal communication). Hence, whether Dbs or another predicted pH-sensitive GEF mediates NHE1-dependent activation of Cdc42 remains to be determined.
It also remains to be determined whether pH-dependent PIP2 binding by GEFs contributes to NHE1-dependent activation of Cdc42. PIP2 binding to Sos2, a Ras GEF, inhibits nucleotide exchange activity, possibly by retaining a cis inhibition of the DH domain by the adjacent PH domain (; ). The functional significance of PIP2 binding to Rho family GEFs, however, is less clear. PIP2 binding to recombinant DH-PH domains in vitro is reported to stimulate (), inhibit (; ), or not affect (; ) activity. Additionally, we cannot rule out other pH-dependent mechanisms, such as scaffolding or conformation changes independent of phosphoinositide binding, because attenuated GEF activity is retained in lysates of E266I cells () and for some proteins conformational changes, ligand-binding affinities, and macromolecular assemblies are sensitive to small changes in physiological pH (). A suggested pH-dependent scaffolding by NHE1 () is also a putative mechanism because NHE1 binds PIP2 () and the ezrin-radixin-moesin protein ezrin (), and ezrin is suggested to sequester Dbl to plasma membrane microdomains () and to regulate Dbl activation of Cdc42 (). Moreover, H efflux by NHE1 could regulate an upstream activator of Cdc42-GEFs, although activity of Rap1B, which is necessary and sufficient to initiate polarity in neurons via activation of Cdc42 (), was not impaired in E266I cells compared with WT cells (Fig. S3, available at ).
Our data indicate positive feedback signaling between Cdc42 and NHE1 activity that is likely critical for polarity in migrating cells by asymmetrically amplifying both signals at the leading edge. Our findings also suggest that RhoGDI dissociation and membrane recruitment of Cdc42 are distinct signaling events that can be regulated independently of guanine nucleotide exchange. Beyond our current focus on regulated Cdc42 activity our data raise the possibility that activity of other GTPases and GEFs, and the affinities of protein modules for binding phosphoinositides, might be pH sensitive and regulated by NHE1 activity.
Activated Cdc42-GTP was determined by precipitation with a GST fusion of the PBD domain of PAK (GST-PAK-CRIB) as previously described (). CCL39, PS120, WT, and E266I fibroblasts were maintained as previously described (). For monolayer wounding, confluent cells were wounded with a multichannel pipette. Detached cells were removed by medium exchange, and lysates were prepared from adherent cells. For experiments with integrin activation, quiescent cells maintained in DME containing 0.2% FBS for 18–24 h were trypsinized, incubated for 10 min with 0.5 mg/ml of soybean trypsin inhibitor (Sigma-Aldrich), collected by centrifugation, and resuspended in serum-free DME. Resuspended cells were plated on dishes coated with 10 μg/ml of bovine plasma fibronectin (Sigma-Aldrich). Cells treated with 50 ng/ml PDGF-BB (Roche Diagnostics) were grown to 70% confluence and maintained for 18–24 h in DME containing 0.2% FBS (). At the indicated times, cells were lysed in 500 μl of lysis buffer () and lysates were clarified by centrifugation. A 20-μl aliquot of the supernatant was saved for determining total Cdc42, and the remaining lysate was incubated with 20 μg of GST-PAK-CRIB bound to Sepharose beads. Precipitated proteins were separated by 12% SDS-PAGE and immunoblotted with anti-Cdc42 antibodies (1:200; BD Biosciences). Immunoblots were analyzed by densitometry using Image (National Institutes of Health). Because of variations in cell density between preparations and GST-PAK-CRIB, which was freshly generated for each experiment, the absolute values of Cdc42-GTP and total Cdc42 for each cell preparation, determined from the immunoblotting signal by Image, were expressed as a ratio of Cdc42-GTP/total Cdc42. The ratios from each cell preparation were expressed relative to the ratio of control WT cells. Hence, although there was variability in Cdc42-GTP/total Cdc42 ratios between cell preparations, the cell-type and condition-specific relative changes in ratios were consistent.
Cell migration was determined on nucleopore filters (8-μm pore; Costar; Corning Inc.) coated on the lower side of the membrane with 10 μg/ml fibronectin or BSA, and chambers were filled with growth medium. Cells were trypsinized and resuspended at a final concentration of 10 cells/ml. A 100-μl aliquot of cell suspension was added to the upper chamber and incubated at 37°C. At the indicated times, cells were washed and the upper surface was wiped to remove nonmigrating cells. The membranes were fixed in 4% paraformaldehyde, washed with PBS, and stained for 5 min in crystal violet at 0.1% in PBS. After three washes in water, membranes were dried overnight, and the crystal violet was extracted in 1 ml acetic acid at 10%. Dye amount was quantified on a spectrofluorometer at 600 nm.
To determine integrin affinity for fibronectin, cells in suspension were mixed with the indicated concentrations of FITC-labeled fibronectin (FluoReporter; Invitrogen) for 60 min. FITC-positive cells were determined using a cell sorter (FACS Vantage SE; Becton Dickinson), and data were analyzed using CellQuest Pro 4.0.1 software (Becton Dickinson). To determine β1 integrin expression, cells were biotinylated on ice for 90 min and lysed in RIPA buffer, and the lysate was incubated with antibodies against β1 integrin (9EG7; BD Biosciences) and with protein A–Sepharose beads. Eluted proteins were probed for biotin with streptavidin-HRP.
WT and E266I cells grown to confluency on glass coverslips were wounded with a pipette tip and, after 15 h, cells at the wound edge were microinjected with the Cdc42 biosensor MeroCBD as previously described (). Images were collected with an inverted microscope (Axiovert 100TV; Carl Zeiss MicroImaging, Inc.) using a camera (cooled CCD; Quantix) and an oil-immersion objective (40× 1.3 NA). The exposure times were 30–300 ms for EGFP and 90–900 ms for the ISO dye. Image analysis was performed using Metamorph software (Molecular Devices) as described in . The processed ISO dye images were divided by the corresponding EGFP images, producing ratio pictures that represent activation patterns of Cdc42. The qualitative assessment of differences in localized activity was performed using line scans and visual inspection. Mean Cdc42 activity in individual cells was calculated using Metamorph and analyzed with Excel (Microsoft).
WT cells were transfected by electroporation (Nucleofector kit; Amaxa Biosystems) with Cdc42-N17, porcine cytomegalovirus, or empty vector, plated on glass coverslips with cherry-red histone pJAG 285, and grown to confluence for wounding or used at 70% confluence for treating with PDGF. NHE1 activity and intracellular pH were determined in cells loaded with the fluorescent pH-sensitive dye BCECF (Invitrogen) as previously described (; ).
The abundance of Cdc42 complexed with RhoGDI was determined as previously described () using total cell lysates incubated with antibodies to RhoGDI (1:50; Invitrogen) conjugated to protein A– Sepharose. Proteins in the immune complex were separated by 12% SDS-PAGE and immunoblotted with antibodies to Cdc42 (1:200; BD Biosciences) or RhoGDI (1:200; Invitrogen). The abundance of Cdc42 and RhoGDI in immune complexes was determined by densitometry.
Quiescent cells at 70% confluence were untreated or treated with 50 ng/ml PDGF and lysed by sonication. 100 μg of protein in postnuclear supernatants was centrifuged at 100,000 for 20 min to obtain soluble (S100) and particulate (P100) fractions. Proteins were separated by 12% SDS-PAGE and immunoblotted with anti-Cdc42 antibodies. Immunoblots were analyzed by densitometry by using Image.
GEF activity in cell lysates was determined as previously described (). 500 μl of lysates from subconfluent quiescent cells untreated or treated with 50 ng/ml PDGF were added to 1 μg α-[P]GTP-GST-Cdc42 (Cytoskeleton, Inc.) in the presence of 2 mM of cold GTP and 10 mM MgCl at room temperature. Samples were removed at the indicated times and diluted with ice cold termination buffer. After centrifugation and washing, radioactivity was quantified by scintillation counting.
Activity of recombinant DH-PH domains of Dbs and intersectin (provided by J. Sondek, University of North Carolina, Chapel Hill, NC) was determined by determining incorporation of fluorescent -mant-GDP into GST-Cdc42 as described previously (). 200 nM of recombinant DH-PH domain and 100 μM GTP were added and guanine nucleotide exchange was determined by measuring the decrease in fluorescence (excitation, 360 nm; emission, 440 nm) with release of mant-GDP from Cdc42 using a spectrofluorometer (SpectraMax M5; Invitrogen).
Lipid micelles were prepared as previously described () using a Mini-Extruder (Avanti Polar Lipids) and contained phosphatidyl choline/PI/PIP2 (86:10:4 molar ratio; Avanti Polar Lipids). Vesicle suspensions adjusted to the indicated pH with KOH or HCl were incubated with 10 μg (3 μM) of recombinant DH/PH protein for 15 min at room temperature and then collected by centrifugation at 100,000 for 60 min. Supernatants and pellets were analyzed by SDS-PAGE and Coomassie staining. The amount of protein on the gel was determined by densitometry analysis using Image. Specific binding was calculated as the abundance of peptide bound to vesicles containing PIP2 minus binding to vesicles in the absence of PIP2. The abundance of peptide bound to vesicles in the absence of PIP2 was minimal and pH independent. To correct for variations in lipid vesicle preparations, data were expressed relative to binding at pH 6.5 for each determination.
Activated Rap1B-GTP was determined in subconfluent cells by precipitation with a GST fusion of the Rho binding domain of RalGDS (GST-RalGDS-RBD) as previously described (). At the indicated times after treating with 50 ng/ml PDGF (Roche Diagnostics), cells were washed in ice cold PBS and lysed in 500 μl of lysis buffer (), and then 20 μl of lysates was saved for determining total Rap1B. The remaining lysate was incubated for 1 h at 4°C with 10 μg of GST-RalGDS-RBD bound to Sepharose beads. Precipitated proteins were separated by 12% SDS-PAGE and immunoblotted with anti-Rap1B antibodies (1:200; Santa Cruz Biotechnology, Inc.). Immunoblots were analyzed by densitometry using Image.
Fig. S1 shows that H efflux by NHE1 is necessary for haptokinetic migration toward fibronectin but not for integrin affinity or expression. Fig. S2 shows that activity of the DH-PH domain of intersectin is pH insensitive. Fig. S3 shows that H efflux by NHE1 does not regulate activity of Rap1B. Online supplemental material is available at . |
The centromeres of eukaryotic chromosomes are flanked by pericentric heterochromatin that is highly variable between species in size and repetitive DNA sequence composition but remarkably conserved in chromatin protein composition and structure from fission yeast to humans (). Pericentric heterochromatin structure is essential for accurate chromosome segregation during mitosis (; ) and is similar in composition to constitutive heterochromatin found at other chromosome regions that also contain repetitive sequences and transposable elements, where it functions to silence transcription, reduce the frequency of recombination, and promote long-range chromatin interactions (; ). Heterochromatin is composed of regular tightly packed arrays of hypoacetylated nucleosomes that are methylated at lysine 9 of histone H3 (MeK9H3), mediated by the Su(VAR)3-9 histone methyltransferases (Clr4 in fission yeast and Suv39h1,2 in mammals). MeK9H3 recruits the heterochromatin protein 1 (HP1) family of proteins (Swi6 in fission yeast), which in turn recruit Su(VAR)3-9 as part of a complex self-reinforcing network of proteins that are enriched at heterochromatic loci (; ; ). Although species-specific differences exist for some components of this network, the overall conservation of heterochromatin structure and function suggests that detailed mechanistic insights gained from experiments in fission yeast and flies will also apply to mammals.
Paradoxically, although constitutive heterochromatin functions to silence transcription, in fission yeast it has been shown that transcription from within pericentric heterochromatin is required for the formation and maintenance of heterochromatin and for sister chromatid cohesion (; ). Transcripts generated by RNA polymerase II are processed into siRNA that is in turn recognized by an RNAi-induced transcriptional silencing complex that is recruited to and required for heterochromatin assembly and gene silencing (; ; ). The RNAi pathway is also required for the formation of heterochromatin and silencing of repetitive sequences in (). In mammalian cells, an unidentified RNA component is required for the association of HP1 with pericentric heterochromatin (; ). However, mammalian homologues to certain key components of the fission yeast transcription–mediated gene silencing network have not been identified (; ). Moreover, attempts to detect transcription from mammalian pericentric heterochromatin have met with varied levels of success, with discrepancies found both in the ability to detect such transcripts and the sizes of any transcripts detected (; ; ; ; ; ; ; ; ; ; ; ; ).
One possible explanation for these inconsistencies is that the transcription of satellite DNA could be cell cycle regulated, making it difficult to detect in asynchronously growing cells or tissues in which most cells are not cycling. In fact, cell cycle regulation of heterochromatin transcription could provide a logical means to drive the reassembly of heterochromatin after the disruptive processes of DNA replication and mitosis, which might not be necessary in a quiescent cell. Here, we show that different types of RNA polymerase II–transcribed RNA species are synthesized from the AT-rich mouse γ (major) satellite repeat sequences at different times during the cell cycle: a small species induced specifically during mitosis and a large heterogeneous set of RNAs induced during late G1 and early S phase. Both were short lived and dependent on the passage of cells through the restriction point.
To examine satellite transcription during the cell cycle, mouse C127 cells were synchronized by selective detachment during mitosis and released into G1 phase for up to 7 h, at which time 5–10% of cells begin to enter S phase (; ). To monitor S phase progression, a portion of mitotic cells were arrested at the G1/S boundary in the presence of the DNA polymerase inhibitor aphidicolin for 10–12 h and released into S phase for an additional 20 h. Total RNA from various time points was then isolated and analyzed by Northern blot hybridization using a mouse γ satellite DNA probe. As shown in , molecules smaller than 200 nt were detected specifically in mitotic cells and were undetectable by 1 h after mitosis. These are smaller than the size of the γ satellite repeat (234 bp). When small RNAs were selectively enriched before Northern hybridization, hybridization signals were detected almost exclusively during mitosis (). Later in G1 phase, a more heterogeneous set of RNAs were detected that were mainly larger than 1 kb, which is consistent with previous papers (; ). These accumulated gradually during the course of G1, reaching a peak in late G1/early S phase, after which the amount of detectable RNA was substantially reduced but still higher than during early G1.
To confirm the short half-life of these transcripts, we examined their sensitivity to the RNA polymerase II inhibitor 5,6-dichloro-1-β--ribofuranosylbenzimidazole (DRB; ). DRB added for as little as 1 h strongly reduced levels of both the small transcripts during mitosis and the large transcripts at 7 h into G1 phase, confirming that both species have a relatively short half-life. To determine whether the relative abundance of these transcripts at different times during the cell cycle reflects their de novo transcription rates, we evaluated the synthesis of nascent transcripts from preengaged RNA polymerase complexes using nuclear run-on assays (). Transcription from γ satellite DNA was strongest during late G1 phase and was also detected in late S phase and mitosis, but was virtually undetectable during early G1 phase. In contrast, transcription from the β-actin gene could be detected at all times except mitosis. In fact, most transcription is silenced during mitosis by phosphorylation and the eviction of transcription factors (). Together, these results demonstrate that small heterochromatic RNAs are synthesized de novo during mitosis and not processed from transcripts synthesized before mitosis, although they could be processed from larger transcripts synthesized during mitosis.
The short half-life for detection of both the mitotic and late G1/early S phase transcripts could be caused by the rapid degradation of the RNA or rapid modification of the transcripts in ways that prevent their detection by hybridization, such as RNA editing (; ). The adenosine-rich and potentially double-stranded () transcripts produced from γ satellite DNA would make excellent substrates for hydrolytic deamination of adenosine residues to inosine residues by double-stranded RNA–specific adenosine deaminases. In fact, vigilin, a component of an adenosine deaminase acting on RNA complex, appears to colocalize with dense chromatin in monkey COS7 cells and, when overexpressed, associates with pericentric satellite sequences in human HEK293T cells (). However, immunolocalization of vigilin with two independent antibodies revealed no colocalization of vigilin with DAPI-dense pericentric heterochromatin clusters (chromocenters) at any time during the cell cycle (unpublished data). Moreover, we sequenced RT-PCR products amplified with degenerate primers or primers designed against γ satellite regions that were unlikely to be affected by editing (). 10 different products from M phase, G1/S phase, and asynchronous cells were identical to the original γ satellite sequence (unpublished data). From these experiments, we conclude that A-to-I editing of γ satellite transcripts is not a major contributor to the rapid loss in detection of the mitotic transcripts.
To confirm these results using an alternative method, we used RNA-FISH. RNA-FISH detects nascent transcripts as they are produced at the site of transcription (; ) and accurately reflects results obtained with the more laborious nuclear run-on method (). RNA-FISH signals hybridizing to the mouse satellite probe were detected on the outer surface of chromocenters (, i–iii), which are easily visualized with a DAPI stain (). No sites were detected with a control probe that did not contain γ satellite sequences (unpublished data). Detection of these sites was completely abolished by treatment of nuclei with RNaseA (, iv), demonstrating that they did not result from unintentional DNA denaturation. Treatment of cells with DRB for 1 h before collection resulted in a complete inhibition of detectable RNA-FISH signals (, v). These controls demonstrate that the signals detected by RNA-FISH represent nascent RNA transcripts originating from γ satellite DNA within pericentric heterochromatin.
The number of transcription sites detected per cell was highly heterogeneous, ranging from 0 to >15. Hence, we quantified both the percentage of positive cells as well as the number of transcription sites per cell at each cell cycle stage (). During mitosis (; M, metaphase; P/M, prophase and metaphase), ∼90% of cells had one to three sites of transcription. This could be an underestimate because the signal intensity per site was weaker in mitotic cells (relative to later times in the cell cycle), possibly caused by the small size of the RNA during mitosis (). The percentage of positive cells dropped considerably during mitotic exit (; A/T, anaphase and telophase), and by early G1 phase, <10% of cells displayed one to four intermediate intensity transcription sites, which is consistent with the lack of detectable transcripts by Northern analysis (). The percentage of positive cells, the number of transcription sites per cell, and the intensity of each site all increased in late G1 and early S phase, followed by a dramatic drop by 4 h in S phase, with only ∼15% cells showing a strong FISH signal. As cells progressed toward the end of S phase, the number of positive cells began to increase again, but with fewer numbers of sites per cell, indicating that a low level of de novo transcription continues into late S phase. The variable increase in detectable sites per cell at 20 h may represent entry of cells into the subsequent cell cycle.
To simplify the distinction between high and low levels of transcription, we estimated the number of cells carrying out the late G1/early S phase mode of satellite DNA transcription by quantifying cells that have an early S phase number of detectable transcription foci (five or greater). This plot (, pink) resembles the Northern quantification shown in . To monitor the progression of these cells through S phase, cells were labeled with BrdU just before collection for RNA-FISH, and aliquots were stained with anti-BrdU antibodies (and DAPI). These results revealed that transcriptional induction clearly occurred before the onset of S phase and was down-regulated during mid S phase (, yellow).
We have previously shown that replication of mouse chromocenters takes place during mid S phase (), close to the time at which satellite DNA transcription decreases. Cells engaged in chromocenter replication can be easily scored because of the prominent intranuclear appearance of the DAPI-stained chromocenters (). Replication begins at the periphery of the chromocenters (, III) followed by a period during which virtually all DNA synthesis in the cell consists of chromocenter replication (IV; ; ). When the percentage of BrdU-positive cells engaged in the replication of chromocenters (, III and IV) was quantified in the same cell populations used for (yellow), a sharp increase in their number was seen within the same 4-h period as the decrease in transcription of γ satellite DNA within the chromocenters (, blue).
The results in suggest that γ satellite transcription may be down-regulated upon chromocenter replication. To investigate this possibility, we repeated the experiments shown in with more precise S phase time points, starting from the G1/S border through 7 h into S phase. These results () revealed a sharp decrease in the percentage of cells positive for transcription between 3 and 4 h, which coincides with a sharp increase in cells replicating chromocenters. However, there were two concerns with these BrdU/RNA-FISH experiments. First, because the denaturation step necessary to reveal BrdU incorporation is incompatible with RNA-FISH detection, it was necessary to quantify each property in separate cell samples. Second, we wanted to rule out the possibility that the cell-synchronizing agent aphidicolin may have affected the results. Hence, to visualize replication of pericentric heterochromatin and transcription of satellite RNA simultaneously within individual asynchronously growing cells, we combined RNA-FISH with immunolocalization of the replication fork protein proliferating cell nuclear antigen (PCNA). After elimination of the soluble pool of PCNA that is not engaged in DNA synthesis (), PCNA staining patterns resembled BrdU patterns throughout S phase (), as was expected (). Hence, cells in G1 phase could be identified by their small, PCNA-negative nuclei, cells at different stages of S phase could be identified by their PCNA staining pattern (, I–VI), and cells in G2 phase could be identified as large PCNA-negative cells. PCNA and RNA-FISH signals did not colocalize throughout almost the entire duration of S phase (), with the exception of 16% of cells in very late S phase (VI), for reasons that are not understood.
As shown in , transcription of γ satellite is considerably higher in early S phase and decreases starting with the onset of chromocenter replication (III). Moreover, the percentage of cells with more than five sites of γ satellite transcription increases from G1 to early S phase and then decreases at the time of chromocenter replication. These results confirm a general incompatibility between γ satellite transcription and replication during S phase, similar to what has been observed for individual sites of replication and transcription throughout S phase (). It is possible that the reduction in transcription is exclusively caused by interference of replication with transcription. However, only a subset of pericentric regions are engaged in replication at any particular moment in time (), so it is unlikely that replication is simultaneously interfering with transcription of all pericentric regions.
Suv39h1,2 is responsible for the trimethylation of lysine 9 of histone H3 (MeK9H3) at pericentric heterochromatin in mice (). In Suv39h1,2 double knockout mouse embryonic fibroblast (MEF) cells, MeK9H3 is lost, HP1 dissociates, DNA methylation is drastically reduced, and the trimethylation of histone H4 lysine 20 (MeK20H4) is lost (; ; ; ). These cells show karyotypic instability and elevated steady-state levels of γ satellite transcripts (). Because these prior experiments were performed on asynchronously growing cells, the accumulation of γ satellite transcripts could have resulted either from elevated transcription rates or a disruption of cell cycle regulation resulting in transcription throughout the entire cell cycle.
To distinguish between these possibilities, we performed PCNA/RNA-FISH staining in wild-type (WT) versus Suv39h1,2 double knockout (D15) MEFs, as described in . Although D15 had a substantially higher percentage of cells transcribing γ satellite DNA from considerably more sites than WT cells, both cell lines showed an increase in transcription transitioning from G1 to early S phase and a decline in transcription upon replication of chromocenters (), which is similar to C127 cells (). Mitotic transcription was also elevated in D15 (). To compare the percentage of cells transcribing γ satellite transcripts at late G1/early S phase levels, as was done for C127 cells in , we adjusted our criteria for the number of RNA-FISH foci per cell to reflect the relatively low level of transcription in WT MEFs (more than one site per nucleus) and the higher level of transcription in the Suv39dn1,2, double knockout cells (more than seven sites per nucleus). When the percentage of cells meeting these criteria was scored, it revealed a clear reduction in the number of highly transcribing cells upon chromocenter replication ().
The very low levels of transcription during early G1 phase raised the intriguing possibility that transcription of pericentric heterochromatin might require passage through the restriction point and commitment to cell division. Hence, we examined cells that were arrested in G0 by contact inhibition. For all cell lines (C127, WT, and D15), very little transcription could be detected in arrested cells (). To distinguish whether long-term arrest in quiescence resulted in transcription down-regulation or if transcription was not induced because cells were prevented from passing through the restriction point, C127 cells were synchronized in mitosis as in and released into G1 phase in the presence of various concentrations of serum in the medium or into a complete medium to which the Cdk inhibitor roscovatine was added 2 h after release into G1 phase. All cell populations were then allowed to proceed to 7 h after mitosis, when substantial up-regulation of γ satellite transcription was observed in control cells (). Both serum deprivation and roscovitine treatment severely inhibited γ satellite transcription. We conclude from this experiment that transcription of mouse pericentric heterochromatin is dependent on passage through the restriction point.
s
h
o
w
t
h
a
t
a
t
l
e
a
s
t
t
w
o
d
i
f
f
e
r
e
n
t
p
o
p
u
l
a
t
i
o
n
s
o
f
R
N
A
m
o
l
e
c
u
l
e
s
a
r
e
e
x
p
r
e
s
s
e
d
f
r
o
m
m
o
u
s
e
p
e
r
i
c
e
n
t
r
i
c
h
e
t
e
r
o
c
h
r
o
m
a
t
i
n
a
t
d
i
f
f
e
r
e
n
t
t
i
m
e
s
d
u
r
i
n
g
t
h
e
c
e
l
l
c
y
c
l
e
.
T
r
a
n
s
c
r
i
p
t
i
o
n
w
a
s
C
d
k
d
e
p
e
n
d
e
n
t
,
i
n
d
i
c
a
t
i
n
g
t
h
a
t
c
e
l
l
s
d
o
n
o
t
s
y
n
t
h
e
s
i
z
e
t
h
e
s
e
t
r
a
n
s
c
r
i
p
t
s
u
n
t
i
l
a
f
t
e
r
t
h
e
y
c
o
m
m
i
t
t
o
p
r
o
l
i
f
e
r
a
t
i
o
n
.
M
o
r
e
o
v
e
r
,
t
h
e
t
r
a
n
s
c
r
i
p
t
s
w
e
r
e
s
h
o
r
t
-
l
i
v
e
d
.
T
o
g
e
t
h
e
r
,
o
u
r
r
e
s
u
l
t
s
p
r
o
v
i
d
e
a
s
a
t
i
s
f
y
i
n
g
e
x
p
l
a
n
a
t
i
o
n
f
o
r
w
h
y
s
u
c
h
t
r
a
n
s
c
r
i
p
t
s
w
e
r
e
n
o
t
d
e
t
e
c
t
e
d
i
n
m
a
n
y
s
t
u
d
i
e
s
t
h
a
t
e
x
a
m
i
n
e
d
q
u
i
e
s
c
e
n
t
o
r
s
l
o
w
l
y
g
r
o
w
i
n
g
t
i
s
s
u
e
b
u
t
w
e
r
e
f
o
u
n
d
i
n
t
i
s
s
u
e
s
t
h
a
t
c
o
n
t
a
i
n
p
r
o
l
i
f
e
r
a
t
i
n
g
c
e
l
l
s
.
M
o
r
e
o
v
e
r
,
t
h
e
y
p
r
o
v
i
d
e
e
v
i
d
e
n
c
e
f
o
r
p
r
o
v
o
c
a
t
i
v
e
l
i
n
k
s
b
e
t
w
e
e
n
h
e
t
e
r
o
c
h
r
o
m
a
t
i
n
a
n
d
c
e
l
l
u
l
a
r
p
r
o
l
i
f
e
r
a
t
i
o
n
t
h
a
t
w
a
r
r
a
n
t
f
u
r
t
h
e
r
i
n
v
e
s
t
i
g
a
t
i
o
n
.
Mouse C127 cells were synchronized in mitosis by mechanical shakeoff after a brief and fully reversible nocodazole treatment (Sigma-Aldrich) as described previously (). Similar results were obtained in experiments repeated without the use of nocodazole. For G1/S synchronization, 10 μg/ml aphidicolin (Calbiochem) was added 5 h after release from mitosis for an additional 10–12 h. Where roscovitine (Calbiochem) was used, 40 μM was added at 2 h after mitosis. For serum deprivation, mitotic cells were plated directly into a medium containing either 0.1% or no serum. For contact inhibition, cells were further cultured for 7 d after reaching confluence with fresh media every day. For BrdU pulse labeling, 15 μg/ml BrdU (Sigma-Aldrich) was added to medium for 30 min before fixation.
Total RNA was prepared using an microRNA isolation kit (Ambion) and treated with DNase (Promega). In parallel, <200 nt RNA (small) fractions were separated from total RNA using the same kit. To rule out any possibility of DNA contamination in our samples, we performed RT-PCR analysis using γ satellite–specific primers. Only reverse transcribed samples gave ladderlike PCR bands, and RNase A (Sigma-Aldrich) treatment completely eliminated the product. Total RNA was resolved via electrophoresis with a denaturing agarose gel, whereas <200 nt fractions were resolved with a denaturing 15% polyacrylamide gel. RNAs were then transferred to a nylon membrane. The γ satellite probe was plasmid pγSat () containing eight copies of the 234-bp repeat as a template (provided by N. Dillon, Imperial College London, London, UK), which was labeled with α-[P]dATP using a random labeling kit (Invitrogen). Total and small RNA hybridization was done at 60 and 25°C, respectively. Nuclear run-on with equal numbers of cells (10 million) was performed as described previously (), except that cells were permeabilized with digitonin (Sigma-Aldrich) as described previously () to maintain the integrity of mitotic chromosomes and allow detection of transcription during mitosis.
The RNA-FISH procedure was performed as described previously (), using cells that were either grown on coverslips or centrifuged from suspension onto coverslips using a cytocentrifuge (Cytospin 2; Shandon). In brief, cells were washed with CSK buffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl2, and 10 mM Pipes, pH 6.8), followed by CSK + 0.5% Triton X-100 (Sigma-Aldrich) permeablization for 5 min and 4% PFA fixation for 10 min on ice, and then stored in 70% ethanol at −20°C for no longer than 2 d. Slides were hybridized with a digoxigenin (Roche Applied Science) nick-translated γ satellite probe overnight at 37°C followed by fluorescent antibody detection as described previously (). In the case of immuno–RNA-FISH, PCNA staining, and RNA-FISH, PCNA immunostaining was performed using a monoclonal PCNA antibody (Oncogene Research Products) after RNA-FISH detection. For RNase A treatment, cells were treated with RNase A after permeabilization and before fixation.
Images were captured with an image restoration microscope system (DeltaVision; Applied Precision) attached to a fluorescence microscope (IX-71; Olympus) equipped with an oil objective lens (PlanApo 60×, 1.42 NA; Olympus) using a charge-coupled device camera (Coolsnap HQ; Photometrics) at RT. Approximately 40 optical sections (with 0.2-μm spacing) were taken and enhanced using the SoftWorx (Applied Precision) constrained iterative deconvolution process.
BrdU staining was performed as described previously (). For vigilin staining, cells grown on coverslips were fixed with cold 70% ethanol. After blocking with 10% normal goat serum in phosphate buffer for 30 min, cells were then incubated with polyclonal antibodies against N and C termini of vigilin (gift of G. Neu-yilik, University of Heidelberg, Heidelberg, Germany) for 1 h at RT, followed by incubation with FITC-conjugated secondary antibodies.
For conventional RT-PCR, RNA samples were reverse transcribed using either poly-dT, major satellite-specific, or random primer and subjected to PCR with γ satellite primers (5′-CATATTCCAGGTCCTTCAGTGTGC-3′ and 5′-GACGACTTGAAAAATGACGAAATC-3′). For the attempt to detect A-to-I–edited RNA, we used a degenerate primer (5′-CGGAATTCGAAAAY [A/C]GAGAAAC-3′) or primers from unlikely-to-be-edited regions (5′-GGAAAATGAGAAACATCCAC-3′) for reverse transcription and secondary primers (5′-CGGGATCCGTTTTCTCGCC-3′ or 5′-TTTTCAGTTTTCTCGCC-3′) for amplification. |
The budding yeast spindle pole body (SPB) is the functional equivalent of the mammalian centrosome. The mitotic exit network (MEN) is an SPB-associated signaling cascade that controls mitotic exit, which is the transition from mitosis into G1 phase of the cell cycle (; ; ). The Ras-like GTPase Tem1 functions at the top of the MEN (). The putative guanine nucleotide exchange factor Lte1 (an activator of the MEN) and the GTPase-activating protein (GAP) complex Bfa1–Bub2 (a MEN inhibitor) regulate Tem1 (; ; ). Tem1 interacts with the Pak-like kinase Cdc15 (), which, in turn, activates the Dbf2–Mob1 kinase complex (). Ultimately, the MEN controls the activity of the conserved phosphatase Cdc14 () and, thereby, mitotic exit ().
In yeast cells, the mother-bud junction determines the site of cytokinesis (). Therefore, cells with an anaphase spindle that is inappropriately positioned within the mother cell would cause cytokinesis to occur parallel to the plane of the spindle and, thus, result in aneuploidy. To prevent this from happening, the spindle orientation checkpoint (SPOC) senses (in an unknown manner) spindle orientation defects and actively inhibits the MEN of cells with a misaligned anaphase spindle. In cells with a correctly aligned anaphase spindle, phosphorylation of Bfa1 by Cdc5 polo kinase reduces Bfa1–Bub2 GAP activity to promote mitotic exit. However, when the spindle is misplaced, the SPOC prevents the Cdc5-dependent phosphorylation of Bfa1. Therefore, the Bfa1–Bub2 GAP complex remains active, and cells fail to exit mitosis and arrest in anaphase instead (; ).
The protein kinase Kin4 is an additional component of the SPOC. On the basis of genetic data, it would appear that functions upstream of and (; ). A striking feature of Kin4 is its SPB distribution in relationship to the Bfa1–Bub2 complex. In cells with a correctly aligned spindle, the Bfa1–Bub2 GAP complex binds preferentially to the budward-directed SPB (, ), whereas Kin4 associates with the SPB that faces the mother cell body (). In contrast, Kin4 and the Bfa1–Bub2 GAP colocalize at both SPBs when the anaphase spindle becomes mispositioned. This recruitment of Kin4 and Bfa1–Bub2 to the same SPBs may be important for the cell cycle arrest response to spindle alignment defects (). The observation that the targeting of Bub2 to both SPBs causes defects in mitotic exit even when the anaphase spindle is correctly positioned is consistent with this notion ().
How the SPOC senses spindle alignment defects and the molecular role of Kin4 in this process are currently unclear. In this study, we present evidence that the γ-tubulin complex receptor protein Spc72 provides a regulated binding site that recruits Kin4 to both SPBs whenever the anaphase spindle is mispositioned. This relocalization enables Kin4 to phosphorylate Bfa1, thereby protecting the Bfa1–Bub2 complex from inactivation by Cdc5 kinase. Thus, the SPB component Spc72 links cytoplasmic microtubules (MTs) with SPOC components and, therefore, could function as part of the sensors of spindle orientation defects.
The SPOC prevents the phosphorylation of Bfa1 by Cdc5 pololike kinase when the anaphase spindle becomes misaligned (). This regulation may occur at SPBs because both Bfa1 and Cdc5 associate with this structure (; ). If this was the case, it could occur at two levels. It could arise from a reduction in the amount of Cdc5 that associates with the SPB or from a specific impairment of the capability of Cdc5 to phosphorylate Bfa1 at SPBs.
To address whether the amount of Cdc5 at SPBs is regulated by orientation of the anaphase spindle, we compared the relative fluorescence intensity of Cdc5-GFP signal at SPBs in wild-type cells with a correctly aligned anaphase spindle with that of dynein-deficient cells ( cells) in which the anaphase spindle is misaligned within the mother cell body (; ). Because SPOC proteins show a polarized association with the SPBs (; ), we measured the Cdc5-GFP signal at both SPBs. In cells with a misaligned anaphase spindle, the Cdc5-GFP signal at both SPBs of a cell was of similar intensity (1,646 ± 836 [old SPB] and 1,535 ± 758 [new SPB]; P > 0.05 by test; ). In contrast, in wild-type cells, the Cdc5-GFP signal from the SPB in the bud was slightly stronger than that from the SPB in the mother cell (1,498 ± 880 vs. 1,086 ± 823; P < 0.05 by test). Importantly, the mean intensity of the Cdc5-GFP signal at the SPBs of misaligned spindles was not significantly different (P > 0.05 by test) from that at the SPBs of correctly aligned spindles (). Thus, spindle misalignment did not lower the amount of Cdc5 at SPBs.
Next, we determined whether the activity of Cdc5 at SPBs is regulated by the SPOC. For this analysis, we wanted to compare the ability of Cdc5 to phosphorylate SPB components, including Bfa1, in cells in which the spindle was correctly aligned with its ability to phosphorylate these substrates when anaphase spindles were misaligned. To perform this experiment, we sought SPB components in addition to Bfa1 that are Cdc5 kinase substrates. The SPB components Nud1 and Spc72 have both been shown to be phosphoproteins that bind to Cdc5 (; ). As such, they are excellent candidates for being targets of Cdc5. Consistently, phosphorylation of both Nud1 and Spc72 was dependent on in vivo, as the slower migrating phosphobands of each protein seen during mitosis of wild-type cells were missing from blots of SDS-PAGE gels from cells (; cell cycle profiles of cells of this experiment are shown in Fig. S1, A–C; available at ). In addition, Cdc5 was able to phosphorylate Bfa1, Nud1, and Spc72 directly in vitro (, P). In contrast, Cdc5 did not strongly modify maltose- binding protein (MBP), GST, Bub2, or Tem1 (). Together, these data suggest that Nud1 and Spc72 join Bfa1 as being substrates of Cdc5.
Although Spc72, Nud1, and Bfa1 are clearly targets of Cdc5, the aforementioned data do not determine whether their phosphorylation by Cdc5 takes place at SPBs or in the cytoplasm. To distinguish between these possibilities, we compared the phosphorylation pattern of Bfa1, Nud1, and Spc72 in and cells. Both cell types arrest at the restrictive temperature in anaphase (; ), but the two strains differ in one important respect: the cytoplasmic side of the SPB persists in cells and plays host to Bfa1, Nud1, and Spc72 (), whereas this SPB substructure has disassembled and, thus, is absent from cells (). Bfa1, Nud1, and Spc72 were hyperphosphorylated in cells (). In contrast, all three proteins were hypophosphorylated in cells. This correlation between the phosphorylation status and presence of a cytoplasmic SPB structure suggests that Bfa1, Nud1, and Spc72 become phosphorylated by Cdc5 at SPBs.
We determined whether the ability of Cdc5 to phosphorylate proteins at SPBs is affected by orientation of the anaphase spindle. For this experiment, we used synchronized cells and cells, both of which arrest at the restrictive temperature in anaphase (>97% and ∼85%) because of an inactive MEN (; ). However, in cells, the anaphase spindle was correctly aligned along the mother-bud axis, whereas it was misaligned in cells because of the failure of the redundant Kar9- and dynein-dependent spindle orientation pathways (for phenotypes, see Fig. S1 D; ). In both and cells, Nud1 and Spc72 became phosphorylated as soon as the B-type cyclin Clb2 accumulated in mitosis (, after 2 h). Thus, orientation of the mitotic spindle did not affect the activity of Cdc5 toward Nud1 and Spc72. This was different for Bfa1. In cells, Bfa1 became hyperphosphorylated (, lane 4). In stark contrast, Bfa1 failed to acquire any modifications in cells and persisted in a hypophosphorylated isoform throughout the experiment (, lane 8; ). These data suggest that the overall kinase activity of Cdc5 at SPBs is not regulated by the SPOC. Rather, a spindle orientation–dependent mechanism located at SPBs specifically protects Bfa1 from Cdc5 kinase when the anaphase spindle is misaligned.
Kin4 kinase functions upstream of the Bfa1–Bub2 complex (; ) and, therefore, is ideally positioned to act as a regulator of Bfa1 function. To address the molecular role of Kin4 in detail, we first tested whether Kin4 kinase activity was required for SPOC signaling. Cells in which Kin4 was mutated to abolish its kinase activity (Kin4; ) had no SPOC checkpoint (Fig. S2, available at ). However, this lack of checkpoint function was not caused by deficiencies in Kin4 expression or localization, as both were indistinguishable from those of wild-type active Kin4. Thus, Kin4 recruitment to the cell cortex and SPB did not require its kinase activity, but the SPOC did (Fig. S2).
To identify Kin4 targets in SPOC signaling, we adopted a candidate approach and focused on SPB-associated proteins because Kin4 binding to SPBs turned out to be essential for SPOC function (see ). We tested whether Kin4 could phosphorylate the MEN scaffold protein Nud1 (), the Bfa1–Bub2 complex, pololike kinase Cdc5, a catalytically inactive version of Cdc5, Cdc5 (), the γ-tubulin complex receptor protein Spc72 (), and the GTPase Tem1 (). In all assays, a catalytically inactive kinase-dead Kin4 was used as a control to exclude the possibility that a contaminating kinase was responsible for the phosphorylation of a candidate protein ( and Fig. S3, A and B, right; available at ).
Kin4 but not the kinase-dead Kin4 phosphorylated both full-length recombinant Bfa1 (, arrows) and a C-terminal degradation product of Bfa1 (, asterisks). Cdc5 polo kinase was weakly phosphorylated when incubated with Kin4 or Kin4 (Fig. S3 B). However, the kinase-dead Cdc5 was not phosphorylated by Kin4 (). This latter result indicates that at least some of the modifications of Cdc5 (Fig. S3) were caused by autophosphorylation. The other candidates that we tested were not strongly phosphorylated by Kin4 (Fig. S3). Together, this experiment identified Bfa1 as a clear in vitro substrate of Kin4.
We next mapped the sites on Bfa1 that were phosphorylated by Kin4 as a step toward understanding the functional significance of this modification. We used Kin4 that had been purified from yeast and phosphorylated recombinant MBP-Bfa1 in vitro. Serine 150 and 180 of Bfa1 were identified by matrix-assisted laser desorption ionization–time of flight (MALDI-TOF)/TOF as Kin4-directed phosphorylation sites (). We mutated both of these serine residues to alanine () to prevent phosphorylation. Phosphorylation of MBP-Bfa1 by purified Kin4 was reduced by ∼90% when compared with the levels seen with wild-type Bfa1 (), suggesting that both serine residues are indeed the major Kin4 kinase sites in our in vitro assay.
To determine whether Kin4 phosphorylates Bfa1 residues 150 and 180 in vivo, we raised phosphospecific antibodies against P-S150 and P-S180. The affinity-purified anti–P-S150 and –P-S180 antibodies only detected Bfa1 after it had been incubated with Kin4 but not after incubation with kinase-dead Kin4 (). This demonstrated that the two antibodies specifically recognized Kin4-phosphorylated Bfa1 and were unable to recognize unmodified Bfa1.
With these antibodies in hand, we could show that Kin4 phosphorylates Bfa1 in vivo. cells were arrested in anaphase by addition of the Cdc15-as inhibitor PP1. In these cells, Bfa1 accumulated in its hyperphosphorylated form. In contrast, in cells overexpressing , the overall phosphorylation of Bfa1 was reduced (, lane 3; ). In anti-Bfa1 immunoprecipitates of cells with expression, the anti–P-S150 and –P-S180 antibodies detected Bfa1 (, lanes 6 and 9). This was not the case without expression (, lanes 5 and 8). Thus, Kin4 phosphorylates S150 and S180 of Bfa1 in vivo.
We next compared the properties of cells expressing with those expressing and were integrated into the locus, where they were expressed as sole copies of from the endogenous promoter. and accumulated to similar levels (Fig. S4 A, 0 h; available at ), both proteins localized in a polar manner preferentially with the budward-directed SPB (; ), both responded to MT depolymerization as Bfa1 and Bfa1 were recruited to both SPBs in the presence of nocodazole (; ), and both proteins recruited the partner protein Bub2 to the SPB (Fig. S4 B). Thus, in terms of SPB localization and the ability to recruit Bub2, the Bfa1 protein behaved as the wild-type Bfa1 molecule.
The spindle assembly checkpoint regulates the metaphase-anaphase transition because it inhibits the anaphase-promoting complex, when chromosome association with the spindle is defective (). For unknown reasons, it requires but not to prevent mitotic exit (; ). Therefore, if Bfa1 is only defective in its ability to be regulated by Kin4, it should sustain the metaphase arrest of cells in response to spindle assembly checkpoint activation to a similar degree to that achieved by cells containing wild-type Bfa1. To test this notion, we analyzed the ability of synchronized cells to maintain a metaphase arrest in the presence of the MT-depolymerizing drug nocodazole. , , and cells were included as controls for cells with and without Bfa1–Bub2 function. and cells arrested in metaphase as large-budded cells with high Clb2 levels (). The cell cycle arrest was robust and lasted for at least 4 h. In contrast, the and cells were unable to maintain a metaphase arrest, as indicated by the rebudding of cells and the degradation of Clb2 ( and not depicted). These data indicate that the Bfa1 mutations do not compromise the transmission of the spindle assembly checkpoint signal by Bfa1 to inhibit mitotic exit in response to MT depolymerization.
Overexpression of is toxic to cells. This growth defect is suppressed by the deletion of either or (). Thus, if Kin4 acts via the phosphorylation of serine residues 150 and 180 of Bfa1, cells should be resistant to overexpression. As reported previously (), - cells failed to grow on galactose plates (, rows 2 and 4). In contrast, - cells grew (, row 3). Importantly, cells were resistant to overexpression (, compare row 5 with rows 3 and 4). This implies that acts by phosphorylating serines 150 and 180 of Bfa1.
If cells are insensitive to Kin4 regulation, cells should show a similar SPOC defect to that exhibited by cells. This prediction was tested by the introduction of a background mutation. In 30% of large-budded cells, the anaphase spindle was misaligned in the mother cell body (). Only 6% of cells with a misaligned spindle progressed through the cell cycle, as indicated by the accumulation of abnormal cell types with three DAPI-staining regions (). The number of large-budded cells with an SPOC-deficient phenotype increased in and cells to ∼40% and in cells to ∼20% (). In our hands, cells consistently showed a less pronounced SPOC defect than either or cells. Importantly, cells behaved in an identical manner to cells (). Together, Kin4 functions in SPOC regulation through modifying serines 150 and 180 of Bfa1.
Phosphorylation of Bfa1 by Kin4 may inhibit the ability of Cdc5 to modify Bfa1 and may stop Cdc5 from inducing mitotic exit (). To test this model, we incubated and cells with nocodazole. The ensuing MT depolymerization brought Bfa1, Kin4, and Cdc5 together at both SPBs. Kin4 could then be able to phosphorylate Bfa1, which could then prevent subsequent modifications of Bfa1 by Cdc5. In contrast, Bfa1 should behave differently in this model. Specifically, even in the presence of Kin4, Bfa1 should still become phosphorylated by Cdc5 because the Bfa1 mutant protein would no longer be able to act as a substrate for Kin4 kinase and would not be able to confer the protection from Cdc5 that phosphorylation on these sites normally imparts. Consistent with this model, Bfa1 of wild-type cells was hypophosphorylated after nocodazole treatment (, lane 2), whereas Bfa1 (, lane 4) accumulated as hyperphosphorylated protein in a similar manner to the accumulation of Bfa1 in cells (). In the absence of Cdc5, Bfa1 remained hypophosphorylated (, compare lane 4 with lane 6), demonstrating that Cdc5 was the kinase that was responsible for the reduced mobility of Bfa1. We conclude that phosphorylation by Kin4 kinase protects Bfa1 from inhibitory phosphorylation by Cdc5.
Kin4 associates with the cortex of the mother cell in mid- anaphase and with the SPB in the mother cell body but not the SPB in the bud (; ). However, after SPOC activation, Kin4 associates with both SPBs (). Thus, the control of Bfa1 by Kin4 that we outline here could arise from a modification of Bfa1 at either the SPB, the cell cortex, or at both locations. Therefore, we asked whether Kin4 at the plasma membrane is able to promote SPOC function. For this analysis, was fused to the C terminus of (amino acids 301–322; ). This domain of Ras2 can target proteins to the plasma membrane through prenylation of cysteine 318 (). To exclude the possibility that the fusion protein itself affects the function of Kin4, cysteines 318 and 319 of the membrane-targeting element of Ras2 were mutated to serine ().
First, we analyzed the localization of Kin4-pr and Kin4- pr-SS. To simplify the analysis, we arrested cells with nocodazole in metaphase and briefly induced expression of the - derivatives by the addition of galactose. GFP-Kin4 associated with the cortex of the mother cell and, as expected for nocodazole-treated cells, associated with both SPBs (, arrows; ). GFP–Kin4-pr-SS behaved as GFP-Kin4. In contrast, the entire cell cortex, including that of the bud, was decorated by GFP–Kin4-pr. This is probably a reflection of the fusion of the membrane-targeting sequence of Ras2. It is noteworthy that the GFP–Kin4-pr signal at the mother cortex was comparable with, or even stronger than, that of GFP-Kin4 or GFP–Kin4-pr-SS (). In addition, anchorage to the cortex effectively blocked the SPB association of GFP–Kin4-pr (). A weak GFP–Kin4-pr SPB signal could be detected in only ∼4% of cells (), and, when this signal was seen, it had an intensity that was <20% of the SPB signal observed in cells. Thus, in contrast to Kin4, Kin4-pr is restricted to the cell cortex and does not accumulate on SPBs.
In vitro measurements using Bfa1 as a substrate (, Coomassie brilliant blue [CBB]) demonstrated that the kinase activity of Kin4, Kin4-pr, and Kin4–pr-SS were similar (, P). Therefore, modification of the C terminus of Kin4 did not affect kinase activity.
Measurements using cells showed that was SPOC proficient. In contrast, behaved in the same checkpoint-deficient manner as cells (). This indicates that membrane-bound Kin4 is unable to sustain the SPOC. Thus, either continuous turnover at the cell cortex or association with the SPB is essential for the SPOC function of Kin4.
To target Kin4 to both SPBs, the region coding for amino acids 177–622 of was fused with (hereafter, we refer to this gene as ). was used as a fusion module because the C-terminal domain of Spc72 is able to target proteins efficiently to the cytoplasmic side of the SPB (; ). was expressed from the endogenous promoter as a sole copy of . Kin4-SPB protein localized to SPBs throughout the cell cycle (). It is important to note that Kin4-SPB associated with both the budward- and motherward-directed SPBs even when these cells had correctly aligned anaphase spindles (). This symmetric SPB localization was never observed for the wild-type Kin4 protein in cells with a correctly aligned spindle ().
Anaphase progression was analyzed in α factor–synchronized cells carrying . In cells, the correctly aligned anaphase spindles persisted for a longer time than in wild-type cells (). This phenotype was suppressed by the deletion of , suggesting that Kin4-SPB delays mitotic exit via modification of Bfa1 activity (). To further confirm this observation, we measured the duration of anaphase in cells by live cell imaging. The duration of anaphase was extended by a mean of 10 min when the timing in cells was compared with wild-type cells (wild type, 20.5 ± 4.0 min, = 34; , 30.8 ± 8.8 min, = 25). In contrast, the duration of anaphase in cells (21.4 ± 4.8 min, = 32) was indistinguishable from wild type (). Consistently, degradation of the mitotic cyclin Clb2 that accompanies mitotic exit was delayed in KIN4-SPB cells (unpublished data). These results suggest that if Kin4 is permanently recruited to both SPBs, it is capable of inhibiting mitotic exit regardless of the orientation of the spindle. This highlights the significance of the restricted association of Kin4 to the motherward SPB in undisturbed wild-type cells and further supports the notion that Kin4 as Cdc5 regulates Bfa1 at SPBs.
To better understand the role of the SPB in mediating the regulation of Bfa1 by Kin4, we sought SPB proteins that could physically interact with Kin4. Kin4 showed a strong two-hybrid interaction with Spc72 (unpublished data). In addition, Kin4 bound to Spc72 and Spc72 in pull-down experiments but was unable to associate with truncated Spc72 (). This suggested that Spc72 targets Kin4 to SPBs.
cells, whereas the cell cortex association was indistinguishable from that of wild-type cells (). Thus, Kin4 binding to the SPB requires amino acids 176–230 of Spc72.
cells, which are devoid of Kin4 at SPBs (), to ask whether the association of Kin4 with the SPB is important for SPOC function.
was combined with the mutation.
cells displayed a cold-sensitive growth defect (unpublished data). Only a small proportion of the population of cells with misaligned anaphase spindles continued to progress through the cell cycle, as indicated by the accumulation of cell types (5.5%) characteristic for SPOC failure (, ).
cells also had misaligned anaphase spindles (∼10%) as a result of defective cytoplasmic MTs ().
cells failed to arrest cell cycle progression in anaphase. Instead, 18.4% of large-budded cells accumulated with a SPOC-deficient phenotype ().
as well as and cells exhibited similar degrees of SPOC defects.
cells to arrest in anaphase clearly demonstrates a functionalinput from Spc72 in SPOC regulation, probably through the recruitment of Kin4 to SPBs.
Finally, we asked whether Spc72 influences SPOC function solely via its interaction with Kin4 or whether it exerts an additional influence on SPOC function. To this end, we analyzed cells at the semipermissive temperature of 30°C. When grown at 30°C, a considerable fraction of cells enter anaphase with misaligned spindles and accumulate large-budded cells with multiple DAPI-staining regions, which are a hallmark of SPOC-deficient cells (). This phenotype was not enhanced by the deletion of , suggesting that functions downstream of in the SPOC pathway. Importantly, cells with misaligned anaphase spindles carried Kin4-GFP at both SPBs (), indicating that the absence of Kin4 at SPBs was not responsible for the SPOC defect of cells.
allele, and by a separate mechanism that is defective in cells grown at 30°C.
#text
Strains and plasmids are listed in Table S1 (available at ). Yeast strains were derivatives of YPH499 and were constructed by PCR-based methods (). The red fluorescent eqFP611 from the sea anemone () was used to mark SPBs through a fusion with (). Alternatively, four copies of the far-red fluorescent protein HcRed, which was discovered through site-directed and random mutagenesis efforts on a nonfluorescent chromoprotein (hcCP) isolated from the Indo-Pacific species, was used in some experiments (). The Cherry-tubulin construct was described previously ().
For synchronization, yeast cells were grown in YPAD (yeast extract, peptone, adenine, and dextrose) medium and arrested in G1 by treatment with 10 μg/ml α factor for 2.5 h or 2 h at 23°C or 30°C, respectively, until >95% of cells showed a mating projection. Cells were then washed with prewarmed growth medium to remove α factor and were resuspended in YPAD medium at the indicated temperatures. Cells with under the control of the promoter were grown in YPA + 3% raffinose and 2% galactose medium. For the depletion of Cdc5, cells were synchronized with α factor and released into YPAD medium to repress the promoter. Synthetic complete medium was used for live cell imaging experiments. To depolymerize MTs, cells were incubated in YPAD medium containing 15 μg/ml nocodazole at 30°C for 2.5 h. Cells carrying were incubated at 14°C for 20 h before examination.
GST and GST-fused proteins were expressed in Rosetta (DE3) by adding 0.1 mM IPTG for 2 h at 37°C. MBP and all MBP-fused proteins were expressed in Rosetta (DE3) and purified according to the manufacturer's protocol (New England Biolabs, Inc.). After intensive washing steps, GST and GST fusion proteins bound to glutathione–Sepharose beads were used for in vitro pull-down assays. For in vitro kinase assays, all recombinant proteins were eluted either with glutathione in the case of GST fusion proteins or with maltose for MBP fusion proteins. Proteins were then dialyzed against a buffer containing 20 mM Hepes, pH 7.4, 0.5 mM EDTA, and 10% glycerol. All recombinant proteins were kept at −20°C except for MBP-Bfa1, which was stored at −80°C.
Yeast total cell extracts were prepared from logarithmically growing cells in immunoprecipitation buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 5% glycerol, 0.2 mM NaVO 100 mM β-glycerophosphate, 50 mM NaF, 1 mM PMSF, 1 mM DTT, 0.5% Triton X-100, and Complete EDTA-free protease inhibitor cocktail [Roche]). 300 μg of total cell extract was incubated with GST or GST-Spc72–bound beads for 2 h at 4°C. The beads were washed three times with immunoprecipitation buffer. Kin4-3HA was detected with the mouse monoclonal antibody 12CA5.
GST-Cdc5 and GST-Cdc5 used in and were expressed in yeast strains provided by S. Sedgwick (National Institute for Medical Research, London, UK) and were purified as described previously (). GST-Kin4 and GST-Kin4 used in were purified from the yeast strains listed in Table S1 using the same protocol. Kin4-3HA, Kin4-3HA, GFP-Kin4, and GFP-Kin4 derivatives in Figs. S3 and were enriched as follows. Total cell extracts were prepared in immunoprecipitation buffer (50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 5% glycerol, 0.2 mM NaVO, 100 mM β-glycerophosphate, 50 mM NaF, 1 mM PMSF, 1 mM DTT, 1% NP40, and Complete EDTA-free protease inhibitor cocktail [Roche]). Kin4 kinase was immunoprecipitated with mouse monoclonal antibody 12CA5 or affinity-purified rabbit anti-GFP antibody bound to protein A–coupled beads. Kin4 and Cdc5 kinase assays were performed as previously described (; ). Radioactivity was either detected with BioMax MS films (Kodak) or determined by a phosphoimager (FLA-300; Fujifilm) and quantified with Image Gauge version 3.45 (Fujifilm). To determine Kin4-dependent phosphorylation sites by MALDI-TOF/TOF, MBP-Bfa1 was incubated with GST-Kin4 purified from the strain provided by S. Sedwick in the same buffer as described for the other Kin4 kinase assays but in the presence of 6.7 mM ATP and was incubated at room temperature for 1 h.
Affinity-purified sheep anti-Cdc5, -Bfa1, -Clb2, and -Tub2 antibodies were described previously (). Anti-Nud1 and -Spc72 antibodies have been described previously (; ). Anti-Sic1 antibodies were raised in guinea pigs. Mouse monoclonal antibody 12CA5 was used to detect Bfa1-3HA and Kin4-3HA. Anti–P-S150 and anti–P-S180 antibodies of Bfa1 were raised in rabbits against peptides RLKQPRS()MMELK and VRFKKS()MPNL, respectively. The antibodies were purified using the nonphosphorylated peptide as preadsorption matrix followed by affinity purification using the immobilized phosphopeptides.
Yeast cells with Kin4-GFP grown in YPAD were immediately analyzed by fluorescence microscopy without washing or fixation. Other cells were fixed with 70% ethanol, washed with PBS, and incubated in PBS containing DAPI to visualize DNA. , , and cells were analyzed by fluorescence microscopy after fixation with 4% PFA in 150 mM phosphate buffer, pH 6.5, for 10 min at room temperature. PFA-fixed cells were incubated with 1 μg/ml DAPI for 15 min at room temperature. Z-series images of 0.35-μm steps were captured with a microscope (Axiophot; Carl Zeiss MicroImaging, Inc.) equipped with a 100× NA 1.45 Plan-Fluar oil immersion objective (Carl Zeiss MicroImaging, Inc.) and a camera (Cascade:1K; Photometrics) and were processed with MetaMorph software (Universal Imaging Corp.; ; ; ; ; S2 A; and S4 B), or images were captured with DeltaVision (Applied Precision) equipped with GFP and TRITC filters (Chroma Technology Corp.), a 100× NA 1.4 plan Apo oil immersion objective (IX70; Olympus), and a camera (CoolSNAP HQ; Roper Scientific) and were quantified/processed with SoftWoRx 3.5.0 (Applied Precision; ; ; and ). Projected images are shown. The fluorescence intensity of Cdc5-GFP was measured on a plane that has an SPB in focus with SoftWoRx 3.5.0. Time-lapse experiments were carried out in synthetic complete medium on a concanavalin A–coated glass-bottom dish (MatTek) with DeltaVision at 30°C. Z series at 0.4-μm steps (2 × 2 binning) were acquired every 1 min. Photoshop (Adobe) was used to mount the images and to produce merged color images. No manipulations other than contrast and brightness adjustments were used.
Fig. S1 shows the cell cycle progression of cells in Fig. S2 presents an analysis of cells. Fig. S3 shows an in vitro kinase assay of Kin4. Fig. S4 presents an analysis of cells. Table S1 presents the strains and plasmids used in this study. Video 1 shows anaphase of a wild-type cell, Video 2 shows anaphase of a cell, and Video 3 shows anaphase of a cell. Online supplemental material is available at . |
An important aspect of the control of eukaryotic gene expression is the regulation of the mRNA translation and degradation, which are often intertwined. The control of translation and mRNA degradation can involve conserved cytoplasmic RNA granules, which are defined as microscopically visible granules containing mRNA and proteins (). One class of cytoplasmic RNA granule is termed a processing body (P-body). P-bodies are dynamic aggregates of untranslating mRNA in conjunction with translation repressors and mRNA degradation components (; ). P-bodies and the mRNPs assembled within them are of interest because they have been implicated in translation repression (; ), normal mRNA decay (; ), nonsense-mediated decay (; ), microRNA (miRNA)-mediated repression (; ), and mRNA storage (; ). In addition, P-bodies are related to cytoplasmic RNA granules containing translationally repressed mRNPs found in germ cells and neurons (; ).
The mechanisms by which P-bodies form are largely unknown. For example, the specific protein–protein and protein–RNA interactions that mediate the formation of an mRNP that is capable of being incorporated into a P-body are unclear. Similarly, the interactions that mediate the aggregation of mRNPs together into microscopically visible structures are unknown. An important issue is to understand how P-bodies assemble and what the consequences of different scales of assembly are to the translation and degradation of mRNA.
P-bodies in yeast through mammals contain a conserved core of proteins (; ; ). Some of these core proteins are involved in mRNA degradation, including the decapping enzyme Dcp1p/Dcp2p, the exonuclease Xrn1p, the Ccr4p–Pop2–Notp deadenylase complex, and the decapping activators Dhh1p, Pat1p, Lsm1-7p, and Edc3p. Some of these proteins, like Dhh1p and Pat1p, have also been implicated in translation repression (; ). Metazoan P-bodies contain additional components, including proteins involved in miRNA-mediated repression such as argonaute, GW182, and Mov10 (; ; ).
Several proteins have been implicated in P-body assembly. For example, in mammalian cells, P-bodies are greatly reduced by knockdown of GW182, RCK/p54, RAP55, Lsm4, Lsm1, Hedls/Ge-1, 4E-T, or miRNA biogenesis in general (; ; ; ; ). However, because translation and P-body formation compete with each other, the lack of a specific protein can affect P-body formation by either reducing the pool of nontranslating mRNA or decreasing the aggregation of nontranslating mRNPs into P-bodies. Strikingly, P-bodies are restored when translation initiation is inhibited by arsenite in mammalian cells depleted of Lsm4 (). This observation argues that Lsm4, and possibly some of the other mammalian factors as well, is not required for P-body assembly per se, but instead contributes to P-body formation by increasing the pool of translationally repressed mRNA. Thus, to understand how P-bodies form, it is important to distinguish whether a component functions in translation repression or in the actual aggregation of mRNPs with each other.
Genetic analyses of P-body assembly in yeast revealed that certain components are dependent on one another for their association with P-bodies and suggested that P-body formation was redundant, with partial roles being played by Dcp2p and Pat1p (). However, the role of the P-body component Edc3p was not investigated. By analyzing the role of Edc3p in P-body assembly, we identified direct physical interactions between components of yeast P-bodies that, in combination with previous observations, lead to a model for the core mRNP structure within P-bodies. We also demonstrated that these core mRNPs, or small mRNP complexes, can be aggregated into microscopically visible P-bodies either through Edc3p acting as a scaffold protein and interacting with itself to interconnect mRNPs or by an aggregation mechanism dependent on the glutamine/asparagine (Q/N)-rich prionlike domain in Lsm4p. These results suggest a stepwise model for P-body assembly with the initial formation of a core mRNA–protein complex that then aggregates through multiple specific mechanisms.
To test the role of Edc3p in P-body assembly, we examined the ability of strains to form P-bodies. In strains, Dcp1pGFP and Dcp2pGFP were no longer seen in the small P-bodies normally observed in mid-log cultures in wild-type strains (, left). Moreover, during glucose deprivation, where P-bodies are large (), we observed that strains showed a strong reduction in the accumulation of multiple proteins into microscopically visible P-bodies (, left), although some P-bodies could still form (see the following paragraph). The strain also showed reduced P-body assembly in response to osmotic stress and in the stationary phase (not depicted). It should be noted that we were unable to determine if P-bodies too small to be detected in the light microscope were forming (see Discussion). Nevertheless, the strong reduction of P-bodies, as judged by six different proteins and under different growth conditions, indicates that Edc3p affects the formation of P-bodies.
We observed that the magnitude of the reduction of P-bodies in the strain was dependent on the experimental conditions used. Most strikingly, if aeration was limited during glucose deprivation, Edc3p was required for P-body accumulation (). In contrast, if cells were deprived of glucose while being strongly aerated, strains still formed P-bodies, although at reduced levels compared with wild-type cells (). These results suggested that there is also an Edc3-independent mode for forming P-bodies that is dependent on mitochondrial respiration. To test this possibility, we isolated and examined petite versions of wild-type and strains that are respiratory deficient (). P-bodies increased in response to glucose deprivation in respiratory-deficient, but otherwise wild-type, cells, as judged by the increase in Dcp2GFP and Dhh1GFP foci (). In contrast, no Dcp2GFP or Dhh1GFP foci were observed after glucose removal in the strain that was respiratory deficient (). These results indicate that Edc3p plays a role in P-body assembly, although there is a second respiration-dependent mechanism during glucose deprivation that can contribute to P-body assembly.
Additional evidence that Edc3p affects, but is not required for, P-body assembly is that the large P-bodies that form in or strains because of the resulting defects in decapping and 5′-to-3′ degradation (; ) are reduced but not eliminated in and strains ().
The strong reduction of P-bodies in the respiratory-deficient strain during glucose deprivation suggests that Edc3p could either be required for the aggregation of mRNPs into P-bodies or for translational repression during glucose deprivation. To distinguish these two possibilities, we examined whether the respiratory-deficient strain could repress translation during glucose deprivation, as judged by polysome analysis.
We observed that both respiratory-deficient wild-type or cells showed a similar reduction in polysomes during glucose deprivation (). Because Edc3p is required for P-body formation under these conditions, this observation indicates that Edc3p primarily functions in the aggregation of nontranslating mRNPs into P-bodies. This result also implies that aggregation of mRNPs into microscopically visible P-bodies is not required for translation repression, at least in response to glucose deprivation.
These results identify Edc3p as a factor that contributes to the assembly of microscopically visible P-bodies. Edc3p is a member of the Lsm16 family of proteins and contains at its N terminus a divergent Lsm, or like-Sm, domain, a central FDF domain of unknown function containing the amino acid motif FDF, and a C-terminal YjeF-N domain that resembles the N-terminal domain of the Yjef protein (; ). To determine how these different Edc3 domains contribute to P-body assembly, we examined the interactions of each of these domains in two-hybrid tests with other P-body components and verified interactions by direct binding experiments with recombinant proteins.
In two-hybrid assays (), full-length Edc3p interacted with Edc3p, Dcp1p, the C-terminal structural domain of Dhh1p (), and the conserved region of Dcp2p (amino acids 1–300). The conserved region of Dcp2p is composed of two distinct structural domains: an N-terminal domain that interacts with Dcp1p and a C-terminal domain that contains the catalytic site (). The C-terminal catalytic domain of Dcp2p encompassing amino acids 102–300 was sufficient to interact with full-length Edc3p. Subsequent experiments revealed that each of these interactions involved a specific domain of Edc3p.
Two observations indicate that the catalytic domain of Dcp2p (residues 102–300) primarily interacts with the Lsm domain of Edc3p. First, the Lsm domain is sufficient in the two-hybrid assay to interact with residues 102–300 of Dcp2p (). Second, a His-tagged Lsm domain binds to recombinant Flag-tagged Dcp2p (either 1–300 or 102–300) in an in vitro pull-down assay from bacterial lysates (). Lysates were used because the Lsm domain became insoluble when concentrated during purification. The Lsm domain was not pelleted by the anti-Flag resin alone () or by GST-tagged Dhh1p (), indicating that the Lsm domain–Dcp2p interaction is specific. These results identify the Lsm domain as a region of Edc3p that directly binds the catalytic domain of Dcp2p.
Two observations indicate that Dcp2p also interacts with the FDF domain of Edc3p. First, although the FDF domain is not sufficient for a two-hybrid interaction with Dcp2p, a combined Lsm–FDF construct shows a stronger interaction with the catalytic domain of Dcp2p than just the Lsm domain (). Second, the purified FDF domain could be affinity purified with both the Flag-tagged 1–300–amino acid fragment and the 102–300–amino acid fragment of Dcp2p, although this interaction is sensitive to moderate salt (). These results identify the FDF domain as a region that can directly bind Dcp2p, albeit weakly.
Two observations indicate that the FDF domain can directly interact with the C-terminal domain of Dhh1p as well. First, the FDF domain was sufficient in two-hybrid experiments to interact with this portion of Dhh1p (). Second, recombinant Dhh1p interacted with recombinant FDF domain in an in vitro pull-down assay (). Moreover, the in vitro interaction between the FDF domain and Dhh1p only required the C-terminal domain of Dhh1p (). In contrast, purified Dhh1p failed to interact with the Lsm domain of Edc3p in pull-down experiments (). These results identify the FDF domain as a site that directly binds Dhh1p.
The YjeF-N domain of Edc3p can interact with itself and was sufficient to interact with Edc3p (). This is consistent with genome-wide analyses that have identified the C terminus of Edc3p as being sufficient to interact with itself by two-hybrid or phage display experiments (; ).
We also obtained evidence that the C-terminal domain of Dhh1p binds directly to the catalytic domain of Dcp2p. Specifically, we observed a two-hybrid interaction between these domains of Dhh1p and Dcp2p (unpublished data). In addition, in binding experiments with purified recombinant proteins, GST-tagged Dhh1p pulled down Dcp2p (), and reciprocally GST-tagged Dhh1p was pulled down by Flag-tagged Dcp2p (not depicted). In these experiments, the catalytic domain of Dcp2p (residues 102–300) was sufficient to bind to Dhh1p (). In addition, the C-terminal domain, but not the N-terminal domain, of Dhh1p was sufficient to bind to Dcp2p (unpublished data). These data indicate that the C-terminal domain of Dhh1p directly interacts with the catalytic domain of Dcp2p.
These results identify three different domains of Edc3p that interact with specific P-body components. To determine how these domains contribute to P-body formation, we deleted each domain of Edc3p and asked if the mutant protein could rescue the P-body aggregation defect of the respiratory-deficient strain, where P-body formation is strongly dependent on Edc3p. We examined the formation of Dcp2GFP foci in both exponentially growing cells and during glucose deprivation. As expected, expression of full-length Edc3p complemented the P-body aggregation defect of the strain under both conditions (). Edc3p lacking the FDF domain functioned in P-body assembly (). In contrast, deletion of either the Lsm or Yjef-N domains caused defects in the formation of Dcp2GFP () and Dhh1GFP foci (Fig. S1, available at ) both in exponentially growing cells and under glucose deprivation. The defect in P-body aggregation in cells expressing the ΔLsm and ΔYjef-N proteins was not simply caused by decreased expression because the amount of these mutant proteins was not substantially decreased relative to full-length Edc3p (Fig. S2, A and B). Moreover, the decrease in Dcp2GFP foci was not caused by the deletion of the Lsm or Yjef-N domains of Edc3p reducing the levels of Dcp2 protein (Fig. S2 C). These results identify the Lsm and Yjef-N domains of Edc3p as being required for P-body aggregation and suggest that Edc3p acts as a scaffold protein that links multiple mRNPs together.
As described earlier, our analysis indicated that there is also an Edc3-independent mechanism by which microscopically visible P-bodies can form. Interestingly, a component of the Lsm1-7p complex, Lsm4p, has a C-terminal Q/N-rich domain, which has been categorized as a prionlike domain (). Moreover, Lsm4p can function in a similar manner to known yeast prion proteins in that it can induce the [PSI+] prion when overexpressed (). Thus, we hypothesized that the Edc3p-independent mode of P-body aggregation might involve the Lsm4p Q/N-rich domain. To test this model, we deleted the C-terminal 97 amino acids encompassing the Q/N-rich sequence within Lsm4p and assayed the effect of this deletion on P-body formation in the absence of Edc3p. The and strains were viable, indicating that the truncated protein is being expressed because Lsm4p is essential.
An important result was that the strain was defective in P-body formation during glucose deprivation (, right). In contrast, deletion of the C terminus of Lsm4 protein by itself did not reduce P-bodies during mid-log growth or have a strong effect on P-body formation in response to glucose deprivation (). This result indicates that either Edc3p or the prionlike domain of Lsm4p is sufficient for P-body aggregation during glucose deprivation. It should be noted that small Dcp2GFP foci were occasionally observed in the strain, indicating that microscopically visible P-bodies can assemble, though inefficiently, in the absence of both Edc3p and the C terminus of Lsm4p. Nevertheless, the severe defect in P-body formation in the mutant identifies the Q/N-rich domain of Lsm4p as a second protein domain that can affect the formation of microscopically visible P-bodies.
Because Dcp2GFP was used as a marker for P-bodies, it was formally possible that the Q/N-rich domain of Lsm4p was solely required for the localization of Dcp2p to P-bodies in the absence of Edc3p. To verify that the deletion mutant was defective for P-body formation, we examined the localization of the reporter mRNA MFA2P-U1A, which contains binding sites for the U1A protein by coexpressing U1A protein fused to GFP. This reporter mRNA accumulates in P-bodies after glucose deprivation (). As shown in , the mutant has a severe defect in the accumulation of the MFA2P-U1A reporter mRNA in response to glucose deprivation. This observation supports the interpretation that both the Q/N-rich domains of Lsm4p and Edc3p contribute to P-body formation.
The Q/N domain of Lsm4p could affect P-body formation by being required for translation repression during glucose deprivation in the strain or the aggregation of translationally repressed mRNPs in the absence of Edc3p. To distinguish these two possibilities, we examined the extent of translation repression in the strain during glucose deprivation. We observed that deletion of the Lsm4 prionlike domain in combination with deletion of Edc3p did not interfere with translation repression under these conditions (). This indicates that the defect in P-body formation in the strain during glucose deprivation is not caused by a failure to repress translation. Thus, these results demonstrate that a second mechanism for P-body aggregation is based on the Lsm4 Q/N-rich domain.
The simplest model for how the Lsm4 Q/N-rich domain contributes to P-body formation is that it aggregates with itself, or other Q/N-rich domains, in a manner analogous to the assembly mechanism for Q/N-rich domains in known prions. However, given the dynamic and reversible nature of P-body assembly, such interactions mediated by the Lsm4 Q/N-rich domain would not be heritable like a classical prion. To test whether the Q/N-rich domain of Lsm4 functions analogously to a prion, we asked whether it could be substituted by the Q/N-rich prion domain of a known yeast prion protein, Rnq1p (). The C terminus of Lsm4p was replaced by the prion domain of Rnq1p, and the fusion protein was tested for its ability to form Dcp2GFP foci in cells lacking both Lsm4p and Edc3p. In the strain expressing the Lsm4Rnq1 fusion protein, Dcp2GFP foci were observed in exponentially growing cells and were increased after glucose deprivation (, bottom). Thus, the prion domain of Rnq1p was able to functionally replace the Q/N-rich domain of Lsm4p for the assembly of P-bodies, which is consistent with Lsm4p promoting P-body assembly via a prionlike aggregation mechanism. Interestingly, the Dcp2GFP foci seen in exponentially growing cells expressing the Lsm4Rnq1 fusion protein were more intense than those seen in cells expressing wild-type Lsm4 protein. The ability of the Rnq1 Q/N-rich domain to enhance P-body formation during exponential growth argues that specific features of the Lsm4 Q/N-rich domain are required for the highly reversible nature of P-body formation.
Our results define two major mechanisms that allow the formation of microscopically visible P-bodies. This provides an opportunity to examine the functional significance of aggregation of mRNPs into larger P-body structures. Because the strain represses translation during glucose deprivation, but is severely defective in forming visible P-bodies, we conclude that the efficient assembly of large visible P-bodies is not required for translation repression during glucose deprivation. To examine how the formation of large P-bodies contributes to mRNA decapping, we examined the decay of two reporter mRNAs, MFA2pG and PGK1pG (), in , , and strains as compared with the wild-type strain. Deletion of the Q/N-rich domain of Lsm4 protein and the Edc3 protein did not substantially alter the decay rate of the unstable MFA2pG mRNA () or the stable PGK1pG mRNA (not depicted), indicating that mRNA turnover is not strongly affected in cells lacking both P-body aggregation mechanisms. This result indicates that the aggregation of mRNPs together into microscopically visible P-bodies is not necessary for the basal control of mRNA decay rates. This observation indicates that assembly of a transcript into an individual mRNP containing the mRNA decapping proteins or into small complexes not detected by light microscopy is sufficient for mRNA decapping under normal conditions (see Discussion).
The Edc3 protein is known to enhance decapping under conditions where decapping rates are compromised by the conditional allele of dcp2-7 (). Given this, it was possible that the ability of Edc3p to enhance decapping by the dcp2-7 protein could result from its function in aggregating P-body components together, and thus concentrating decapping factors and substrate mRNA molecules together. Alternatively, because Dcp2p can interact directly with the Lsm and FDF domains of Edc3p, Edc3p might stimulate Dcp2p function by direct physical interactions.
To test whether P-body aggregation by Edc3p enhances decapping, we examined the ability of the different domain deletion mutants of Edc3p to complement the decapping defect seen in strains. As observed previously, exacerbates a partial defect in the decay of the MFA2pG mRNA at a low temperature caused by the temperature-sensitive allele dcp2-7 (). Deletion of also slows growth in cells (), in which 3′-to-5′ mRNA decay is inactive and decapping is partially defective because of the growth temperature (). An important result is that the Yjef-N deletion mutant complemented both the decay and growth defects caused by the absence of Edc3p (). Because the Yjef-N domain is necessary for P-body aggregation in exponentially growing cells, these results indicate that Edc3p's role in enhancing decay under these conditions is not dependent on its ability to form large P-body aggregates. In contrast, decay and growth were defective when either the ΔLsm or ΔFDF Edc3p variants were expressed (), which indicates that both the Lsm and FDF domains are necessary for Edc3p to enhance decapping. Because these are the domains that interact with Dcp2p and Dhh1p, this result suggests that Edc3p may enhance decapping by binding to and directly affecting the function of the decapping machinery.
e
f
o
r
m
a
t
i
o
n
o
f
c
y
t
o
p
l
a
s
m
i
c
P
-
b
o
d
i
e
s
c
a
n
b
e
a
s
s
o
c
i
a
t
e
d
w
i
t
h
t
r
a
n
s
l
a
t
i
o
n
r
e
p
r
e
s
s
i
o
n
a
n
d
/
o
r
m
R
N
A
d
e
c
a
p
p
i
n
g
a
n
d
d
e
g
r
a
d
a
t
i
o
n
.
A
n
u
n
r
e
s
o
l
v
e
d
i
s
s
u
e
i
s
h
o
w
P
-
b
o
d
i
e
s
f
o
r
m
a
n
d
t
h
e
f
u
n
c
t
i
o
n
a
l
s
i
g
n
i
f
i
c
a
n
c
e
o
f
a
s
s
e
m
b
l
y
i
n
t
o
l
a
r
g
e
r
s
c
a
l
e
a
g
g
r
e
g
a
t
e
s
.
I
n
t
h
e
s
i
m
p
l
e
s
t
m
o
d
e
l
,
t
h
e
f
o
r
m
a
t
i
o
n
o
f
P
-
b
o
d
i
e
s
c
a
n
b
e
c
o
n
s
i
d
e
r
e
d
t
o
o
c
c
u
r
i
n
a
t
l
e
a
s
t
t
w
o
d
i
s
t
i
n
c
t
s
t
e
p
s
:
a
n
i
n
i
t
i
a
l
s
t
e
p
w
h
e
r
e
i
n
a
n
i
n
d
i
v
i
d
u
a
l
m
R
N
A
i
s
a
s
s
o
c
i
a
t
e
d
w
i
t
h
v
a
r
i
o
u
s
p
r
o
t
e
i
n
s
t
o
f
o
r
m
a
n
m
R
N
P
c
a
p
a
b
l
e
o
f
a
g
g
r
e
g
a
t
i
o
n
i
n
t
o
a
l
a
r
g
e
r
P
-
b
o
d
y
,
a
n
d
a
s
e
c
o
n
d
s
t
e
p
w
h
e
r
e
b
y
i
n
d
i
v
i
d
u
a
l
m
R
N
P
s
a
r
e
t
h
e
n
a
g
g
r
e
g
a
t
e
d
t
o
g
e
t
h
e
r
t
o
f
o
r
m
P
-
b
o
d
i
e
s
o
f
s
u
f
f
i
c
i
e
n
t
s
i
z
e
t
o
b
e
v
i
s
i
b
l
e
i
n
t
h
e
l
i
g
h
t
m
i
c
r
o
s
c
o
p
e
.
H
o
w
e
v
e
r
,
i
t
a
l
s
o
r
e
m
a
i
n
s
p
o
s
s
i
b
l
e
t
h
a
t
t
h
e
r
e
a
r
e
i
n
t
e
r
m
e
d
i
a
t
e
s
t
e
p
s
i
n
t
h
e
a
g
g
r
e
g
a
t
i
o
n
o
f
m
R
N
P
s
t
h
a
t
c
a
n
a
s
s
e
m
b
l
e
c
o
l
l
e
c
t
i
o
n
s
o
f
m
R
N
P
s
t
o
o
s
m
a
l
l
t
o
b
e
s
e
e
n
i
n
t
h
e
l
i
g
h
t
m
i
c
r
o
s
c
o
p
e
.
B
y
e
x
a
m
i
n
a
t
i
o
n
o
f
P
-
b
o
d
y
f
o
r
m
a
t
i
o
n
i
n
y
e
a
s
t
,
w
e
w
e
r
e
a
b
l
e
t
o
i
d
e
n
t
i
f
y
i
m
p
o
r
t
a
n
t
i
n
t
e
r
a
c
t
i
o
n
s
t
h
a
t
p
r
o
m
o
t
e
P
-
b
o
d
y
f
o
r
m
a
t
i
o
n
.
T
h
e
s
e
i
n
t
e
r
a
c
t
i
o
n
s
s
u
g
g
e
s
t
a
n
i
n
i
t
i
a
l
m
o
d
e
l
f
o
r
P
-
b
o
d
y
a
s
s
e
m
b
l
y
i
n
v
o
l
v
i
n
g
i
n
t
e
r
a
c
t
i
o
n
s
o
f
s
u
b
c
o
m
p
l
e
x
e
s
o
f
c
o
r
e
P
-
b
o
d
y
c
o
m
p
o
n
e
n
t
s
w
i
t
h
t
h
e
m
R
N
A
a
n
d
e
a
c
h
o
t
h
e
r
,
f
o
l
l
o
w
e
d
b
y
a
g
g
r
e
g
a
t
i
o
n
o
f
i
n
d
i
v
i
d
u
a
l
m
R
N
P
s
,
o
r
s
m
a
l
l
c
o
m
p
l
e
x
e
s
o
f
m
R
N
P
s
,
i
n
t
o
l
a
r
g
e
r
P
-
b
o
d
i
e
s
b
y
a
s
e
l
f
-
i
n
t
e
r
a
c
t
i
o
n
d
o
m
a
i
n
o
f
E
d
c
3
p
a
n
d
t
h
e
Q
/
N
-
r
i
c
h
C
-
t
e
r
m
i
n
a
l
t
a
i
l
o
f
L
s
m
4
p
.
Strains used in this study are listed in Table S2 (available at ). Strains with C-terminal GFP fusion proteins were derived from strains described previously (; ). C-terminal deletion of LSM4 was constructed using a PCR-based protocol (). The Dcp2GFP strain was constructed from a single cross between yRP2164 and a haploid strain derived from a heterozygous diploid strain (Open Biosystems). Strains were grown at 30°C unless otherwise stated in either yeast extract/peptone medium (YP) or synthetic medium (SC) supplemented with appropriate amino acids and 2% glucose. Respiratory-deficient (petite) derivatives of wild-type or strains were obtained by screening for colonies that had spontaneously lost the ability to grow on glycerol. Plasmids and oligonucleotides used in this study are listed in Table S3.
Cultures were grown to OD 600 of 0.3–0.4, collected by centrifugation, washed, and resuspended in SC plus amino acids supplemented with glucose. For glucose depletion in , after washing and resuspension in YP without glucose, cells were incubated in a flask in a shaking water bath for 10 min and washed with SC without glucose. For glucose depletion without aeration, cells were washed and resuspended in SC without glucose and incubated in a microcentrifuge tube. For glucose depletion with aeration, after washing and resuspension in SC without glucose, cells were incubated in a flask in a shaking water bath for 10 min. Images for were acquired with a confocal microscope (PCM 2000; Nikon) using an objective (Plan Apo 100× 1.4 NA; Nikon) with 3× magnification using a photomultiplier tube (R928; Hamamatsu Photonics) and software (Compix Media). All images are a maximum intensity projection of a z series compilation of 6–10 images performed by Compix software. Images for , , and were acquired with a deconvolution microscope (Deltavision RT; Applied Precision) using an objective (UPlan Sapo 100× 1.4 NA; Olympus). They were collected using software (softWoRx) as 512 × 512-pixel files with a camera (CoolSNAP HQ; Photometrics) using 1 × 1 binning. Images are maximum intensity projections of a z series compilation of 15 images made using Image J and have been adjusted to the same contrast range except in .
Cultures at OD 600 of 0.4–0.6 were split, washed, and resuspended with YP with or without glucose, and then incubated in a shaking water bath at 30°C for 10 min. Cells were harvested by centrifugation at 4,000 rpm for 1 min at 4°C over ice, washed in lysis buffer (10 mM Tris-HCl, pH 7.4, 100 mM NaCl, and 30 mM MgCl2), and stored at −80°C. Lysates were prepared by vortexing cells with glass beads at 4°C in lysis buffer containing 0.5 mg/ml heparin and 1mM DTT and clarified by centrifugation for 2 min at 4,000 rpm at 4°C. 10 A254 units of clarified lysate were loaded on 15–50% sucrose gradients containing 50 mM Tris-HCl, pH 7.0, 50 mM NHCL, 12 mM MgCl, and 1 mM DTT, sedimented in a rotor (SW41; Beckman Coulter) using an ultracentrifuge (L8-M; Beckman Coulter) at 4°C for 2.5 h at 39,000 rpm, and collected while the A254 value was monitored using a continuous flow cell UV detector (UA-6; Teledyne Isco).
Two-hybrid fusion plasmids were constructed by homologous recombination of PCR products into pOAD or pOBD-2 in yeast strains PJ694a and PJ694α () as described previously (). Two-hybrid plasmids and strains were obtained from the Yeast Resource Center (provided by S. Fields, University of Washington, Seattle, WA). Interactions were measured by β-galactosidase plate assays in diploids containing pOAD and pOBD-2 derivatives.
Purification of His-Dcp2(1–300)-Flag and His-Dcp2(102–300)-Flag was performed as described previously (). GST-tagged versions of Dhh1p were expressed in BL-21 and purified as described previously (), except that only the first purification step on glutathione-Sepharose was performed. The GST-Dhh (46–461) plasmid was provided by H. Song (Institute of Molecular and Cell Biology, Proteos, Singapore). The His-tagged FDF domain of Edc3p was purified from BL-21 (DE3) using His-bind resin (Novagen) after incubation with 1 mg/ml lysozyme for 30 min on ice in 20 mM Tris-HCl, pH 7.9, 500 mM NaCl, and 5 mM imidazole containing complete protease inhibitors without EDTA (Pierce Chemical Co.) followed by sonication. Lysates containing a His-tagged Lsm domain of Edc3p were prepared from BL-21 (DE3) by lysing cells in 20 mM Tris-HCl, pH 7.9, 100 mM NaCl, and 10% glycerol containing 1 mg/ml lysozyme for 30 min followed by sonication and clarification.
Binding reactions were performed at 4°C in binding buffer (50 mM Hepes, pH 7.0, 100 mM NaCl, 1 mM DTT, 2 mM MnCl, 2 mM MgCl, 1% Triton X-100 (Fisher Scientific), 10% glycerol, and 10 mg/ml BSA) containing ∼20 ng/μl His-Lsm or His-FDF and 25–67.5 ng/μl GST-Dhh1 or His-Dcp2-Flag proteins. Because of the low concentration of His-Lsm in the extracts, the NaCl concentration was increased to 225 mM. Glutathione-Sepharose 4B (GE Healthcare) and anti-Flag M2 agarose (Sigma-Aldrich) was used to pull down GST-Dhh1 and His-Dcp2-Flag, respectively. Western analysis was performed using polyclonal anti-His antibody (Santa Cruz Biotechnology, Inc.).
Western analysis of Edc3 and Dcp2GFP proteins was performed by preparing extracts from wild-type and strains expressing Flag-tagged versions of Edc3p. Flag-tagged full-length ΔLsm and ΔFDF Edc3 proteins were detected using anti-Flag M2 monoclonal antibody (Sigma-Aldrich). The ΔYjef-N Edc3 protein was detected using a polyclonal antibody raised in rabbits to a purified His-tagged FDF domain (Cocalico Biologicals) because it co-migrated with a nonspecific band detected by the anti-Flag antibody. Dcp2GFP was detected using anti-GFP antibody (Covance).
Cells were grown to OD 600 of 0.3–0.4 in media containing 2% galactose at 30 or 24°C for experiments using , and then transcription was repressed by resuspending in media with 4% glucose. Total RNA was extracted (), and the amount of MFA2pG mRNA was quantified by Northern analysis with oRP140 () using a phosphorimager (Molecular Dynamics). Loading corrections were performed as described previously ().
Table S1 lists Q/N-rich regions in proteins involved in RNA metabolism and location of conserved Q/N-rich regions in metazoan P-body proteins. Table S2 lists yeast strains used in this paper. Table S3 lists plasmids and oligonucleotides used in this paper. Fig. S1 shows the effect of Edc3p domain deletions on Dhh1GFP foci. Fig. S2 shows levels of mutant Edc3p proteins and Dcp2GFP. Online supplemental material is available at . |
Removal of introns from primary RNA transcripts (splicing) takes place in specialized complexes called spliceosomes, in which factors needed for splicing of pre-mRNAs are enriched. Currently, >150 different proteins and several small RNAs have been identified as part of spliceosomes, which are organized in distinct subcomplexes. The most prominent spliceosome subunits are the uridine-rich small nuclear RNPs (U snRNPs) of the Sm class. They consist of an RNA component (uridine-rich small nuclear RNA [U snRNA]) and numerous proteins that are either common for all or specific for one particle (for review see ).
Even though splicing occurs in the nucleus, major parts of the biogenesis of U snRNPs take place in the cytoplasm. The nuclear-encoded mG-capped U snRNA is transiently exported to the cytoplasm to allow binding of the common (Sm) proteins. This leads to the formation of the Sm core domain, the structural framework of all spliceosomal U snRNPs of the Sm class (). Formation of the Sm core is required for cap hypermethylation and the subsequent nuclear import of U snRNPs (). Within the nucleus, U snRNPs are first targeted to subnuclear domains termed Cajal bodies (CBs), where additional modifications on the RNA occur and at least some specific proteins are added. Eventually, the mature U snRNPs migrate to perichromatin fibrils, the sites of transcription and splicing (for reviews see ; ).
Interestingly, recent studies indicated that several aspects of the biogenesis cycle of U snRNPs are factor mediated and regulated in vivo. The most prominent factor in this process is the survival motor neuron (SMN) complex, a macromolecular entity that actively mediates the binding of the common Sm proteins onto U snRNAs. This complex consists of nine major proteins, including the SMN gene product, Gemin2–8, and the unr-interacting protein (unrip; for reviews see ; ; ) (; ). The SMN complex is controlled by another complex, whose name-giving component is the type-II protein arginine methyltransferase 5. This unit, possibly in conjunction with other factors, converts arginine residues in some Sm proteins into symmetrical dimethylarginines, thereby enhancing their affinity for the SMN complex and stimulating U snRNP assembly (; ; ; ). Furthermore, it has been shown that the SMN complex (or parts thereof) also participate in the subsequent nuclear import of U snRNPs (; ). Once in the nucleus, both units migrate to CBs, where the SMN complex accumulates and U snRNPs are released to sites of splicing after additional maturation steps (). These observations suggest that U snRNPs dissociate from SMN complexes in CBs and that the SMN complex returns as a separate unit to the cytoplasm at later stages.
Although the cytoplasmic role of the SMN complex is understood in some detail, its functions in the nucleus are only poorly characterized. Thus, it is still unclear how U snRNPs are separated from the SMN complex after nuclear import and how the return of the SMN complex to the cytoplasm is facilitated. An important player in this process might be unrip, which interacts with the SMN complex primarily in the cytoplasm. Knockdown of this factor leads to enhanced accumulation of SMN in nuclear bodies (), suggesting a role of unrip in the intracellular distribution of the SMN complex. The biogenesis of U snRNPs appears also to be influenced by phosphorylation of different components of the assembly machinery. Thus, it has been shown that SMN is highly phosphorylated when it is in the cytoplasm, whereas the nuclear pool is hypophosphorylated (). Compartment-specific determinants and the phosphorylation status of SMN (and potentially other SMN-complex components) could hence influence the biogenesis pathway of U snRNPs in the cytoplasm and in the nucleus.
Here, we show that the SMN complex specifically interacts with the nuclear phosphatase PPM1G/PP2Cγ (here referred to as PPM1G). This C-type, Mg/Mn-dependent phosphatase has previously been shown to be specifically involved in nuclear functions, including splicing (; ) and nucleosome assembly (). Knockdown of PPM1G led to reduced dephosphorylation of SMN-complex proteins and abolished accumulation of SMN proteins in CBs. Rescue experiments demonstrate that the catalytic activity of the phosphatase is necessary to maintain the specific localization of the SMN proteins in the nucleus. Our data support the idea that PPM1G is a major regulator of nuclear functions of the SMN complex: it determines the SMN complex's compartment-specific phosphorylation pattern and is required for its correct localization and stability.
As published recently, the compartment-specific phosphorylation pattern of the SMN protein pointed to the activity of a nuclear phosphatase, which dephosphorylates proteins of the SMN complex (). One potential candidate that may govern SMN modifications in the nucleus is PPM1G, which has been suggested to localize to the nucleus and to function in splicing (). Consistent with this, antibodies generated against the human () or the orthologue of PPM1G (xPPM1G; ) detected this protein in the nucleoplasm in both human HeLa and XL177 cells when used for indirect immunofluorescence (). Sequence analysis revealed two stretches of basic amino acids in the C terminus of human PPM1G (hPPM1G), which could serve as an NLS. The region of the potential C-terminal NLS is conserved throughout most vertebrates (). Indeed, when we expressed the full-length or a truncated version of hPPM1G lacking the C-terminal NLS as EYFP fusions in human HeLa cells, only the full-length protein showed nuclear accumulation, whereas the truncated version did not (). Likewise, accumulation of fluorescently labeled xPPM1G in in vitro reconstituted nuclei was not seen for analogous C-terminal truncations of xPPM1G, although the truncation did not affect the catalytic activity of xPPM1G (unpublished data). Moreover, when we manually dissected oocytes into cytosolic and nuclear fractions, we exclusively detected xPPM1G in the nuclear fraction using our xPPM1G antibody (). Using quantitative immunoblotting, we estimated the abundance of PPM1G in oocyte nuclei to be in the range of 5 μM and at least 1 μM in HeLa cell nuclei, which is consistent with recently published data (). This shows that PPM1G is an abundant nuclear phosphatase in vertebrates.
To address whether PPM1G might regulate nuclear functions of the SMN complex, we tested if PPM1G was directly associated with the SMN complex. We affinity-purified SMN-complex proteins from total cellular extracts using a monoclonal antibody directed against the N terminus of human SMN. Bound proteins were eluted and compared with an eluate of a control column by SDS-PAGE and silver staining. Known SMN-complex components (i.e., SMN, Gemin3–5, and unrip) were eluted from this column as determined by mass spectrometry and/or Western blotting (, compare lanes 2 and 3). Moreover, among the proteins coeluting with the SMN complex in the 70-kD range was the protein phosphatase PPM1G, which we identified on the basis of 21 peptides covering 32.9% of its sequence (not depicted). Immunoblot analysis with antibodies against hPPM1G confirmed that PPM1G specifically bound to the SMN complex but not to the control column (, compare lanes 2 and 3).
To further analyze binding of PPM1G to the SMN complex, we affinity purified the SMN complex, as described in the previous paragraph, and incubated the complex with recombinant His-tagged human wild-type (wt) PPM1G (rec.hPPM1Gwt). In parallel, we generated a mutant PPM1G (rec.PPM1Gmut) in which aa 496 was changed from an aspartate into an alanine (). The recombinant mutant protein showed dramatically reduced catalytic activity (unpublished data). After incubation and removal of unbound proteins by extensive washing, the SMN containing immunocomplexes were analyzed by SDS-PAGE and silver staining () and immunoblotting (). Detection of the SMN protein verified the specificity of the purification ( [bottom]; note that SMN could only be detected upon immunopurification with the specific antibodies [IP SMN] but not in controls [IP control]). PPM1Gwt and catalytically inactive PPM1G were equally detectable in the bound fraction of purified SMN complexes but not in controls ( [top]). Although the wt protein actively dephosphorylated the SMN protein, as indicated by a slightly changed mobility of SMN ( [SMN] and D [bottom]), interaction of either wt or catalytically inactive PPM1G with the SMN complex was equally efficient but did not change the overall composition of the complex as judged by silver staining of immunocomplexes and immunoblot detection of SMN and unrip (). Collectively, these data identify the nuclear phosphatase PPM1G as a novel interaction partner of the SMN complex.
We next asked whether PPM1G was required for proper localization of the SMN complex. Based on immunofluorescence experiments, it had been reported that SMN is diffusely distributed in the cytoplasm but localized in the nucleus, where it is found to be highly enriched in specific bodies termed gems (). However, later studies showed that these bodies are in most, but not all, cell lines equivalent to the well-known nuclear CBs (; ). In agreement with previous data, we observed strong accumulation of SMN in CBs of HeLa CCL2 cells (unpublished data). To analyze the role of PPM1G in the localization of SMN proteins, we knocked down PPM1G expression in HeLa CCL-2 cells by RNA interference using three different synthetic double-stranded siRNA oligonucleotides. As judged by immunoblotting, all three siRNAs efficiently reduced the protein amount of PPM1G (; the knockdown efficiency was typically 75–95%). Oligo pairs #1 and #3, however, were reproducibly the most efficient and therefore were used for further phenotypic analysis (). When SMN-complex localization was tested in siRNA-treated cells, we observed substantial reduction of the nuclear PPM1G in the majority of treated cells (). This was correlated with SMN disappearing from CBs (). Likewise Gemin2, a direct interaction partner of SMN, disappeared from CBs upon PPM1G knockdown (). Concomitantly, in ∼50% of all cells the diffuse nucleoplasmic signal of SMN or Gemin2 was increased (), whereas low nucleoplasmic SMN signals were observed in other cells (; and ). Similarly, upon overexpression of PPM1G lacking the C-terminal NLS () we frequently observed reduced cytoplasmic and increased diffuse nuclear staining of SMN. However, accumulation of SMN in CBs was not affected (Fig. S1, available at ). Collectively, these data suggest that nuclear PPM1G is required for correct nuclear-cytoplasmic distribution and subcellular localization of the SMN complex.
To confirm correlation of the efficiency of PPM1G knockdown and the loss of the localized nuclear SMN signal, we transfected cells with control siRNA oligos or siRNA oligos to knockdown PPM1G and mixed them after transfection. Indeed, when residual staining of PPM1G was hardly detectable, no accumulation of SMN in nuclear bodies was observed. In contrast, PPM1G was still clearly detectable in cells, which regularly showed normal SMN staining in the nucleus ().
Accumulation of SMN in CBs depends on the interaction of SMN with coilin (, ), one of the central molecular components of CBs (). We therefore determined whether down-regulation of PPM1G affected coilin localization. Knockdown of PPM1G frequently resulted in the accumulation of additional small coilin-positive structures (). However, coilin still strongly accumulated in CBs upon PPM1G knockdown (). Importantly, this was observed in cells in which loss of SMN from CBs was verified by direct SMN detection (). Collectively, these data suggest that PPM1G is required for the accumulation of SMN and Gemin2 in CBs but is not a major determinant of overall CB integrity as judged by coilin staining.
In the course of our experiments, we noted that the expression levels of SMN and Gemin2 were reduced upon knockdown of PPM1G, which is consistent with a destabilization of SMN-complex proteins. We therefore determined the amounts of different SMN proteins by quantitative immunoblotting after PPM1G knockdown with the two most effective oligo pairs. When PPM1G was depleted by ∼90%, the levels of SMN, Gemin2, and Gemin3 were also considerably reduced. This effect was specific, as the level of the nonrelated protein regulator of chromosome condensation 1 (RCC1) was not affected (). To determine where SMN was primarily destabilized, we knocked down PPM1G (, left), fractionated the cells, and determined amounts of SMN-complex proteins in the nuclear and cytosolic fractions by immunoblotting. Although levels of both Gemin2 and SMN were reduced in both fractions, protein loss was more pronounced in nuclear fractions. Interestingly, levels of the cytoplasmic interaction partner of the SMN-complex unrip were not substantially changed (, right). When we followed the knockdown of PPM1G and the destabilization of SMN proteins over time, we observed that PPM1G preceded the depletion of SMN and Gemin2, suggesting that its knockdown is causative for the reduced amounts of SMN proteins (). To determine whether decreased SMN levels led to impaired SMN function, we tested assembly of U1 snRNPs by band-shift experiments (). As expected, reduced snRNP assembly was observed upon PPM1G knockdown (; 48.9 ± 10% compared with control extracts), but after normalizing SMN protein levels similar activities were again observed (, normalized SMN; 107 ± 29.4%), indicating that residual SMN complex after PPM1G knockdown was equally functional to that in control lysates.
Previously published data describing a function of PPM1G in cell-free splicing assays () suggested an important role of PPM1G for the assembly of the complete spliceosome machinery. Defects in the integrity of the splicing machinery led to dramatic changes in the characteristic speckle-like localization of the arginine/serine-rich splicing factor 2 (SC35) protein (), an integral component of spliceosomes. Although PPM1G knockdown caused extensive or complete loss of the localized nuclear accumulation of Gemin2 as a marker for the SMN complex (), no obvious change in the overall pattern of SC35 staining was observed. This suggests that the overall structure of the splicing machinery was not readily affected after knockdown of PPM1G, whereas Gemin2 was lost from CBs.
A possible explanation for the loss of SMN proteins from CBs is that PPM1G directly dephosphorylates a protein of the SMN complex and that this dephosphorylation is instrumental for efficient targeting or accumulation of the SMN complex in CBs. This hypothesis assumes that the loss of the catalytic activity of PPM1G is responsible for the observed phenotype. To directly test whether the catalytic activity of PPM1G is required for the stability of the SMN complex in nuclear CBs, we used the mouse orthologue of the human cDNA to complement the siRNA-induced phenotype. The mouse mRNA mismatches in two positions with the corresponding human mRNA, making it much less sensitive to the human siRNAs. Overexpression of the mouse HA-tagged wt protein upon transfection with siRNA oligos considerably increased the number of cells showing the characteristic localization of the SMN complex in CBs compared with cells treated with the siRNA only (; and not depicted). To analyze whether the catalytic activity of PPM1G was essential for rescue activity, we generated a catalytically inactive mutant of mouse PPM1G (mPPM1G) in which aspartate in position 493 was replaced by alanine (D493A) corresponding to the D496A mutant in hPPM1G (; ). Upon expression of the mutant protein, the number of cells showing SMN accumulation in CBs was not substantially different from the number in nonrescued controls (). This indicates that the catalytic activity of PPM1G promotes accumulation of the SMN complex in CBs and suggests that PPM1G is a major determinant of the SMN-complex localization in the nucleus.
Given that phosphatase activity of PPM1G was required for SMN accumulation in CBs, we next analyzed the dephosphorylation of the major phosphoproteins of the SMN complex, SMN and Gemin3 (). We prepared control extracts, which had been incubated with immobilized unspecific IgG, and extracts from which PPM1G was efficiently immunodepleted using immobilized anti-PPM1G antibodies (). To measure the efficiency of dephosphorylation of SMN-complex proteins, we first immunopurified the SMN complex and labeled SMN-complex proteins by the addition of γ-[P]ATP and a complete HeLa extract. Next, we compared the dephosphorylation of purified SMN-complex components in a chase reaction in extracts without ATP. Indeed, we observed considerably less efficient dephosphorylation upon depletion of PPM1G from extracts used for the dephosphorylation reaction (, compare ΔC and ΔP after 60 min). Moreover, the mobility of Gemin3 on SDS gels was reproducibly increased upon dephosphorylation in control extracts but not in PPM1G-depleted extracts, whereas the mobility of Gemin4 remained unchanged (). This suggested that PPM1G dephosphorylated both SMN and Gemin3 in cell-free extracts. To confirm this result in living cells, we determined the phosphorylation pattern of SMN- complex proteins upon knockdown of PPM1G. PPM1G knockdown did not change the overall phosphorylation pattern of cellular proteins as judged by immunoblotting using total lysates and an antibody against phosphoserine/threonine epitopes (). To determine the phosphorylation states of Gemin3 and SMN, we separated total proteins from lysates of control cells or cells after PPM1G knockdown on conventional 2D gels. Immunoblot analysis using antibodies against SMN and Gemin3 revealed that both proteins were phosphorylated as indicated by populations of more acidic forms of the proteins (). Strikingly, for both SMN and Gemin3, more acidic forms accumulated upon knockdown of PPM1G. Despite the overall reduction of SMN and Gemin3, relative levels of the less acidic forms of SMN (sector 2) were ∼3.5-fold decreased, whereas the relative levels of the more acidic forms (sector 1) were increased (). Similarly, more acidic forms of Gemin3 in sector 1 were increased approximately threefold (). Collectively, this strongly suggested that PPM1G directly dephosphorylates SMN-complex proteins in the nucleus.
Unrip associates with the SMN complex primarily in the cytoplasm, and knockdown of unrip leads to the enhanced accumulation of the SMN complex in CBs in the nucleus (; ). This raised the possibility that PPM1G and unrip act antagonistically on the SMN complex. To test this, we knocked down PPM1G, unrip alone, or both gene products together by siRNA treatment, and then we analyzed SMN localization in CBs (). The efficiencies of the single or double knockdown were compared with controls and quantified by immunoblotting (). As published previously (), knockdown of unrip alone led to an increase in the number of nuclear bodies in which SMN accumulated (unpublished data). In addition, we observed the formation of cytoplasmic SMN aggregates in 10–20% of all cells (). Knockdown of PPM1G alone (oligo pair #1; , ΔP) led to the loss of the nuclear, CB-localized SMN signal in the majority of cells (, SMN signal in PPM1G-depleted cells). In contrast, knockdown of PPM1G and unrip together restored apparently normal SMN localization in CBs (). A similar number of cells displayed normal, punctate SMN staining in the nucleus (1–5 bright spots, little diffuse nucleoplasmic signal; ) compared with untreated controls. Interestingly, cells showing cytoplasmic SMN aggregates upon knockdown of unrip alone (see above) were not observed anymore after double knockdown. However, the amount of SMN was reduced upon knockdown of PPM1G alone and of PPM1G together with unrip (, ΔU, ΔP, and ΔUΔP). Collectively, this suggested that unrip and PPM1G have compartment-specific and antagonistic activities toward the SMN complex.
The biogenesis of U snRNPs is a complicated process and requires activities both in the nucleus and in the cytoplasm. Newly transcribed and exported U snRNAs recruit Sm proteins to assemble the Sm core and are subsequently hypermethylated in the cytoplasm. The partially assembled and modified RNPs are then imported into the nucleus, where additional maturation steps lead to the completion of mature U snRNPs (). It is clear from several studies that the SMN complex acts as a key assembler of U snRNPs in the cytoplasm (; ; ). Thus, this entity not only facilitates Sm core formation but also serves as a binding platform for the cap-methyltransferase TGS1/PIMT () and nuclear import factors (). In fact, it has been shown that the SMN complex travels into the nucleus together with the U snRNP (; ). In the initial phase in the nucleus, the U snRNPs appear to stay associated with SMN complexes until they reach CBs. At that point, U snRNPs and SMN complexes are believed to dissociate. Whereas U snRNPs disperse in nuclear interchromatin granules and perichromatin fibrils (the sites of storage and splicing, respectively), the SMN complex is likely to return to the cytoplasm, although this has not been shown directly. Hence, the question arises of how the different functions of the SMN complex, its interactions with different partners, and the dynamic equilibrium of its subcellular localizations are regulated. Several components of the SMN complex were shown to be phosphorylated, particularly SMN, which displayed a compartment-specific phosphorylation pattern. This suggested that posttranslational modifications could be involved in the regulation of SMN-complex functions ().
Here, we have shown that the nuclear phosphatase PPM1G specifically copurifies with the SMN complex. We observed that reduction of PPM1G levels in cell-free extracts, as well as in living cells, led to reduced dephosphorylation of SMN and Gemin3. Collectively, this strongly suggests that PPM1G directly dephosphorylates SMN and Gemin3 in the nucleus. Moreover, decreased levels of PPM1G cause dramatic changes in the subcellular localization of SMN as shown by RNA-interference studies. These studies revealed that the loss of PPM1G was strictly correlated with the loss of SMN accumulation in CBs. PPM1G knockdown caused a similar phenotype for the SMN-interacting protein Gemin2, suggesting that accumulation of the whole SMN complex in CBs was affected. A partial rescue of this phenotype was observed when we subsequently overexpressed mPPM1Gwt, but not upon overexpression of a catalytically inactive variant. Interestingly, the loss of SMN accumulation in CBs was not accompanied by the accumulation of SMN in other nuclear bodies (i.e., gems). This might suggest that reversible phosphorylation controls the ability of the SMN complex to accumulate in nuclear bodies in general. We therefore propose that PPM1G phosphatase activity is needed to maintain compartment-specific phosphorylation patterns of SMN-complex proteins in the nucleus, which are required for accumulation of the SMN complex in nuclear bodies.
It has been shown previously that accumulation of the SMN complex in CBs depends on the direct interaction of SMN and coilin (, ). Consistent with that, knockdown of the SMN protein inhibits targeting of other proteins of the SMN complex to CBs. At the same time, however, it induces some fragmentation of CBs into numerous smaller foci likely because of compromised ongoing U snRNP biogenesis (; ; ). Although the complete loss of localized SMN signal from CBs in our experiments was not correlated with the loss of coilin accumulation in CBs, we still observed an increased number of small coilin-positive structures. Moreover, PPM1G knockdown in our experiments also resulted in decreased levels of SMN, Gemin2, and Gemin3, both in the cytoplasm and in the nucleus. PPM1G knockdown might therefore generate a mild version of the phenotype seen upon direct SMN knockdown. When we determined the pattern of U snRNPs using the Y12 antibody after PPM1G knockdown, no obvious differences to control cells were observed (unpublished data). Also, colocalization of the U2 snRNP–specific factor U2B with coilin in CBs could readily be detected in the vast majority of PPM1G knockdown cells in which SMN did not accumulate in CBs (unpublished data). As PPM1G knockdown does not cause changes in the overall pattern of U snRNP localization and has only mild effects on coilin accumulation, this raises the possibility that accumulation of SMN in nuclear bodies is not required for U snRNP biogenesis.
In many PPM1G knockdown cells, the loss of SMN from CBs was accompanied by an increase in the levels of diffuse nucleoplasmic SMN. This suggests that deregulated SMN localization in the nucleus is not a consequence of reduced levels of SMN, but that it might actually cause reduced SMN levels. Reduced cytoplasmic SMN levels explain the reduced capacity of extracts from PPM1G knockdown cells in U1 snRNP assembly. Yet, the specific activity of the SMN complex in cytosolic extracts from PPM1G knockdown cells was not affected and localization of U snRNPs was not changed. If PPM1G knockdown did not affect U snRNP biogenesis, it would likely also not influence the capacity of cells in splicing of pre-mRNAs. However, it had been demonstrated previously that PPM1G plays a role in splicing if tested in cell-free extracts from HeLa nuclei. Still, PPM1G seemed not to be directly essential for splicing in the complete system, suggesting the presence of a redundant activity in complete splicing extracts (). Consistent with that, specific knockdown of PPM1G from a cellular system in our hands did not induce differences in the morphology of splicing speckles as judged by immunofluorescence with antibodies against SC35. However, we did not directly test the efficiency of splicing in living cells with reduced levels of PPM1G. It is conceivable that after PPM1G knockdown, with time, general splicing defects become apparent or that splicing of a certain subset of substrates is immediately impaired, as has now been shown for alternative splicing of the CD44 pre-mRNA ().
Our results on changed phosphorylation patterns of SMN components implicate the existence of a kinase, which phosphorylates proteins of the SMN complex. Like unrip, this kinase is supposedly a cytoplasmic interaction partner of the SMN complex, which is spatially separated from the activity of PPM1G. To date, such an activity has not been identified. However, we showed that delocalization of SMN from CBs upon knockdown of PPM1G was compensated upon concomitant knockdown of unrip. Conversely, generations of additional SMN-positive aggregates in the nucleus and in the cytoplasm seen upon unrip knockdown were not observed when PPM1G was knocked down in parallel. Given an antagonistic subcellular localization of unrip and PPM1G, this not only indicates that proteins of the SMN complex can exchange between the nucleus and the cytoplasm, but it might also suggest that unrip is involved in cytoplasmic SMN phosphorylation.
Previously, PPM1G has been reported to regulate diverse nuclear functions in splicing (; ), histone maturation (), and cell cycle progression (; ). Our data now demonstrate that PPM1G dephosphorylates proteins of the SMN complex in the nucleus and is an important regulator of the subcellular localization of SMN-complex proteins. Many steps in SMN localization and dynamics demand regulation, such as timing of import and export of components, interactions with binding partners in the specific subcellular domains, and time of residence in a particular compartment.
We consider dephosphorylation in the nucleus likely to follow import of the SMN complex into the nucleus to precede nuclear accumulation and targeting to CBs, dissociation from U snRNPs, and potentially retransport to the cytoplasm. It will be interesting in the future to determine if the dephosphorylation of SMN-complex proteins mediates efficiency, order, and timing of these events.
His-hPPM1G and His-xPPM1G were cloned into pQE32 (SphI–KpnI; QIAGEN). A PCR template for hPPM1G constructs, clone FLEXO833D0920D (RZPD), was used. His-xPPM1G was amplified by PCR from complete cDNA of eggs. His-hPPM1G D496A was generated by site-directed mutagenesis. EYFP-hPPM1G was generated by PCR and cloned into a monomeric pEYFP-C1 vector (BglII–KpnI; gift from J. Ellenberg, European Molecular Biology Laboratory, Heidelberg, Germany). EYFP-hPPM1G-NLS was generated by cutting the EYFP-hPPM1G full-length construct with BglII and SacI and cloning into the corresponding sites of the same vector. For the HA-mPPM1G construct, mPPM1G was amplified from IRAKp961L079Q (RZPD) and cloned into the pCMV-HA vector (SalI–BglII; CLONTECH Laboratories, Inc.). HA-mPPM1G D493A was generated using site-directed mutagenesis.
PPM1G fusion proteins were expressed in BL21 pRep4 (QIAGEN) at 37°C and purified via Ni-NTA (QIAGEN) in 20 mM Hepes, pH 7.5, 500 mM KCl, 8 mM imidazole, 5% glycerol, 2 mM MgCl, and protease inhibitor mix (1 μg/ml of pepstatin A, aprotinin, and leupeptin [PPM1G buffer]). Bound proteins were eluted in PPM1G buffer containing an additional 250 mM imidazole and further purified on a poly-Lysine column (bound in PPM1G buffer with 650 mM KCl and eluted with increasing KCl concentrations). Purified PPM1G was dialyzed to 100 mM KCl in PPM1G buffer (without imidazol and protease inhibitors).
Rabbits were immunized using purified 6× histidine-tagged xPPM1G or hPPM1G. Antibodies were affinity purified via matrix-coupled (Affi-Gel; Bio-Rad Laboratories) antigens. Antibodies against Gemin2 were raised in rabbit against full-length 6× histidine-tagged human Gemin2 protein, affinity purified using the respective GST-tagged antigen, and covalently linked to a glutathione-Sepharose matrix (GE Healthcare).
Final oligonucleotide concentration was 75 nM in all experiments except those shown in . Here, final oligonucleotide concentration was 50 nM for oligonucleotides against each target, unrip and PPM1G mRNA. siRNA oligonucleotides were transfected with a mixture of RNA oligonucleotides and Lipofectamine 2000 (Invitrogen) in opti-MEM (1 μl Lipofectamine to 0.8 μg RNA).
The following double-stranded RNA oligonucleotides against human mRNA sequences were used: control, agacgcauugucaacauccugucug (vimentin; Invitrogen); PPM1G #1, ucacauuccagaugccaucacaggc (Invitrogen); #2, uuaaagagcuggcacagaudtdt (Dharmacon); and #3, aggcuaccaugacuauugadtdt (Dharmacon). Against unrip, a 1:1 mixture of two oligonucleotides was used ().
HeLa CCL2 (American Type Culture Collection) and XL177 cells were used for immunofluorescence experiments. Cells on glass coverslips were washed and fixed with 3% paraformaldehyde. The following primary antibodies were used: monoclonal SMN antibody (clone 7B10; ); affinity-purified polyclonal Gemin2 antibody (this study); affinity-purified polyclonal coilin antibody (a gift from M. Platani, European Molecular Biology Laboratory; ); polyclonal affinity-purified anti-HA tag antibody (ab9110; Abcam); polyclonal affinity-purified anti-GFP antibody (a gift from D. Görlich, Zentrum für Molekulare Biologie, Heidelberg, Germany); polyclonal affinity-purified antibodies against xPPM1G and hPPM1G (this study); polyclonal antibody against unrip (); monoclonal antibody against SC35 (Novus Biologicals, Inc.); monoclonal antibody against coilin (5P10; gift from K. Neugebauer, Max Planck Institute of Cell Biology and Genetics, Dresden, Germany; ); and a monoclonal antibody against Sm proteins (Y12; gift from I. Mattaj, European Molecular Biology Laboratory).
Samples were separated by SDS gel electrophoresis () and transferred to nitrocellulose membranes. The following primary antibodies were used: against PPM1G, SMN, and Gemin2 as described for immunofluorescence, monoclonal Gemin3 antibody from rat (a gift from F.A. Grässer, Uniklinik, Homburg/Saar, Germany), anti-RCC1 sera from rabbit (), affinity-purified anti-snurportin antibody from rabbit (), and monoclonal anti-Histidine antibody (clone HIS-1; Sigma-Aldrich).
Alexa Fluor 680 goat anti–rabbit IgG (Invitrogen) and IRDye 800 anti–mouse IgG (Rockland Immunochemicals, Inc.) were used as secondary antibodies. Fluorescence signals from immunoblots were detected and quantified using an infrared scanner (LI-COR; Odyssey). Anti-xPPM1G () was detected with horseradish peroxidase–coupled anti–rabbit antibody (Sigma-Aldrich). Gemin3 antibody from rat was detected with horseradish peroxidase–coupled anti–rat antibody (Sigma-Aldrich), and secondary antibody signals were detected using ECL reagent (Pierce Chemical Co.). For quantification of Gemin3 signals, Advanced Image Data Analyzer software (Raytest) was used.
HeLa cells were lysed in 20 mM Hepes, pH 7.5, 500 mM KCl, 8.7% glycerol, 2 mM MgCl, 1% NP-40, and protease inhibitor mix (1 μg/ml of pepstatin A, aprotinin, and leupeptin) by dounce homogenization.
For immunoprecipitation of the SMN complex, complete extract was incubated with anti-SMN (7B10) antibody prebound to protein G Dynabeads (Invitrogen). Beads were washed with buffer, as for dialysis of extract, but containing 0.01% Triton X-100 and 200 mM KCl. Beads were incubated at 32°C for 30 min in extract supplied with a mixture of cold and γ-[P]ATP, washed again, divided, and incubated in control-depleted extract or PPM1G-depleted extract without ATP at 32°C for 60 min. Beads were washed with buffer as for dialysis of extract, but containing 5 mM EDTA instead of MgCl and incubated in SDS-PAGE sample buffer. Samples were analyzed by immunoblotting and autoradiography.
After siRNA treatment, cell pellets were resuspended in isoelectric focusing buffer containing 6 M urea, 2 M thiourea, 2% CHAPS, 0.15% DTT, and 0.5% ampholytes (Pharmalytes; GE Healthcare). In this buffer, cells were sonified and centrifuged in a rotor (TLA 100.3) for 30 min at 40,000 rpm. Supernatant was used for isoelectric focusing with Immobiline Dry Strips, pH 3–10 (nonlinear; GE Healthcare), followed by SDS-PAGE and immunoblot.
Immunoblotting of dilution series of recombinant-purified xPPM1G and hPPM1G compared with complete oocyte extract or complete HeLa lysates was used to estimate PPM1G concentrations to be 700 nM in oocytes, 5 μM in oocyte nuclei, 300 nM in HeLa, and at least 1 μM in HeLa nuclei.
Approximately 2 × 10 HeLa CCL2 cells were transfected with a mixture of 0.7 μg DNA and 1.4 μl Lipofectamine 2000 (Invitrogen) in OPTI-MEM medium (Invitrogen) according to the manufacturer's protocol.
Immunoprecipitations were performed with 7B10 monoclonal anti-SMN antibody () or normal mouse serum covalently linked to protein G–Sepharose. After washing, immunoprecipitated proteins were eluted by boiling in 2× SDS sample buffer.
The protein–protein interaction assay consisted of the following: immunoprecipitated SMN complex from HeLa cell extract was incubated with 5 μg of recombinant hPPM1G in 50 μl of buffer containing 20 mM Hepes, 100 mM KCl, 5 mM MgCl, 2 mM MnCl, and 2% glycerol, pH 7.5. After washing, interacting proteins were eluted from the Sepharose matrix by boiling in 2× SDS sample buffer.
Fig. S1 shows that expression of EYFP-PPM1G without NLS sequence (EYFP-PPM1G-NLS), but not expression of PPM1Gwt (EYFP-PPM1G), changes localization of the SMN complex in HeLa cells. Online supplemental material is available at . |
Although the distributions of intracellular organelles in each cell type are highly complex, very few mechanisms have been discovered by which a cell might sense and respond to the incorrect or correct position of an organelle (). The budding yeast is an excellent model to study the distribution of organelles because they move into the bud and duplicate in a predictable manner coordinated with the cell cycle. In G1, polarized growth commences after bud site selection; in S phase, small buds grow by the accumulation of organelles and other components made in the mother cell. In G2, the actin cytoskeleton depolarizes, leading to a switch from apical to isotropic (equal in all directions) growth with autonomous production of organelles in the bud. Finally, in M phase, buds acquire a copy of the genome and participate in cytokinesis. Progress through budding is monitored by checkpoints analogous to nuclear checkpoints that relay information to the nucleus, delaying cell cycle progression if bud formation is defective. To date, aspects known to be monitored include cell wall deposition (), the actin cytoskeleton (), and the septin collar at the bud neck (; ). The latter two pathways both activate Swe1 (the wee1 homologue), which inhibits Cdc28 (the cdk1 homologue) to delay the G2→M transition, a mechanism that has been called the morphogenesis checkpoint ().
The ER in yeast consists of the nuclear envelope and a network lying just beneath the plasma membrane called the cortical ER (cER), with a few cytoplasmic ER tubules linking these two domains (). Similar cER exists in all higher eukaryotic cells, with specific functions in calcium signaling and lipid traffic (). In yeast, the plasma membrane has multiple focal attachments to a portion of the cER that is biochemically specialized for synthesizing plasma membrane lipids (). ER inheritance can be divided into three distinct phases: first, cytoplasmic ER tubules move into small buds along actin cables over the relatively long distance of the mother-bud axis; second, the first domain of cER forms by attachment to plasma membrane at the bud tip; and third, cER spreads around the entire bud to form a polygonal tubular network (; ). The attachment step is potentially facilitated by the interaction of translocon components (Sbh1 and Sbh2) and reticulons (Rtn1, Rtn2, and Yop1) on the ER with exocyst components (Sec3, Sec6, and Sec8) on the plasma membrane, without which cER inheritance is delayed (; ; ). Other proteins implicated in ER inheritance include Swa2 (also called Aux1), which unfolds clathrin (); Ypt11, a Rab protein on the ER (); Ptc1 and Nbp2, which are regulators of MAPK signaling (); Ice2, an integral ER membrane protein of unknown function (); and the motor Myo4 and its adaptor She3, which are thought to carry ER tubules into the bud ().
VAP (vesicle-associated membrane protein–associated protein) is a small, highly conserved integral ER protein (). The major VAP homologue in yeast, Scs2 (), interacts with a large number of other proteins (; ). Some of these, including the sterol transfer proteins Osh2 and Osh3, are on the plasma membrane (), where they are restricted by Scs2 to those parts of the plasma membrane with subjacent cER (). This suggests that Scs2 complexes bridge from cER to the plasma membrane and led us to ask whether Scs2 has a role in forming cER. We report now that the amount of cER is reduced ∼50% in cells lacking Scs2, with buds more affected than mother cells. Scs2 interacts with an unidentified receptor localized to sites of polarized growth, indicating a role for Scs2 in attaching cER to bud tips. In addition, we found that interacts genetically with and that defects in either gene, and especially both in combination, disrupt septins at the bud neck, which triggers the morphogenesis checkpoint.
To examine the role of VAP on ER structure, we examined the effect of scs2 deletion on a fluorescent ER marker in live cells. Confocal sections of wild-type cells expressing the reporter showed the typical pattern of nuclear envelope, occasional cytoplasmic strands, and an extensive network of cER ( and Fig. S1, available at ). Qualitatively cER formed an incomplete circle, and quantitatively cER was subjacent to 72% of plasma membrane (Fig. S1). In cells, qualitatively cER rarely formed circles, and quantitatively cER was present in 37% of the periphery ( and S1). The reporter used, RFP-ER, contains the transmembrane domain (TMD) of Scs2. In case this had unforeseen interactions, we performed similar studies with two other ER reporters tagged with GFP. Are2, an ER resident enzyme with multiple TMDs (), was present at 39% of the periphery in cells compared with 77% in wild type, and the C-terminal domain of Sec12, which includes a single TMD specifying ER localization (), was at 33% of the periphery in cells compared with 57% in wild type (Fig. S1). In addition, we used an automated method to process images of cells coexpressing RFP-ER with plasma membrane–targeted GFP. The proportion of total ER colocalized with the plasma membrane fell from 50.3% in wild-type cells to 25.6% in cells, which is a relative decrease of 49% (Fig. S2, A and B). Thus, three different ER-targeted reporters showed reductions of ∼50% in the amount of cER upon the loss of Scs2.
We next assessed whether affects cER inheritance, which is indicated by a stronger phenotype in buds than mothers (). In mother cells with identifiable buds, the deletion of disproportionately affected the levels of cER in buds compared with mother cells (). This effect of was found in different yeast strains (unpublished data) and was not enhanced by further deletion of the homologue (unpublished data).
To validate these results, we performed thin section EM on a strain in which Scs2 levels are regulated by carbon source (). Repression of (here referred to as ) reduced cER in unbudded profiles by 53% (), verifying the findings with fluorescent markers. This excludes an alternate possibility that in , cER is present in normal quantity but with reduced access. Looking next at budded profiles, we found that had a far stronger effect in buds (31% of the amount of cER compared with wild type) than in mothers (55%; ), confirming that Scs2 has a role in cER inheritance. We also examined two other aspects of ER morphology in cells. First, where cER formed, its morphology as a tubular network was essentially preserved (Fig. S2 C), ruling out an effect similar to the overexpression of reticulons (; ). Second, the overall amount of ER in buds (i.e., including cytoplasmic tubules) was not affected by (Fig. S2 D). Thus, Scs2 is required for formation of the correct attachment of cytoplasmic ER to the periphery to make cER but not for transport of cytoplasmic ER into buds or for microanatomy of the cER network.
We next compared cells with strains carrying one of four other mutations that affect cER inheritance: , , , and (; ; ). cER in our , , , and strains appeared normal (Fig. S3 A, available at ). This lack of effect is compatible with the original studies implicating these genes in cER inheritance because there the gene deletions caused only minor delays in cER inheritance, which are strain dependent (; ). We also combined with the other deletions to indicate genes acting in common pathways. cER in , , and cells was not distinguishable from cells (Fig. S3 B and not depicted). In comparison, we failed to introduce directly into a strain by PCR, which is consistent with the aggravating genetic interaction noted previously in a large-scale genetic study (). To examine this interaction, we introduced repressible (; ) into a strain. cells, which grew slowly (not depicted), contained less cER than either or , particularly in buds (), where cytoplasmic ER often accumulated in a punctum at some distance from the bud tip (). A similar pattern was seen rarely in cells but not in (unpublished data).
Synthetic genetic array (SGA) analysis for both and with known nonessential genes implicated in cER inheritance identified a genetic interaction between and (). Aggravating genetic interactions were also found between these genes and and , regulators of MAPK pathways that are important for moving ER tubules into the bud (). This provides additional evidence that Scs2 and Ice2 function in attachment of the ER to the cortex. Conversely, the growth defect of was alleviated by the deletion of reticulons, chiefly Rtn1; an alleviating interaction between and could not have been reported in our experiment because neither single deletion reduced the growth rate. The interaction between and was confirmed by tetrad analysis and growth assay of the double mutant on synthetic defined medium (). Interestingly, this phenotype was less severe on rich medium, an effect we have yet to understand. On both media, these cells had reduced cER similar to cells (unpublished data). Overall, these results indicate that Ice2 and Scs2 act in parallel pathways, without which cytoplasmic ER enters buds but fails to attach to the bud tip.
If Scs2, an integral ER protein, mediates the attachment of ER to the bud tip, is it possible that Scs2 interacts with a component at the bud tip? To examine this possibility, we expressed Scs2 lacking its C-terminal TMD and tagged with GFP (Scs2ΔTMD-GFP), which might be expected to be uniformly cytosolic, as is GFP. Instead, Scs2ΔTMD-GFP targeted specific extranuclear sites, including tips of small buds and sites of recent cell division (the incipient bud site), which are both sites of polarized growth, in a manner that changed throughout the cell cycle (). In addition, there was weak cortical targeting in some cells and targeting to the nucleolus. Targeting was stronger in cells lacking endogenous Scs2 (unpublished data), suggesting competition with endogenous Scs2 for a saturable receptor. Importantly, the same targeting was also seen with the human proteins VAP-A and -B and was not affected by the P56S mutation of VAP-B associated with amyotrophic lateral sclerosis ( and not depicted; ). To determine whether full-length Scs2 also shows polarized targeting, we expressed GFP-Scs2 at low levels. This revealed weak focal targeting to the tips of small buds and sites of incipient budding (). Such a minor accumulation at sites of polarized growth was not seen with other ER markers and was not seen when GFP-Scs2 was more highly expressed (unpublished data), explaining how we overlooked it previously (). These results show that Scs2 and other VAP homologues have a conserved interaction at sites of polarized growth.
We next studied the relationship between polarized targeting by Scs2, its role in forming cER, and the previously described interactions of Scs2. The only interaction of Scs2 to be mapped in molecular detail is with short linear FFAT motifs (two phenylalanines [FF] in an acidic tract; ; ; ). To test for the role of Scs2–FFAT interactions, we used a panel of eight Scs2 mutants obtained from mapping the FFAT-binding site (). Within the panel, the affinity of binding to FFAT in vitro and the inhibition of Opi1 in vivo correlated perfectly with each other (). These read-outs correlated approximately with targeting of mutant ΔTMD-GFP constructs to the bud tip and septum, but there were two mutants (K40A and K84N) that targeted poorly relative to their interaction with FFAT (Table S2, available at ). In particular, K40A had normal FFAT binding () and the same reversion of the ino phenotype as wild-type Scs2 (Fig. S4 A), but K40AΔTMD-GFP localized weakly compared with wild-type Scs2ΔTMD-GFP (), suggesting that FFAT binding is not important for polarized targeting. Because this approach cannot completely exclude a role for FFAT, we localized Scs2ΔTMD-GFP in strains missing combinations of the four yeast proteins with FFAT motifs (Osh1, Osh2, Osh3, and Opi1) and found no defect in polarized targeting or cER structure (Table S1 and unpublished data). In addition to interacting with FFAT motifs, Scs2 has been suggested to have five FFAT-negative binding partners (Stt4, Pil1, Num1, Fks1, and Rpn10; ), but deletion or inactivation of these, in turn, did not inhibit the polarized targeting of Scs2 (Table S1) or the formation of cER in buds (unpublished data). Overall, these results show that Scs2 has a novel, conserved interaction targeting sites of polarized growth, which might be related to its role in the formation of cER.
We next examined how mutations in Scs2 affect its function, comparing K40A (see previous paragraph) with T42A, which has no polarized targeting and does not bind FFAT (Table S2). For cER formation, both mutants rescued cER partially, K40A slightly more than T42A (). For complementation of the growth defect, K40A rescued partially, and T42A was inactive (). Thus, the rescue of cells correlates with cER rescue and the degree of polarized targeting but not with FFAT binding. The only other activity described for Scs2 is in gene silencing (; ), which is likely related to its presence on the inner nuclear envelope (). We excluded a role for this pool of Scs2 in cER inheritance by comparing the activities of intranuclear and extranuclear variants of Scs2: the rescue of cER was greater with extranuclear Scs2 (Fig. S4, B and C). Together, these data suggest that Scs2 has a novel interaction at sites of polarized growth that is required for cER formation and the rescue of cells.
Targeting of Scs2 to the bud tip might reflect either active delivery by an actin-mediated process with continuous recycling () or binding to a more static bud tip component. We tested this by depolymerizing actin with a latrunculin A treatment for 20 min, which had no effect on Scs2ΔTMD- GFP targeting but did delocalize Sac6-RFP (unpublished data). Similarly, inhibition of membrane fusion in a strain did not affect targeting (unpublished data). This suggests that the receptor for Scs2 at the bud tip is not rapidly cycling (for example, on secretory vesicles; ). Longer treatment with latrunculin A (60 min) did reduce targeting, suggesting that the receptor for Scs2 is not completely static (a property of many bud tip components; ).
We next used a candidate approach to look for the receptor for Scs2 at sites of polarized growth. 57 deletion strains missing known bud tip proteins were tested for targeting of Scs2ΔTMD-GFP, but in all of these strains there was at least some polarized localization, indicating that none of the genes tested code for the sole Scs2 receptor (Table S1). 10 of the deletion strains showed variant targeting: two were better localized than wild type ( and ), five were less well localized (, , , , and ), and three had punctate Scs2ΔTMD-GFP at the bud tip (, , and ). The clearest conclusion from this is that targeting requires the polarisome, which is a 12S complex of Bni1, Bud6, Pea2, and Spa2 that establishes polarity in yeast (; ). Interestingly, some of these components are also partially mobilized by long-term treatment with latrunculin A (). To confirm that Scs2 targeting is mediated by the polarisome, we compared the distributions of RFP-Scs2ΔTMD and GFP-tagged Pea2 and found that the two proteins were superimposed or closely adjacent (). Among the candidates for Scs2 receptors that we excluded were Sho1 (the yeast homologue of occludin, which binds VAP; ) and the known polarisome interactors Msb3/Msb4 and Sph1 (tested in a strain; Table S1; ; ). These data suggest that the receptor for Scs2 is a currently undefined bud tip component acting downstream of the polarisome.
If the polarisome is important in targeting Scs2, there might be cER abnormalities in polarisome mutants. Using our fluorescent reporter, we found that , , and cells had normal ER architecture (unpublished data), indicating that the modest delocalization of Scs2ΔTM-GFP in these mutants does not affect cER inheritance. In contrast, cER in buds was abnormal, often failing to reach the bud tip (). Instead, cER formed a cup shape around the bud neck, a unique phenotype suggesting that Bni1 affects more than just Scs2 targeting. This can be understood in light of previous findings that Bni1 is required for the assembly of actin cables in the bud (), which are used by Myo4 to transport ER (). We next examined the effect on cER of combining with the deletion of polarisome components. cells and cells both showed cER defects similar to cells ( and S3 C). cells showed an additional phenotype, with multiple cytoplasmic ER tubules originating from the bud neck and very little cER at the bud tip (). cells showed a phenotype similar to but far stronger than cells, with cER absent from most bud tips and cup-shaped cER emanating from the bud neck not only in buds but also in many mother cells (). Videos show cER creeping through the bud neck and along the cell cortex into the proximal bud (Videos 1 and 2, available at ), with no cytoplasmic ER tubules crossing to the bud tip as in wild-type cells (; ). Because Bni1 is only partially responsible for targeting Scs2 to the bud tip, the stronger phenotype of cells suggests that is epistatic to . If Bni1 (and its role in the polarisome) cooperates with Scs2 in attaching cER to the bud tip, the polarisome functions in a pathway parallel to and should show genetic interactions with but not with . Indeed, SGA analysis showed that (but not ) has strong aggravating genetic interactions with , , and (). Therefore, in the absence of Ice2, recruitment of ER to the bud tip by the polarisome becomes essential.
mutants have been reported as one of the most elongated yeast deletion strains, their elongation putting them in the 99th centile of 4,800 strains measured in a genome-wide study of cell shape (). We found the same elongation in mutants of both BY4741 () and RS453B strains with deleted or repressed (not depicted). We tested which aspects of the Scs2 protein are important for rescue of cell shape. First, there was no restoration of cell shape if Scs2 was either made soluble by deleting its TMD or if full-length Scs2 was retained on the inner nuclear membrane (Fig. S4 D), indicating that Scs2 needs to be anchored in the extranuclear ER to rescue cell shape. Second, K40A and T42A mutants both partially rescued cell shape, although slightly more so for K40A (). Finally, the known interactors of Scs2 (Opi1, Osh1/Osh2/Osh3 [singly and together], Fks1, Num1, Pil1, Rpn10, and Stt4) were not important in cell shape, as the lack of any of these proteins did not cause elongation (unpublished data). Thus, rescue of elongation appeared to correlate with the factors we found to be required both for polarized targeting of Scs2ΔTMD and for cER formation, suggesting that cellular elongation results from defective cER inheritance. Although this link has not been reported previously, , , , , and strains are elongated (albeit less than ), whereas and have normal shape and and are among the most rounded of all strains, similar to (; ; and unpublished data). Thus, although all genes implicated in cER inheritance and structure do not appear to act similarly, cellular elongation is a common feature.
The morphogenesis checkpoint results from an imbalance between Swe1 and Mih1 (a Cdc25 homologue), which respectively inhibit and activate Clb2–Cdc28 complexes that control the apical isotropic growth switch in G2 (; ; ). Overactivation of Swe1 delays the switch, leading to cellular elongation; in contrast, deleting hastens the switch, producing rounder cells. The strain was significantly rounded up by the introduction of (), indicating that the effect of on cell shape is caused by an imbalance between Swe1 and Mih1. In contrast, the milder elongation of was not reverted in cells. To confirm the suggestion that Swe1 overactivity is responsible for shape changes in the strain, we determined the effect on shape of overexpressing Hsl7, an adaptor protein that brings Swe1 to the septin ring to facilitate its phosphorylation by Hsl1 and subsequent degradation (; ; ). Excess Hsl7 considerably rounded up the strain, indicating that its elongation is caused by excess Swe1 activity.
Because these results indicate increased Swe1 activity in cells, we investigated Swe1 levels. We found that levels of Swe1-myc were increased in both and strains, either in unsynchronized cells () or in cells synchronized in S phase by incubation with hydroxyurea (HU; ). When samples from cells with HU were separated on gels without SDS to exaggerate the retarding effect of hyperphosphorylation (; ; ), we found that phosphorylation of Swe1-myc was affected by and . Both mutations caused an increase in partially hyperphosphorylated Swe1-myc (, lanes b and c), the most active species, and a lack of the smear of maximally hyperphosphorylated Swe1-myc (found in wild-type cells; , lane a), which is less active and is the substrate for ubiquitination and degradation (). Alongside the changes in phosphorylation, myc-positive degradation products appeared in both and strains (), indicating altered degradation. These results demonstrate that loss of both and leads to increased Swe1 activity even though we found that the inactivation of Swe1 rescues the shape of only, not .
We next tested the effect of loss on the viability of strains with defective cER (). The deletion of in a wild-type background slightly increased the growth rate as reported previously (). The deletion of did not affect the growth rate, nor did deletion of in the strain, indicating that the main effect of in cells is on shape. The deletion of resulted in a mild growth defect that was considerably exacerbated by the deletion of . However, the loss of in cells, which led to considerable rounding up (axial ratio of 1.18 compared with 1.41 in cells), strongly impaired growth (doubling time of 385 min compared with 128 min), indicating that the loss of cER renders cells dependent on Swe1 for cell survival, with having an additive effect beyond alone.
Many pathways increase Swe1 activity. One is ER stress, which causes cell death in the absence of signals through Mpk1 in the MAPK to calcineurin pathway (). Because was not essential for cell survival in or strains (unpublished data), it appears that cER inheritance defects do not activate this ER stress pathway. Two defects in the cytoskeleton of the bud signal to Swe1 via the morphogenesis checkpoint: perturbed actin and septins (). For actin, one pathway following its depolymerization signals via Mpk1 to inhibit Mih1, leaving Swe1 unopposed (). However, the combination of with or did not reduce elongation, nor did the overexpression of Mih1 (unpublished data), implying that and do not require Mpk1 for the elongation. Actin defects may also activate the high osmolarity glycerol pathway, which lies upstream of Swe1 activation by delocalizing Hsl7 (). However, Hsl7-GFP was not delocalized in and strains (unpublished data), suggesting that the high osmolarity glycerol pathway is not activated by cER defects.
Septin disorganization results in budding defects, bud neck deformities, and an increase in Swe1 levels (). Therefore, we looked for signs of altered septin organization in and cells and found that a small minority had multiple buds compared with this being undetectable in wild-type cells (). This led us to look for the same phenotype in the strain, in which >40% of cells had multiple buds, indicative of septin defects. To directly visualize septins, we used a GFP-tagged version of the septin Cdc10. cells formed defective septin rings and mislocalized Cdc10-GFP mainly to bud tips, phenotypes that were rare in single mutants but completely absent in wild-type cells (). We also performed SGA analysis of and with bud neck kinases that affect septin function, which identified interactions of both and with , a key kinase in septin ring assembly (; ; ; ), and also between and (). This is evidence for the specific involvement of cER in septin assembly. We confirmed the interactions with by tetrad analysis () and examined septin assembly in the double mutants. considerably disrupted septins (), an effect intensified in both and , particularly leading to chains of unseparated cells () that were multinucleate (not depicted). In some cases, septin collars were completely absent or present only as puncta at the erstwhile bud neck (). The cellular phenotype of the double mutants suggests the improper assembly of septin collars, leading to a failure in the completion of cytokinesis, and explains the growth defects.
In this study, we find that the ER transmembrane protein Scs2 is required for cER inheritance. Scs2 appears to act in the attachment phase, as ER tubules lacking Scs2 are delivered into buds but not to bud tips. Scs2 interacts with an unidentified bud tip component, suggesting that it bridges directly from the ER to the plasma membrane. Bud tip targeting of Scs2ΔTMD is not indirect via another component of cER because it occurs where cER is absent from the bud tip (unpublished data). We propose that bridging by Scs2 acts at a similar stage in cER inheritance as the translocon–exocyst interaction, although with some differences. Loss of translocon components had no discernible effects on cER inheritance in our strains (unpublished data), and, in other studies, the loss of exocyst components does not affect cER in mother cells (; ), which is considerably reduced in cells lacking Scs2.
showed aggravating genetic interactions with , whose protein product is also integral to the ER and, when deleted, results in cER defects in buds and mothers. The variation of the growth phenotype of with medium is opposite to that reported previously for mutants, which fair worse on rich medium (), and in neither case is the reason yet understood. As one might predict, mutations of the polarisome, which lies upstream of Scs2 targeting, also show aggravating genetic interactions with , although we cannot exclude the possibility that these interactions involve regulation of the exocyst by the polarisome. The alleviating interaction between and reticulons, especially , might result from the conversion of cER into large sheets caused by the loss of Rtn1 (and Rtn2 less so; ; ), thus reversing the effect of mutations that decrease cER. Together, the genetic interactions and the extreme loss of cER in cells imply that Scs2 and Ice2 function in parallel or partially redundant pathways attaching ER to the cortex.
The involvement of the polarisome in Scs2ΔTMD targeting led to the finding that Bni1 is required for ER tubule movement in buds, a defect greatly exaggerated in , where cER also failed to reach the distal pole of mother cells. The effects of losing Bni1 might be predicted from its function in nucleating actin filaments at bud tips and the presence of some Bni1 at the distal pole of mothers (; ). The slow movement of cER through the bud neck in cells is reminiscent of the actin-dependent short-range dynamic reorganization of cER reported in both yeast and higher eukaryotic cells (; ). This is a compensatory mechanism by which buds receive cER, which was hypothesized to exist previously (). Because Scs2 and Bni1 function in the preferred long-distance pathway of cER inheritance, the genetic interactions of with both and suggest that Ice2 acts in this second pathway.
The finding that two genes involved in cER inheritance are also required for septin assembly indicates that there is a causal link, although the molecular basis remains unknown. Septins form filamentous scaffolds for various cellular functions, act as diffusion barriers, and rearrange in cellular events such as cytokinesis (). Within the bud neck, the septin ring and cER both interact with the cortex, and they also interact with each other, as septins were recently shown to create a diffusion barrier in cER at the bud neck (). During budding, not only does the ER have to rearrange but the septin ring grows into the bud to form an hourglass that then splits into two rings (). Therefore, it is possible that the effects of defective cER inheritance on septins result from local interactions with altered ER at the bud neck. However, because we found no gross alteration in ER morphology at the bud neck (it does not accumulate there, for example), we favor the possibility that a feature of ER within buds regulates septins. Normal bud cER may function to promote septin assembly, and it is also possible that excess cytoplasmic ER in buds impairs septin organization. Relevant cER functions include the recruitment of specific septin regulators (for example, Cdc28 is found on the ER; ; ), or, alternatively, the influence may be indirect via a generic function of cER. No specific essential function has been ascribed to cER rather than cytoplasmic ER, but apposition of ER to the bud plasma membrane allows the direct, nonvesicular traffic of lipids, calcium, and even proteins (; ; ).
Because septins scaffold multiple kinases that phosphorylate Swe1, the bud neck is the physical location where many cellular inputs are integrated to regulate G2→M progression (). This close interrelationship means that even a minor defect in septin architecture is amplified by the activation of Swe1, which, in turn, inhibits septin accumulation at the bud neck (). Thus, the abnormalities in septin assembly seen in and mutants, the first reported for any primary defect in the ER, may be subtle (), but they are also functionally important, as shown both by synergy with the deletion of (a regulator of septins and Swe1) and by the strong synthetic effect in the double mutant. cER inheritance has properties of a classic checkpoint in that overriding the checkpoint is strongly detrimental to growth (). In cells, the switch from apical to isotropic growth occurs, but cells are then very slow to divide. In comparison, the activation of Swe1 in cells aids growth, presumably by allowing the bud extra time to acquire ER before actin depolarizes, even though cortical attachment is slow thereafter.
cER distribution is now the second example in which the nuclear cell cycle machinery responds to the distribution of extranuclear membranes, in addition to breakdown of the Golgi ribbon in higher eukaryotic cells in transit through G2/M (). Given the precise way in which organelles are arranged in individual cell types, it is not entirely surprising that cells have adapted to monitor their distribution. Although our results focus on VAP in yeast, the identical polarized targeting of human VAP-A/B suggests that VAP plays a part in polarized functions of the ER in higher eukaryotic cells. Among the most polarized cell types are neurons, which have highly asymmetric processes containing specialized subdomains of ER (; ). The position of this ER, both pre- and postsynaptically, generates signals detected locally (; ; ), and it is possible that these signals also reach the nucleus. The relevance of VAP in this process is underlined by the finding that mutations in human VAP-B cause motor neuron disease ().
All plasmids were based on the pRS series and contained the constitutive portion of the promoter except when indicated. Plasmids used to visualize ER in living cells by confocal microscopy were as follows: the TMD of Scs2 (residues 220–244; sequence ENESSSMGIFILVALLILVLGWFYR) placed after the tandem repeat of dimeric dsRed (RFP-ER) cloned into pRS416 () or pRS405 (; because of its minimal size [25 amino acids], this targeting motif minimizes the possible effect of the septin-mediated diffusion barrier at the bud neck []; the C terminus of Sec12, including its TMD (residues 355–471) after GFP cloned into pRS406 (; gift of H. Pelham, Laboratory of Molecular Biology, Cambridge, UK; ); and the sterol ester synthase Are2 tagged with GFP (gift of G. Daum, Technical University Graz, Graz, Austria; ). For simultaneous visualization of ER and plasma membrane by confocal microscopy, both RFP (RFP-ER) and a plasma membrane–targeting reporter, Psr1(N-term) (residues 1–28; MGFISSILCCSSETTQSNSNSAYRQQQS; ), followed by GFP were cloned into the vector YCp50 (). Plasmids used in Scs2 rescue experiments, including K40A and T42A mutants, were previously described () based on pRS416 and include a myc tag (MEQKLISEEDL) before the Scs2 sequence. The plasmid Scs2-Prm3 is similar to the rescue plasmids, but, after residue 219 of Scs2, the TMD is replaced by the C-terminal 65 residues of Prm3 (69–133). The plasmid Scs2-Prm3ΔNLS omits the NLS of Prm3 (i.e., residues 74–133; ). GFP-Scs2-Prm3 plasmids are constructed similarly with GFP at the N terminus of the myc tag. Plasmids expressing Scs2ΔTMD-GFP and variants were based on pRS416 and contain the coding region of Scs2 missing the TMD (1–224 + linker RGAGAGAPVEK) followed by GFP. VAP-BΔTMD-GFP is the same but contains the human VAP-B cytoplasmic domain (1–222 + linker PVEK). Plasmid RFP-Scs2ΔTMD contains dimeric RFP and residues 1–225 of Scs2 in pRS416. YCpLG-GFP-HSL7 (gift of J. Thorner; University of California, Berkeley, Berkeley, CA) expresses GFP-tagged Hsl7 from the promoter (). Induction to examine Hsl7 localization was for 90 min in minimal medium, and induction for the effect of Hsl7 on cell shape was for 8 h. Plasmid DLB2258 (gift from D. Lew; Duke University Medical Center, Durham, NC) is derived from pJM1102 () and integrates at the locus to introduce a 12-myc cassette after the Swe1 coding region without generating an untagged adjacent copy. Overexpression of C terminal–tagged Mih1 was achieved from a 2μ plasmid containing the promoter (Open Biosystems) and was checked by Western blotting.
Deletion strains were obtained from freezer stocks of the yeast deletion collection (BY4741, Mat a, and KanMX) unless otherwise stated. Other gene deletions were constructed by the PCR method with the heterologous markers , , or NatR. Strain CLY3 is a - derivative of the wild-type strain RS453B (). TLY251 is a strain with the inducible expression of Scs2 achieved by replacing the promoter with the inducible/repressible promoter, as previously reported (). Repressible was also introduced into . These strains were grown either in galactose for the induction of or switched to dextrose for >16 h to repress function (; , ). Double deletion strains with and were constructed by replacement of the ORF with in the corresponding single-deletion strains based on BY4741. and single- and double-deletion strains as well as all and derivatives were also derived from the meiotic products of heterozygous diploids, with multiple spores of each genotype being compared. The strain was made by replacement of the ORF with in . A strain dependent on a conditional allele of has been described previously (). , , , and strains were gifts from Y. Ohya (University of Tokyo, Tokyo, Japan), H. Pelham, E. Bi (University of Pennsylvania School of Medicine, Philadelphia, PA), and R. Arkowitz (Universite de Nice, Nice, France). In each experiment, comparable wild-type parental strains were included. To test the function of temperature-sensitive alleles, mutant and parent cells were grown at 25°C and transferred to 37°C for 30 min before visualization. Strains isogenic to BY4741 expressing Pea2-GFP and Cdc10-GFP were purchased from Invitrogen. Mutant strains with Cdc10-GFP were made by mating with deletion strains in the same background and sporulation. Growth assays were performed on isogenic strains made by mating and tetrad analysis. For quantitation of doubling times, cultures were diluted to OD = 0.1 and assayed over 7 h. Because strains and derivatives showed strong cell shape phenotypes on minimal medium, this was done in rich medium (yeast extract/peptone). Doubling times were calculated by linear regression of log plots of growth curves. R values were >0.95 for all regressions.
Yeast growing in log phase were examined with a confocal microscopy system (AOBS SP2; Leica) at room temperature (63× NA 1.4 objective) using LCS software (Leica) for acquisition and Photoshop (Adobe) to increase levels uniformly within experiments. For unbiased imaging of cER in mutant strains, transmission imaging was used to focus on the central plane of cells before capturing fluorescence images. To test a strain for polarized localization of Scs2ΔTMD-GFP, nine random fields of cells (∼100 cells per field) were examined, and polarized fluorescence was compared with the wild-type parental strain within each experiment.
Cells growing in log phase were studied ultrastructurally using permanganate fixation and uranyl acetate to highlight membranes, including cER, as previously described (; ). Thin sections were counterstained and viewed on a transmission electron microscope (model 1010; JEOL).
To determine the axial ratio for a particular strain, multiple differential interference contrast images were taken of the strain, using the same microscope magnification throughout. For every whole cell profile included, the long axis was chosen and marked, and the short axis was then drawn at right angles at the broadest part of the cell. Mothers and unbudded cells were grouped together, and buds were analyzed separately. Mean axial ratio was typically calculated for 200–300 cells.
Strains transformed with plasmid DLB2258 to add 12xmyc tags to the C terminus of the genomic copy of Swe1 were grown to mid-log growth phase. Where indicated, cells were supplemented with 200 mM HU for a further 4 h. 1–2 × 10 cells were pelleted by centrifugation, lysed directly into 2× sample buffer for SDS-PAGE, and heated at 95°C for 5 min before vortexing with 400–600 μm of glass beads (acid washed). Samples were separated on 10% polyacrylamide gels for Western blot analysis both to determine overall Swe1-myc levels with anti-myc (monoclonal 9E10) and to control for loading by probing with antiphosphoglycerate kinase antibody (monoclonal; Invitrogen). In addition, samples were separated on 7% SDS-PAGE without SDS in the separating gel to examine slowly migrating phosphorylated forms of Swe1 (). The Gel function of ImageJ was used to quantify the main full-length Swe1-myc bands relative to Pgk1.
SGA analysis was performed as described previously (). In brief, and query strains were generated by PCR-mediated homologous recombination and mated to each strain of the deletion array of nonessential genes using a Singer RoToR high density array robot. After selecting for heterozygous diploids, cells were sporulated, and haploid cells were allowed to germinate. Subsequent rounds of selection allowed the growth of double-deletion strains or single-deletion strains as controls. Digital images of plates were obtained, and colony sizes were measured and normalized as previously described (). Potential interactions were scored by comparing the area of double mutant colonies with single mutants, and a t statistic was derived for each interaction in which more negative numbers indicate an aggravating interaction and more positive numbers indicate an alleviating interaction. In the analysis of septin-related kinases (), the two nonessential septins Shs1 and Cdc10 could not be assayed because of germination defects inherent to these deletion strains. All experiments were performed in triplicate.
Fig. S1 shows the effect of on cER in fields of cells expressing three different markers. Fig. S2 uses a semiautomated method to assess the same effect. Fig. S3 shows the lack of effect of many mutations other than that are implicated in cER inheritance on overall cER levels. Fig. S4 shows that K40A is active in vivo and that Scs2 forms cER and rounds up cells only when it is embedded in the extranuclear ER. Table S1 shows the effect of mutating candidate genes on the localization of Scs2ΔTM-GFP. Table S2 correlates FFAT binding with bud tip localization in Scs2 mutants. Videos 1 and 2 show the slow movement of cER into the proximal portion of cells. Online supplemental material is available at . |
Neurodegenerative disorders such as Huntington's disease (HD) and Parkinson's disease are characterized by the accumulation of intracellular ubiquitin-containing protein aggregates. Although polyubiquitin is a well-known signal for degradation by the ubiquitin–proteasome system, several aggregation-prone proteins have been shown to disrupt the function of the proteasome (; ; ; ). Activation of an alternative lysosomal mechanism of protein degradation, known as macroautophagy (hereafter referred to as autophagy), is often observed in protein aggregation diseases (; ; ). Autophagy is a dynamic process whereby cytoplasmic material is sequestered within double membrane-enclosed vesicles (autophagosomes), which eventually fuse with lysosomes where the encapsulated material is degraded (). This pathway is known to be important in developmental processes and human disease and for meeting amino acid poor growth conditions (). Recently, it was found that loss of autophagy causes neurodegeneration even in the absence of any disease-associated mutant proteins (; ), suggesting that the continuous clearance of cellular proteins through basal autophagy prevents their accumulation, and in turn prevents the disruption of neural function and subsequent neurodegeneration.
Autophagy is generally considered a ubiquitous bulk degradation mechanism for long-lived proteins and organelles. The molecular mechanisms underlying the autophagic process have been extensively studied in yeast, using genetic screens to identify autophagy-defective (atg) and vacuolar protein sorting (vps) mutants (). Subsequent inactivation of atg orthologues in higher eukaryotes has revealed that the autophagic machinery is highly conserved. Nonetheless, the molecular mechanisms and signals involved in the recognition of autophagic substrates and trafficking of autophagosomes are poorly understood.
The endosomal sorting complexes required for transport (ESCRTs), first identified by characterization of yeast vps class E mutants, have proven important for recognition of ubiquitinated endocytosed integral membrane proteins, their sorting into the intralumenal vesicles (ILVs) of the multivesicular body (MVB) and subsequent degradation in the lysosome/vacuole. Ubiquitinated cargo is first recognized by the Vps27p/Hrs (hepatocyte growth factor–regulated tyrosine kinase substrate)–Hse1p/STAM (signal transducing adaptor molecule) complex. Vps27p/Hrs then recruits the ESCRT-I complex (Vps23p/Tsg101, Vps28p, Vps37p, Mvb12p) to the endosome membrane by binding Vps23p/Tsg101 (tumor susceptibility gene 101). The ubiquitinated cargo is further assumed to be delivered to ESCRT-II (Vps22p, Vps25p, Vps36p) before it gets internalized into MVBs through the activity of ESCRT-III (Vps2p, Vps20p, Vps24p, Vps32p). Ubiquitin is removed proteolytically and the ESCRT machinery is finally dissociated from the endosomal membrane by activity of the ATPase Vps4p/SKD1 (for review see ; ; ).
Depletion of ESCRT subunits results in MVBs with abnormal morphology, called the “class E compartment” in yeast. The effects of depleting different ESCRT subunits on internalization and degradation of integral membrane proteins such as the epidermal growth factor receptor (EGFR) have been well characterized (). However, little is known about the role of functional MVBs and ESCRT subunits for autophagic degradation. The classical view has been that the autophagic and endocytic pathways converge at the lysosomal level, but autophagosomes have also been found to undergo fusions with earlier parts of the endocytic pathway (; ; ; ). The term “amphisome” is used to describe pre-autolysosomal compartments containing both autophagic and endocytic material (), but the specificity of amphisome formation and the molecular mechanism involved are poorly understood.
Recently, the ESCRT-III subunit CHMP2B (charged multivesicular body protein 2B, also known as chromatin-modifying protein 2B)/Vps2B was found to be mutated in a large Danish pedigree with frontotemporal dementia (FTD) (), and in patients with amyotrophic lateral sclerosis (ALS) (). FTD is the second most common form of presenile dementia after Alzheimer's disease (; ) and is characterized neuropathologically by the presence of either tau pathology or ubiquitin pathology, which is termed frontotemporal lobar degeneration with ubiquitin-immunoreactive inclusions (FTLD-U) (). The cellular pathologies of both FTLD-U and ALS demonstrate accumulation of ubiquitin-positive protein deposits that are also positive for p62/Sequestosome-1, a common component of protein inclusions associated with neurodegenerative disease (). p62 can bind polyubiquitin through its UBA domain () and interacts with the autophagic protein Atg8/LC3 (; ), thus providing a possible link between protein accumulation and aggregation with autophagy-mediated clearance.
In light of the mutations found in CHMP2B, we asked whether the ESCRT machinery is required for autophagic degradation and prevention of formation of protein aggregates associated with neurodegenerative disease. In this study we show that autophagic degradation is inhibited in cells depleted of ESCRT subunits and in cells overexpressing CHMP2B mutants, leading to accumulation of protein aggregates containing ubiquitinated proteins, p62 and Alfy (autophagy-linked FYVE protein). Moreover, we show that TAR DNA-binding protein 43 (TDP-43), recently identified as the major ubiquitinated protein in FTLD-U and ALS, accumulates in ubiquitin-positive inclusions in ESCRT-depleted cells. Using conditional cell-based systems for HD, we find that functional MVBs are also required for efficient clearance of the expanded polyglutamine aggregates. Collectively, our data indicate that efficient autophagic degradation requires functional MVBs and provide a possible explanation to the observed neurodegenerative phenotype seen in patients with CHMP2B mutations.
To investigate whether depletion of ESCRT subunits leads to accumulation of ubiquitinated proteins, HeLa cells transfected with siRNA against Hrs, ESCRT-I (Tsg101), -II (Vps22), or -III (Vps24) were labeled with antibodies recognizing ubiquitin (Ub), the early endosome antigen 1 (EEA1) and the late endosomal/lysosomal marker Lamp2 and analyzed by confocal immunofluorescence microscopy. In contrast to siRNA controls (; and Fig. S1, available at ), depletion of Hrs resulted in accumulation of ubiquitin on EEA1-positive early endosomes ( and Fig. S1). This was also the case in cells depleted of Tsg101, but in addition large ubiquitin-positive EEA1-negative structures were seen in close proximity to Lamp2-positive membranes ( and Fig. S1). Cells depleted of Vps22 generally had a similar phenotype as Tsg101-depleted cells ( and Fig. S1), but the penetrance of the Vps22-depleted phenotype was weaker than in cells lacking Tsg101. This could be due to different siRNA-mediated knockdown efficiency (). There was no accumulation of ubiquitin on early endosomes in Vps24-depleted cells, but large ubiquitin-positive structures that either were devoid of endosomal membranes or localized close to Lamp2-positive structures could be detected ( and Fig. S1). Similar results were obtained using at least two different siRNA oligonucleotides for each protein. We used siRNA against the Vps24 subunit of the ESCRT-III complex, found to form a sub-complex with Vps2/CHMP2 (; ), as two isoforms of Vps2/CHMP2 are expressed in HeLa cells (unpublished data). Depletion of one ESCRT subunit did not affect the expression levels or stability of subunits of other ESCRT complexes (unpublished data).
To further investigate the nature of the large ubiquitin-positive EEA1-negative structures found in ESCRT-depleted cells, we used cells treated with siRNA against Tsg101 and Vps24 for further studies. We first asked whether p62 and Alfy, proteins known to associate with cytoplasmic ubiquitin-positive structures (; ), accumulate in cells depleted of Tsg101 or Vps24. As can be seen in , both p62 () and Alfy () were found to associate with the ubiquitin-positive structures that accumulate in ESCRT-depleted cells. Both the number and size of p62-positive structures were dramatically increased in Tsg101- and Vps24-depleted cells, as quantified from more than 300 cells using ImageJ software. The p62-positive particles were grouped according to their size, and the numbers per 100 cells are presented in . Control cells had an average of 355p62-positive structures per 100 cells, with only 29 having an area above 2 μm. In contrast, cells depleted of Tsg101 or Vps24 had on average 1,251 and 1,605 p62-positive structures, with 263 and 332 above 2 μm, respectively. Thus, depletion of ESCRT subunits leads to both increased numbers and sizes of ubiquitin-, p62- and Alfy-positive aggregates or inclusions.
Formation of ubiquitin-positive aggregates, as seen in Tsg101- and Vps24-depleted cells, could be caused either by increased protein synthesis or decreased protein degradation. The ubiquitin–proteasome system is the principal means of eliminating polyubiquitinated damaged or misfolded proteins, but also loss of autophagy was recently found to lead to accumulation of ubiquitin-positive inclusions and cause neurodegeneration in mouse models (; ). We therefore asked whether protein synthesis, proteasome activity, or autophagic degradation was affected by knockdown of ESCRT subunits. The total level of protein synthesis, as measured by incorporation of [H]-leucine, was not considerably changed in ESCRT-depleted cells (Fig. S2 A, available at ). Neither the relative mRNA levels of p62 nor LC3-B were significantly increased in ESCRT-depleted cells, as measured by quantitative real-time PCR (qRT-PCR) (Fig. S2 B). Proteasome activity was analyzed in the absence or presence of the proteasome inhibitor PSI in control and ESCRT-depleted cells, and, interestingly, proteasome activity was increased rather than decreased in ESCRT-depleted cells (Fig. S2 C). This could be due to proteolytic cross-talk, induced by inhibition of autophagic degradation in ESCRT knockdown cells, as described in the next section.
LC3/Atg8 is a widely used marker for autophagy, as it binds specifically to autophagic membranes and remains bound throughout the pathway (; ). p62, known to interact with LC3 and to be degraded by autophagy (; ), is another commonly used autophagy marker. HeLa cells stably expressing LC3 fused to GFP () were transfected with control, Tsg101, or Vps24 siRNA, and the levels of GFP-LC3 and p62 were analyzed by confocal microscopy. Although p62 and GFP-LC3 colocalized on small cytoplasmic structures in control cells (, inset), a massive accumulation of these two proteins were seen in cells depleted of Tsg101 or Vps24 (, insets). Both the overall levels of p62 and GFP-LC3 and their degree of colocalization were increased in Tsg101- and Vps24-depleted cells, as quantified from three independent experiments (). p62 accumulation was also seen upon immunoblotting of cell lysates from ESCRT-depleted HeLa (), HepII, and U2OS cells (unpublished data). Using differential detergent extraction, a technique often used as a quantitative measure of protein inclusion formation (; ), we observed a shift in p62 from less stringent extraction buffers (1% Triton-X-100, soluble) into 2% SDS (insoluble) in ESCRT-depleted cells, representing a shift into a more aggregated conformation (). p62 accumulation was also seen in cells depleted of Atg5 (Fig. S3, available at ), and although the mechanism responsible for p62 accumulation is clearly different in Atg5- and ESCRT-depleted cells, these data further demonstrate that p62-positive aggregates form when autophagic turnover is inhibited HeLa cells.
As GFP-LC3 has the potential to associate with membrane-free aggregates (), probably due to its complex formation with p62 (; ), a better way to analyze cellular autophagy levels is to measure the ratio between the 18-kD cytosolic and 16-kD lipidated autophagosome-bound form of LC3, LC3-I, and LC3-II, respectively (, ). As can be seen in , increased levels of LC3-II were detected in cells depleted of Tsg101 or Vps24. This could in principle be caused either by elevated levels of autophagy or decreased degradation. However, the endogenous LC3-II level was similar in control and ESCRT-depleted cells treated with the proton ATPase inhibitor Bafilomycin A, known to inhibit lysosomal degradation (). If ESCRT knock-down lead to increased levels of autophagy, we would expect the LC3-II levels to be higher in ESCRT-depleted than control cells treated with Bafilomycin A. Moreover, the fact that LC3-B mRNA levels were unaffected in ESCRT-depleted cells compared with control cells (Fig. S2 B) also indicates, although indirectly, that LC3 synthesis is unchanged.
To further investigate whether autophagic degradation is inhibited in cells depleted for ESCRTs we took advantage of a recently developed double-tagged mCherry-GFP-LC3 construct (dtLC3) (), which is detected as yellow fluorescent (green merged with red) in nonacidic structures (autophagosomes and amphisomes) and as red only in autolysosomes due to quenching of GFP in these acidic structures (see , cartoon). Whereas 50% of the total dtLC3 signal was red in control cells (), only 20% red was detected in Tsg101- () and Vps24-depleted cells (). This indicates that transport of mCherry-GFP-LC3 to acidic lysosomes, i.e., formation of autolysosomes, is inhibited in ESCRT-depleted cells. Depletion of Tsg101 and Vps24 also inhibited starvation-induced degradation of long-lived proteins in HeLa cells (). Compared with control cells, there was an average 60 and 43% reduction of the level of starvation-induced degradation in Tsg101- and Vps24-depleted cells, respectively. Collectively, our data strongly indicate a general requirement for functional MVBs in autophagy.
Our data indicate that ubiquitin-positive, p62-positive aggregates accumulate in ESCRT-depleted cells due to defective autophagic degradation, but at what step is this pathway impeded? Although the classical view is that autophagosomes fuse directly with lysosomes, autophagosomes have also been found to undergo fusions with earlier parts of the endocytic pathway (; ; ; ) and the term amphisome is used to describe pre-autolysosomal compartments containing both autophagic and endocytic material (). Because the ESCRT-complexes are required for proper formation of MVBs, it is likely that loss of ESCRT subunits inhibits either the fusion of autophagosomes with MVBs or fusion of amphisomes with lysosomes.
To address this issue we used immunoelectron microscopy (EM). To perform double-labeling experiments with autophagic and endosomal markers, we used HeLa cells stably expressing GFP-LC3 (). Cryosections were incubated with antibodies against GFP and the MVB/late endosome marker lyso-phosphatidic acid (LBPA), and the presence of amphisomes could then be scored based on both LC3-LBPA colocalization and morphology (). GFP-LC3 and LBPA positive amphisomes were detected in control cells (), but more frequently in cells lacking Tsg101 (42% increase) () and Vps24 (29% increase) (). In addition, clusters of double-membrane structures, consisting of autophagosomes and tubular structures which might represent phagophores, all labeling strongly for GFP-LC3, are typically found in Tsg101-depleted cells (seen in ∼25% of the cells) (), but were not so prominent in Vps24-depleted cells and never seen in control cells. Increased levels of amphisomes in ESCRT-depleted cells, visualized as colocalization between Alfy and LBPA, was also found using confocal immunofluorescence (IF) microscopy (Fig. S4, available at ). In some cells Alfy positive structures seemed to be surrounded by LBPA-positive membranes (Fig. S4 D), suggesting that protein aggregates may accumulate in amphisomes when the MVB pathway is impeded. However, due to the lower resolution of the confocal microscope we cannot distinguish fused (amphisomes) from docked autophagosome–endosome vesicles, and EM analysis is thus a better way of addressing this issue.
To determine whether p62-positive structures could be found within autophagic vesicles, cryosections of HeLa cells treated with control, Tsg101, or Vps24 siRNA were incubated with antibodies against p62, followed by colloidal gold-labeled secondary antibodies. p62-positive structures were detected within amphisomes both in control cells and in cells depleted of Tsg101 or Vps24 (). In addition, clusters of small vesicular–tubular elements and larger vesicles of typical endosomal morphology were often found within p62-labeled areas in Tsg101- and Vps24-depleted cells (). Membrane-free dense p62-positive cytosolic aggregates were also frequently found in these cells (, C and D; and ), but never in control cells ().
Mutations in the ESCRT-III subunit CHPM2B were recently linked to FTD () and ALS (). Sequencing of CHMP2B in a Danish pedigree with autosomal-dominant FTD identified a G-to-C transition in the acceptor splice site of exon 6 in CHMP2B in affected individuals, generating two aberrant transcripts. One transcript contained the 201-bp intronic sequence in between exon 5 and 6 (CHMP2B), resulting in a premature stop codon, and thus a 36-amino acid C-terminal truncation. The other had a 10-bp deletion due to use of a cryptic splice site located 10 bp from the 5′ end of exon 6 (CHMP2B), leading to the final 36 amino acids of CHMP2B being replaced with an abnormal 29-amino acid C terminus ().
The patient brains from this family contain ubiquitin and p62-positive inclusions () and therefore, one hypothesis that could explain the neurodegenerative phenotype seen in these patients is that mutations in CHMP2B result in repressed autophagic degradation. To investigate this hypothesis, HeLa cells were transfected with myc-tagged wild-type or mutant CHMP2B, stained with antibodies against myc, ubiquitin, and p62, and analyzed by confocal microscopy. As can be seen in , untransfected cells and cells expressing wild-type CHMP2B showed weak staining for ubiquitin and p62 on small cytoplasmic structures. In contrast, the levels of ubiquitin and p62 were strongly increased (20–40 times) in cells expressing CHMP2B () and CHMP2B (). Increased p62 levels were also detected by immunoblotting in cells expressing CHMP2B compared with mock-transfected cells or cells transfected with wild-type CHMP2B (). Although the transfection efficiency was below 30%, the total expression levels of CHMP2B and CHMP2B were equal (). p62 was also found to accumulate in the Triton-X-100 insoluble fraction in CHMP2B-expressing cells, as determined by differential detergent extraction (unpublished data). As in ESCRT-depleted cells, increased levels of GFP-LC3 were seen in HeLa GFP-LC3 cells expressing CHMP2B mutants (Fig. S5, B and C, available at ) compared with untransfected cells and cells expressing wild-type CHMP2B (Fig. S5 A).
We then used immuno-EM to further characterize the morphology of p62-positive structures formed in cells expressing CHMP2B mutants. In cells depleted of Tsg101 and Vps24, p62-positive autophagosomes, amphisomes, clusters of small vesicular–tubular elements, and membrane-free dense p62-positive cytosolic aggregates were found (). Similar p62-positive structures were detected in cells expressing the CHMP2B mutant and a p62-positive cluster of small vesicular–tubular elements can be seen in . Collectively, our results indicate that expression of CHMP2B mutants inhibit autophagic degradation, leading to accumulation of ubiquitin, p62, and GFP-LC3.
FTLD-U and ALS are characterized by abnormal accumulation of p62- and ubiquitin-positive, tau- and α-synuclein-negative neuronal cytoplasmic inclusions, and TDP-43 was recently identified as the major ubiquitinated protein of these diseases (; ). We therefore asked whether TDP-43 accumulates in cytoplasmic aggregates in HeLa cells depleted of Tsg101 or Vps24. In control cells, TDP-43 was mainly detected in the nucleus and in a few cells also in small cytoplasmic structures positive for p62 (). In contrast, in cells depleted of Tsg101 () or Vps24 (), TDP-43 accumulated in aggregates that also stained positive for p62 and ubiquitin. Thus, our data show, for the first time, a link between TDP-43–positive inclusions and depletion of proteins required for MVB formation and autophagic degradation.
We did not detect TDP-43 in the cytoplasmic ubiquitin-, p62-positive structures found in cells expressing CHMP2B and CHMP2B (Fig. S5, E and F). This is in line with recent data showing that ubiquitin-positive inclusions in patients from the Danish family with the CHMP2B mutation are TDP-43 negative (). This is in contrast to other cases of FTLD-U, and might suggest that two distinct effects of MVBs may be occurring in TDP-43–positive and–negative FTLD-U.
Our data indicate that functional MVBs play an important role in preventing formation of ubiquitin-positive inclusions, and we next asked whether clearance of expanded polyQ inclusions associated with HD also requires functional MVBs. Conditional expression of exon 1 of Huntingtin with a pathogenic polyglutamine stretch of 65 or 103 repeats fused to monomeric CFP (Htt 65Q- or 103Q-mCFP) in HeLa and Neuro2a (N2a) cell lines was recently shown to lead to formation of inclusions that were readily cleared by autophagy within 5 d after shutting off protein expression (). The time required for aggregate clearance was comparable to the clearance observed in primary neurons generated from an inducible mouse model of HD (; ). We therefore used these cells to determine whether inclusion clearance still occurred in cells depleted of Vps24. This was analyzed by filter-trap experiments (amount of SDS-insoluble material) () or confocal analysis (number of cells having aggregates) (). Although 25–30% of HeLa Htt103Q-mCFP transfected with control siRNA had inclusions in the absence of doxycycline (control-dox), only 5% of the control cells had visible inclusions after 3 d of dox treatment (control+dox) (). Aggregate clearance was severely reduced in cells depleted of Vps24, as 15–20% of the cells had visible Htt103Q-mCFP inclusions after 3 d of dox treatment (Vps24+dox) (). Similar results were obtained using HeLa Htt65Q-mCFP cells, as analyzed by filter-trap experiments (). Moreover, depletion of Vps24 in mouse neuronal cells (N2a Htt103Q-mCFP) also inhibited inclusion clearance compared with control cells after 3 d of dox treatment ().
Together, our data indicate that clearance of Htt-positive inclusions depends on functional MVBs. This adds to previous reports showing that pathogenic Htt proteins are cleared by autophagy (; ; ; ) and reveals an important role of ESCRT complexes in autophagic degradation of protein aggregates.
The ESCRT complexes are involved in sorting of endocytosed ubiquitinated integral membrane proteins into the ILVs of MVBs. Their depletion in mammalian cells has been found to inhibit degradation of proteins like the EGF-R and results in MVBs with abnormal morphology (). However, little is known about the role of MVBs, and the consequence of depleting ESCRT subunits, for autophagic degradation. In this study we show that siRNA-mediated depletion of ESCRT subunits inhibits autophagic degradation, leading to accumulation of large ubiquitin-positive protein aggregates, also containing Alfy and p62, proteins known to closely associate with cytoplasmic ubiquitin-positive structures and autophagic membranes (; ). We show that degradation of the autophagic membrane protein Atg8/LC3-II and the autophagic substrate p62 is inhibited and that there is a shift toward p62 insolubility, indicative of a more aggregated conformation, in cells depleted of Tsg101 and Vps24. Autophagosomes and amphisomes are formed in ESCRT-depleted cells, but the formation of autolysosomes is inhibited. This is in line with previous studies, showing that mutations in Vps class C proteins (; ) and the class E Vps proteins Vps4p/SKD1 (; ; ) and CeVPS-27 () impede the formation of autolysosomes. In addition, we observed clusters of small vesicular–tubular elements and membrane-free dense p62-positive cytosolic aggregates in ESCRT-depleted cells. Although we cannot exclude the possibility that ESCRTs also have non-MVB related functions in autophagy, our data strongly indicate that dysfunctional MVBs inhibit autophagic degradation, leading to formation of large aggregates that eventually may cause neurodegenerative disease.
The ESCRT-III subunit CHMP2B, forming a complex with Vps24 (; ), was recently found to be mutated in a large Danish family with familial FTD () and in patients with ALS (). CHMP2B mutations are not a common cause of FTD, as several studies have failed to identify CHMP2B mutations in FTD patients (; ). We show here that cells expressing CHMP2B mutants corresponding to the mutation found in the Danish FTD patients are characterized by accumulation of ubiquitin, p62, and LC3, indicating that autophagic degradation is impeded in these cells. Our data thus suggest a possible explanation to the observed neurodegenerative phenotypes seen in CHMP2B mutant patients. Ubiquitin and p62 are common components of protein inclusions associated with neurodegenerative disease () and are found in the brains of the Danish FTD patients (). However, the ubiquitin- and p62-positive inclusions observed in the Danish FTD patient brains with the CHMP2B mutation occur at low frequency as compared with other cases of FTLD-U and are generally observed mostly in the hippocampus, which is not a site of the neurodegenerative pathology (). This suggests that although autophagy may be impaired globally, leading to cell death, the formation of ubiquitin- and p62-positive inclusions occurs only in a subset of cells in vivo. This could be because the majority of cells with such inclusions have degenerated by the end stage of the disease that is observed in material, or that the inclusions are a protective mechanism against impaired autophagy.
Ubiquitin-positive tau-negative neuronal cytoplasmic inclusions are common pathological features in FTLD-U and ALS, and TDP-43 was recently identified as the major ubiquitinated protein in these disorders (; ). We here show that TDP-43 accumulates in cells depleted of Tsg101 and Vps24, suggesting that impaired MVB function could have a role in TDP-43 aggregate formation in FTLD-U and ALS. It is not clear from our experiments if TDP-43 itself is degraded via autophagy, although this could be one explanation for its accumulation. In contrast to other cases of FTLD-U, the ubiquitin-positive inclusions found in patients from the Danish CHMP2B mutant family are TDP-43 negative (). We also failed to detect TDP-43 in the cytoplasmic ubiquitin-, p62-positive structures that accumulate in cells expressing mutant CMHP2B, suggesting that the molecular mechanisms responsible for accumulation of p62 and TDP-43 differ.
Using a cell-based system for HD, we show that ESCRTs (Vps24) are required also for efficient clearance of the Htt polyQ aggregates, both in human HeLa and mouse neuronal cells. It has previously been shown that the autophagic pathway is responsible for clearance of Htt polyQ aggregates (; ; ; ), but our results show, for the first time, that functional MVBs and a Vps class E protein are required for efficient clearance of Htt inclusions.
Recently, it was found that mice with neuronal-specific deficiencies for Atg5 or Atg7, proteins known to be essential for autophagy, are characterized by accumulation of cytoplasmic inclusion bodies and a neurodegenerative phenotype (; ). Our results indicate that depletion of ESCRT proteins likely will result in a similar phenotype, but do not allow us to conclude about when these proteins become ubiquitinated. There is little or no evidence to date showing that polyubiquitinated proteins are normally degraded by autophagy and it is probably more likely that proteins that are normally turned over by autophagy become polyubiquitinated when autophagic degradation is inhibited. However, it was recently found that induction of autophagy leads to enhanced delivery of ubiquitin to lysosomes and that this correlates with enhanced lysosomal bactericidal capacity (), which might support the idea that polyubiquitinated proteins are autophagic substrates. p62, containing a ubiquitin-binding UBA domain (), would be an excellent candidate for recognition and targeting of polyubiquitin linked proteins to the autophagic pathway ().
In conclusion, we have shown that depletion of ESCRT subunits or overexpression of CHMP2B mutant proteins inhibit autophagic degradation, leading to accumulation of ubiquitin-positive aggregates that contain proteins associated with neurodegenerative disease. Our data indicate that functional MVBs are required to prevent accumulation of abnormal proteins that can disrupt neural function and ultimately lead to neurodegeneration.
HeLa cell cultures were maintained as recommended by American Type Culture Collection (Manassas, VA). HeLa and N2a cells stably expressing a Tet-off inducible exon1 of Htt carrying a polyQ expansion of 65 or 103 residues, fused to monomeric enhanced CFP at the C termini (Htt 65Q- or 103Q-mCFP), were cultured as described () and 100 ng/ml doxycycline (dox) was used to shut off production of new protein. HeLa cells stably expressing GFP-LC3 were a gift from Aviva Tolkovsky (University of Cambridge, Cambridge, UK; ).
Rabbit antibodies against Hrs (), Vps24 (), and Alfy () have been described before. An antibody recognizing human Vps22 was made by injecting rabbits with recombinant Vps22 as a fusion with maltose binding protein (MBP) (Eurogenetec). The antiserum was affinity purified on Vps22-MBP Affi-Gel beads (Bio-Rad Laboratories). A mouse monoclonal antibody against Tsg101 was obtained from GeneTex. Rabbit anti-LC3 antibody was a gift from Tamotsu Yoshimori (Osaka University, Osaka, Japan). Human anti-early endosomal antigen (EEA)1 antiserum was a gift from Ban-Hock Toh (Monash University, Melbourne, Australia). Rabbit anti-lysosomal-associated membrane protein (LAMP)2 was a gift from Gillian Griffiths (University of Oxford, Oxford, UK). Mouse monoclonal anti-lyso-phosphatidic acid (LBPA) was provided by Jean Gruenberg (University of Geneva, Geneva, Switzerland). Guinea pig anti-p62 C-terminal antibody was from PROGEN Biotechnik GmbH. Mouse monoclonal antibodies against α-tubulin and against conjugated mono- and polyubiquitin (FK2) were from Sigma-Aldrich and Affiniti Research Products, respectively. Anti-GFP antibody was from AbCam. Rabbit anti-TDP-43 antibody was from ProteinTech Group. Rabbit anti- c-Myc antibody was from Santa Cruz Biotechnology, Inc. Cy2-, Cy3-, and Cy5-labeled secondary antibodies were from Jackson ImmunoResearch Laboratories.
The following previously described siRNA oligonucleotides were used: Hrs (), Tgs101 (), Vps24 (), and control (). Vps22 was depleted by the siRNA duplex: sense 5′-CUUGCAGAGGCCAAGUAUA-3′ and antisense 5′-UAUACUUGGCCUCUGCAAG-3′ (MWG-Biotech). Results were confirmed by the use of ON-TARGETplus SMARTpool siRNA (Dharmacon) against human Hrs, Tsg101, and Vps22. ON-TARGETplus SMARTpool siRNA against human Atg5 and against mouse Vps24 were also used. Transfection of HeLa cells with siRNA oligonucleotides was performed as described previously (). In brief, the cells were transfected with 40–100 nM siRNA using Oligofectamine (Invitrogen) for 3 d; the cells were then replated and left for another 2 d before experiments were performed. Specific protein knockdown was demonstrated by running equal amounts of cell lysate on SDS-PAGE, followed by Western blotting using antibodies against Hrs, Tsg101, Vps22, or Vps24.
HeLa cells were transfected with mCherry-GFP-LC3 after 4 d of siRNA transfection and incubated for another 24 h before analysis by confocal microscopy. Transfection of HeLa cells with cDNA encoding myc-tagged wild-type CHMP2B, CHMP2B, or CHMP2B in pLNCX2 (CLONTECH Laboratories, Inc.), or mCherry-GFP-LC3 was performed using FuGene6 (Roche), according to the manufacturer's instruction.
To analyze the cellular levels of different proteins and their solubility, cells were first extracted in ice-cold lysis buffer (50 mM NaCl, 10 mM Tris, 5 mM EDTA, 0.1% SDS, and 1% Triton X-100 + protease and phosphatase inhibitor cocktails), centrifuged (14,000 rpm) for 10 min, and the supernatants (soluble fraction) collected. The remaining protein pellets were washed with phosphate-buffered saline (PBS) before extraction with 2% SDS-containing sample buffer (insoluble fraction). Protein concentrations in the soluble fractions were determined and ∼20 μg of protein per sample was loaded and resolved on 15% or 4–20% gradient gels (Pierce Chemical Co.) followed by electro-blotting to Immobilon-P membranes (Millipore). The blots were probed with specific antibodies, which were detected using standard ECL reagents. The intensities of the of the different bands obtained were quantified using the software provided by the ChemiGenius imaging system (Syngene) and relative amounts quantified using tubulin as a loading control. 200 nM Bafilomycin A was added for 8 h to inhibit lysosomal degradation where indicated.
HeLa cells grown on coverslips, transfected or not with the indicated siRNA or plasmid, were fixed in 3% paraformaldehyde, permeabilized with 0.05% saponin, and stained for fluorescence microscopy as described previously (). Coverslips were examined using a microscope (LSM 510 META; Carl Zeiss MicroImaging, Inc.) equipped with a Neo-Fluar 100×/1.45 oil immersion objective. Image processing and analysis were done with Zeiss LSM 510 software version 3.2, ImageJ (National Institutes of Health, Bethesda, MD; , 1997–2007), and Adobe Photoshop 7.0.
HeLa cells, transfected or not with the indicated siRNA or plasmid, were fixed in 4% formaldehyde/0.2% gluteraldehyde in 0.1 M phosphate buffer at room temperature for 40 min, washed, scraped, and pelleted in 12% gelatin at 10,000 rpm. Specimens were infiltrated with 2.3 M sucrose, mounted on silver pins, and frozen in liquid nitrogen. Ultrathin cryosections were cut at −110°C (EM FCS ultramicrotome; Leica) and collected with a 1:1 mixture of 2% methyl cellulose and 2.3 M sucrose. Sections were transferred to formvar/carbon-coated grids and labeled with antibodies against p62 or GFP and LBPA, followed by Protein A conjugates essentially as described (). Sections were observed at 60–80 kV using a Philips CM10 and a JEOL JEM-1230 electron microscope, equipped with a SIS Megaview 3 or Morada camera, respectively. Quantification of labeling was performed by counting gold particles on randomly chosen cell profiles. The number of p62-positive aggregates was counted in ∼200 cells per group (control, Tsg101-, and Vps24-siRNA) from three different experiments.
HeLa cells treated with control or siRNA against Tsg101 or hVps24 were incubated for 24 h with 0.25 μCi/ml -[C] valine–supplemented media. Cells were rinsed three times with PBS to remove unincorporated radioisotopes and then chased in fresh complete media containing 10 mM cold valine for 2 h to allow degradation of short-lived proteins. Cells were rinsed in HBSS + 10 mM Hepes and incubated for 4 h with either complete media or HBSS + 10 mM Hepes + 10 mM Valine ± 10 mM 3-methyl adenine (3-MA) (Sigma-Aldrich). Cells were then scraped and, using TCA, protein was precipitated from both the incubation media and the cells. Proteolysis was assessed as the acid-soluble radioactivity divided by the radioactivity maintained in the precipitate.
48 h after siRNA transfection, HeLa Htt65Q- or 103Q-mCFP cells or N2a Htt103Q-mCFP cells were exposed to 100 ng/ml dox for another 3 d to shut down production of new Htt65Q/103Q-mCFP protein and permit more than 50% of clearance to occur (). The effect of siRNA treatment on polyQ aggregate clearance was analyzed by using the membrane filter assay for detection of amyloid-like polyglutamine-containing protein aggregates, according to the published protocol (; ). An anti-GFP antibody was used to detect 65Q/103QmCFP. The total protein load was normalized to the volume of the soluble fraction. Alternatively, clearance of Htt103QmCFP aggregates was analyzed by confocal immunofluorescence microscopy. The percentage of cell having aggregates was quantified by counting 300 cells for each condition from three independent experiments.
Cells were incubated for 2 h in Hepes medium lacking leucine. The cells were then incubated in Hepes medium containing 2 μCi ml [H]-leucine for 20 min at 37°C. Cells were extracted with 5% trichloroacetic acid (TCA) for 10 min, followed by a wash (5 min) in 5% TCA and subsequently dissolved in 0.1 M KOH. The cell-associated radioactivity was measured.
Proteasome activity was analyzed using the Proteasome-Glo Cell-based assay (Promega), which measures the chymotrypsin-like protease activity associated with the proteasome complex, according to the manufacturer's instruction. The proteasome inhibitor PSI (50 μM) was added 5 h before the analysis.
Total RNA was extracted from 5 × 10 HeLa cells using the Aurum Total RNA mini kit (Bio-Rad Laboratories), according to the manufacturer's instructions. Purity and quantity were measured by optical density. 1 μg total RNA was used for cDNA synthesis using the iScript cDNA Synthesis kit (Bio-Rad Laboratories). Real-time PCR was performed in parallel 20-μl reactions containing 10 μl 2× QuantiTect SYBR Green PCR master mix (QIAGEN), 2 μl 10× QuantiTect Primer Assay (QIAGEN), and 20 ng cDNA (2 ng were used for actin) in 96-well optical plates. The cycling conditions for the LightCycler480 (Roche) were 95°C for 15 min, 40 cycles of 94°C 15 s, 55°C 20 s, and 72°C 20 s The following prevalidated QuantiTect Primer Assays were used: Hs_SQSTM1_1_SG, Hs_MAP1LC3B_1_SG, Hs_ACTB_1_SG, Hs_TBP_1_SG. Real-time efficiencies were calculated from the slopes of the standard dilution curves.
values (
> 35 rejected) and relative quantities calculated by the ΔΔ
equation or transformed into linear form by 2. Transcripts were normalized to the quantity of actin and TBP for each condition.
Fig. S1 shows single channel images of the insets in Fig. S2 shows that neither protein synthesis, transcription, nor proteasome activity is drastically affected by depletion of ESCRT subunits. Fig. S3 shows that both the number and size of p62-positive structures increase strongly in cells depleted of Atg5. Fig. S4 shows increased colocalization of Alfy and LBPA in HeLa cells depleted of ESCRT subunits. Fig. S5 shows that p62 and GFP-LC3, but not TDP-43 accumulate in cells expressing CHMP2B mutants. Online supplemental material is available at . |
Polycystin-2, also known as TRPP2, is a member of the transient receptor potential (TRP) family of channels that are present in organisms from yeast () to humans (; ; ; ; ). Characteristic of TPR channels, polycystin-2 contains six transmembrane domains and functions as a Ca-permeable, nonselective cation channel (; ; ; ). This channel is found predominantly in the ER, but also in the basolateral plasma membrane, mitotic spindles (), and, notably, primary cilia (; ). Polycystin-2 is of particular interest because mutations affecting it or polycystin-1, with which it interacts, are the primary cause of autosomal dominant polycystic kidney disease (PKD), a genetic disorder characterized by the formation of cysts, particularly in the ducts of the kidney and liver (; ).
Polycystin-2 located on primary cilia in the kidney is a pivotal factor in the etiology of PKD. The first clue linking PKD and cilia came from the characterization of intraflagellar transport (IFT) subunits in (). IFT is a transport mechanism responsible for the assembly and maintenance of eukaryotic flagella (; ; ). The IFT machinery consists of the anterograde motor kinesin-2, the retrograde motor dynein 1b (dynein 2 in mammals), and IFT particles comprised of ∼17 proteins. These IFT components are conserved in organisms with cilia including humans, mice, zebrafish, , and (; ; ). The link between PKD and cilia emerged when the mouse homologue of one IFT subunit, IFT88, was found to be encoded by (), a gene previously shown to cause kidney defects very similar to those seen in humans with autosomal recessive PKD (). The tie between PKD and cilia became clearer as homologues of polycystin-1 and -2 were found on the cilia of sensory neurons of () and the primary cilia of mouse kidney (; ). These and subsequent studies have shown that almost all the proteins associated with the PKD phenotype (polycystin-1, polycystin-2, fibrocystin, nephrocystins, inversin, cystin, IFT88/polaris/Tg737, and Bardet-Biedl syndrome proteins) are present on cilia ().
Understanding the importance of cilia in maintaining homeostasis of the kidney has stimulated the study of the sensory function and signaling pathways of kidney cilia. These studies support a model in which polycystin-1 and -2 comprise a mechanosensory complex that translates deflection of the cilium into signals associated with the control of the cell cycle (). Ciliary bending results in an increase in intracellular Ca () that is dependent on polycystin-1 and -2 (). In the absence of fluid flow and ciliary bending, the C terminus of polycystin-1 is cleaved and moves to the nucleus, where it activates signaling pathways known to be involved in cell cycle control (; ). How the signals generated by ciliary PKD2 are transferred to the cell body and whether the many PKD- related proteins located in the cilium interact with PKD1 and 2 to participate in signal transduction is unclear.
Most studies of PKD have been performed in mice, vertebrate kidney cell lines in culture, or the nonmammalian model (; ). In none of these systems is it practical to isolate sufficient cilia for biochemical analysis. For this reason we chose to work on polycystin-2 in the biflagellate alga , where biochemical quantities of cilia can be easily isolated. This organism also has excellent, well-established genetics to complement the biochemistry, as well as a sequenced genome and proteomic data from both the cilia and basal bodies (; ). Furthermore, homologues of polycystin-2 and other PKD-related proteins were identified by comparative genomic and flagellar proteomic analyses (; ). In this paper, we cloned a gene encoding a member of the PKD2 family. The protein is cleaved in the cell body before entering flagella and is enriched in the flagellar membrane. Most of the PKD2 (CrPKD2) remains static in the flagella, but a fraction (<10%) moves at a rate similar to IFT. RNAi knockdown of CrPKD2 results in a decrease in the efficiency of mating, a process initiated by the contact of flagella of gametes of opposite mating types (mts), which activates a calcium signaling pathway required for the fusion of the gametes. These observations provide new insights into the function and dynamics of PKD2 in cilia.
A predicted protein in the genome sequence database (C_590099 in assembly two) showed significant similarity to human polycystin-2 (e = 6 × 10), and 10 peptide fragments of this protein were found in the flagellar proteome (). The available EST clones corresponding to this protein () were sequenced, and one, AV395567, encoded most of the protein as well as the 3′ untranslated region (UTR), including a polyadenylation signal (TGTAA) typical for a gene. No EST clone encoded the putative transcriptional start site of this gene, so several sets of primers were designed based on genomic sequence to be used for RT-PCR. According to the sequence of these PCR products, the cDNA begins 410 bp upstream from the transcriptional site predicted by the C_590099 gene model (the cDNA sequence is available from GenBank/EMBL/DDBJ under accession no. ). The gene (Fig. S1 A, available at ) encodes a protein of 1,626 aa (Fig. S1 B) with a molecular mass of 181 kD and a theoretical isoelectric point of 4.86.
A comparison of the predicted gene product with a variety of TRP channels shows that it clusters with the TRPP subfamily (Fig. S2, available at ) and exhibits the signatures typical of PKD2: it has six transmembrane domains that show homology to ion transport proteins, a coiled-coil domain at the C terminus, and an EF hand domain (). The six residues in the CrPKD2 sequence predicted to coordinate Ca binding in the EF hand (positions 1, 3, 5, 7, 9, and 12) are allowable (): X (), Y (NS), Z (DNSTG), −Y {GP}, −X (DNQSTAGC), and −Z (D) (CrPKD2 residues are underlined; Prosite documentation PDOC00018). Moreover, the C terminus of CrPKD2 contains two acidic clusters; one has the casein kinase II phosphorylation motif (EGDDKDDSPEVREE), where S is the phosphorylation site predicted by NetPhosK 1.0 (), and the other one does not (DEDEDDD; and S1). Acidic clusters bind phosphofurin acidic cluster sorting proteins, which are involved in routing PKD2 between the ER, Golgi apparatus, and plasma membrane in mammalian cells (). Because acidic clusters are characteristic of mammalian PKD2, but not PKD2-like proteins (), CrPKD2 may be an orthologue of PKD2; however, this trait alone cannot unambiguously distinguish CrPKD2 from other PKD2 family members (i.e., PKD2L1 and PKD2L2). In addition, the casein kinase II phosphorylation site that controls targeting and function of PKD2 in the sensory cilia of () is also present in CrPKD2.
Unlike other PKD2s, however, CrPKD2 is predicted to have a coiled-coil domain at its N terminus. Another distinguishing characteristic of CrPKD2 is that its first extracellular loop is atypically long (1,085 aa). Whereas two subfamilies of TRP channels, TRPP and TRPML, have an extracellular loop between the first two transmembrane domains, this loop is typically only ∼200 aa residues long ().
Two polyclonal antibodies were generated against CrPKD2. One antibody (L1-PKD2) was directed against the first predicted loop (aa 358–1,019) expressed in () and the other (C-PKD2) was generated against a C-terminal peptide (aa 1,444–1,462). Immunoblots of whole cells probed with the L1-PKD2 revealed three proteins of 210, 120, and 90 kD (, lane 1). Only the two smaller fragments (120 and 90 kD) were present in flagella (, lane 2). The smaller forms of CrPKD2 do not appear to arise from alternate mRNA splicing because only one 6-kbp transcript, corresponding to the 210-kD form, hybridizes to probes (Fig. S3, available at ). When immunoblots were probed with C-PKD2, two forms, 210 and 90 kD, were detected in whole cells (, lane 3) and only the 90-kD form was seen in flagella (, lane 4). C-PKD2 did not recognize the 120-kD protein, suggesting that this species lacks the C terminus. The simplest explanation of the multiple forms of CrPKD2 is that posttranslational cleavage of 210-kD full-length CrPKD2 produces a 120-kD N-terminal fragment and a 90-kD C-terminal fragment. The epitope recognized by L1-PKD2 spans the cleavage site, so this antibody recognizes both fragments, whereas the C-PKD2 epitope is unique to the C-terminal fragment (). To test this explanation, an N-terminal peptide of the first loop (aa 358–583) was used to affinity purify an antibody from the L1-PKD2 antiserum. This antibody (N-PKD2) was predicted to only react with the 210- and 120-kD forms of CrPKD2. As expected, only these two bands were seen in immunoblots of whole cells and only the 120-kD band appeared in flagella (, lanes 5 and 6).
The identities of the 210- and 90-kD bands were further corroborated by tagging the C terminus of CrPKD2 with GFP and expressing the fusion protein in along with endogenous CrPKD2. Probing immunoblots of cells expressing CrPKD2–GFP with an antibody against GFP revealed two bands of 240 and 120 kD (, lane 1). These represent the 210-kD full-length CrPKD2 plus the 30-kD GFP tag and the 90-kD C-terminal fragment plus the 30-kD GFP tag. Only the 120-kD band was present in flagella (, lane 2). The identity of the GFP-tagged bands in whole cells was confirmed with the C-PKD2 antibody, which detected both GFP-tagged bands along with the two endogenous untagged forms (, lane 3). These data fit the model that the 210-kD initial CrPKD2 product is cleaved near the C end of the first extracellular loop (aa 583–1,019) to produce a 120-kD N-terminal fragment and a 90-kD C-terminal fragment. A more detailed examination of immunoblots of the CrPKD2–GFP cells is presented in Fig. S4 (available at ). Further evidence that the 210-, 120-, and 90-kD bands are all derived from CrPKD2 came from analysis of CrPKD2 RNAi knockdown strains, in which the levels of all three bands were reduced (see Function of CrPKD2 in the mating of ).
To determine whether the cleavage occurs in the cell body or flagella, we isolated extracts from cells, which have only short flagellar stubs because of a mutation in the gene encoding IFT52, a component of IFT complex B (; ), and cells, which form basal bodies consisting of microtubule singlets instead of triplets because of a deficiency in -tubulin (). These aberrant basal bodies do not attach to the cell surface, form transitional fibers, nor nucleate flagellar microtubules (). We also isolated extracts from wild-type cell bodies, i.e., cell bodies from which the flagella had been detached. Both full-length and cleaved CrPKD2 were present in , , and wild-type cell bodies (), indicating that CrPKD2 can be cleaved in the cell body before entering the flagella.
When cells were stained with the N-PKD2 or C-PKD2 antibody, CrPKD2 was detected both in the cell body and flagella of vegetative cells and gametes (). No difference was detected in the distribution of CrPKD2 with the two antibodies. Immunoblot analysis of the flagella of cells undergoing gametogenesis showed that a fourfold increase in the amount of CrPKD2 gradually occurred during gametic differentiation induced when cells are starved for nitrogen ().
In the flagellar proteomic analysis, most of the CrPKD2 peptides were obtained from the axonemal fraction rather than the membrane and matrix fraction (). This was an unexpected result for a transmembrane protein, which raised the possibility that CrPKD2 might be a membrane protein tethered to the axonemal microtubules. To investigate this further, we developed methods for the isolation and purification of flagellar membranes. Membranes were released from isolated flagella with 0.1% NP-40 (4°C for 30 min), and the axonemes were removed by centrifugation. The resultant membrane and matrix fraction contained membrane vesicles in addition to soluble flagellar proteins. Membranes either were sedimented directly from the membrane and matrix fraction to recover all the membranes present or further purified by density gradient centrifugation using Optiprep to obtain a band of uniform density (see Materials and methods). EM () and SDS-PAGE analysis () of the resultant membrane fractions showed they contained little axonemal contamination and were enriched in the major flagellar membrane glycoprotein (FMG) 1 (). CrPKD2 was concentrated in the flagellar membrane fraction (), confirming that it is a membrane protein. Interestingly, IFT proteins also were found in these membrane fractions.
Because some CrPKD2 remained on the axonemes after detergent extraction, we used more stringent conditions in an attempt to solubilize more CrPKD2. Freezing flagella in liquid nitrogen and thawing releases the matrix and a small amount of membrane, but much of the membrane remains in the axonemal pellet (not depicted). Such axonemes were extracted twice with 0.1 or 1% NP-40 at room temperature for 30 min, and after low speed centrifugation, the soluble fraction contained the membrane and the pellet contained the axonemes. No vesicles or other membrane remained associated with the axonemes after extraction in 1% NP-40 (), and all the membrane glycoprotein FMG-1 was released (); however, 20–25% of the CrPKD2 still remained in the axonemal fraction. Thus, these two membrane proteins show different patterns of solubilization in the flagella: FMG-1 is completely released by nonionic detergent, whereas flagellar CrPKD2 is present in two pools, one that is easily released by nonionic detergent and one that is not.
To examine the movement of CrPKD2 in the flagellum, we made a construct designed to express CrPKD2 fused to GFP via a flexible linker of 10 aa. After transformation of cells with the construct, the fluorescence of PKD–GFP was detected in flagella. Like the immunofluorescence localization of CrPKD2 (), flagellar CrPKD2–GFP was concentrated in puncta (), which may represent clustered channels. When visualized by time-lapse imaging, <10% of the PKD–GFP particles moved continuously to the tip of the flagella at a velocity of ∼1.6 μm/s, even though most of the puncta did not move ( and Video 1, available at ). Movement of CrPKD2– GFP toward the base was not detected, perhaps because the faint fluorescence of CrPKD2–GFP was bleached as the protein moved to the tip and became too faint to see as it returned to the cell body.
To assay the movement of CrPKD2–GFP in the absence of IFT, we generated a strain that expressed CrPKD2–GFP in a background. Because is a temperature-sensitive mutant of the IFT anterograde motor, we could monitor CrPKD2–GFP movement in the presence and absence of IFT at permissive and restrictive temperatures, respectively. 50 videos were generated of flagella at permissive and nonpermissive temperatures, and kymographs of these were analyzed for movement. In 10 sequences taken at permissive temperature, movement was observed; however, none was seen in flagella of cells incubated at 32°C. These results suggest the movement of visible puncta of CrPKD2 is dependent on IFT.
Movement of small puncta is difficult to detect in kymographic analysis of videos, so FRAP was used as a more sensitive assay for the effect of IFT on CrPKD2 movement. To this end, a 2-μm section in the center of a flagellum of cells expressing CrPKD2–GFP was bleached at permissive temperature, the area was allowed to recover for 1 or 2 min without illumination, and the fluorescence intensity was measured again. The fluorescence recovered 10 and 19% of its original intensity after 1 and 2 min of recovery, respectively (). This result is consistent with the observation that most of the flagellar PKD remains stationary, whereas only a small fraction of it moves.
FRAP also was performed on cells expressing CrPKD2–GFP at restrictive temperature. After 2 min of recovery, the fluorescence in the bleached region recovered only 9.4% at nonpermissive temperature (). The reduction of FRAP at 32°C was not simply a temperature effect, because no such reduction () was seen in control cells. The cells were used because they have paralyzed flagella, facilitating measurement of CrPKD2 FRAP, but are not deficient in FLA10 or IFT (). These data suggest that IFT is important for the movement of CrPKD2–GFP in flagella, but because some recovery of fluorescence occurred in the absence of IFT, other mechanisms may be involved.
Further evidence for a role of IFT in the movement of CrPKD2 was obtained from biochemical analysis of the flagella of cells after incubation at 32°C for 1 or 1.5 h in 10 mM Hepes, pH 7.2. At these times, IFT proteins were dramatically decreased (), no IFT particles could be observed moving by differential interference contrast (DIC) microscopy (), and the flagella had shortened to about half length (Table S1, available at ). Under these conditions, the concentration of flagellar CrPKD2 increased relative to total flagellar protein (). To determine if this increase was caused by the absence of IFT or simply flagellar shortening, the amount of CrPKD2 was examined in flagella induced to shorten by the addition of sodium pyrophosphate (NaPPi; ), which causes the flagella to shorten as IFT continues unabated (; ). In this case, the concentration of flagellar CrPKD2 did not change () as the flagella shortened. Thus, flagellar shortening does not per se affect the concentration of flagellar CrPKD2 when IFT is functioning properly; however, when IFT is defective, the concentration of CrPKD2 in the flagella increases, perhaps because it cannot be removed from the flagella in the absence of IFT.
The observation that CrPKD2 increases in the flagella of gametes () suggests that CrPKD2 may have some function in the calcium-dependent steps of mating. The sexual cycle of begins with gametogenesis (). Vegetative (haploid) cells differentiate into gametes when they are starved for nitrogen. When gametes of mt+ and − strains are mixed, the flagella of gametes of opposite mts adhere to each other, mediated by mt-specific agglutinins on the flagellar membrane (). Flagellar adhesion is followed by a cascade of signaling events that result in the shedding of the cell wall, formation of mating structures on mt+ and − gametes, and ultimately cell fusion (; ; ; ; ). The early stages of this cascade are blocked by drugs that interfere with Ca signaling (; ; ), suggesting that an influx of Ca is required as an early step in mating. This influx is followed by the phosphorylation and activation of a cyclic GMP-dependent protein kinase (CrPKG), which in turn increases the concentration of flagellar cAMP by up-regulating an adenylate cyclase in the flagellar membrane (; ; ; ). Ultimately, the cAMP concentration in the cell body rises and the mt+ and − gametes elaborate mating structures that touch to initiate cell fusion.
We took advantage of RNAi to knockdown the endogenous CrPKD2 and examine the function of CrPKD2 in mating. The RNAi construct consisted of genomic CrPKD2 DNA including its native promoter with the corresponding cDNA in reverse orientation (). This strategy has been shown to efficiently knockdown endogenous proteins in (; ). Transformants were obtained in which CrPKD2 was decreased to 10% of wild-type levels (). Mating efficiency was measured as the percentage of gametes fused to form quadriflagellates after mixing mt+ gametes of wild-type or CrPKD2-depleted strains with mt− wild-type gametes (see Materials and methods). Compared with wild-type gametes, the RNAi cells showed up to a 75% reduction in mating efficiency, and the severity of the defect correlated with the extent of CrPKD2 depletion ().
Inefficient mating could be caused by a failure of cells to form gametes when starved for nitrogen. To determine whether CrPKD2-depleted cells expressed agglutinins on their flagellar surface during gametogenesis, aggregation of mt+ and − gametes was observed. RNAi knockdown gametes appeared to adhere normally to wild-type gametes of the complementary mt, demonstrating that there was no obvious defect in gametogenesis and that the flagellar agglutinins were expressed normally on the flagella of the CrPKD2-depleted gametes. For a more complete assessment of the ability of the CrPKD2-depleted gametes to mate, mating was done in the presence of cAMP. The addition of cAMP to the mating reaction bypasses the initial stages of the signaling pathway where a calcium influx is required (; ). The mating efficiency was restored in CrPKD2-depleted gametes by the addition of 15 mM dibutyryl-cAMP and 0.15 mM papaverine, indicating that these gametes were competent to mate. To investigate the effect of CrPKD2 depletion on signaling events upstream of the increase in cAMP, the phosphorylation state of a CrPKG, a protein known to be phosphorylated after flagellar adhesion in wild-type cells (; ), was assayed. Phosphorylation of CrPKG was reduced in matings of the RNAi46 strain to the wild type (), suggesting that CrPKD2 functions in mating between flagellar adhesion and CrPKG phosphorylation. The residual phosphorylation of CrPKG in these matings was caused by its phosphorylation in the wild-type gametes used to mate with the CrPKD2-depleted gametes.
xref
#text
The wild-type strains CC-125 (mt+) and CC-124 (mt−), paralyzed flagella strain (CC-1297, mt−), temperature-sensitive flagellar assembly mutant ( allele, CC-1919, mt−), and
strain CC-3681 (, mt−) were obtained from the Chlamydomonas Genetics Center. The double mutant was generated by K. Kozminski (University of Virginia, Charlottesville, VA). Cells were grown on solid media supplemented with 1.5% agar or in a liquid minimal medium, MI (), MI-N (MI medium without nitrogen to induce gametogenesis), or Tris-acetate-phosphate media () at 22–23°C with a 14/10-h light/dark cycle and constant aeration.
The EST clone AV395567 () corresponding to the predicted CrPKD2 protein C_590099 (Joint Genome Institute version 2 draft assembly of the genome) was sequenced. It included the sequence from bp 1,223 of the cDNA through the 3′ UTR of . We extended the cDNA to the 5′ UTR by sequencing the two RT-PCR products generated using two sets of primers: TCTTGGGAGTGCTGTATGAGCAGCGC (bp 318–343 of the cDNA) and CGGTGGTGTTGTACACGATGC (bp 1,696–1,716 of the cDNA); and AAGTAGCATGTCACATTAATGCATG (bp 1–25 of the cDNA) and CCGACACCGACTGGTTAAGGT (bp 1,125–1,145 of the cDNA), which were generated according to the genome sequence database (). The mRNA was isolated and reverse transcribed as described previously (). The RT-PCR products were cloned into PCRII-TOPO (Invitrogen), generating plasmids pHK5 and pHK15, respectively. The intron coded by genomic DNA was not present in the PCR products.
Loop1 CrPKD2 antibodies were generated against a His-tagged fragment of CrPKD2 (aa 358–1,019) expressed in and purified by affinity chromatography using nickel–nitrilotriacetic acid agarose according to the manufacturer's instructions (QIAGEN). The purified protein was used for antibody production in rabbits at Pocono Rabbit Farm and Laboratory Inc. A peptide (AGEGDDKDDSPEVREEKRK, corresponding to aa 1,444–1,462 of CrPKD2) was synthesized and used to produce a second antibody by the same company. The N-PKD2 antibody was affinity purified from the Loop1 antiserum using a peptide containing aa 358–583 of CrPKD2, which was expressed in and purified using nickel–nitrilotriacetic acid agarose.
Flagella were isolated from by pH shock as described previously () and modified by . The flagella from 40–48 liters of wild-type cells were resuspended in 6 ml HMDEK buffer (10 mM Hepes, pH 7.2, 5 mM MgSO, 1 mM DTT, 0.5 mM EDTA, and 25 mM KCl, including the protease inhibitors 1.0 mM PMSF, 50 μg/ml soybean trypsin inhibitor, 1 μg/ml pepstatin A, 2 μg/ml aprotinin, and 1 μg/ml leupeptin) for a protein concentration of ∼3–4 μg/μl, and NP-40 (Calbiochem) was added to a final concentration of 0.1%. The mixture was incubated at 4°C for 30 min with shaking. The membrane vesicles were separated from the axonemes by centrifugation at 16,000 for 10 min at 4°C, the supernatant was centrifuged again to remove any remaining axonemes, and the membranes were harvested by centrifugation at 228,000 for 30 min (TLA 120.2 rotor, Optima Ultracentrifuge; Beckman Coulter). The pellet was resuspended in 1 ml of 0.1% NP-40 HMDEK buffer and centrifuged again to obtain the final pellet. Alternatively, membranes were further purified by density gradient sedimentation. For this, 1/3 volume of 60% iodixanol (Optiprep density gradient medium; Sigma-Aldrich) was added to the membrane-containing supernatant for a final concentration of 15%, and 0.9 ml of the mixture was added to a 1.5-ml tube. After underlaying with 0.3 ml of 30% iodixanol in HMDEK + 0.1% NP-40 buffer, the sample was centrifuged for 1 h and 11 min at 431,000 . A white band of membrane vesicles in the middle of the tube was collected and diluted 10 times with 0.1% NP-40 HMDEK buffer. The mixture was split and centrifuged as before, sedimenting the membrane vesicles. One membrane pellet was used for EM and the other was resuspended in 100–200 μl HMDEK buffer with 0.1% NP-40 for immunoblotting.
To remove all the flagellar membrane from the axoneme, the flagella were freeze–thawed twice to release the matrix and a small amount of membrane. Much of the membrane remained in the axonemal pellet. The axonemes were exacted twice with 0.1 or 1% NP-40 at room temperature for 30 min, and after centrifugation at 16,000 for 10 min, the soluble fractions contained the membrane and the pellet contained axonemes. The membranes were sedimented at 280,000 for 30 min.
Preparation of the whole cell extract, determination of the protein concentration, PAGE, and immunoblotting were performed as described previously (). Immunoblots were scanned and the relative protein concentrations were determined using ImageJ (National Institutes of Health).
To make the CrPKD2–GFP fusion, the bacterial artificial chromosome clone 18K16 () was cut with EcoRV and SpeI, generating an ∼18-kb fragment, which included the entire gene and its promoter. This fragment was subcloned into the EcoRV and SpeI sites of pBluescript II KS+ (Stratagene), generating a plasmid named pHK25. This plasmid was cut with HindIII, generating three fragments. One of these fragments, an 8-kb fragment containing the promoter and most of the genomic DNA encoding CrPKD2, was cloned into the HindIII site of the pBluescript KS+, generating pHK28. A second fragment, a 5.8-kb fragment containing the last two introns, exons, the 3′ UTR of , and the KS+ vector, was religated to produce pHK26.
To tag the gene, we cloned the gene into a unique EcoRI site in intron 11, flanked by the first intron of RBCS2 (). For this, the two ends of the intron were subcloned as follows: using the genomic DNA as template, the 3′ end of this intron was amplified with the primers CGAGTCGACGAGCAAGCC and GTTCCTGCAAATGGAAAC. EcoRI and BamHI sites (added sites in bold) were added to the primers and the PCR product was cut with EcoRI and BamHI and cloned into the same sites of pBluescript II KS+ vector (pHK31). Another set of primers was used to amplify the 5′ end of the first intron of the RBCS2: CACCAGGTGAGTTCGACGAGCAAG including a BamHI site and CAAATGGAAACGGCGACG including XbaI and EcoRI sites. The PCR product was cut with BamHI and XbaI and cloned into pHK31 (BamHI and XbaI sites), generating pHK32. The BamHI fragment of the pCrGFP (Entelechon GmbH), which contains the codon-adapted GFP ending with a stop codon, was inserted into the BamHI site of pHK32, generating pHK35. The sense orientation of this and following inserts was identified by restriction enzyme digests and confirmed by sequencing. The EcoRI fragment from pHK35 was inserted into the EcoRI site in intron 11 of the gene of pHK26 in the sense orientation, generating pHK37. The 8-kb HindIII fragment that contains the promoter and 5′ end of was inserted in the HindIII site of pHK37 in the sense orientation, generating pHK38.
To remove the stop codon at the end of from pHK38, two primers, AGGTCGACTCTAGAC and TTTGTACAGCTCGTCCATGCCG, both containing a BamHI site, were used to amplify the fragment. The PCR product, which did not have the stop codon, was cloned into the TOPOII-PCR vector (Invitrogen), generating plasmid pHK34. The BamHI fragment of the pHK34 was exchanged with the BamHI fragment of pHK37 (which has a stop codon at the end of the gene), generating plasmid pHK39. The HindIII fragment of pHK28 was inserted into pHK39 in the sense orientation to generate pHK41. Plasmids pHK38 and pHK41 were linearized with SpeI before transformation.
Because we did not observe fluorescence in transformants that harbor pHK38 or 41, these cells were only used for immunoblots, and the overlap PCR method was used to make another construct in which GFP was fused to the end of CrPKD2 through a flexible linker. Using the left (GGAGAGTGTTTTGG, HindIII) and right primers (CGCGCCGGAGGCGCCCTGGCCGGAGGCGCCCTGGGGCGGGGTCTCATTCATCA) and pHK26 as the template to amplify the C terminus of CrPKD2, the PCR product containing the fragment from the HindIII site in the last intron to the stop codon of CrPKD2 was obtained. The nucleotide sequence GGCGCCTCCGGCCAGGGCGCCTCCGGCGCG, which corresponds to a flexible protein linker GASGQGASGA, was included in the right primer. Another set of primers (GGCGCCTCCGGCCAGGGCGCCTCCGGCGCGAAGGGCGAGGAGCTGTTCACC and CTCCGCTTCAATACG, KpnI) was used to amplify and the 3′ UTR of RBCS2 from the plasmid pCrGFP. The sequence encoding the protein linker was included in the left primer. Both PCR products were purified using a PCR purification kit (QIAGEN), and equal molar amounts of the products were used as templates to fuse these two PCR products together. For this purpose, primers (GGAGAGTGTTTTGG [HindIII] and CTCCGCTTCAATACG [KpnI]) were used, and the new PCR product was cut with HindIII and KpnI and cloned into the pBluescript II KS+ vector (HindIII and KpnI sites). The resulting plasmid was named pHK49. The HindIII fragment from pHK28 was inserted into the HindIII site of pHK49 in the sense orientation, generating pHK52. pHK52 was linearized with SpeI before transformation.
To prepare cells (CC-3681, , and for transformation, cell walls were removed with autolysin as described previously (). Linearized plasmids pHK38, pHK41, and pHK52, together with pCB412, which harbors an gene as a selectable marker (gift from C.F. Beck, University of Freiburg, Freiburg, Germany), were introduced into the cells using the glass bead method. Transformants were selected on Tris-acetate-phosphate medium plates and tested for GFP by Western blot analysis. GFP antibody was obtained from Roche Applied Science (clones 7.1 and 13.1).
To construct the (mt−) and (mt−) strains expressing CrPKD2–GFP, the original transformant (mt−) was crossed to CC-125 (mt+) to produce an mt+ strain. This strain was crossed with and , and the progeny were analyzed to obtain the desired phenotype.
To make the genomic and cDNA fusion construct for RNAi, primers (GGACATGGTTCGTAGCGTTTAATGCC [700 bp in front of the translation start site of gene] and CGACACCGACTGGTTAAGGT [corresponding to the cDNA 1,145–1,125]), including a SalI site, were used to amplify the promoter and 5′ region of the gene using genomic DNA as a template. The 2.5-kb PCR product was cloned into TOPO TA vector with the 3′ SalI site near the EcoRV site of the vector, generating plasmid pHK3. Another set of primers (AAGTAGCATGTCACATTAATGCATG [corresponding to bp 1–25 of the cDNA] and CGACACCGACTGGTTAAGGT [corresponding to the cDNA 1,145–1,125]), including a SalI site, were used to amplify the cDNA fragment, which corresponds to the genomic fragment in pHK3. The PCR product was cloned into the TOPO TA vector, and a clone with the 5′ end near the EcoRV site was identified and named pHK15. The EcoRV and SalI fragment from the plasmid pHK15 was inserted into the same sites in pHK3, creating pHK19. The PvuII fragment of pSI103 (), which includes the aphVIII gene driven by the HSP70A-RBCS2 fusion promoter, was inserted into the EcoRV site of the pHK19, generating pHK22 (). pHK22 was linearized with ScaI before transformation using the glass bead method. Transformants were selected on plates with 10 μg/ml paromomycin (Sigma-Aldrich) and tested for PKD2 levels by Western blot analysis. 95 transformants were screened and 15 clones were picked for further analysis. Eventually, four clones showed a stable reduction of CrPKD2. Two of these (Ri22 and Ri39) showed only a modest decrease, and only one of these is included. The other two (Ri45 and Ri46) showed a more substantial depletion of CrPKD2.
Wild-type strains were prepared for immunofluorescence light microscopy using methanol fixation, and were stained with primary and Alexa fluor–conjugated secondary antibodies (Invitrogen) as described previously (), modified by the addition of 0.05% glutaraldehyde directly to the medium for primary fixation before fixing with methanol at −20°C. Images were recorded with a microscope (Eclipse TE2000; Nikon) equipped with a Plan Apo 100×, 1.4 NA objective lens and a forced-air–cooled camera (Cascade 512B; Photometrics). Photoshop (Adobe) was used to adjust brightness and contrast and crop images.
For photobleaching, cells were immobilized with 0.02-M LiCl () and embedded in 0.75% low-melt agarose. First, a DIC picture was taken of the flagellum, and the GFP fluorescence of the selected region was recorded for 2 s using laser illumination at 32 mW (F). Next, the selected region of the flagellum was bleached for 3 s with the laser at 300 mW. GFP fluorescence of the selected region was recorded again for 2 s using the laser at 32 mW (F). After 1 or 2 min of recovery, the GFP fluorescence of the selected region was recorded for 2 s at low power (F). Finally, another DIC picture of the flagellum was taken. By comparing the two DIC pictures of the same flagellum, videos were selected for further analysis in which the flagellum did not move and the focus did not change. The mean fluorescence of each parameter, F, F, and F, were used to calculate recovery: % recovery = (F − F)/F. For performing the experiment at 32°C, cells were incubated at 32°C in a water bath and observed for a maximum of 10 min at room temperature on the microscope. Alternatively, cells were observed in a glass-bottom Petri dish (WillCo Wells B.V.) in a stage-mounted dish heater (DH-35; Warner Instruments). Data were analyzed using MetaMorph (Universal Imaging Corp).
cells were maintained at 22 or 32°C to induce flagellar resorption. Wild-type cells were induced to resorb their flagella with 20 mM NaPPi. Aliquots of these cells were fixed with 1% glutaraldehyde. Cells were imaged as described in the previous section and the lengths of individual flagella were measured using the MetaMorph software package.
Gametes were generated by resuspending vegetative cells in MI-N medium at a density of 1–2 × 10 cells/ml. The cells were incubated under continuous light for 16–24 h and gametes (CC-125 and RNAi strains) were adjusted to the same cell density. The gametes to be assayed were mixed with a twofold excess of gametes of the opposite mt (CC-124) to ensure that mating efficiency of the test strains was not limited by depletion of the mating partner. After completion of the mating reaction in 1 h, the cells were fixed by addition of glutaraldehyde, and the number of the quadriflagellate zygotes and biflagellate cells were counted by phase-contrast microscopy (; ). The percentage of cells that had mated was then calculated according to the method of .
To assay flagellar protein tyrosine kinase activity, gametes of CC-125 or Ri46 were mixed with the same number of CC-124 gametes, and flagella were isolated 3 min after mixing. The concentration of the flagellar protein was measured using the Amido black method (), and equivalent amounts of protein were used to assay the protein tyrosine kinase activity by immunoblotting for phosphorylated CrPKG, the substrate of protein tyrosine kinase ().
Axonemes and membrane pellets were treated sequentially at room temperature for 1 h each with 2.5% glutaraldehyde in HMDEK, 1% osmium tetroxide in HMDEK, and 1% uranyl acetate in water with brief rinses in between. The pellets were dehydrated through ethanol and propylene oxide and embedded in epoxy resin according to standard procedures. The final steps of dehydration and the initial steps of resin infiltration were performed at −20°C for the membrane samples. Silver sections were observed with an electron microscope (1230; JEOL) equipped with a digital camera (Orca HR; Hamamatsu Photonics).
Gametes were deflagellated using the pH shock method and vigorously aerated under light. Total RNA was isolated from cells before deflagellation and after 10 or 15 min of flagellar regeneration. RNA was blotted and hybridization was performed as described previously ().
Table S1 lists the lengths of flagella measured from cells induced to resorb their flagella. Fig. S1 shows a diagram of the gene and the amino acid sequence of the protein. Fig. S2 shows a phylogenetic tree of TRP channels. The Northern blots in Fig. S3 demonstrate that only one 6-kbp transcript is present encoding CrPKD2. Additional immunoblots of the CrPKD2 fragments from cells expressing CrPKD2–GFP are shown in Fig. S4. Video 1 shows the movement of CrPKD2 in the flagella of gametes at permissive temperature. The online version of this article is available at . |
For eukaryotic cells that use cilia and flagella for motile functions, motility is commonly modulated in response to extracellular cues; this modulation may include changes in waveform (; ; ; ), beat frequency (), or direction of the effective stroke (; ). Despite the diversity of responses between cell types, changes in motility are often preceded and mediated by changes in the intraflagellar concentrations of the second messengers calcium and cAMP. Ciliary and flagellar beating results from the spatial regulation of dynein activity along the axonemal microtubules (). Our goal is to understand how changes in intraflagellar calcium concentrations are converted to changes in dynein-driven microtubule sliding to modulate motility.
Substantial evidence from our laboratory and others indicates that the central apparatus and radial spokes form a signal transduction pathway that modulates ciliary and flagellar beating in response to second messengers (for reviews see ; ). Using both functional and structural approaches, our previous studies demonstrated that calcium control of motility involves the regulation of dynein-driven microtubule sliding and that CaM is a key axonemal calcium sensor (,; ; ). Based on these results, we postulate that the calcium sensor regulates the activity of specific dynein subforms and/or dynein arms attached to specific subsets of doublet microtubules, thus modulating the size and shape of ciliary/flagellar bends.
Understanding how CaM might regulate dynein activity to modulate ciliary motility requires the localization of CaM within the axoneme as well as the identification of CaM binding partners. have reported that a fraction but not all of the axonemal CaM associates with the radial spokes in . We have identified and characterized additional CaM-binding proteins in flagella using an immunoprecipitation approach. We developed antibodies against a peptide antigen unique to the C terminus of CaM and used these antibodies to precipitate CaM from extracted axonemal proteins. We previously reported that eight polypeptides precipitate with CaM and that these polypeptides form at least two different protein complexes (). One complex is comprised of five polypeptides in addition to CaM and is associated with the C1 microtubule of the axonemal central apparatus (). Here, we report the identities and localization of three polypeptides comprising the second CaM-containing complex and provide data supporting the hypothesis that this complex plays an important role in modulating the activity of specific subsets of dynein arms.
To identify CaM-containing complexes within the axoneme, we used anti-CaM antibodies in immunoprecipitation experiments (). Using extracts isolated from mutant axonemes lacking the radial spokes (), our anti-CaM antibodies precipitated a total of eight polypeptides in addition to CaM (; see in ). Five polypeptides form a complex that includes PF6 and localizes to the C1a projection of the central apparatus (). However, the three additional polypeptides that are precipitated (designated CaM-IP2, -IP3, and -IP4; ) are present in all central apparatus–defective mutants, including . Therefore, these three polypeptides are not components of the radial spokes or central apparatus. These three polypeptides are also precipitated from extracts isolated from mutant axonemes lacking the outer dynein arms and inner arm I1 (; ) as well as the inner arm–defective strains and (not depicted), indicating that they do not localize to these dynein subforms. All three polypeptides are also precipitated from the move backward–only strains and (unpublished data), indicating that the assembly of these polypeptides is unaffected in these mutant strains.
To determine the identities of CaM-IP2, -IP3, and -IP4, these polypeptides were excised from gels and analyzed by mass spectrometry (see Materials and methods; data summarized in ). The identity of CaM-IP2 was determined from four peptide sequences. Based on searches of the genome database (version 3.0; ), the gene is located on the same contig as Zsp1; therefore, the gene is most likely on linkage group VII. In searches of the proteome database, CaM-IP2 is flagellar-associated protein 91 (FAP91; ). BLAST searches reveal that CaM-IP2 (FAP91) is most similar to the human homologue of AAT-1 (C3orf15; GenBank/EMBL/DDBJ accession no. ; E value of E = 7e-49). AAT-1 was originally identified as a testis-specific protein in mice that forms a quaternary complex with AMY-1 (a c-myc–binding protein), an A-kinase anchoring protein (AKAP), and two regulatory subunits of PKA (cAMP- dependent protein kinase; ). More recently, a total of seven alternatively spliced isoforms of AAT-1 have been identified, and some of these isoforms are expressed in a variety of human tissues (). The largest isoform of AAT-1, AAT-1L, is an ∼90-kD protein. The testis-specific isoform AAT-1α is reported to contain only the C-terminal 98 amino acids of AAT-1L.
CaM-IP2 is 32% identical and 49% similar to AAT-1 over a stretch of 423 amino acids at the N terminus (Fig. S1, available at ). Sequence similarity with AAT-1 may extend beyond the N terminus of the protein, but the C-terminal half of the CAM-IP2 sequence is not represented in the genome. We have tried to complete the CaM-IP2 coding sequence using a variety of approaches, including 5′ and 3′ rapid amplification of cDNA ends, obtaining small genomic fragments of CaM-IP2 for sequencing, and optimizing our sequencing reactions for GC-rich sequences. We have confirmed the predicted amino acid sequence for the N-terminal one third of the protein but were unsuccessful in obtaining the complete coding sequence. As judged by SDS-PAGE, the apparent molecular mass of CaM-IP2 is 183 kD, which is almost double that of AAT-1L. However, Northern blots of RNA isolated from wild-type cells indicate that the CaM-IP2 transcript is ∼4.4 kb (Fig. S2). Based on our experience comparing genomic and coding sequences, the true molecular weight of this protein is most likely no greater than 120 kD.
The identity of CaM-IP3 was determined from the amino acid sequence of five peptides. Based on database searches, CaM-IP3 is located on linkage group III and corresponds to FAP61. Using corresponding cDNA sequences in the expressed sequence tag (EST) database and RT-PCR, we determined that the gene encodes a 1,115–amino acid protein that contains a predicted NADH dehydrogenase domain. This domain is found in both class I and II oxidoreductases as well as NADH oxidases and peroxidases. The domain includes a small NADH-binding domain within a flavin adenine dinucleotide– binding domain and, thus, is thought to be involved in energy conversion. Searches of sequence databases reveal that the predicted human protein C20orf26 is most similar to CaM-IP3 (E value of E = 6.8e-56), sharing 25% identity and 43% amino acid similarity along the entire length of the protein. The predicted molecular mass of CaM-IP3 is 118 kD, although the protein has an apparent molecular mass of ∼140 kD as judged by SDS-PAGE. The acidic isoelectric point of CaM-IP3 (4.76) may partially account for this discrepancy.
The identity of CaM-IP4 was determined from the amino acid sequence of five peptides. Based on database searches, CaM-IP4 is also located on linkage group III and corresponds to FAP251. CaM-IP4 contains seven WD-40 repeats that comprise much of the N terminus of the protein. WD-40 repeats form propeller-like structures that serve as an interface for protein–protein interactions (). In searches of sequence databases, a protein predicted from the database (GenBank/EMBL/DDBJ accession no. ; E value of E = 9e-53) is most similar to CaM-IP4 with 29% amino acid identity and 43% similarity along the entire length of this protein. CaM-IP4 is also 24% similar and 39% identical to a protein predicted from the human genome sequence (GenBank/EMBL/DDBJ accession no. ; E value of E = 6e-37). All of these polypeptides show similarity to microtubule-associated protein–like 5 from echinoderms. In addition to the WD-40 repeats, each of these proteins has a putative EF-hand domain at the C terminus. Although a motifs search does not predict an EF hand at the C terminus of CaM-IP4, the InterProScan algorithm provided by the European Molecular Biology Laboratory predicts an EF hand–like domain.
Based on CaM immunoprecipitation, it is possible that CaM-IP2, -IP3, and -IP4 bind CaM individually, in subsets, or together to form a single complex. To differentiate among these possibilities, we developed polyclonal antibodies in rabbits against either bacterially expressed protein fragments or synthetic peptides for each of CaM-IP2, -IP3, and -IP4 (see Materials and methods). CaM-IP4 proved to be nonantigenic. However, we were successful in obtaining antibodies that specifically recognize either CaM-IP2 or CaM-IP3 on Western blots ().
We then used these antibodies in immunoprecipitation experiments. Antibodies generated against CaM-IP3 failed to precipitate CaM-IP3. On the other hand, the CaM-IP2 antibody specifically precipitated CaM-IP2 as well as -IP3 and -IP4 from extracts isolated from radial spokeless axonemes (, middle lane). For extracts isolated from wild-type and axonemes, additional polypeptides are precipitated by the CaM-IP2 antibodies that are not precipitated from extracts of the spokeless mutant . Based on the molecular weights of these polypeptides () as well as their absence from precipitates of axonemal extracts, we suspected that these polypeptides were radial spoke components. Corresponding Western blots using antibodies generated against radial spoke protein 2 (RSP2) and RSP3 confirm that RSPs are precipitated by the CaM-IP2 antibodies. The simplest interpretation of these results is that CaM and CaM-IP2, -IP3, and -IP4 form a single complex that is associated with the radial spokes.
To provide further evidence that the CaM-IP2, -IP3, and -IP4 complex is associated with the radial spokes, we investigated whether this complex cosediments with the radial spokes using sucrose density gradient centrifugation (). For axonemal extracts isolated from wild-type flagella, the polypeptides comprising the radial spokes cosediment at ∼20S (). Western blots of corresponding gradient fractions reveal that CaM-IP2 and -IP3 cosediment with the radial spokes. For axonemal extracts isolated from flagella that lack the spoke heads (), the polypeptides comprising the spoke stalks cosediment at ∼15S (). Western blots of corresponding gradient fractions from extracts reveal that CaM-IP2 and -IP3 also cosediment with the radial spoke stalks (). For axonemal extracts isolated from flagella that completely lack the radial spokes (), CaM-IP2 and -IP3 cosediment as a smaller complex at ∼11S (). Although we do not have antibodies that recognize CaM-IP4, immunoprecipitation experiments using sucrose gradient fractions confirm that CaM-IP4 also cosediments with CaM-IP2, -IP3, and -IP4 (Fig. S3, available at ). These results support the hypothesis that CaM, CaM-IP2, -IP3, and -IP4 form a single complex that is associated with the spoke stalk.
As noted in our CaM-IP2 sequence analysis, CaM-IP2 is most similar to AAT-1, which is thought to form a quaternary complex with an AKAP-binding protein. Based on previous reports that RSP3 is an AKAP (), that RSP3 is located at the base of the radial spoke stalk (), and our results that CaM-IP2 cosediments with the spoke stalk (this study), we hypothesized that CaM-IP2 binds to RSP3. To test this hypothesis, we used a gel overlay assay. Bacterially expressed proteins (CaM-IP2, CaM-IP3, and RSP3) were resolved by SDS-PAGE and transferred to nitrocellulose membrane. The membrane was then incubated with bacterially expressed RSP3, and the overlain protein was detected by immunoblotting. As shown in , RSP3 binds only to CaM-IP2. These combined results indicate that CaM-IP2, -IP3, and -IP4 are associated with the radial spoke stalk via interactions between CaM-IP2 and RSP3.
The association of CaM with specific proteins is regulated by the calcium-binding state of CaM. All immunoprecipitation experiments were performed using low calcium conditions. Therefore, the eight polypeptides we precipitated do not require calcium to form protein complexes with CaM. To investigate whether CaM binding of these polypeptides is calcium sensitive, we first immunoprecipitated CaM from axonemal extracts under low calcium conditions; the protein A beads were then washed using high calcium buffer (see Materials and methods). CaM remained associated with the beads, whereas CaM-IP2, -IP3, and -IP4 were selectively extracted in the presence of high calcium (). These results demonstrate that the binding of CaM-IP2, -IP3, and -IP4 to CaM is calcium sensitive. We also performed immunoprecipitation experiments using the CaM-IP2 antibody under high calcium conditions. All three members of the complex are precipitated by the CaM-IP2 antibody in extracts from spokeless axonemes, indicating that the complex does not dissociate under high calcium conditions (). In addition, the CaM-IP2 antibodies precipitate the spokes from axonemal extracts isolated from either wild-type or central pairless axonemes, indicating that the complex does not dissociate from the spokes under high calcium conditions ().
Our data are consistent with the localization of a CaM-containing complex at the base of the spokes. Previous studies have indicated that the components of the dynein regulatory complex (DRC) localize to the doublet microtubules near the spokes (; ; ). Therefore, we hypothesized that the CaM-IP2, -IP3, and -IP4 complex may be disrupted in mutants lacking subsets of DRC components. To investigate this possibility, we precipitated the complex from axonemal extracts isolated from DRC mutants ( and ) using either our anti-CaM or anti–CaM-IP2 antibodies. The anti-CaM antibodies precipitate substantially less CaM-IP2, -IP3, and -IP4 from and axonemal extracts compared with that precipitated from wild-type extracts even though axonemes isolated from these mutants appear to have wild-type levels of CaM as judged by Western blotting (). These results suggest that the assembly of CaM-IP2, -IP3, and -IP4 is defective in DRC mutants. However, when we precipitate the complex using our anti–CaM-IP2 antibodies, the precipitates from and are indistinguishable from that of wild type (). These results indicate that CaM-IP2, -IP3, and -IP4 assemble normally in DRC mutants but that their association with CaM is disrupted.
Localization of the CaM-IP2, -IP3, and -IP4 complex at the base of the radial spokes, potentially near DRC components, suggested a possible role in modulating dynein-driven microtubule sliding. Based on this localization, we hypothesized that regulation of dynein activity by this complex would occur downstream of regulatory signals derived from the radial spokes or central apparatus. To test this hypothesis, we conducted the microtubule sliding assay in the presence of the anti–CaM-IP2 antibody. We initially tested whether the antibodies possessed any function-blocking activity using wild-type axonemes and a microtubule sliding assay (see Materials and methods). The anti–CaM-IP2 antibodies had no effect on microtubule sliding velocity ().
We then examined dynein activity in mutant axonemes. The first mutant tested was the central pairless strain ; axonemes from this mutant have reduced dynein activity compared with wild-type (). To our surprise, the anti–CaM-IP2 antibody significantly increases microtubule sliding velocity (P < 0.001 by test; ). In addition, the increase in velocity in central pairless axonemes is dependent on the concentration of the antibody, with maximal effect at 0.2 μM CaM-IP2 antibody (). Importantly, the increase in dynein activity is not observed upon the addition of anti–CaM-IP3 or anti-C1a32 antibodies (), further indicating that the effect is specific for binding of the anti–CaM-IP2 antibody. Based on these results and the hypothesis that the complex affects dynein activity downstream of regulatory cues from the radial spokes, we predicted that addition of our antibodies to radial spokeless axonemes would also increase dynein activity. Indeed, addition of the CaM-IP2 antibodies to radial spokeless axonemes () significantly increases dynein activity (P < 0.001 by test; ). The simplest interpretation of these results is that the CaM-IP2, -IP3, and -IP4 complex is involved in modulating the activity of the dynein arms.
To determine whether specific subforms of axonemal dynein are the targets for modulation by this complex, we repeated the sliding microtubule experiments using mutants that lack specific dynein subforms. The CaM-IP2 antibodies have no effect on the sliding velocities of a mutant lacking the outer dynein arms (; ). Double mutants lacking both the central apparatus and outer dynein arms () or radial spokes and outer dynein arms () have shorter length flagella than wild type, and many of the isolated axonemes fail to undergo microtubule sliding in this assay. However, for those axonemes in which microtubule sliding occurred, velocities doubled upon addition of the CaM-IP2 antibodies (). These velocities are significantly lower than those of (P < 0.001 by test). The CaM-IP2 antibodies had no effect on the sliding velocity of axonemes. These results indicate that the outer dynein arms may be one of but not the exclusive targets of regulation by the CaM complex. Therefore, we also compared dynein activity in double mutants lacking various inner dynein arm subforms.
Addition of the CaM-IP2 antibody restored dynein activity in mutant axonemes lacking the central apparatus and the I2 inner arms (; ; ), which are also known as fast protein liquid chromatography fractions a, c, and d (; ). The CaM-IP2 antibodies increased sliding velocities of axonemes to levels not significantly different (P > 0.22 by test) from those of axonemes incubated with CaM-IP2 antibody. These results suggest that the I2 inner arms are not the targets of regulation by this complex.
The and mutants lack the inner arm I1 or fraction f heavy chains (; ; ). Addition of the CaM-IP2 antibodies to axonemes isolated from the mutant that lack the spokes and the I1 inner dynein arm failed to increase dynein activity (). These results indicate that I1 is a target for regulation by this complex. We previously reported that double mutants lacking I1 and the central apparatus () have microtubule sliding velocities that are significantly greater than those of axonemes (P < 0.001 by test; ). The loss of I1 from central pairless mutants partially relieves the inhibition of microtubule sliding in central apparatus–defective mutants. However, sliding velocities of are also significantly lower than those of axonemes exposed to the CaM-IP2 antibodies (P < 0.001 by test; ). Addition of the CaM-IP2 antibodies to mutant axonemes failed to increase dynein activity to wild-type levels. These combined results support the hypothesis that the CaM-IP2, -IP3, and -IP4 complex plays a role in modulating the activity of inner dynein arm I1.
Previous studies have indicated that reduced sliding velocities in central pairless and radial spokeless mutants is correlated with the phosphorylation state of the IC138 intermediate chain of the I1 inner dynein arm (, ; ; ; ). Reducing the phosphorylation of IC138 using kinase inhibitors results in increased sliding velocities in these mutants (; ; ). Therefore, we hypothesized that addition of the CaM-IP2 antibodies to mutant axonemes may result in reduced IC138 phosphorylation and, therefore, increased dynein activity. Using antibodies generated against IC138, we performed Western blots of isolated axonemes in the presence or absence of CaM-IP2 antibodies (). Phosphorylated forms of IC138 have been detected on Western blots of radial spoke– and central apparatus–defective axonemes as slower migrating species that are not present when samples are treated with phosphatase (). We do not detect any decrease in the overall phosphorylation state of IC138 in either radial spokeless () or central pairless () axonemes upon the addition of anti–CaM-IP2 antibodies. These results suggest that the CaM-IP2 antibodies do not increase dynein activity by affecting the phosphorylation state of I1 IC138. Alternatively, the phosphorylation state of specific residues may be affected that cannot be detected by Western blotting.
#text
Strain A54-e18 (, , , +), the wild-type strain for motility and axoneme structure, was obtained from P. Lefebvre (University of Minnesota, St Paul, MN). The central pair–defective strain , the radial spoke–defective strains and , and the dynein-defective strains , , , , and were obtained from the Genetics Center (Duke University). The , , and strains were obtained from W. Sale (Emory University, Atlanta, GA). Generation of the double mutants and have been previously described (). The double mutant was selected from nonparental ditype tetrads. All cells were grown in constant light in Tris acetate phosphate media ().
Flagella were severed from cell bodies by the dibucaine method () and isolated by differential centrifugation in NaLow (10 mM Hepes, pH 7.4, 5 mM MgSO, 1 mM DTT, 0.5 mM EDTA, and 30 mM NaCl). Axonemes were isolated by adding NP-40 (Calbiochem) to flagella for a final concentration of 0.5% (wt/vol) to remove flagellar membranes. Axonemes were initially extracted in NaHigh (10 mM Hepes, pH 7.4, 5 mM MgSO, 1 mM DTT, 0.5 mM EDTA, and 0.6 M NaCl) at a concentration of 6 mg/ml on ice for 20 min. The axonemes were pelleted, resuspended in NaHigh, and immediately pelleted again. The supernatant was discarded, and the pellet was extracted with KI (10 mM Hepes, pH 7.4, 5 mM MgSO, 1 mM DTT, 0.5 mM EDTA, 30 mM NaCl, and 0.5 M KI) at a concentration of 12 mg/ml for 30 min on ice. This extract is referred to as the KI extract or high salt extract. KI extracts were dialyzed against NaLow buffer and clarified by centrifugation at 12,000 relative centrifugal force for 10 min. For some experiments, the clarified extracts were loaded onto 5–20% sucrose gradients and subjected to ultracentrifugation at 35,000 rpm for 16 h in an SW41Ti rotor (Beckman Coulter). 0.5-ml fractions were collected from the bottom of the tube and prepared for SDS-PAGE.
Immunoprecipitation was performed according to with the following modifications. 150 μl KI axonemal extract was incubated with ∼40 μg of affinity-purified anti-CaM or anti-IP2 antibody. After four washes with TBST (150 mM NaCl, 50 mM Tris-HCl, and 0.5 mM EDTA, pH 7.5), beads were resuspended in 90 μl TBST and 50 μl of 5× SDS-PAGE sample buffer. For experiments performed in high calcium, a final concentration of 1 mM CaCl was included in the TBST buffer, and all washes were performed with TBST + 1 mM CaCl. For immunoprecipitation performed in low calcium conditions followed by a final high calcium wash, the beads were washed as before with TBST, and then the beads were suspended in 100 μl TBST with 2 mM CaCl. The beads were rotated for 5 min at RT and briefly spun, and the supernatant was transferred to a new tube. 100 μl TBST was added to the remaining beads, and all samples were prepared for SDS-PAGE gels. For immunoprecipitation using sucrose gradient fractions, 50–80 μg of antibody was used with 1,200 μl of pooled sucrose gradient fractions. The sample was mixed for 5–8 h and washed with four 1-ml vol of TBS-T.
Gel bands were excised from Coomassie-stained gels. These bands were analyzed by matrix-assisted laser desorption/ionization–time of flight mass spectrometry with postsource decay conducted at the University of Massachusetts Medical School or liquid chromatography/liquid chromatography–electrospray ionization mass spectrometry performed at the Harvard Microchemistry and Proteomics Facility. Comparisons of peptide masses with translated genomic or EST sequences were made using the BLAST algorithm at the National Center for Biotechnology Information and the genome database (version 3.0; ). Searches of the flagellar proteome () were performed using .
The coding sequence for each polypeptide was confirmed by RT-PCR. RNA was isolated from wild-type cells before and after pH shock–induced deflagellation according to . PolyA enrichment was performed using the Oligotex mRNA kit (QIAGEN) according to the manufacturer's instructions. Reverse transcription was performed using either Superscript II or III (Invitrogen) according to the manufacturer's instructions. Subsequent PCRs were performed using primers based on the known or predicted coding regions as determined by EST, EST contig, or mass spectrometry data.
The CaM peptide and corresponding antibodies as well as the anti-C1a32 antibodies were generated as described previously (). For CaM-IP2, the first 680 bp of the coding sequence was cloned into the pET30 vector and transformed into BL21 (DE3) pLysS cells. Protein expression was induced by the addition of IPTG, and the expressed protein was purified from bacterial cell lysates using a Ni-resin column according to the manufacturer's protocol (Novagen). For CaM-IP3 antibodies, a C-terminal peptide (CZTHAQDAVLEFARAHAAE) was synthesized and conjugated to keyhole limpet hemocyanin (Aves Labs, Inc.). For both CaM-IP2 and -IP3, polyclonal antibodies were generated in rabbits against the purified protein or peptide at Spring Valley Laboratories. Antibodies were affinity purified on Sulfolink columns (Pierce Chemical Co.) according to the manufacturer's protocol. Columns contained purified expressed protein (CaM-IP2) or purified peptide (CaM-IP3).
For immunoblots, equivalent loads of flagella (8 mg/ml), axonemes, extracts, and/or extracted axonemes were subjected to SDS-PAGE using 7% polyacrylamide gels. Gels were transferred to polyvinylidene difluoride (Immobilon P; Millipore). CaM Western blots were performed as previously described (). For CaM-IP2 and -IP3 Western analyses, 7% polyacrylamide gels were subject to SDS-PAGE and transferred for 1 h to polyvinylidene difluoride (Immobilon P; Millipore). Membranes were blocked for 1 h in 2% BSA (regular fraction V; A-7906; Sigma- Aldrich) in TBST (0.1% Tween and TBS, pH 7.5). For primary antibody incubations, affinity-purified anti–CaM-IP2 or anti–CaM-IP3 antibodies were diluted 1:100 or 1:5,000, respectively, and anti-RSP2 or anti-RSP3 antibodies (provided by D. Diener, Yale University, New Haven, CT) were diluted 1:5,000 in TBST. Primary antibody incubations were conducted for 2 h at RT or overnight at 4°C. Membranes were washed three times for 5 min with TBST, incubated with anti–rabbit HRP secondary antibodies (GE Healthcare), and diluted 1:30,000 in TBST. After four 5-min washes with TBST, the ECL Plus Western Blotting kit (GE Healthcare) was used for chemiluminescent detection. Gels were silver stained according to the methods described in .
For anti-IC138 Western blots, axonemes isolated from wild type, , and were resuspended at 1 mg/ml in pCa8 or pCa4 buffer containing 1 mM ATP. For some samples, 50 μg of axonemes were treated with 4 μg of affinity pure anti–CaM-IP2 or 4 μg of affinity pure anti–C1a-32 antibodies (). For samples that were treated with calf intestine alkaline phosphatase (Roche), the axonemes were incubated with 0.75 U of calf intestine alkaline phosphatase for 30 min at RT. 10 μg axonemal proteins were resolved on a 5% polyacrylamide gel and transferred to nitrocellulose. Blots were incubated with anti-IC138 (1:10,000 in TBST) overnight at 4°C. After washing, blots were incubated for 45 min with anti–rabbit HRP in TBST, washed, and incubated with ECL Plus for detection. The IC138 antibodies were provided by W. Sale.
RSP3-GST (provided by W. Sale) was expressed and purified on a GST-resin column (Novagen) according to the manufacturer's protocol. The N terminus of CaM-IP2 was expressed and purified as described in Peptide and antibody production. For CaM-IP3, DNA sequence encoding the C-terminal 220 amino acids was cloned into pET30 and transformed into BL21 (DE3) pLysS cells. For CaM-IP4, 1.8 kb of DNA sequence encoding amino acids 1–580 was cloned into the pET30 vector. Expression was induced with IPTG, and proteins were purified from bacterial cell lysates using a Ni-resin column (Novagen). For SDS-PAGE, 15 μg CaM-IP2 (37 kD), 15 μg CaM-IP3 (32 kD), 15 μg CaM-IP4 (62 kD), and 25 μg RSP3-GST (110 kD) were resolved on a 12% polyacrylamide gel and transferred for 40 min to nitrocellulose membrane. Membranes were blocked overnight at 4°C in 5% milk/TBST (0.1% Tween) and were incubated with 40 μg/ml RSP3-GST in 1% BSA/TBST for 2 h at RT. After three 5-min TBST washes, membranes were incubated with anti-RSP3 (1:5,000) in TBST for 1 h and with secondary antibody (anti–rabbit HRP; GE Healthcare) diluted 1:30,000 in TBST for 30 min. The ECL Plus Western Blotting kit (GE Healthcare) was used for detection.
Flagella were severed from cell bodies by the dibucaine method () and isolated by differential centrifugation in buffer A (10 mM Hepes, pH 7.4, 5 mM MgSO, 1 mM DTT, 0.5 mM EDTA, and 50 mM potassium acetate). Axonemes were isolated by adding NP-40 (Calbiochem) to flagella for a final concentration of 0.5% (wt/vol) to remove flagellar membranes. Measurement of sliding velocity between doublet microtubules was based on the methods of . Microtubule sliding was initiated with buffer A containing 1 mM ATP and 2 μg/ml type VIII protease (Sigma-Aldrich) and was recorded as described previously (). All data are presented as mean ± SEM. The test was used to determine the significance of differences between means. For some experiments, axonemes were incubated with affinity-purified anti–CaM-IP2 antibodies for 15 min at room temperature (22°C) before the induction of sliding. Antibody concentrations are noted on corresponding figures and are estimates based on the conversion factor of IgG = 150,000 g/mol.
Fig. S1 shows an amino acid sequence comparison of CaM-IP2 and human AAT-1. Fig. S2 shows Northern blots comparing transcript levels of CaM-IP2 and RSP3 before and after deflagellation. Fig. S3 shows Western blots of sucrose gradient fractions and immunoprecipitation experiments of corresponding pooled fractions. Online supplemental material is available at . |
Endocrine tissues and neurons release hormones and neurotransmitters through a process of regulated vesicular secretion. Abnormalities in regulated secretion cause a variety of diseases, including hypertension, diabetes, and neurological disorders. Although not widely appreciated, the heart functions as an endocrine organ by regulating blood volume and natriuresis through the secretion of natriuretic peptides (NPs) (). Atrial NP (ANP) is secreted exclusively by atrial myocytes, whereas brain NP (BNP) is secreted predominantly by ventricular myocytes, albeit at a low rate. In response to acute and chronic stress, ventricular myocytes up-regulate the synthesis and secretion of ANP and BNP. As such, BNP is a well-accepted clinical marker of ventricular wall stress and directly correlates with the severity of heart failure ().
ANP and BNP bind to the same guanylyl cyclase-linked receptor, NP receptor-A, causing vasodilation, natriuresis, and diuresis, as well as inhibition of endothelin-1 release and the renin–angiotensin system (). These effects lead to a decrease in blood volume and total peripheral resistance, thereby reducing pre- and afterload on the heart and blood pressure (). BNP also prevents pathological cardiac hypertrophy and fibrosis () and has emerged as a promising therapy for heart failure ().
Considering their important roles in cardiovascular physiology, the stimuli that control NP synthesis have been under intense investigation over the past 30 years and extensive information has been amassed (). In contrast, the mechanisms involved in NP packaging into vesicles, trafficking of vesicles to the cardiomyocyte cell membrane, and exocytosis remain largely unexplored. In this regard, several proteins involved in vesicle formation, transport, docking, and fusion are expressed in the heart and are contained in ANP-containing LDCVs (; ; ; ). However, little is known about the possible involvement of these proteins in ANP secretion.
Here, we describe a novel Ras-related protein, called RRP17, which is expressed in the heart and neuroendocrine tissues. RRP17 interacts with Ca-activated protein for secretion-1 (CAPS1), a mediator of LDCV secretion (), and influences the storage and secretion of ANP in cardiomyocytes in vivo and in vitro. Consistent with a possible role in regulating cardiac endocrine functions, mice lacking RRP17 display abnormalities in cardiac ANP secretion and blood pressure regulation in unconscious animals. The interaction of RRP17 with CAPS1 provides insights into the molecular basis of ANP secretion and endocrine functions of the heart.
To identify novel signal transduction molecules involved in the regulation of cardiac function, we used microarray analysis to compare mRNA pools derived from NIH-3T3 fibroblasts and NkL-Tag cells, a mouse cardiac cell line derived from the ventricle of a transgenic mouse expressing large T antigen under control of a cardiac-specific promoter (). We initially selected a pool of ESTs expressed specifically in the NkL-Tag cells and subjected these ESTs to a secondary screen through NCBI, EMBL, and Celera databases to identify uncharacterized transcripts. One EST, AA624579, belonged to the mouse UniGene cluster Mm.66275, which was classified by NCBI as encoding a novel putative Ras-like protein. Computer analysis of the mouse and human genomes revealed that the Mm.66275 cluster contains an open reading frame (ORF) encoding a putative 203-amino acid protein. The Mm.66275 transcript was confirmed by sequencing of RT-PCR products generated using heart and brain cDNAs as template. An orthologue of this protein exists in all sequenced representatives including , , and (), but not in or . There is high homology between orthologues of this protein, such that the mouse orthologue is 99% identical to the human protein and 67% identical to that of .
The protein encoded by the Mm.66275 UniGene cluster has a predicted secondary structure characteristic of the small G-protein superfamily and includes a guanine nucleotide-binding (GNB) domain and a prenylation signal at the C terminus, known as a CAAX-box (). This putative protein is most similar to a recently described member of the small G protein superfamily Ras-related protein on human chromosome 22, RRP22 () (). Based on the nomenclature of RRP22, we named this newly identified protein RRP17 (Ras-related protein located on human chromosome 17).
Although RRP17 clearly belongs to the Ras family of small G proteins, it possesses two important features that distinguish it from other members of the Ras family. First, similar to RheB and in contrast to H-Ras, RRP17 has an arginine instead of the usual glycine at the third amino acid position of the P-loop (GAGVGK(S/T)) (). The substitution of Gly for a charged or bulky amino acid, such as Arg, results in a decrease of GTPase activity of Ras proteins due to steric interference with the γ-phosphate of GTP and ensures Ras oncogenicity (). A second difference is the 13-amino acid insertion (from D59 to E74) within the G3 loop of the Switch II region (DXXG). The glycine residue within this loop forms a hydrogen bond with the γ-phosphate of GTP, and is universally conserved through the small GTPase family (; ). In RRP17, this glycine is replaced with a hydrophobic nonflexible proline residue. RRP22 is the only other small G protein known to possess a similar size insertion that disrupts the G3 loop (see supplemental material, available at ). Although the amino acid composition of that insertion is different from the insertion of RRP17, there is a conserved proline residue within this region of both proteins, corresponding to amino acid 67 and 69 of RRP17 and RRP22, respectively. Due to this 12-amino acid insertion, the Switch II region of RRP17 is longer than the corresponding region of any known GTPase and potentially has a distinct and more rigid 3D structure compared with RheB, the closest homologue with a known crystal structure (). Indeed, 3D prediction analysis shows an additional α-helix within the Switch II region of RRP17 (). A dendrogram reveals that RRP17 and RRP22 form a distinct family within the Ras-superfamily ().
Northern blot analysis revealed a single 3.6-kb RRP17 transcript in mouse and human heart and brain, as well as in human skeletal muscle and pancreas (). RRP17 mRNA was detected at embryonic day (E) 9.5 in the developing heart, skeletal myotomes, and dorsal portion of the neural tube by in situ hybridization (). At E13.5 and E15.5, RRP17 mRNA was detected in the heart and neural tissue including the central and peripheral nervous system. Expression of RRP17 mRNA in the embryonic brain is not uniform but restricted to more mature neurons and absent from the progenitor neuronal cells of the ventricular zone (). RRP17 mRNA is also present in mature peripheral neurons, such as the neurons of the dorsal root ganglia () and the Auerbach's and Meissner's plexi of the intestinal tract (unpublished data). Although RRP17 expression was detected in human skeletal muscle, we did not detect it in mouse skeletal muscle, even when we examined expression in individually isolated mouse muscles, such as extensor digitorum longus, white vastus longus and soleus (unpublished data). The basis for this species-specific difference of RRP17 expression is unclear.
In an initial effort to elucidate the functions of RRP17, we sought to identify proteins that interact with RRP17 in a yeast two-hybrid screen using brain and heart cDNA libraries and CAAX-box-deleted RRP17 as bait. 13 of the 186 positive clones encoded the C-terminal region of CAPS1, a regulator of LDCV secretion (). CAPS1 contains several functional domains, including a C2 domain that binds calcium, a pleckstrin homology (PH) domain that binds phosphatidylinositol-bisphosphate, a Munc homology domain (MHD) found in Munc family proteins, a family of proteins which are involved in secretion, and a dense core vesicle-binding domain (DCVD) at the C terminus ().
The longest cDNA identified by the yeast two-hybrid screen started immediately after the PH domain at amino acid 648 and the shortest cDNA started at amino acid 700 (). These results were confirmed by pull-down assays, which showed the association of CAPS1 with GST-RRP17 (). The specificity of interaction was demonstrated by the absence of binding of CAPS1 and GST-RhoA, a close relative of RRP17 ( and ), suggesting that CAPS1 does not bind every member of the Ras family. We observed that individually expressed CAPS1(670–1372) or RRP17 localized to the supernatant or pellet fraction, respectively. Therefore, we designed a cosedimentation assay which showed that coexpression of RRP17 and CAPS(670–1372) targets CAPS1 to the RRP17-containing pellet fraction (). Of note, we consistently observed that the presence of RRP17 increased the total amount of CAPS1 protein in the cell, although we do not know the mechanism of this effect.
Using immunocytochemistry, the interaction of CAPS1 and RRP17 was further confirmed in transfected HeLa cells. Exogenous RRP17 was distributed in a distinct perinuclear pattern whereas CAPS1 was distributed throughout the cytoplasm (). Upon coexpression of CAPS1 and RRP17, CAPS1 protein was redistributed to colocalize with RRP17, consistent with the interaction of the two proteins (). Residues 670–1372 of CAPS colocalized with RRP17, whereas CAPS1 deletion mutants lacking this region did not colocalize with RRP17 (unpublished data).
The interaction of RRP17 with CAPS1 suggested the involvement of RRP17 in regulated secretion. The published expression pattern of mouse CAPS1 is similar to that of RRP17, except that CAPS1 was not previously detected in heart (). Given that the normal adult heart secretes NPs only from the atria, we isolated RNA from the cardiac atria and ventricles independently and analyzed CAPS1 expression. CAPS1 mRNA was detected in atrial, but not ventricular RNA ().
We speculated that if CAPS1 participated in the secretion of cardiac NPs, then it should be expressed in ventricular cardiomyocytes during conditions that enhance NP secretion. To test this hypothesis, we analyzed ventricular myocytes from two different mouse models of cardiac hypertrophy; using the α-MHC-CnA transgenic mouse line that overexpresses activated calcineurin in the heart () and surgical partial occlusion of the thoracic aorta (TAB) (). Indeed, the level of CAPS1 mRNA was significantly increased in the ventricles of mice with calcineurin-induced cardiac hypertrophy (). Similarly, cardiac hypertrophy induced by increased afterload due to thoracic aorta banding (TAB) caused substantial up-regulation of CAPS1 mRNA in parallel with ANP mRNA (). These findings show that CAPS1 and RRP17 are coexpressed in the atria and that expression of CAPS1 mRNA in the ventricles is augmented by hypertrophic stimuli, suggesting that RRP17 and CAPS1 may participate in regulated secretion of NPs in hypertrophic ventricular myocytes. Interestingly, the expression level of RRP17 mRNA was unaffected by these hypertrophic stimulus (, and unpublished data).
To further explore the potential involvement of RRP17 in NP secretion, we infected primary rat neonatal cardiomyocytes with recombinant adenovirus expressing either Flag-RRP17 protein or β-galactosidase as a control. 16 h after infection, medium supplemented with serum was replaced with serum-free medium and the amount of secreted ANP was measured 3, 12, and 35 h later (). Cardiomyocytes infected with adenovirus expressing RRP17 secreted 2.5- and 3.5-fold more ANP compared with uninfected or lacZ expressing cardiomyocytes, respectively (). The enhancement of ANP secretion by RRP17 was not associated with an elevation of ANP mRNA expression (). These results suggest that RRP17 facilitates release of ANP in cardiomyocytes.
To determine the function of RRP17 in vivo, we generated RRP17-deficient mice by targeted disruption of the gene. The mouse gene spans ≈12 kb on chromosome 11 and consists of three protein-coding exons and two alternative exons in the 5′ untranslated region (UTR) (). The ORF starts within exon 2 and is not affected by alternative splicing of the 5′ UTR. To generate mice lacking RRP17, exon 3 and portions of exons 2 and 4 were replaced with a promoterless nuclear LacZ gene and a neomycin-resistance gene such that the targeted allele lacked almost the entire RRP17 ORF (). The targeting vector was electroporated into 129SV/Ev ES cells and targeted clones were identified by Southern blot analysis of genomic DNA (). Embryonic stem cells heterozygous for the RRP17 deletion were injected into C57BL/6 blastocysts, and chimeric mice transmitted the mutant allele through the germ line.
intercrosses. There was no obvious phenotype observed in null animals when left undisturbed. RT-PCR analysis of heart and brain RNA showed diminished expression of the RRP17 transcript in heterozygous mice and its complete absence in homozygous mutant mice (, and unpublished data).
The expression pattern of β-galactosidase from the mutant allele recapitulated the pattern of mRNA expression in the central nervous system and heart (). In adult brain, the expression of β-galactosidase from the locus was restricted to neurons of the cortex, hippocampus, and cerebellum, but was sparse in the brainstem and almost absent in the hypothalamic region (). The expression of nuclear lacZ from the gene was confined to neurons and cardiomyocytes and not detected in interstitial tissue or vascular cells ().
mice showed that LDCVs were smaller in the mutant than in wild-type cardiomyocytes (), although the difference in the mean of the vesicle size did not reach statistical significance (unpublished data). Passive stretch, the major stimulus for ANP secretion in the heart (), can be simulated in cell culture by challenging isolated cardiomyocytes with hypotonic buffers. A decrease in the extracellular osmolarity from 300 to 200 mOsm/kg HO causes an increase in cardiomyocyte volume by 50% and induces secretion of prestored ANP granules (; ; ). Normally, passive mechanical stretch causes prompt translocation of LDCVs from the perinuclear area of atrial cardiomyocytes toward the cellular membrane ().
Challenging atrial cardiomyocytes isolated from wild-type mice with hypotonic buffer resulted in secretion of 50 to 60% of stored ANP during the first 10 min, with no appreciable secretion of ANP for the next 50 min (). RRP17-deficient myocytes also responded within the first 10 min; however, they secreted only 10 to 20% of stored ANP ().
mice displayed a significantly higher systolic and diastolic blood pressure, reflected in the mean arterial pressure, and a higher heart rate ().
mice compared with wild-type mice ().
LDCVs mediate secretion of many biologically active substances from cardiomyocytes, neurons, and endocrine cells. Small GTPases are known to be involved in regulation of multiple steps of LDCV secretion with functions ranging from vesicle formation to vesicle transport, tethering to the cellular membrane and fusion steps (; ). In this study, we describe a previously unknown Ras-like protein, RRP17, which regulates secretion of ANP from cardiomyocytes and interacts with CAPS1, a protein involved in biogenesis and storage of LDCV.
Classification of members of the Ras superfamily is based on similarity of amino acid sequence (). At a level of identity of 30%, five major Ras families have been identified, Ras, Rho, Arf, Ran, and Rab. RRP17 and RRP22 proteins differ from other members of the Ras-superfamily within the Switch I and II regions, which bind and hydrolyze GTP and interact with regulators and effectors of G protein signaling (). The glycine residue of the Switch II region, which coordinates the positioning of Mg ion and γ-phosphate of GTP, is substituted by a proline residue in RRP17. In addition, the Switch II region of RRP17, predicted to adopt an α-helix structure, is 12 amino acids longer than the corresponding region of other members of the Ras superfamily (). These differences suggest that RRP17 utilizes a distinct mechanism of coordination and hydrolysis of GTP, and imply that RRP17 interacts with a unique set of effectors and regulators.
RRP17 binds to CAPS1, a MUN domain protein involved in biogenesis and storage of LDCVs in neuronal and endocrine cells (; ). Interestingly, Munc13-4, another MUN domain–containing protein, interacts with Rab27 to regulate dense core granule secretion in platelets (). Although the exact domain of Munc13-4 that binds Rab27 has not been identified, the RRP17-interacting domain of CAPS1 spans most of the sequence between the PH and DCVD domains (unpublished data) and roughly corresponds to the MUN domain of the Munc13 protein ().
The MUN domains of Munc13 () and CAPS1 (unpublished data) have a predicted tertiary structure similar to that of importin β. Of note, importin β interacts with Ran GTPase through three binding regions, Switch I, Switch II, and the C-terminal α-helix. Co-crystallization of importin β and Ran-GTPase showed that importin β forms a right-handed superhelix that serves as a binding pocket for RanGTP (). Based on structural analysis, RanGTP has been proposed to insert itself into a cylinder formed by the importin β superhelix, thereby discharging importin β's cargo. Further studies are needed to determine whether RRP17 interacts with CAPS1 in a similar fashion.
We show that RRP17 is expressed in atria and ventricles of mice, whereas CAPS1 is present only in the atria. However, upon induction of cardiac hypertrophy, CAPS1 appears in ventricular myocytes. This finding suggests that ventricular myocytes may use the regulated secretion pathway during hypertrophy. It is known that atrial and ventricular myocytes secrete ANP and BNP via the regulated and constitutive secretion pathways, respectively (). However, several reports suggest that ventricular myocytes concurrently display constitutive and regulated secretion pathways (). Our findings suggest that during cardiac hypertrophy ventricular myocytes activate the regulated secretion pathway, in addition to the constitutive secretion pathway.
Although our results demonstrate a clear role for RRP17 in LDCV secretion, we acknowledge that RRP17 is also expressed in tissues that do not possess LDCV secretion (e.g., skeletal muscle). These findings suggest that RRP17 may have additional functions yet to be determined.
mice were defective in ANP release.
mice is not entirely clear and may be explained by a reduced rate of secretion or a packaging defect.
atrial myocytes. Notably, deletion of one allele of CAPS1 also results in perturbation of LDCV intracellular distribution (), which may reflect a defect in vesicle trafficking ().
ANP plays an important role in regulation of blood pressure via direct action on renal function and the peripheral vasculature, as well as via inhibition of the renin–angiotensin system and suppression of release of endothelin-1 (). Therefore, dysregulation of ANP release may cause elevation of blood pressure, as seen in mice lacking the ANP receptor (). Deletion of RRP17 in mice increased atrial ANP content and was associated with arterial hypertension and tachycardia in anesthetized mice.
mice is in the same range as seen in the
mice ().
mice is, in all likelihood, more complex than in
mice due to expression of RRP17 in other endocrine organs and the central nervous system.
The discovery of RRP17 and its role in the secretion of cardiac NPs provides new insights into the molecular mechanisms underlying the endocrine influences of the heart and may ultimately contribute to an understanding of the pathophysiology of cardiovascular diseases as well as the molecular basis of unresolved heterogeneity of BNP levels among different individuals with heart failure.
EST (AA624579), obtained previously by microarray analysis (), was analyzed using the NCBI BLAST server. This EST, shown to belong to the UniGene Mm.66275 cluster, was designated as a putative novel Ras-like protein-10B, Rasl10B. The complete cDNA of the gene was constructed by alignment of partial human and mouse cDNAs and ESTs deposited in NCBI, followed by RT-PCR verification. The structure of the gene was determined via alignment of cDNA sequences (three splice variants of cDNA were observed) to the mouse and human genomes in the NCBI database followed by human to mouse comparison using VISTA server ().
The prediction of the secondary structure of RRP17 protein and alignment to other Ras-like proteins were performed using ClustalW Alignment Algorithm on MacVector 6.5.3 software, and PBIL server (). A phylogenetic tree was constructed using Cluster and Topological algorithms via GeneBee Database Service (). Three-dimensional prediction and alignment of RRP17 to small G proteins of determined crystal structure were performed using 3D-PSSM (), Cn3D 4.1 software (NCBI), and DeepView/Swiss-Pdb Viewer software (GlaxoSmithKline).
For semi-quantitative PCR, cDNA was generated from 2 μg of total RNA with random-hexamer primers using the Superscript III kit (Invitrogen). To perform semi-quantitative PCR, 0.1 μl of P-dCTP (10 μCi/μl) was added per 25-μl reaction and the PCR product was resolved on an acrylamide gel. The radioactive signal was detected and quantified using a PhosphorImager (GE Healthcare) and X-ray film (Kodak). To ensure that equal amounts of cDNA were added to each reaction sample, L7 primers were initially used to determine the amount of cDNA per sample. Amplifications with L7 primers were performed at different numbers of cycles to ensure that densitometry was performed within linear range. Based on these results, calibrated amounts of cDNA were used in the PCR reactions. Primer sequences are available upon request.
Mouse embryos at ages ranging from E7.5 to E15.5 were dissected and fixed in 4% paraformaldehyde in PBS treated with diethylpyrocarbonate. Whole-mount and section in situ hybridizations were performed as described previously (), using sense and antisense probes prepared from RRP17 cDNA. Images of whole-mount in situ hybridization were captured by a microscope (M420; Leica) using a 3CCD camera (C5810; Hamamatsu). Images of radioactive in situ hybridization were captured using a microscope (DM2000; Leica) with a camera (VI470; Optronic) using 2.5× and 10× objectives. Images were processed using Adobe Photoshop 7.0.
LacZ staining was performed on heart and brain sections of adult mice using β-gal staining solution (5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 2 mM MgCl, and 1 mg/ml X-gal in PBS) overnight at room temperature. Images were captured on a Stemi SV11 microscope (Carl Zeiss MicroImaging, Inc.) with Apochromatic optics and a Macrofair camera (Optronics) () and a DM2000 microscope (Leica) with VI470 camera (Optronics) () using 10× and 40× objectives. Captured images were exported as 8-bit TIFF files and processed using Adobe Photoshop 7.0.
RRP17 expression constructs were generated by PCR-based cloning using mouse RRP17 cDNA as a template and were subcloned into pcDNA3 vectors (Invitrogen) with N-terminal Flag tag. Full-length rat HA-CAPS1 was provided by Thomas F.J. Martin (University of Wisconsin). C-terminal fragments of CAPS1 (amino acid residues are indicated in the results section) were generated by PCR using the Y2H prey as a template. All PCR products were verified by sequencing.
To generate “bait” for a yeast two-hybrid screen, the wild-type cDNA lacking the sequence encoding the CAAX box was subcloned into pGBKT7 vector (CLONTECH Laboratories, Inc.) in frame with the GAL4 DNA-binding domain at the C-terminus. The CAAX box of RRP17 was omitted from the fusion protein to prevent targeting of the protein to the cellular membrane due to prenylation. Adult heart and brain cDNA libraries, subcloned into pACT2 vector, were fused to the C terminus of the GAL4 trans-activation domain (CLONTECH Laboratories, Inc.). The “bait” was cotransfected with the indicated cDNA library into AH109 yeast cells and the assay was performed according to the MATCHMAKER GAL4 Two-Hybrid System 3 protocol (CLONTECH Laboratories, Inc.).
GST-tagged RRP17 cDNA was subcloned into pGEX-KG vector (GE Healthcare) and expressed in BL21(LysS) grown in LB supplemented with 0.5 M sorbitol and 2.5 mM betaine. Protein expression was induced by the addition of 250 mM isopropylthio-β--galactoside (IPTG) for 4 h at room temperature. Protein purification was performed according to published protocols with some modifications (Novagen Manual).
CAPS1 protein was in vitro translated using TNT kit (Promega) in the presence of S-methionine followed by incubation with equal amounts of recombinant GST-RRP17, GST-RhoA, or GST in 50 mM Hepes-KOH (pH 7.4), 100 mM NaCl, 1.5 mM MgCl, 0.1% NP-40, 10 μM GDP, and complete EDTA-free protease inhibitors (Roche) for 1 h at 4°C. Concentrations of recombinant proteins were estimated on SDS-PAGE using serial dilutions. Glutathione Sepharose (GE Healthcare) was added and incubated for 15 min followed by four washes with centrifugation at 2,000 . The pellet resulting from the final wash was boiled in SDS-loading buffer and resolved on a 4–15% gradient SDS-PAGE gel. The gels were dried and exposed to X-ray film for autoradiography.
Cells (HeLa and COS-7) were cultured in DME supplemented with 10% FBS, -glutamine, and penicillin/streptomycin at 37°C. Cell transfections were performed using FuGene6 reagent (Roche) according to the manufacturer's instructions.
Rat neonatal cardiomyocytes were isolated and infected with recombinant adenovirus as previously described (). Samples of media were collected at indicated intervals and processed by radioimmunoassay (RIA). At the end of the experiment, one half of plates were used for protein isolation and other half for RNA isolation to determine the level of ANP peptide and mRNA, respectively.
Recombinant adenovirus expressing full-length RRP17, and β-galactosidase proteins were generated using a Cre-loxP in vitro recombination system as previously described (). Cardiomyocytes were infected at a multiplicity of infection (MOI) of 10 for 3 h in plating medium, after which the medium was replaced with fresh growth medium.
To measure ANP in atria, isolated atria were snap frozen in liquid nitrogen, weighed, and homogenized in 10× volume of 0.1 M CHCOOH using a Dounce homogenizer. Homogenates were centrifuged at 30,000 for 30 min at 40°C. Supernatants were stored at −80°C until processed by RIA. Each sample was measured in duplicate.
To measure ANP in plasma, whole blood samples were collected via direct cardiac puncture on 3.8% EDTA (9:1 blood/EDTA) and aprotinin (50 KIU/μl). Blood samples were kept on ice until the completion of the study and then centrifuged at 1,000 for 10 min at 4°C. Plasma was then removed and then snap frozen in liquid nitrogen.
Cells were fixed for 10 min with either −20°C methanol or 4% paraformaldehyde, blocked with 10% normal goat serum (NGS) in PBS containing 0.1% Triton X-100 for 30 min followed by incubation with primary antibodies in 5% NGS/PBS for 1 h at room temperature. Working dilutions of primary antibodies used were as follows: polyclonal anti-Flag (1:300; Sigma-Aldrich), and monoclonal anti-CAPS (1:100; BD Transduction Laboratories). The cells were washed three times with 0.1% Triton X-100/PBS before the addition of secondary antibodies conjugated to either FITC or Cy3 (1:400, Jackson ImmunoResearch Laboratories) and counterstained with Hoechst 33238. Coverslips were mounted on glass slides using Vectashield (Vector Laboratories). Images were captured on a confocal laser scanning microscope with 63× oil immersion objective lens (model LSM 510-V3.2; Carl Zeiss MicroImaging, Inc). Captured images were exported as 8-bit TIFF files and processed using Adobe Photoshop 7.0.
An targeting vector was constructed to delete most of the coding region, including the starting ATG codon, by using a pN-Z-TK2 vector (a gift of R. Palmiter, University of Washington, Seattle, WA). The targeting vector contains a nuclear () cassette and a gene under the control of the RNA polymerase II promoter flanked by cloning sites and () gene cassettes. Both arms of the targeting vector were obtained by PCR from a mouse 129SvEv genomic library using High Fidelity PCR kit (Roche), sequenced and verified against the NCBI database. The short arm comprises sequences from 1009 to 80 bp upstream of the first ATG codon. The long arm comprises sequences starting from the nucleotide coding amino acid 145 and continuing 6 kb downstream. The targeting vector was electroporated into 129SvEv-derived ES cells, and targeted cells were selected with G-418 and FIAU. 500 ES clones were isolated and analyzed by Southern blot analysis for homologous recombination. Three targeted clones were injected into 3.5-d C57BL/6 mouse blastocysts. The resulting chimeric mice were bred to C57BL/6 females to achieve germline transmission of the mutant allele. Only mice of the 129SV/Ev × C57Bl6 mixed background were characterized in this study.
Primary atrial cardiomyocytes were isolated essentially as described for rat neonatal cardiomyocytes, with some modifications. In brief, the left and right atria from 10 to 15, 2–4 day-old mice were isolated under a dissection scope. Dissected tissues were minced and digested with 0.08% Collagenase Type II and 0.025% DNase Type I (Worthington). Cells were plated onto 24-well Primaria plates (Fisher Scientific) with fresh DMEM/F12 medium containing 5% horse serum (measured osmolarity ∼315 mOsm/kgHO) being replaced the next day. The cells were used 2 d after isolation. The medium was replaced with hypotonic buffer (100 mM NaCl, 5 mM KCl, 1 mM CaCl, 1.5 mM MgCl, 10 mM glucose, and 10 mM Hepes-NaOH, pH 7.3) at time “0” followed by 10-, 20-, 30-, and 60-min time points. Osmolarity of buffer (∼232 mOsm/kgHO) was confirmed by freezing-point depression osmometry using a Fiske Micro-Osmometer Model 210. All fractions were collected into siliconized Eppendorf tubes for further analysis by RIA. Cells remaining on the plates were harvested into hypotonic buffer (as above), freeze-opened, and used for ANP RIA and DNA content estimation (for normalization). To estimate the amount of DNA in each well, aliquots of cell lysates were mixed with 0.1 mg/ml Hoechst 33238 solution at a 1:4 ratio and the samples were analyzed using a FLUOstar Optima plate reader (excitation at 365 nm, emission at 458 nm). The amount of ANP secreted into the medium was expressed as a fraction of the total amount of ANP in the medium and cells. Data are expressed as means ± SE, with each experiment (consisting of 5–6 samples) being presented as an individual line on the graph.
To induce left ventricular hypertrophy, 6-wk-old mice were subjected to either thoracotomy (sham) or thoracotomy with partial occlusion of transverse aortic arch using a 27-gauge needle as previously described ().
To measure systemic hemodynamics, mice were anesthetized via isoflurane inhalation and transferred to a heat-controlled pad to maintain constant body temperature. The right carotid artery was exposed via direct midline neck incision, the distal end ligated, and the proximal end of the artery was catheterized with 1.4F Millar catheter (Millar Instruments). After allowing the animal a minimum of 30 min for stabilization, a minimum of 15 min of steady-state data were obtained.
The supplement shows the multiple alignment of GTP binding domains of representatives of the Ras family. The blue box highlights members of the RRP17 family. Online supplemental material is available at . |
Dendrite morphogenesis requires an intrinsic differentiation program that is guided by extracellular cues and electrical activity. These different forces converge to control the gene expression and cytoskeletal dynamics that specify dendrite growth and branching, as well as dendritic spine formation (; ; ; ; ). Various small GTPases act as key regulators in dendrite development (; ). The Rho family GTPases, such as RhoA, Rac1, and Cdc42, regulate actin dynamics and profoundly influence dendrite morphogenesis (; ; ). Another small GTPase Ras family is also important for activity-dependent filopodia formation and dendritic complexity via the activation of a MAPK signaling pathway (; ; ; ; ). Recent studies indicate that the Ras–phosphatidylinositol 3-kinase–Akt–mammalian target of the rapamycin signaling pathway has a pivotal role in regulating dendrite formation (; ). These small GTPases undergo structural changes in response to alternative binding of GDP and GTP: activation by GTP requires guanine nucleotide exchange factor (GEF), and inactivation requires GTPase-activating protein (GAP) (). The GTP-bound active form interacts with downstream effectors to activate cell signaling pathways, including that in dendrite morphogenesis (; ).
Various signaling pathways underlying dendrite development eventually affect the dynamics of actin filaments or microtubules (MTs) or both during dendrite growth. The high molecular weight MAP2 (referred to hereafter as MAP2) family proteins are an abundant group of MT-associated proteins that are predominantly expressed in neurons () and are selectively enriched in the soma and dendrites (). In dendrites, MAP2 is a major component of the cross-bridges between MTs, or between MTs and other cytoskeletal components (; ). In MAP2-deficient mice, the dendrites have a reduced MT density, and are shorter in hippocampal neurons (). Moreover, MAP2 can be highly phosphorylated and is thought to act as a phosphorylation- dependent modulator of MTs during dendrite development (). The details of the molecular mechanisms underlying these cytoskeletal regulations, however, are not fully understood.
Very-KIND protein (v-KIND) was first characterized as a nervous system–specific protein carrying two kinase noncatalytic C-lobe domains (KINDs), a RasGEF N-terminal domain (RasN), and a putative RasGEF domain (). The functional role of this new class of RasGEF v-KIND is largely unknown. In this study, we identified v-KIND as a transcript with up-regulated, brain-specific, and cerebellum- dominant expression profiles by transcriptomic analysis of postnatal mouse cerebellum (Cerebellar Development Transcriptome Database [CDT-DB] project, ). v-KIND was expressed during mouse brain development and interacted with Ras and MAP2. Overexpression of v-KIND suppressed dendrite growth, leading to simple branching patterns. In contrast, knockdown of v-KIND expression by RNA interference (RNAi) led to more complicated dendrite branching patterns. The results of the present study suggest that v-KIND is important for the control of dendrite growth. v-KIND is the first RasGEF known to bind to MAP2, thereby providing a new mechanism to link the Ras GTPase signaling pathway and MAP2-MT cytoskeletal organization during dendrite morphogenesis.
We first identified a transcript with an up-regulated and brain-specific expression profile (CDT-DB clone number CD01268; GenBank/EMBL/DDBJ accession no. ) in the transcriptomic analysis of cerebellar development, and cloned it as a mouse v-KIND gene (GenBank accession no.: ). v-KIND encoded a 1,742-amino acid protein that included two KIND domains (KIND1 and KIND2), a coiled-coil region, a Ras GEF N-terminal (RasN) domain, and a RasGEF domain ().
Analysis of eight postnatal day (P) 21 mouse tissues showed that v-KIND mRNA was predominantly expressed in the brain (). Expression of v-KIND mRNA in P7 mouse brain was very low and barely detected in the other tissues (unpublished data). Temporal expression analysis in developing mouse cerebellum revealed that the expression of v-KIND mRNA was sharply up-regulated between P7 and P12 ().
Cellular localization of v-KIND mRNA in P7 and P21 mouse brain was analyzed by in situ hybridization (). At P7, v-KIND mRNA expression levels were generally low throughout the brain, but high in hippocampus, thalamus, and the cerebellar white matter (). At P21, v-KIND mRNA was detected in nearly all fields at varying intensities (): it was predominantly localized in the cerebellum, and was observed at intermediate levels in the hippocampus and thalamus (). In the cerebellum, v-KIND mRNA was concentrated in the internal granular layer (IGL) and expression was higher in the anterior lobe than in the posterior lobe (). v-KIND mRNA was predominantly observed in granule cells, and was expressed in cells within the white matter (). These developmentally regulated and region-specific expression patterns suggest that v-KIND is responsible for specific developmental events in postnatal mouse brain.
We investigated v-KIND protein expression in developing mouse brains using an antibody that specifically recognizes v-KIND protein. The specificity of the antibody is shown in Fig. S1 (available at ). The v-KIND protein level was greatly increased by P21 (), which almost paralleled the mRNA expression pattern (). Subcellular fractionation analysis of P7 and P21 mouse cerebella shows that v-KIND protein is a soluble protein but appears to associate with precipitable components, such as cell membranes or cytoskeletal elements (Fig. S1 B).
The cellular distribution of v-KIND protein in mouse brain was analyzed using immunohistochemistry (). Widespread immunolabeling patterns of v-KIND were observed in the P21 mouse brain (). Intense immunolabeling signals were detected in the IGL of the cerebellum () and moderate immunolabeling signals were observed in the hippocampus (), which was consistent with the Western blotting data, showing that v-KIND is more abundant in cerebella than hippocampus (Fig. S1 C). In the cerebellum, there was more v-KIND protein in the anterior lobe than in the posterior lobe (), which was similar to the mRNA distribution pattern (). In the IGL, v-KIND protein was predominantly observed around the periphery of the granule cells and in the glomerular rosette areas, indicating that v-KIND protein expressed in granule cells is largely localized in the somatodendritic region (). In the hippocampus, v-KIND protein was distributed in the CA1-CA2-CA3 region () and was localized in the soma and dendrites of CA1 pyramidal neurons ().
To investigate the function of the RasGEF domain of v-KIND, we analyzed the interaction between v-KIND and Ras. v-KIND coimmunoprecipitated with Ras from cerebellar lysates (). This interaction was verified by an exogenous expression system in COS7 cells (). The expressed FLAG-tagged v-KIND was coimmunoprecipitated with endogenous Ras by the anti-Ras antibody, whereas the mutant lacking the RasGEF domain (ΔGEF), failed to be precipitated. These results suggest that v-KIND is associated with Ras through the RasGEF domain.
The RasGEF activity of v-KIND was then investigated by the pull-down assay using the GST-fused Raf-1 Ras binding domain (RBD), which specifically binds to the activated GTP-bound form of Ras (Ras-GTP). Coexpression of H-Ras with v-KIND in COS7 cells showed increased immunoreactivity for H-Ras-GTP pulled-down by the Raf-1 RBD in comparison with expression of H-Ras alone (), indicating that coexpressed v-KIND has potential RasGEF activity.
The RasGEF domain has binding affinity for both Ras-GDP and Ras-GTP (). We investigated the binding of v-KIND to Ras in hippocampal lysates in the presence of either GDP-β-S or GTP-γ-S (). In general, GDP-β-S inhibits the production of Ras-GTP, whereas GTP-γ-S activates it (). The quantity of Ras-GTP was measured by Raf-1 RBD pull-down assay. As expected, the immunoreactivity for Ras-GTP pulled-down by the Raf-1 RBD was higher in the presence of GTP-γ-S than in the presence of GDP-β-S (, bottom). In addition, the presence of GTP-γ-S increased the amount of v-KIND coimmunoprecipitated with anti-Ras antibody in comparison with that of GDP-β-S (, top), indicating that v-KIND binds to both Ras-GDP and Ras-GTP, but binds more preferentially to Ras-GTP.
We compared the binding specificity of v-KIND and ΔGEF to H-Ras by the Raf-1 RBD pull-down assay. v-KIND together with H-Ras was pulled down by Raf-1 RBD from lysates of COS7 cells coexpressing H-Ras and v-KIND, whereas ΔGEF was not pulled down by Raf-1 RBD from lysates of cells coexpressing H-Ras and ΔGEF (). These results suggest that the RasGEF domain of v-KIND has RasGEF activity and a major region responsible for the interaction with the activated Ras.
We verified the in vivo association of endogenous v-KIND and activated Ras in mouse hippocampal lysates (). In contrast to GST alone (as a negative control), GST-Raf-1 RBD pulled down v-KIND as well as Ras. Similarly, in contrast to control mouse IgG, the anti-Ras antibody coimmunoprecipitated v-KIND and Ras. Collectively, these results suggest that v-KIND acts as a RasGEF in mouse brain.
Subcellular localization of v-KIND was analyzed by immunocytochemistry in primary cultured neurons. In cerebellar granule cells at 14 days in vitro (DIV), v-KIND was distributed prominently in the soma and punctately in dendrites as marked by anti-MAP2 antibodies (Fig. S1 D). Because the dendrites of hippocampal neurons are easier to visualize than the dendrites of cerebellar granule cells, we compared the expression patterns of v-KIND with those of MAP2 at various differentiation stages in hippocampal primary cultures (DIV3, 6, 10, 15, and 21). Immunoblotting analysis revealed that the expression level of v-KIND gradually increased and peaked at DIV15, which was similar to that of MAP2 (). Immunocytochemical analysis showed v-KIND immunoreactivity almost overlapped with MAP2 immunoreactivity in soma and dendrites and was not localized to axons throughout differentiation (). Immunostaining patterns around the dendrite shafts differed between v-KIND and MAP2: v-KIND immunoreactivity appeared as many small puncta around the dendrites, although MAP2 immunoreactivity appeared very heavily and smoothly on the dendrites (). In DIV10 cultures, v-KIND immunoreactivity was also observed in the more distal tips of the dendrites, which were MAP2-negative, but F-actin positive ().
Because Ras family proteins regulate dendrite morphogenesis, we investigated the function of v-KIND in dendrite growth by overexpressing v-KIND in hippocampal primary cultures. Based on the fact that peak dendritic outgrowth occurs at DIV6-7 (), and taking into account transfection efficiency and post-transfection expression levels, hippocampal cultures were cotransfected with pCAG-v-KIND plus pCAG-EGFP (as a conventional reporter for cell morphology) at DIV3, and their dendrite morphology was observed at DIV11, DIV15, and DIV21. At DIV11, cells expressing EGFP alone exhibited normal dendrite growth with long branches, whereas v-KIND plus EGFP-overexpressing cells had a severely retarded dendritic pattern with very short protrusions (). At DIV11, v-KIND overexpression significantly reduced the number of cells with dendrites longer than 50 μm long (∼39% of cells with v-KIND overexpression vs ∼70% with EGFP alone; ; P < 0.001) and significantly decreased total dendrite length per cell (378 ± 153 μm with v-KIND vs 999 ± 463 μm with EGFP alone; ; P < 0.001). At DIV15, v-KIND–overexpressing cells had a few long, but less-branched dendrites and many short dendrites, whereas cells expressing EGFP alone had longer, branched dendrites (). Total dendrite length of individual v-KIND-overexpressing cell (599 ± 258 μm) was approximately two times shorter than that of cells expressing EGFP alone (1,156 ± 433 μm; ). At DIV21, cells expressing EGFP alone had almost mature arborization patterns with complex and long branched dendrites, whereas cells overexpressing v-KIND had very simple dendritic arborization patterns with a few long and less-branched dendrites (). Total dendrite length of v-KIND-overexpressing cells (566 ± 216 μm) was less than that of cells expressing EGFP alone (1753 ± 421 μm; ; P < 0.001). In addition, the number of dendritic branch tips per cell (branches with a distance from soma to tips longer than 20 μm were counted) was reduced by v-KIND overexpression (11 ± 5) compared with expression of EGFP alone (17 ± 5; ; P < 0.001). These results demonstrate that overexpression of v-KIND induces significant dendrite growth defects, including reduced elongation and branching.
To further analyze the function of v-KIND, we used RNAi to knockdown v-KIND expression. The knockdown efficiency of three types of stealth v-KIND RNAi constructs (RNAi-1, 2, 3) was confirmed in COS7 cells by cotransfecting cells with pEGFP-v-KIND (Fig. S2, A–D, available at ). Hippocampal cultures were transfected with each v-KIND RNAi construct together with EGFP at DIV3 and the dendritic patterns of EGFP-positive neurons were analyzed at DIV21 (). In contrast to three control RNAi constructs (Control-1, -2, -3), all v-KIND RNAi constructs induced the formation of complex dendritic patterns with significantly increased dendrite branching and length (). The efficiency of v-KIND RNAi knockdown was verified by transfecting with the RNAi construct at DIV3 followed by analyzing endogenous v-KIND levels at DIV7 and DIV17. The representative data of RNAi-2 effects are shown in . In contrast to the Control-2, the RNAi-2 significantly reduced v-KIND immunoreactivity at both DIV7 and DIV17, indicating that the RNAi was effective for v-KIND knockdown for at least 2 wk after transfection. For statistical analyses, we combined all data obtained from the experiments transfecting with each RNAi construct (RNAi-1, RNAi-2, or RNAi-3) and those from the control experiments transfecting with each control construct (Control-1, Control-2, or Control-3). The average total dendrite length per cell transfected with the v-KIND RNAi construct and control RNAi was 2,359 ± 475 and 1,610 ± 316 μm, respectively (). The average number of dendritic branch tips per cell was 35 ± 9.7 for v-KIND RNAi-transfected cells and 24 ± 6 for control RNA-transfected cells (). These data indicated that RNAi-induced knockdown of cellular v-KIND expression enhanced dendrite growth (, P < 0.001) and complexity in hippocampal neurons (, P < 0.001).
To confirm the cellular effect induced by v-KIND knockdown, we next investigated whether v-KIND RNAi rescued the dendritic phenotype induced by v-KIND overexpression. Hippocampal neurons were cotransfected with FLAG-tagged v-KIND plus RNAi-2 (v-KIND+RNAi-2) or with FLAG-tagged v-KIND plus Control-2 (v-KIND+Control-2) at DIV3, and were observed at DIV21 (). v-KIND+Control-2 transfected cells had simple and less branched dendritic patterns (), similar to those of v-KIND-overexpressing cells (). On the other hand, v-KIND+RNAi-2 transfected cells exhibited more complex dendritic patterns (). Exogenous FLAG-tagged v-KIND expression in transfected hippocampal neurons was visualized by immunostaining with anti-FLAG antibody (). In contrast to v-KIND and v-KIND+Control-2 (, respectively), v-KIND+RNAi-2 transfected cells had significantly decreased FLAG immunoreactivity (), indicating that RNAi effectively inhibited exogenous FLAG-tagged v-KIND. Similar results were also observed in RNAi-1 and RNAi-3 transfected cells. Statistical analysis by combining all data of these RNAi expression experiments (RNAi-1, -2, and -3) and their control experiments (Control-1, -2, and -3) indicated that the average total dendrite length per cell transfected with v-KIND, v-KIND+ Control, and v-KIND+RNAi was 1,159 ± 381, 921 ± 317, and 2,280 ± 332 μm, respectively () and that the average number of dendritic branch tips was 15 ± 6, 14 ± 4, and 31 ± 8, respectively (). These data indicated that cotransfection with RNAi rescued retarded dendrite growth by v-KIND overexpression with regard to both length (, P < 0.001) and branching (, P < 0.001). Collectively, these results suggest that v-KIND controls the development of dendritic arborization, probably by limiting dendrite growth, such as dendritic branching and extension.
To clarify the molecular mechanism underlying v-KIND-mediated control of dendrite growth, we analyzed v-KIND associated proteins by coimmunoprecipitation from brain lysates with anti- v-KIND antibody followed by MASS spectrometry analysis (). Silver staining after separating on sodium dodecyl sulfate-PAGE (SDS-PAGE) revealed a high molecular weight band of ∼280 kD specifically precipitated by anti-v-KIND antibody, but not by rabbit IgG. This band was identified as MAP2 by MASS spectrometry analysis ().
We first examined the interaction between v-KIND and MAP2 by coimmunoprecipitation experiments using COS7 cells lysates with exogenously coexpressed MAP2 and FLAG-tagged v-KIND or various deletion mutants: mutants lacking the first KIND domain (ΔKIND1), the second KIND domain (ΔKIND2), RasN (ΔRasN), or RasGEF (ΔGEF). Anti-v-KIND antibody coimmunoprecipitated MAP2 coexpressed with v-KIND, ΔKIND1, ΔRasN, and ΔGEF, but not ΔKIND2 (). These results suggest that v-KIND interacts with MAP2 via the KIND2 domain. To further confirm the MAP2 binding region of v-KIND, we performed pull down assays from mouse brain lysates using GST fusion KIND1 and KIND2 proteins. Immunoblotting and silver staining analyses showed that GST-KIND2, but not GST-KIND1, pulled down endogenous brain MAP2 (, top). These results demonstrated that the KIND2 domain is responsible for interaction with MAP2.
We analyzed changes in the interaction between v-KIND and MAP2 during mouse brain development by coimmunoprecipitation using anti-v-KIND and anti-MAP2 antibodies (; and Fig. S3, A and B, available at ). The quantity of v-KIND that coimmunoprecipitated with MAP2 peaked at P14, and the quantity of MAP2 that coimmunoprecipitated with v-KIND sharply increased at P14 and decreased at P21. Thus, the peak stage for the association between v-KIND and MAP2 appears to parallel the mid-late developmental stage of dendrite development. In fact, colocalization of v-KIND with MAP2 in P21 mouse cerebellum was also detected in the IGL and in neurons of the deep cerebellar nuclei (Fig. S3, C–K).
MAP2 is one of the most phosphorylated proteins in neurons and its phosphorylation state correlates with dendrite development (). We evaluated whether serine (Ser) and Thr phosphorylation levels of MAP2 expressed in COS7 cells were affected by coexpression of v-KIND. Phosphorylated MAP2 was detected by immunoprecipitation with anti-MAP2 antibody followed by immunoblotting with anti-phospho-Ser and anti–phospho-Thr (p-Thr) antibody (). Ser phosphorylation in MAP2 did not change significantly (unpublished data). In contrast, v-KIND coexpression increased p-Thr immunoreactivity of MAP2 (, top), which was confirmed by statistical analysis of quantified immunoreactive band intensity (, P < 0.05). These results suggested that coexpressed v-KIND enhances Thr phosphorylation of MAP2.
Many protein kinases phosphorylate MAP2 during brain development (). ERK and JNK-1, downstream kinases of Ras signaling pathways, are involved in Thr phosphorylation of MAP2 (; ). We investigated the effects of the ERK1/2 inhibitor U0126 () or the JNK-1 inhibitor SP600125 () on v-KIND induced p-Thr immunoreactivity of MAP2 (). U0126 or SP600125 treatment reduced Thr phosphorylation levels of MAP2 augmented by v-KIND coexpression, suggesting that JNK-1 and ERK, downstream targets of Ras signaling, are involved in the v-KIND–mediated MAP2 phosphorylation.
To study the protein structure and functional relationships of v-KIND in dendrite growth, we expressed various FLAG-tagged v-KIND constructs (v-KIND, ΔKIND1, ΔKIND2, ΔRasN, ΔGEF, KIND1 domain, and KIND2 domain) in hippocampal cells and immunocytochemically analyzed by anti-FLAG antibody (). Overexpressed v-KIND was mainly distributed in the soma and dendrite shafts where MAP2 was concentrated (). Similar to v-KIND, ΔKIND1, ΔRasN, and ΔGEF were also concentrated in the soma and dendritic shafts (). In contrast, ΔKIND2 was ubiquitously localized in the soma, dendrites shafts, spines, and axons (). Overexpressed KIND1 domain was widespread in dendrites and dendritic protrusions, including spines (), whereas overexpressed KIND2 domain had a similar localization as v-KIND, and was mainly localized in dendritic shafts, but not in spines (). Because KIND2 strongly binds with MAP2 (), these results indicate that v-KIND is anchored to MTs via the KIND2 domain.
KIND2 binds MAP2 as strongly as v-KIND protein, indicating that KIND2 is sufficient to replace endogenous v-KIND for MAP2 binding ( and ). We transfected the FLAG-tagged v-KIND or KIND2 domain into hippocampal neuronal primary cultures at DIV3, and observed the dendrite morphology at DIV11, DIV15, and DIV21 (). Expression of exogenous v-KIND and KIND2 proteins was confirmed by immunocytochemistry using anti-FLAG antibody (Fig. S4 A, available at ). In contrast to KIND2 overexpression induced significant increase in the total dendrite length (, bottom three panels and ; P < 0.001) and branching (, P < 0.05) compared with no expression of v-KIND constructs. These results demonstrate that the MAP2-binding KIND2 domain acts as a dominant negative form of v-KIND in the neuron, and that MAP2 is a target for v-KIND which controls dendrite growth. Moreover, it is noted that overexpression of the construct containing both KIND1 and KIND2 domains (KIND1+2) in hippocampal neurons induced significant increases in both dendrite length () and branching () compared with overexpression of KIND2 alone (Fig. S4 B). These results suggest another binding partner of v-KIND via KIND1 might also be involved in v-KIND-mediated regulation of dendrite growth and complexity and/or that the functional ability of KIND2 might be enhanced in the presence of KIND1.
We demonstrated the fundamental properties of v-KIND in hippocampal neurons as described above. We then analyzed the role of v-KIND in dendrite growth of cerebellar granule cells was analyzed by a series of v-KIND overexpression and knockdown experiments. Overexpression of v-KIND caused stunted short dendrites (, total dendrite length, P < 0.001) in cerebellar granule cells, whereas that of KIND2 alone induced extended long dendrites (, total dendrite length, P < 0.05). These observations were very similar to those in hippocampal neurons. The effects of v-KIND knockdown on granule cells were analyzed by transfecting with each RNAi construct or each Control construct soon after cell dissociation followed by fixation for EGFP cell morphology imaging at DIV7 (). Western blot analyses showed effective knockdown of v-KIND expression levels in RNAi-2-transfected cultures at DIV7 and DIV14 in comparison with those in Control-2-transfected cultures (). In addition, v-KIND knockdown by RNAi-2 promoted dendrite growth in granule cell cultures compared with that with Control-2 (, P < 0.01), which was similar to hippocampal neurons shown in . These results indicated that v-KIND has an important role in the control of neuron dendrite development in both cerebellar granule cells and hippocampal neurons.
We investigated MAP2 Thr phosphorylation in cerebellar granule cells when endogenous v-KIND expression was knocked down by RNAi. Western blotting analysis and quantitative statistical analysis showed that MAP2 Thr phosphorylation was reduced concomitantly with decreased v-KIND levels by knockdown with RNAi-2 (, P < 0.05). Similar data were also obtained from hippocampal neurons after v-KIND knockdown by RNAi (unpublished data). These results suggest that v-KIND is involved in MAP2 Thr phosphorylation, which is likely important for dendrite development.
#text
A series of first-strand cDNAs was produced by reverse-transcription (RT) from 20 ng of total cerebellar RNAs at varying developmental stages, using an oligo-dT primer. The cDNA sequence corresponding to the nucleotide positions 583 to 1182 (a.a 175–394) of v-KIND (GenBank/EMBL/DDBJ accession no. ) was amplified using the primers 5′-GCAGGTGGGGAAAACTCC-3′ (forward primer) and 5′-CTGGCGTCTCCAGCAAAT-3′ (reverse primer). The cycling conditions were as follows: denaturing at 94°C for 3 min, amplification by 25 cycles at 94°C (15 s), 55°C (30 s), and 72°C (1 min), and extension at 72°C for 5 min. To analyze tissue distribution, total RNAs prepared from various tissues of P7 or P21 mice were used for RT-polymerase chain reaction (PCR). The RT-PCR of glyceraldehyde- 3-phosphate dehydrogenase with primers 5′-GCCATCAACGACCCCTTCATTGACCTC-3′ (forward primer) and 5′-GCCATGTAGGCCATGAGGTCCACCAC-3′ (reverse primer) were used as internal controls.
In situ hybridization (ISH) brain histochemistry was basically performed as described previously (). The cDNA sequence corresponding to nucleotide positions 1962 to 2561 (a.a 654–857) of the v-KIND cDNA was used as a template to prepare the digoxigenin-labeled antisense riboprobes using a digoxigenin-dUTP labeling kit (Roche Diagnostics). Paraffin sections of mouse brain (10 μm thick) were fixed in 4% paraformaldehyde for 5 min, washed twice in phosphate buffered saline (PBS), and treated with freshly prepared 10 μg/ml proteinase K (Invitrogen) at room temperature. After acetylation, the sections were subjected to digoxigenin-based hybridization procedures. In brief, the sections were incubated in hybridization buffer containing 0.2 μg/ml digoxigenin-labeled riboprobes at 60°C overnight in a humid chamber. The hybridized sections were washed by successive immersion in 1 × SSC (150 mM NaCl and 15 mM sodium citrate, pH 7.0; 60°C, 10 min, twice), 2 × SSC (37°C, 10 min), 2 × SSC containing 20 μg/ml RNase A (37°C, 30 min), 2 × SSC (37°C, 10 min), and 0.2 × SSC (60°C, 30 min, twice). The hybridization signals were detected using a digoxigenin detection kit (Roche Diagnostics). ISH images () were captured with a microscope (Olympus BX51) equipped with a 2×/NA 0.08 (PlanApo) or 20x/NA 0.7 (UPlanApo) objective and a charged-coupled device camera (ProgRes C14; JENOPTIK) and processed with Photoshop (Adobe).
COS-7 cells were grown in culture dishes in DME/10% fetal bovine serum at 37°C and 10% CO in a humidified incubator. Hippocampal dissociated primary cultures were prepared from embryonic day 17 Wistar rats (Nippon SLC) as described previously (). In brief, excised hippocampi were treated with 45 U papain (Worthington), 0.01% DNase I (Boehringer-Mannheim), 0.02% DL-cystein, 0.02% bovine serum albumin, and 0.5% glucose in PBS for 20 min at 37°C. After adding 20% bovine serum, cells were dissociated by repeatedly passing them through a 1 ml plastic pipette tip. Dispersed cells were plated at a density of 1.10 cells/cm onto poly-L-lysine-coated (Sigma-Aldrich) glass coverslips (Matsunami) in neurobasal medium (GIBCO BRL, Life Technologies) containing 2% B27 supplement (Invitrogen), 500 μM -glutamine, 0.1 mg/ml streptomycin (Meiji), and 100 U/ml penicillin (Banyu, Tokyo, Japan). Cultures were maintained in a humidified atmosphere of 5% CO in air at 37°C.
Antibodies against mouse v-KIND were generated by immunizing rabbits with a synthetic peptide, CEKHSRKIQDKLRRMKATFQ (aa1724–1742), after conjugation with keyhole limpet hemocyanin. Affinity purification of the antibody was performed using a HiTrap NHS-High Performance column (Amersham Biosciences). The antibody was used at a dilution of 1:2000 for immunoblotting and 1:1,000 for immunohistochemistry and immunocytochemistry. Antibodies against MAP2a/b (M1406) and MAP2 (M4403) were obtained from Sigma-Aldrich; the antibody against GFP (SC-9996) was from Santa Cruz Biotechnology, Inc.; the anti-p-Thr antibody (#9381) was from Cell Signaling Technology; and the antibody against FLAG (F3165) was from Sigma-Aldrich. All antibodies were used at a dilution of 1:1,000 in immunoblotting and 1:200 in immunohistochemistry and immunocytochemistry.
Mouse v-KIND was isolated by PCR from mouse cerebellum and cloned into a pCAG vector with a FLAG-tag (a gift from Dr. Junichi Miyazaki, Div. Stem Cell Regulation Research, Osaka University Graduate School of Medicine, Osaka, Japan). v-KIND mutants were prepared by PCR-based methods and inserted into pCAG vectors with a FLAG-tag. H-Ras cDNA was isolated by PCR from mouse cerebellum and cloned into a pCAG vector with a HA-tag. HMW MAP2 cDNA was a gift from Dr. N. Cowan (NYU Medical Center, New York, NY) and was constructed by PCR-based methods and cloned into pCAG vector with a HA-tag. KIND1 and KIND2 domain fragments were prepared by PCR-based methods, and were inserted into pGEX-4T-2.
COS-7 cells were transfected by Lipofectamine 2000 reagent (Invitrogen) 24 h after plating at a density of 2.0 × 10 cells/cm. After an additional 6 h of incubation, cells were washed and cultured in DME/10% fetal bovine serum for 2 d. Hippocampal neurons were transfected by the Ca-phosphate-mediated method with some modification (; ). For 10 cells plating on each coverslip, 2 μg of plasmid was dissolved in 10 μl 250 mM CaCl, mixed with 10 ml 2×BBS (280 mM NaCl, 1.5 mM NaHPO, 50 mM BES, pH 7.1), and added to 400 μl of the fresh culture medium. After incubation for 1 to 2 h at 37°C with 2.5% CO, cells were washed with HBSS and cultured in conditioned medium. The calcium phosphate method using a CellPhect Transfection kit (Amersham Biosciences) was used to transfect cerebellar neurons on DIV1 in serum-free defined medium on poly- -lysine-coated glass coverslips. To quantify the percentage of cells with elongated dendrites at DIV11, 20 fields were randomly selected from three independent experiments with v-KIND or KIND2 overexpression based on healthy morphology, then analyzed by counting the number of cells with dendrites that were longer than 20 μm. To quantify dendritic length and arbor numbers in neurons overexpressing v-KIND protein or KIND2 domain, dendritic length and the number of arbors of neurons randomly selected from three independent experiments (15 fields total) were measured. Results are presented as the mean ± SD and were analyzed with test using the Excel 2003 software program. P values less than 0.05 were considered significant.
Three types of v-KIND stealth RNAi were obtained from Invitrogen. Their sense sequences were as follows: CAUCCAGGAGGAAUUUGCCUUUGAU (control sequence CAUGGAGGAUAAGUUUCCUUCCGAU), GAGCAGCUGCUAAAGAACCUCUUCA (control sequence GAGGUCGAAUCAAGACUCCUACUCA), and GAGACGGGAGGUUUCACCAUGACUA (control sequence GAGGGAGGAUUUCACUACAGACCUA). These RNAi were transfected into COS cells using the Lipofectamine 2000 reagent according to the manufacture's directions (Invitrogen), and into hippocampal neurons together by pCAG-EGFP by Ca phosphate precipitation at a concentration of 50 nM as described above. For immunoblotting experiments, RNAi was transfected into cerebellar granule cells soon after dissociation using the Mouse Neuron Nucleofector kit and the Nucleofector device as previously described (Amaxa) (; ).
ICR mice were anesthetized with diethyl ether and transcardially perfused with 4% paraformaldehyde in PBS. The excised brains were immersed for 2 h in the same fixative and cryosectioned (20 μm thick). For immunohistochemistry (IHC), fixed brain sections were preincubated with 5% normal goat serum in PBS containing 0.03% Triton X-100 for 1 h and then incubated with primary antibody overnight at 4°C. After washing with PBS, immunodetection was performed using horseradish peroxidase-conjugated secondary antibody (1:500) and diaminobenzidine. IHC images () were captured with a microscope (Olympus BX51) equipped with a 2x/NA 0.08 (PlanApo) or 20x/NA 0.7 (UPlanApo) objective and a charged-coupled device camera (ProgRes C14; JENOPTIK). The acquired images were processed with Photoshop (Adobe). For immunocytochemistry, cultured cells were fixed with 4% paraformaldehyde with 4% sucrose, permeabilized in 0.2% Triton X-100 in PBS for 5 min at room temperature, preincubated with 5% normal donkey serum in PBS for 1 h, and then incubated with each affinity-purified antibody for 16 h at 4°C. Multiple staining was performed using combinations of rabbit anti-v-KIND antibody (1:1,000), mouse anti-MAP2 antibody (1:200; Sigma-Aldrich), mouse anti-Flag antibody (1:200; Sigma-Aldrich), and Alexa 647-phallotoxin (1:50; AlexaFluor 647 phalloidin; Invitrogen). For fluorescent immunostaining, cells were then incubated with AlexaFluor 488-, 555-, or 647-conjugated secondary antibody (Invitrogen) in 5% donkey serum/PBS for 1 h at room temperature. Immunofluorescent images were captured with a confocal microscope (LSM 510 META; Carl Zeiss MicroImaging, Inc.) equipped with a 20x/NA 0.50 (Plan-NEO FLUAR), 40x/NA 0.75 (Plan-NEO FLUAR), or 100x/NA 1.4 oil iris (Plan-APO CHROMAT) objective lens (20×, ; 40×, , , and ; 100×, and ) and LSM 510 META software (Carl Zeiss MicroImaging, Inc). The acquired images were processed with Photoshop (Adobe).
Protein extraction and Western blotting analysis were performed as described previously (). In brief, COS cells or mouse cerebella were lysed and homogenized in 1% Triton X-100 buffer (50 mM Hepes, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl, 1 mM EGTA, 100 mM NaF, 1 mM NaVO, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM phenylmethylsulfonylfluoride). The subcellular fractionation of the P7 and P21 mouse cerebella was performed as described previously (). In brief, the P7 and P21 mouse cerebella were first homogenized in homogenization buffer (0.32 M sucrose, 5 mM Tris·HCl, 1 mM EGTA, 1 mM dithiothreitol, 1mM pepstatin A, 1mM leupeptin, and 1mM NaVO). The protein lysates were centrifuged at 1,000 for 10 min, the pellet was lysed in 1% Triton X-100 buffer (PPt1), and the supernatant was centrifuged at 105,000 for 1 h. The pellet was lysed in 1% Triton X-100 buffer (PPt2+3), and the supernatant was used as Sup3. For immunoprecipitation, after centrifuging at 1,000 for 10 min, protein solutions (containing ∼1 mg protein) were mixed with the primary antibody and incubated for 1 h on ice. The mixtures were rotated with Protein A sepharose or Protein G sepharose (Amersham Biosciences) for 1 h at 4°C. The sepharose was washed four times with 1% Triton X-100 buffer. After boiling in sample buffer (0.4 M Tris-HCl, pH 6.8, 8% SDS, 40% glycerol, and 0.04% bromophenol blue) for 5 min, equal portions of protein solution were separated by SDS-PAGE and probed with diluted primary and horseradish peroxidase-conjugated secondary antibodies (1:2,000). Silver staining was performed using the silver staining kit according to the manufacturer's instructions (Daiichi Pure Chemicals Co.). The GST fusion protein pull-down assay was performed as follows. expressing GST fusion proteins were lysed with lysis buffer (50 mM Tris-HCl, pH 7.4, 25% sucrose, 1% Triton X-100, 5 mM MgCl), and the cell extracts containing ∼10 μg GST-fusion proteins were coupled to glutathione- sepharose (Amersham Biosciences) by rotating for 1 h at 4°C. After washing three times with 1% Triton-X/PBS, the sepharose was coupled to cerebellar protein lysates, which had been precleared with glutathione-sepharose for 1 h at 4°C. After rotating for 1 h at 4°C, the GST-fusion protein complex was washed 5 times with cell lysis buffer and subjected to SDS-PAGE and immunoblotting. In GTP-γ-S and GDP-β-S treatment, protein lysates were treated with 0.1 mM GTP-γ-S (Sigma Chemical Co.) or 1.0 mM GDP-β-S (Sigma-Aldrich) in the presence of 10 mM EDTA, pH 8.0 at 30°C for 15 min (to activate or inactivate Ras). Then the GST pull-down assay was performed as described above. Quantitative analysis of immunoblotting band intensity was performed by Quantity-One software (Bio-Rad Laboratories). Relative intensity to the control was shown (% control). Values are mean ± SD. *, P < 0.05 as compared with the control.
For identification of v-KIND interacting proteins, in-gel digestion was performed following Coomassie staining, bands corresponding to MAP2 were excised, and the gel pieces were destained with 50% CHCN in 50 mM NHHCO solution. After removing the supernatant, cysteine residues were reduced with dithiothreitol, carbamido methylated with iodoacetamide, and the proteins were digested with trypsin at 37°C overnight. The tryptic peptides were recovered by sequentially adding 50% CHCN/1% trifluoroacetic acid, 20% HCOOH/25% CHCN/15% i-PrOH, and 80% CHCN solutions. The supernatants were collected and pooled into one tube, and the volume was reduced in vacuo. The dried tryptic peptides were suspended in 2% CHCN/0.1% trifluoroacetic acid and applied to the following liquid chromatography-tandem mass spectroscopy system. Chromatographic separation was accomplished with the MAGIC 2002 HPLC system (Michrom BioResources, Inc.). Peptide samples were loaded onto a Cadenza C18 custom-packed column (0.2 × 50 mm, Michrom BioResources, Inc.), and eluted using a linear gradient of 5% to 60% CHCN in 0.1% HCOOH for 30 min with a flow of 1 ml/min. Samples were ionized with the Nanoflow-liquid chromatography electrospray ionization, and tandem mass spectrometry data were obtained with an LCQ-Deca XP ion trap mass spectrometer (Thermo Electron Corp.). The Mascot database searching software (Matrix Science Inc.) was used for the identification of acetylated proteins.
Fig. S1: subcellular distribution of v-KIND in mouse cerebellum. Fig. S2: three oligonucleotide Stealth v-KIND RNAi efficiently knocked down exogenously expressed v-KIND in COS7 cells. Fig. S3: association of v-KIND with MAP2. (A-B), Coimmunoprecipitation of v-KIND and MAP2 from brain lysate. Fig. S4: FLAG-tagged v-KIND protein and the mutant overexpressed in hippocampal neurons. Fig. S5. v-KIND links Ras signaling to MTs. v-KIND is anchored to the dendritic shaft through the KIND2 domain by its MAP2 binding affinity. Online supplemental material is available at . |
Integrins are a family of heterodimeric cell adhesion receptors, which not only mediate cell adhesion to extracellular matrix proteins but also transmit signals that are vital for anchorage-dependent cell survival, proliferation, and motility (). Thus, integrins play pivotal roles in physiological processes such as wound healing, immune responses, and hemostasis. Aberrant integrin signaling is also centrally involved in the development of human diseases such as thrombosis, cancer, and autoimmune diseases and is the target of many therapeutic agents (; ). Therefore, understanding the molecular events in the transduction of integrin signals is important in our understanding of many biological processes and has the potential to reveal novel intervention for diseases.
Integrin signaling is bidirectional in that ligand binding function of integrins requires conformational change in the extracellular ligand binding domain induced by intracellular signals (inside-out signaling), and ligand binding transduces outside-in signals that mediate cellular responses (; ; ; ). An early functional outcome of integrin outside-in signaling is cell spreading, which is characterized by the formation of filopodia and lamellipodia and represents the outward movement of cell membrane, polymerizing actin, and cytoskeletal complexes at the leading edge (). Integrin signaling later induces cell retraction, which is the inward movement of the cell membrane and cytoskeletal complexes, often at the rear end of cells. Coordinated spreading and retraction allows cell migration and in blood platelets facilitates stable platelet adhesion, thrombus formation, and consolidation. It has been shown that small GTP binding proteins Rac and cdc42 are involved in signaling mechanisms of cell spreading and that RhoA-mediated signaling is important in cell retraction (). However, it remains unclear how integrin signaling initiates and temporally regulates these two seemingly opposing cellular responses; this is a fundamental question in cell biology. In this study, we have discovered that calpain cleavage of β at Tyr serves as a molecular switch that changes the outcome of the integrin outside-in signals from mediating cell spreading to promoting cell retraction. Furthermore, the switch from cell spreading to retraction is mediated by calpain cleavage of the c-Src binding site at the integrin C terminus, which relieves the c-Src–dependent inhibition of RhoA, and thus facilitates integrin-mediated, RhoA-dependent contractile signaling.
We have previously reported that the calcium-dependent protease calpain cleaves the cytoplasmic domain of the integrin β subunit mainly at Y, thus removing the C-terminal RGT sequence (; ). Calpain cleavage of β is inhibited by tyrosine phosphorylation at Y of the β cytoplasmic domain (). To understand the physiological role of calpain cleavage of β, we developed a calpain-resistant mutant of β by replacing R with a negatively charged glutamic acid (R760E; ). This mutation is intended to mimic the effect of phosphorylation at the adjacent Y by inhibiting calpain cleavage without perturbing the functionally critical NITY motif. The R760E mutant was stably expressed in a CHO cell line () together with the wild-type (WT) integrin α subunit. To determine whether the R760E mutant confers resistance to calpain-mediated proteolysis, lysates from cells expressing WT integrin αβ and the R760E mutant were incubated with or without purified μ-calpain and immunoblotted with Ab762, an antibody recognizing the intact β C-terminal TYRGT sequence, or Ab759, an antibody that recognizes the β TNITY sequence only when β is cleaved by calpain at Y (; ). The specificities of these antibodies have been previously characterized in and . In particular, we have shown that the calpain cleavage–specific antibody Ab759 only reacts with the β that is truncated at the C-terminal side of Y and fails to recognize β that is not cleaved (WT) or truncated at different sites, and that Ab762 reacts only with WT β that is not cleaved and fails to react with β cleaved at Y or other sites (). MAb 15, recognizing the β extracellular domain (which is not disturbed by calpain; ), serves as loading control (). WT β shows a large increase in reactivity with Ab759 after treatment with μ-calpain. This increase is mirrored by the loss of reactivity with Ab762, indicating loss of intact C terminus. In contrast, calpain- treated R760E mutant β showed no decrease in Ab762 reaction and dramatically diminished reactivity with Ab759. The relatively low reactivity of Ab762 with the R760E mutant compared with WT β is caused by R to E mutation but is not caused by reduced R760E expression, as indicated by the reaction with mAb 15. These results demonstrate that R760E mutation is calpain cleavage resistant (). Levels of stable surface expression of the R760E mutant and WT αβ were also analyzed by flow cytometry to ensure comparable expression after repeated cell sorting with an anti-αβ monoclonal antibody, D57 (). We showed previously that the β cytoplasmic domain is cleaved by endogenous calpain in platelets and integrin-expressing CHO cells adherent to fibrinogen (). To test whether R760E is resistant to endogenous calpain during cell adhesion, CHO cells containing either WT cells or R760E mutant β were allowed to adhere and spread on immobilized fibrinogen and were subsequently immunolabeled with affinity-purified cleavage-specific antibody Ab759 to detect integrin cleavage and with mAb 15 to mark β. WT β was cleaved by calpain (). In contrast, the R760E mutant showed marked reduction in calpain cleavage (). Thus, the R760E mutation renders β resistant to cleavage by calpain in cells adherent to fibrinogen.
Fibrin clot retraction, mediated by platelets and β integrin– expressing cells, requires β-dependent outside-in signaling and cell retraction (; ). To determine whether and how the calpain-resistant mutation of β affects integrin signaling, we compared the ability of the WT β- and R760E-expressing cells to mediate clot retraction (). Clot retraction mediated by R760E cells was significantly decreased (P < 0.001). Consistent with this result, calpain cleaved WT β at Y during clot retraction, and this cleavage was impeded in the R760E mutant cells (). It was previously reported that calpain I knockout platelets showed reduced ability to mediate clot retraction (). Similarly, a calpain inhibitor, MDL28170, also reduced clot retraction mediated by β integrin–expressing CHO cells (). However, the calpain substrate responsible for this effect of calpain was unclear. Thus, our data represent a new finding that calpain cleavage of β stimulates clot retraction. This finding is consistent with the data obtained in platelets. Although massive integrin cleavage induced by calcium ionophore or high concentrations of thrombin was not significantly affected in calpain I platelets in a previous study (possibly because of the involvement of Calpain II; ; ), we show in that calpain cleavage of β at Y was considerably reduced in calpain I platelets when the cleavage was induced by a lower concentration of thrombin (0.1 U/ml), which is consistent with the finding that the calpain I mice were defective in clot retraction when clot retraction was stimulated with a low concentration of thrombin ().
To determine whether the calpain-resistant β mutant affects integrin-dependent cell spreading, cells expressing WT or R760E mutant αβ were allowed to adhere to fibrinogen, and spreading kinetics of these cells were observed (). R760E mutation significantly enhanced and accelerated cell spreading compared with WT αβ (P < 0.0001), indicating that calpain cleavage of β negatively regulates cell spreading. This result is consistent with the data that cells expressing a deletion mutant of β mimicking calpain cleavage at Y (Δ759 cells) are defective in integrin-dependent cell spreading (). Thus, calpain cleavage of β negatively regulates the integrin outside-in signals that promote cell spreading while it stimulates the integrin outside-in signaling that activates cellular retractile machinery.
The increase in clot retraction and inhibition of cell spreading by calpain cleavage of β can result either from cleavage-induced activation of cell retractile function or cleavage-induced inhibition of cell spreading signaling. To differentiate these two possibilities, we sought to inhibit the cell retractile signaling mechanisms. Previous studies demonstrate that stress fiber formation in cells adherent to integrin ligands requires activation of RhoA () and Rho-dependent kinases (ROCK), which stimulate myosin light chain phosphorylation (; ) and thus retractile function of the actomyosin complex (). Indeed, the ROCK inhibitors Y-27632 or H-1152, when preincubated with WT β-expressing CHO cells () and platelets (), impeded clot retraction, indicating that the activation of the RhoA-ROCK signaling pathway is important in integrin-dependent cell retraction. To determine whether this pathway is responsible for calpain cleavage–dependent reversal of the direction of cell membrane movement, we studied the effects of ROCK inhibitors on cell spreading of WT β-expressing or calpain cleavage– mimicking β-expressing cells. The CHO cells expressing the calpain-cleaved form of β (Δ759) were defective in spreading on fibrinogen, and this defect was completely corrected by the ROCK inhibitors (). Spreading of CHO cells expressing WT αβ was not significantly enhanced (P > 0.05) by low concentrations of ROCK inhibitors (). However, higher concentrations of either H-1152 or Y-27632 accelerate cell spreading in WT cells (). Interestingly, enhancement of spreading by cells expressing cleavage-resistant integrin mutant R760E was not observed even with higher-dose ROCK inhibition (), suggesting that the inhibitory effect of Rho-kinase signaling on cell spreading is already diminished when β is resistant to calpain cleavage. These results suggest that cleavage of β at Y caused activation of the RhoA-ROCK signaling pathway and cell retraction, thus preventing cell spreading.
RhoA is activated by GTP binding and inactivated when bound GTP is hydrolyzed to GDP (). The Rho binding domain (RBD) of the RhoA effector protein rhotekin specifically binds GTP-bound RhoA and can therefore be used to indicate RhoA activation (). To determine whether calpain cleavage of β stimulates RhoA activation, cells were allowed to spread on fibrinogen-coated surfaces, and then were fixed, permeabilized, and incubated with fluorescent GST-RBD fusion protein (GST-RBD-555). GST-RBD-555 strongly reacted with the WT integrin-expressing cells spreading on fibrinogen (). GST-RBD-555 binding was specific because it was competitively inhibited by unlabeled GST-RBD () and by preincubation of cells with cell-permeable C3 transferase () but was unaffected in the presence of 20 mM glutathione (). The specificity of GST-RBD-555 is further supported by its colocalization with the stain of a monoclonal antibody that recognizes RhoA (). Thus, by using this novel in situ RhoA activation assay, we show that RhoA is activated after WT integrin-mediated cell adhesion and spreading on fibrinogen. In contrast to WT β, cells expressing the calpain-resistant β mutant R760E display reduced reactivity with GST-RBD-555, indicating that the resistance to calpain cleavage by R760E mutation and increased cell spreading in R760E cells are associated with inhibition of integrin-induced RhoA activation. Conversely, Δ759 cells, which are defective in spreading on fibrinogen, showed robust reactivity with GST-RBD-555, indicating that the calpain-cleaved form of β induces RhoA activation without requiring cell spreading. Collectively, these data demonstrate that cleavage of β at Y switches on the RhoA-ROCK signaling pathway, leading to cell retraction and inhibition of cell spreading.
It has been reported that a member of the Src family of protein kinases (SFK), c-Src, interacts with the β cytoplasmic domain and that this interaction requires the YRGT sequence in the C terminus of β (,). To determine whether c-Src interaction with β is affected by the R760E mutation or calpain cleavage–mimicking truncation of β, WT and mutant β, stably expressed in CHO cells, were immunoprecipitated with Ab8053, an anti-β3 antibody, and precipitates were immunoblotted for β-associated c-Src (). We found that c-Src coimmunoprecipitated with both WT and R760E mutant forms of β but failed to interact with the cleavage-mimicking mutant Δ759 (). These data are consistent with previous studies (,), indicating that cleavage of the β C-terminal RGT sequence disrupts c-Src binding to β.
It has been shown that c-Src plays an important role in cell spreading () and that c-Src interaction with β is important for this function (; ). Therefore, we examined whether the loss of β-associated c-Src function is responsible for the activation of cell retraction and inhibition of cell spreading induced by calpain cleavage at Y. PP2, an SFK inhibitor, abolished WT β-mediated cell spreading on fibrinogen (). The inhibitory effect of PP2 bears striking resemblance to the deficiency in cell spreading in Δ759 cells. More importantly, the inhibitory effect of PP2 was reversed by ROCK inhibitors H-1152 and Y-27632, indicating that the RhoA-ROCK–mediated retractile signaling is required for the inhibitory effect of this SFK inhibitor on cell spreading (), which is similar to the requirement for ROCK in β cleavage–mediated inhibition of cell spreading. To exclude the possible nonspecific effect of pharmacological inhibitors, WT β-expressing cells were transfected with cDNA, encoding a dominant-negative mutant of c-Src (K295R/Y527F; ), and cotransfected with 1/6 quantity of GFP cDNA to indicate successful transfection. The control cells expressing transfected GFP exhibited normal cell spreading on fibrinogen (). In contrast, the cells transfected with the dominant-negative c-Src showed a defect in cell spreading similar to that caused by PP2 (). This inhibitory effect of the dominant-negative c-Src was reversed when cells were cotransfected with an equal concentration of dominant-negative RhoA (N19-RhoA; ; ). The data obtained with cDNA transfection is thus consistent with the data using inhibitors of SFK and ROCK, indicating that integrin-dependent cell spreading occurs independent of c-Src when RhoA-mediated retractile signaling is disabled. Because integrin-mediated c-Src activation inhibits RhoA but does not affect cdc42 and Rac functions (), our data indicate that c-Src promotes integrin-dependent cell spreading by its inhibition of RhoA-mediated retractile signals. Furthermore, the SFK inhibitor PP2 corrected the defective RhoA activation in calpain cleavage–resistant R760E cells, restoring RhoA activation and inhibiting the spreading of R760E cells (), which is similar to the phenotype observed in Δ759 cells expressing the calpain-cleaved form of β. These results indicate that without calpain cleavage of β, β-dependent c-Src activity inhibits RhoA activation and thus promotes cell spreading. Conversely, inhibition of SFK with PP2 dramatically enhanced clot retraction (), further demonstrating that SFK inhibits cell retractile machinery. Therefore, cleavage of β at Y relieves the inhibitory effect of c-Src on RhoA activation, switches on RhoA-ROCK–dependent cell retraction, and thus changes the functional outcome of integrin signaling from cell spreading to cell retraction.
Previous studies suggested that activity of c-Src associated with β is increased compared with that of c-Src that is not associated with β (). To investigate the possibility that cleavage of β at Y regulates integrin-dependent overall c-Src activation in adherent cells, we measured the amount of active c-Src in CHO cells expressing WT and calpain cleavage–resistant (R760E) and –mimicking (Δ759) mutants of β. All three cell lines were similarly capable of inducing considerable c-Src phosphorylation at Y (an indicator of c-Src activation) when allowed to adhere and spread onto fibrinogen-coated surfaces (), suggesting that deletion of the c-Src binding site in β3 did not affect the overall integrin-mediated c-Src activation. Furthermore, if c-Src activity alone was sufficient to mediate the inhibition of RhoA, constitutively active c-Src would be able to induce integrin-mediated cell spreading. Thus, Δ759 cells were transfected with cDNA encoding a constitutively active mutant of c-Src (E378G) and cotransfected with 1/6 quantity of GFP cDNA to indicate successful transfection. Increased levels of total c-Src and expression of constitutively active c-Src (Y phosphorylated) in E378G c-Src–transfected Δ759 cells in suspension were verified by Western blot (). The E378G c-Src–transfected Δ759 cells are not different from the GFP-expressing control Δ759 cells in that both spread poorly on fibrinogen (). These data suggest that the inhibitory effect of β cleavage on cell spreading is not because of its effect on overall c-Src activation but is associated with dissociation of c-Src (and therefore c-Src activity) from the β cytoplasmic domain. Thus, only the β-associated c-Src activity is responsible for mediating RhoA inhibition and promoting cell spreading. Collectively, our results strongly suggest that cleavage of β at Y dissociates c-Src from the β C-terminal domain and thus relieves the inhibitory effect of β-associated c-Src on RhoA activation, stimulating the RhoA-ROCK retractile signals, which switches the functional outcome of integrin signaling from mediating cell spreading to promoting cell retraction.
In this study, we have discovered that β cleavage by calpain serves as a molecular switch that changes the outcome of integrin outside-in signaling from mediating cell spreading to promoting cell retraction. The switch in the outcome of integrin signaling from cell spreading to retraction is mediated by β cleavage–dependent activation of the RhoA retractile signaling pathway, and β cleavage–dependent activation of the RhoA signaling pathway is mediated by cleavage-dependent relief of the inhibitory effects of β-associated c-Src on the RhoA signaling pathway. Thus, our study also reveals that the requirement of SFK in integrin-dependent cell spreading is caused by its inhibition of the RhoA-dependent retractile function of the cell. These new findings represent major conceptual advances in our understanding of how integrin signaling controls and regulates the direction of cell membrane movement.
The data supporting our conclusions in this study are mainly obtained using the reconstituted integrin signaling model in CHO cells expressing WT and calpain cleavage–resistant or –mimicking mutants of β. Integrins play critical roles in all mammalian cells that adhere to extracellular matrix. Although different cell types express different members of the integrin family, integrin signaling, particularly outside-in signaling, is generally conserved in adherent cells, including CHO cells. In the past 20 yr, many important findings in β integrin function and signaling have been discovered using the CHO cell model expressing the WT and mutant β integrins, including the finding of the important role of talin in integrin activation (; ; ; ). Furthermore, the results obtained in CHO cells have been consistently verified in native β-expressing cells, such as platelets (, , ; ; ), indicating the validity of the CHO cell model. Importantly, the data obtained in CHO cell models in this study are consistent with the role of calpain I in clot retraction and in β cleavage as indicated in experiments using human and mouse platelets. Nevertheless, it is recognized that CHO cells are not totally identical to native β-expressing cells such as platelets, and thus the results obtained with the CHO cell model will need to be further verified in platelets and other β3-expressing cells in the future.
We have previously shown that the cytoplasmic domain of β is a calpain substrate and that the cleavage occurs mainly at Y, removing the C-terminal RGT sequence (; ). However, although calpain is activated after calcium elevation is induced by ligand binding to the integrin, very little cleavage of β occurs early during cell spreading because of the protection of integrin from calpain cleavage by tyrosine phosphorylation of the β cytoplasmic domain, especially at the periphery of a cell (). In contrast, at a later time during cell spreading when Y of β becomes dephosphorylated, substantial β cleavage occurs (). Also, calpain cleavage of β is readily detectable both during clot retraction in platelets and in the β-expressing CHO cells. The kinetics of integrin cleavage are consistent with our data, indicating a role for calpain in the transition between cell spreading and cell retraction in both CHO cells and in platelets. In fact, the data obtained with cells expressing the calpain-resistant and calpain cleavage–mimicking mutants clearly reveal that cleavage of β at Y generates the change in integrin signaling from mediating cell spreading to promoting cell retraction. Thus, we have discovered a calpain-dependent molecular switch that is regulated temporally and topographically by tyrosine phosphorylation of β and that controls the direction of membrane and cytoskeleton movement induced by integrin outside-in signaling. It is important to note that the cleavage-mimicking mutant of β creates an experimental condition where 100% of β cytoplasmic domain is cleaved at Y, which serves to clearly reveal the cleavage-induced changes in functional outcome. In platelets, >22% of β is cleaved at Y within an hour after platelet activation (), which coincides with the occurrence of marked clot retraction. However, in more delicate situations, such as during cell spreading and migration, local and dynamic calpain activation signals are likely to cause localized cleavage of a small percentage of β molecules and thus dynamically control the local movement of membranes and cytoskeleton.
The β cleavage–mediated inhibition of cell spreading and promotion of cell retraction is not caused by its inhibition of cell spreading mechanisms but is mediated by the integrin cleavage-dependent activation of the RhoA retractile signaling pathway. Previous studies suggest that Rac and cdc42 play important roles in mediating cell spreading (filopodia and lamellipodia formation) and that RhoA is increasingly activated in spread cells (). RhoA activates ROCK, which phosphorylates and inactivates myosin light chain phosphatase, and enhances myosin light chain phosphorylation and retractile function of the actomyosin complex (; ; ), inducing cell retraction. Cleavage of β at Y inhibits integrin-mediated cell spreading (; ). If the inhibitory effects of β cleavage on cell spreading should result from its inhibition of Rac- and cdc42-mediated cell spreading signaling, this inhibitory effect should not be corrected by inhibition of RhoA pathway. Therefore, our data that β cleavage–dependent inhibition of cell spreading is corrected by ROCK inhibitors indicate that calpain cleavage of β activates RhoA-dependent cell retraction, which subsequently inhibits the opposite movement of cell membrane resulting in inhibition of cell spreading.
Previous studies indicate that members of SFK are required for integrin-mediated signaling that leads to cell spreading (). The mechanism responsible for this role of SFK is not clear but is believed to be an early step in cell spreading signaling. However, we show that the inhibitory effect of an SFK inhibitor and a dominant-negative c-Src mutant on β-mediated cell spreading is reversed by inhibition of the RhoA signaling pathway, indicating that cell spreading is independent of c-Src when RhoA signaling is abrogated. These data therefore provide the first direct evidence that the role of SFK in promoting integrin-dependent cell spreading is not that it is required for cdc42- and Rac-dependent cell spreading signaling, but rather is its inhibition of RhoA-dependent retractile signaling and thus cell retraction, which confines the outward movement of the cell membrane and cytoskeleton. Our data are consistent with the previous studies, which show that SFK inhibits RhoA activity but has no effect on the activity of cdc42 and Rac (). Our data are also consistent with the finding that RhoA inactivation by p190RhoGAP promotes membrane protrusion and polarity (). More importantly, we show that cleavage of β at Y abolishes c-Src binding to β but does not significantly affect overall c-Src activation in cells spread on fibrinogen, and that the RhoA-dependent inhibition of cell spreading by β cleavage at Y is not reversed by expression of a constitutively active Src. These data suggest that whereas overall intracellular c-Src activation during cell adhesion does not require c-Src association with β, active c-Src molecules dissociated from β are not capable of inhibiting the RhoA signaling pathway and promoting cell spreading on β ligands, possibly because of the lack of β scaffold. However, the calpain cleavage–resistant mutation in β inhibits RhoA activation in a Src-dependent manner. Thus, our results suggest that only the β-associated c-Src is responsible for the inhibition of the RhoA pathway and facilitation of cell spreading. Conversely, our results suggest that calpain cleavage of β, which abolishes Src binding to β (; ), relieves the inhibitory effects of β-associated c-Src on the RhoA pathway, activates RhoA, and thus switches the outcome of integrin signaling from cell spreading to retraction.
Interestingly, β cleavage–dependent relief of c-Src– mediated RhoA inhibition and activation of RhoA-dependent cell retraction is associated with the cleavage of only a fraction of β molecules (; ) and cannot be prevented by the presence of uncleaved β molecules in the same cell. This apparently dominant effect of β cleavage suggests that c-Src activity at one location in a cell does not cross-inhibit RhoA activation induced by calpain cleavage of β at a different location and is consistent with the knowledge that spreading (such as filopodia and lamellipodia) and retraction can occur simultaneously at different locations within a cell (such as during cell migration). The requirement of c-Src association with β for the c-Src–mediated RhoA inhibition provides a mechanism for the localized role of c-Src in promoting cell spreading. Further studies are required to understand how β-associated c-Src activity locally regulates RhoA and whether additional mechanisms are needed for RhoA activation when β cleavage by calpain relieves the inhibitory effect of β-associated c-Src.
Based on data from this and other studies, we propose a new model of integrin signaling that incorporates a phosphorylation- and protease-mediated mechanism regulating integrin-mediated cell membrane movements. In this model (), initial integrin signaling activates protein tyrosine kinases, which induce phosphorylation of the cytoplasmic domain of β (, ), and elevates intracellular calcium levels, which activate calpain (). Tyrosine phosphorylation at Y of β in the early phase of cell spreading protects the c-Src binding site in the C-terminal domain of β from calpain cleavage (). The interaction of c-Src with β promotes activation of β-associated c-Src by a Csk- () and protein tyrosine phosphatase IB–dependent mechanism () and/or allows c-Src–dependent signaling in the vicinity of the β C terminus, which facilitates cell spreading by inhibiting RhoA-dependent cell retraction. In spread cells, Y of β becomes dephosphorylated (), allowing β cleavage by activated calpain, which removes the C-terminal RGT sequence of β important in c-Src binding. Integrin cleavage by calpain thus relieves the local inhibitory effect of β-associated c-Src on the RhoA-ROCK signaling pathway and locally activates RhoA-dependent cell retractile signals (; ). Thus, calpain cleavage of β switches the direction of integrin outside-in signaling from mediating spreading to promoting cell retraction.
Plasma containing 0.38% sodium citrate and 40 μg/ml aprotinin (Sigma-Aldrich) was depleted of fibronectin as previously reported in . Ab762 and Ab759 were previously described in and . mAb 15, D57, and Ab8053 were provided by M. Ginsberg (University of California, San Diego, La Jolla, CA; ). Phosphospecific anti-Src Y antibody was obtained from Cell Signaling; anti-RhoA and cell permeable C3 transferase were obtained from Cytoskeleton, Inc.; Sudan black was obtained from Sigma-Aldrich; Alexa Fluor 594–conjugated goat anti–rabbit IgG and Alexa Fluor 488– conjugated goat anti–mouse IgG were obtained from Invitrogen; anti–v-Src monoclonal antibody, SFK inhibitor PP2, ROCK inhibitors H-1152 and Y-27632, and calpain inhibitor MDL 28170 were obtained from Calbiochem; and an anti–c-Src polyclonal antibody was obtained from Santa Cruz Biotechnology, Inc. GST-RBD cDNA were a gift from T. Kozasa (University of Illinois, Chicago, IL). GST-RBD fusion protein was purified as previously described in . Purified protein was fluorescently labeled with an Alexa Fluor 555 microscale protein labeling kit (Invitrogen).
The β R760E mutation was performed using PCR. The forward primer has the sequence of AGAGCTTAAGGACAC. The reverse primers (CTCATTAAGTCCCCTCGTAGGTGATATTGG) contain an XhoI digestion site, stop codon, and the 18-nucleotide C-terminal β sequence coding the intended R760E mutation. PCR products were digested with AspI and XhoI and ligated into a β cDNA construct in a modified cDM8 vector containing only one XhoI site at the 3′ end of the multiple cloning site. The construct was verified by DNA sequencing.
CHO cells stably expressing WT human αβ (2b3a), human glycoprotein Ib-IX (GPIb-IX; 1b9), or both αβ and GPIb-IX were previously described in and . CHO cells expressing GPIb-IX and a truncation mutant β, mimicking calpain cleavage at Y, and complexed with WT α (Δ759) were also described previously in and . DNA transfection was performed using Lipofectamine 2000 (Invitrogen). R760E β mutant cDNA was cotransfected with WT α and pcDNA 3.1/Hyg plasmid at a ratio of 5:5:1 into 1b9 cells. Complex formation between α and β subunits was verified by flow cytometry using antibody D57 against αβ complex. Stable R760E mutant cell lines were obtained by antibiotic selection and repeated cell sorting using D57 as previously described (). To ensure comparable expression, R760E and Δ759 cells were sorted using the expression level of the stable WT αβ-expressing cells as a gate (; ). For transient transfections, cells stably coexpressing WT αβ were transfected with either 2 μg GFP, 2 μg GFP plus 12 μg of dominant-negative c-Src (), 2 μg GFP plus 12 μg of dominant-negative N19-RhoA (), or 2 μg GFP plus 12 μg of both dominant-negative c-Src and N19-RhoA, using Lipofectamine 2000. In some experiments, cells were transiently transfected with either 2 μg GFP or 2 μg GFP plus 12 μg of constitutively active c-Src (E378G). Cells were analyzed 24 h after transfection.
10 CHO cells were directly solubilized in buffer containing 2% Triton X-100, 0.15 M NaCl, 1 mM CaCl, and 0.02 M Tris, pH 7.4, in the presence or absence of 10 mM EDTA. Cell lysates were incubated for 40 min at 37°C with or without 1 μg of purified μ-calpain. SDS-PAGE sample buffer containing 10 mM EDTA, 1 mM PMSF, and 0.2 mM E64 was added to stop reactions. For experiments using mice, 6–8-wk-old mice of either sex were anesthetized by intraperitoneal injection of pentobarbital and blood was drawn from the inferior vena cava. Blood from 5–6 (calpain I and calpain I; ) mice was pooled or platelets were isolated by differential centrifugation as described in . 3 × 10/ml washed platelets were resuspended in modified Tyrode's buffer () and incubated at 22°C for 2 h before use. Platelets were stimulated with 0.1 U/ml thrombin in a platelet aggregometer at 37° for 7 min, and then solubilized in SDS-PAGE sample buffer. All samples were analyzed by SDS-PAGE using 4–15% gradient gels and were immunoblotted using various antibodies. Enhanced chemiluminescence (GE Healthcare) was used for visualization of antibody reactions.
Chamber slides (Lab-Tek; Nunc) were precoated with 10 μg/ml fibrinogen and blocked with 5% BSA in PBS. 50 μl CHO cells (expressing WT or mutant integrins; 5 × 10/ml) in Tyrode's buffer was added to the wells and incubated at 37°C for 1 h. After washing, cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100, 0.1 M Tris, 10 mM EGTA, 0.15 M NaCl, 5 mM MgCl, 1 mM PMSF, 0.1 mM E64, and 1% BSA (permeabilization buffer, pH 7.5). Samples were blocked with 5% BSA, incubated with 10 μg/ml mAb 15 and 0.5 μg/ml of purified Ab759, and, after washing, stained with fluorescently labeled secondary antibodies. To detect RhoA activation, cells were fixed in 4% paraformaldehyde in PBS with 10 mM MgCl and incubated with 0.7 μg/ml mAb 15 and Alexa Fluor 555–labeled GST-RBD in permeabilization buffer containing 10 mM MgCl. After further staining with Alexa Fluor 488–conjugated goat anti–mouse IgG and washes, the antibody and GST-RBD staining were detected using a confocal microscope (Carl Zeiss MicroImaging, Inc.). In some experiments, cells were also preincubated with or without 5 μg/ml of cell-permeable C3 transferase to inhibit RhoA, according to the manufacturer's instructions, and then stained with GST-RBD-555. Surface area in spreading cells was estimated by measuring total pixel number per cell in digitally recorded images using Image J software (National Institutes of Health). Quantitation of 555-nm fluorescence in R760E cells was determined by using the area tool in LSM 5 software (Carl Zeiss MicroImaging, Inc.) and expressed as total pixel number. Statistical significance was determined using a test.
400 μl of indicated CHO cell lines (4 × 10/ml), in complete DME in the presence or absence of 5 μM ROCK inhibitors H-1152 or Y-27632 or calpain inhibitor MDL 28170, was mixed with 100 μl of plasma, 2 U/ml thrombin, and 5 mM CaCl in a partially Sigmacote-treated glass cuvette (Chrono-Log). For platelet samples, citrated platelet-rich plasma was mixed with 0.2 U/ml thrombin. The clots were allowed to retract at 37°C and were photographed at various times. The two-dimensional sizes of retracted clots on photographs were quantified using Image J software and were expressed as clot size. Statistical significance was determined using a test.
Clots were fixed with 4% paraformaldehyde, embedded in 50% Tissue-Tek OCT compound (Sakura Finetek), placed in 25 × 20 × 5-mm Tissue-Tek Cryomolds (Sakura Finetek), and snap frozen in liquid nitrogen. Frozen sections were made using a cryostat microtome (Microm HM 505 E; Carl Zeiss MicroImaging, Inc.). Slides were then washed and quenched with 5% Sudan black (wt/vol) in 70% ethanol. Sections were permeabilized with 0.1% Triton X-100, 0.1 M Tris, 10 mM EGTA, 0.15 M NaCl, 5 mM MgCl, 1 mM PMSF, 0.1 mM E64, and 1% BSA, pH 7.5 (permeabilization buffer), blocked with 5% BSA, and immunolabeled with 10 μg/ml mAb 15 and 0.5 μg/ml of purified Ab 759. After washing, cells were stained with Alexa Fluor 594–conjugated anti–rabbit IgG and 488–conjugated anti–mouse IgG. Data was collected using a confocal microscope.
CHO cells expressing WT, R760E mutant, and Δ759 mutant integrins (100 μl/well, 1.5 × 10/ml) in complete DME were incubated in fibrinogen- coated chamber slides at 37°C for 20, 40, and 60 min. At each time, cells were immediately rinsed and fixed with 4% paraformaldehyde. For assays using PP2 and ROCK inhibitors, cells were preincubated at 22°C for 30 min with or without 10 μM PP2 and/or 4 μM H-1152 or Y-27632. Cells were allowed to spread for 1 h at 37°C, rinsed, and fixed with 4% paraformaldehyde. Images were captured using a microscope (DM IRB; Leica). The size of the cells from four random fields for each time and cell line was determined using Image J software. Relative increase in cell size was calculated by dividing the mean size of spread cells by the mean size of suspended cells from each cell line. Statistical significance was determined using a test.
Cells were lysed in lysis buffer containing 1% NP-40, 150 mM NaCl, 50 mM Tris, pH 7.4, 1 mM sodium orthovanadate, 1 mM NaF, and Complete protease inhibitor mixture (Roche Molecular Biochemicals). Lysates were incubated with either rabbit anti-β or an equal amount of rabbit IgG and subsequently with protein A–Sepharose (GE Healthcare). Beads were washed four times with lysis buffer, analyzed by SDS-PAGE using 4–15% gradient gels, and immunoblotted with mAb 15 for β and monoclonal anti-Src (Calbiochem).
10-cm polystyrene dishes (Fisher Scientific) were coated with 10 μg/ml fibrinogen and blocked with 5% BSA in PBS. Equal suspensions containing 5 × 10 CHO cells stably expressing WT, R760E mutant, or Δ759 mutant integrins in modified Tyrode's buffer were either kept in suspension or added to the dishes and incubated at 37°C for 30 min. Plates were rinsed three times with PBS and cells were subsequently lysed with the buffer containing 1% NP-40, 150 mM NaCl, 50 mM Tris, pH 7.4, 1 mM sodium orthovanadate, 1 mM NaF, and Complete protease inhibitor mixture. Lysates were analyzed by SDS-PAGE using 4–15% gradient gels, and then immunoblotted with anti–phospho-Src Y and anti–c-Src. |
Solid tumors and their metastases require the process of angiogenesis to invest themselves with host microvasculature and to undergo pathologic progression (). Blockade of this process leads to tumor arrest and apoptosis (). Effective agents include proteolytic fragments of collagen XVIII (endostatin), platelet factor 4 (anginex), TNP-470 (a fumagillin derivative), antibodies against integrins and proangiogenic growth factors (e.g., anti-VEGF antibodies), small-molecule inhibitors of endothelial growth-factor receptors (e.g., VEGF, FGF, and PDGF receptor tyrosine kinase inhibitors), and matrix metalloproteinase inhibitors (; ; ; ). Many proangiogenic growth factors, in particular, FGF-2 and heparin-binding splice isoforms of VEGF-A, such as VEGF, bind to heparan sulfate proteoglycans expressed on tumor endothelia or in the extracellular matrix, which facilitates formation of signaling complexes composed of the ligands and their receptor tyrosine kinases (; ; ). The dependence of receptor activation on heparan sulfate suggests that endothelial heparan sulfate might represent a therapeutic target and that inhibitors might have synergistic effects because of interference with multiple growth factor signaling pathways important for tumorigenesis.
Binding of both FGF-2 and VEGF depends on the sulfation state of heparan sulfate, in particular -sulfation of glucosamine residues in specific domains along the heparan sulfate chain (; ; ). A family of four -acetylglucosamine -deacetylase/-sulfotransferases (Ndst) catalyze this reaction (), but microvascular endothelial cells only express and (). -deficient mice are essentially normal, aside from defective granule formation in connective tissue-type mast cells (; ). In contrast, mice with a systemic deletion of die perinatally because of forebrain defects, skeletal malformation, and lung hypoplasia (; ; ; ), but vasculogenesis and angiogenesis up to birth appear normal by gross examination and general histology. Here, we show that altering Ndst1 expression in endothelial cells alters growth of syngeneic murine tumors because of a unique alteration in the tumor microvasculature. In contrast, the mutation had no effect on physiological angiogenesis, e.g., during cutaneous wound repair. The change in tumor angiogenesis appears to result from altered binding and signaling by the proangiogenic growth factors FGF-2 and VEGF.
To determine the role of Ndst1 in angiogenesis in adult animals, mice harboring a conditional loxP-flanked allele of (
) were crossed with transgenic mice expressing the bacteriophage recombinase Cre under the control of the angiopoietin receptor () promoter, which is expressed in a pan-endothelial fashion (; ).
(mutant) and
Cre (wild type) mice were recovered at the expected Mendelian frequency and showed the normal pattern of weight gain, indicating that vasculogenesis and physiological angiogenesis were sufficient for normal development. Adult mutant mice were indistinguishable from their wild-type littermates in general health, behavior, and body weight ().
females also reproduced normally and had normal litter sizes, indicating that sufficient reproductive angiogenesis occurred to allow for normal colony propagation.
Physiological angiogenesis during cutaneous wound repair also appeared normal. After a biopsy punch wound of dorsal skin, an intense angiogenic response was detected 3–4 d after injury at the granulating edges of healing wounds. Both mutant and wild-type mice showed a similar response in clustering of CD31-positive processes at the wound margin (). No significant difference in the density of processes at the wound edges was noted between the two groups (). Vascular density fell with distance from the wound margin in a similar fashion in both mutant and wild-type mice as well. The rate of wound contraction over the 4-d period also did not differ (), and in separate experiments, complete resolution of wounds occurred in both groups of mice by 10 d. Analysis of retinal vascular development under normoxic conditions showed only a very minor effect that was not significant (Friedlander, M., personal communication). These findings suggest that physiological angiogenesis was unaffected by endothelial inactivation of .
and
Cre mice. As shown in , growth of subcutaneous tumors in mutant animals was significantly attenuated (P < 0.001) relative to that observed in wild-type littermates at 10 and 15 d after injection. This effect was also seen in a second transplant model in which syngeneic mouse melanoma (B16BL6) tumors were injected subcutaneously. Like LLC tumors, melanoma exhibited a marked reduction in mean tumor volume in mutant mice compared with wild-type littermates at 10 d (52% reduction in mean volume; P < 0.02). Because skin ulceration occurred and the animals showed signs of morbidity after 15 d, all tumor growth experiments were terminated as required by Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) guidelines for the humane treatment of animals.
Because the transgene is expressed in endothelial cells, leukocytes, and marrow-derived endothelial precursors (; ), we performed bone marrow transplantation experiments to determine the relative contributions of these cell populations to the phenotype. Mutant mice transplanted with wild-type marrow showed persistent attenuation in tumor growth, indicating that deletion of in tissue-resident endothelia was sufficient to achieve a marked alteration in tumor growth (, compare a and b, center). Transplantation of wild-type mice with mutant marrow did not alter tumor growth (, right), indicating that alteration of heparan sulfate in marrow-derived cells had no effect. Control transplantations in which the genotype of the marrow donor and the recipient mouse were matched showed the expected results (, left). These experiments exclude factors such as altered tumor immunity because of changes in leukocyte heparan sulfate. Histological examination of the tumors also showed that tumor infiltration by CD45/Gr-1–positive granulocytes was not statistically different in the mutant, suggesting that reduced acute leukocyte infiltration caused by altered endothelial heparan sulfate () did not affect tumor growth. These findings also exclude effects due to marrow-derived endothelial precursors, which can serve as early contributors to vascularization in some tumors (; ).
To determine whether the decrease in tumor mass was caused by a change in tumor microvasculature, we examined the pattern of staining of CD31 (PECAM-1 [platelet/endothelial cell adhesion molecule-1]) in tumor sections (). Microvascular density was significantly reduced in tumors grown on the mutant background (; 26.8 ± 4.7 vs. 35.3 ± 4.7 microvessels per 400× field in mutant and wild-type mice, respectively; P < 0.02). A slight increase in vessel caliber was also apparent when antibody binding was visualized by fluorescence microscopy ().
To obtain greater morphometric detail, intravital microscopy was used to assess microvascular branching, density, skeletal length, and caliber in LLC tumor spheroid preparations implanted on dorsal skinfold chambers. Views of tumor microvasculature at 12 d after implantation revealed a general reduction in vascular density and complexity within tumors grown in the mutant (). Morphometric analysis of digitized images showed a marked reduction in degree of branching, density of linearized vasculature (grid intersection), and skeletal length (). A small, but statistically significant (P < 0.02) increase in vascular caliber was noted as well. These alterations may have resulted from a primary branching defect of tumor neovasculature () and are similar to changes in vascular development after in vivo treatment with anti-VEGF antibodies or VEGF receptor tyrosine kinase inhibitors (; ) and alterations in FGF-dependent bud formation and branching morphogenesis of respiratory epithelium during tracheal development () and vertebrate lung development ().
To gain insight into the mechanism underlying the alteration in tumor vascularization, we attempted to purify endothelial cells from tumors and normal tissues. However, we were unable to obtain sufficient numbers of cells from tumors, and the cultures were overtaken rapidly by small numbers of contaminating tumor cells. In contrast, highly purified cells were obtained from lungs and livers of mutant and wild-type mice. These cells proliferated for two to three passages in culture, and each population maintained their differentiated phenotype based on CD31 expression (), positive staining for von Willebrand Factor, and uptake of DiI-labeled acetylated low-density lipoprotein (). Both mutant and wild-type endothelial cells grew at comparable rates in culture (in medium supplemented with heparin, endothelial cell growth supplement, and 20% FBS; ). Omission of heparin from the growth medium mildly decreased growth of both mutant and wild-type cells. Thus, isolated cells showed no obvious effect of Ndst1 deficiency on proliferation in monolayer culture.
allele in endothelial cells from
mice was >95%, as measured by quantitative PCR, consistent with values obtained by Southern blotting (). Examination of expression of expression by quantitative RT-PCR showed >95% reduction in message in mutant cells, without any significant change in expression of (unpublished data). Analysis of disaccharides liberated from heparan sulfate by heparinase degradation showed a reduction of -sulfated units, leading to an overall 2.3-fold reduction in glucosamine -sulfate (). Residual -sulfation presumably arose from the action of (). Because of the dependence of downstream -sulfotransferases on -sulfation (), the degree of glucosamine 6--sulfation and uronyl-2--sulfation of chains was reduced as well (). The change in heparan sulfate structure led to a decrease in cell-surface binding of FGF-2 and VEGF to mutant endothelia compared with cells derived from wild-type animals (). In contrast, binding of EGF, which does not require heparan sulfate, was unaffected by the mutation. Binding of FGF-2 and VEGF to S-radiolabeled heparan sulfate isolated from the mutant endothelia was reduced as well ().
Under standard culture conditions, wild-type and mutant primary endothelia proliferated at similar rates and to the same extent (), but striking differential growth responses of cells were observed when cells were cultivated on Matrigel and supplemented with FGF-2 or VEGF. Wild-type endothelia proliferated and formed extensive branched networks in response to 1–10 ng/ml FGF-2, whereas mutant cells were unresponsive (). Increasing the concentration of FGF-2 to values in excess of the physiological range (100 ng/ml) was sufficient to stimulate a sprouting response by mutant cells (, right), indicating that they expressed functional FGF receptors that were not activated at lower doses of FGF-2. Treating wild-type cells with heparinase, which degrades heparan sulfate, mimicked the phenotype of the mutant (). The addition of unfractionated heparin also blocked the response, whereas it had no effect on mutant cells (even at low concentration). Mutant cells also failed to respond to VEGF, whereas wild-type cells sprouted robustly in response to this factor ().
Cre mice with Adenoviral-Cre yielded comparable results (unpublished data), indicating that the effect was cell autonomous and not the result of selection of unresponsive cells during their cultivation.
When mutant cells failed to sprout on Matrigel, the cells remained predominantly clustered and harbored frequent apoptotic cells (), consistent with the idea that the alteration in heparan sulfate led to decreased growth responses through the relevant receptor tyrosine kinases (FGFR1, FGFR2, and VEGFR2). Western blotting studies with antibodies showed no change in receptor expression (unpublished data). Addition of FGF-2 to wild-type endothelia resulted in marked phosphorylation of Erk1/2 after 10 min of stimulation, and the signal was sustained for 60 min (, top). Mutant endothelial cells, however, exhibited about a threefold reduction in signaling at both time points. VEGF-dependent Erk1/2 phosphorylation was even more dramatically affected. Wild-type cells responded robustly by 10 min, and the signal declined by 60 min. In contrast, mutant cells were essentially unresponsive to VEGF (, bottom). Treatment of wild-type cells with heparinase caused similar effects (unpublished data). FGF-2 and VEGF had only a modest effect of phosphorylation of PKB/Akt in either wild-type or mutant cells ().
To determine whether the altered responses of endothelial cells in vitro corresponded to changes in growth factor binding in vivo, the distribution of FGF-2 and VEGF was measured in tumor sections. Immunohistochemical analysis of endogenous FGF-2 in tumors showed diffuse staining throughout the tissue regardless of genotype (unpublished data). Incubation of samples with biotinylated FGF-2 also showed a diffuse distribution of binding sites throughout the tumor stroma. However, close inspection of the vasculature revealed intense staining in the perivascular zones in tumors from wild-type mice, a region rich in basement membrane–type heparan sulfate proteoglycans (, solid arrowheads; ). In contrast, tumors from mutant animals showed a marked reduction in staining (, open arrowheads), consistent with decreased binding to endothelial cell–derived heparan sulfate ().
In contrast to endogenous FGF-2, we were able to examine the localization of endogenous VEGF in the tumor microvasculature using GV39M mAb, which binds to complexes between VEGF-A and its major signaling receptor VEGFR-2/flk-1 (; ). Immunofluorescence of tumor sections from wild-type mice demonstrated a subset of vessels that stained strongly (, top). Many of these vessels also stained for laminin, consistent with studies that showed VEGF in association with basement membranes in some vascular beds (Brekken, R., personal communication). Vessel-associated GV39M was localized in a predominantly ablumenal orientation relative to CD31 (), supporting the idea that endothelial VEGFR-2 engages VEGF from the stromal side of the vessel (; ). Mutant tumor sections stained less strongly with GV39M mAb, especially in laminin-associated vessels (, bottom). Interestingly, the distribution of laminin, which also binds heparan sulfate, was not grossly affected. Pericyte investment, as detected by mAb against smooth muscle cell actin, also was not significantly affected by the mutation (). Combined CD31/TUNEL staining of tumor sections from mutant mice revealed several apoptotic nuclei associated with walls of the microvasculature that were not present in wild-type tumors (), consistent with the in vitro data presented in .
Because the Ndst1 deficiency affected vessel-associated VEGF–VEGFR-2 complexes (), we examined whether tumor growth depended on VEGFR-2 expression. When LLC tumors were grown in VEGFR-2 heterozygous mice, mean tumor volumes were reduced by ∼50% compared with the wild type (P < 0.0001; ). CD31-stained tumor sections from VEGFR-2 mutants also exhibited reduced microvascular density (33% reduction relative to wild type; P < 0.03) and increased vessel caliber () similar to that described for tumors grown in
mice ().
We have demonstrated that altering Ndst1 expression in endothelial cells results in decreased tumor angiogenesis and tumor growth. Vascular changes in the tumor included decreased vessel density and branching and increased caliber of the remaining vessels. These changes correlated with diminished endothelial process formation in vitro in response to FGF-2 and VEGF, altered growth factor binding to isolated endothelial cells and heparan sulfate, and attenuation of Erk1/2 signaling. Altering endothelial heparan sulfate decreased the association of FGF-2 and VEGF with the tumor vasculature in vivo and increased endothelial apoptosis in both branching assays as well as the tumor vasculature. These findings correlate changes in endothelial cell growth mediated by FGF-2 and VEGF to decreased vascularization of the tumor, which in turn diminished tumor growth.
Recently, showed that VEGF signaling in endothelial cells and angiogenesis were fully supported by heparan sulfate expressed in trans by adjacent perivascular smooth muscle cells.
mice and the fact that pericyte density was unaltered (), it would appear that pericyte heparan sulfate will not substitute for endothelial heparan sulfate during tumor angiogenesis. Thus, in the context of tumor vasculature, endothelial heparan sulfate behaves in a cell-autonomous manner. Ideally one would like to further potentiate the phenotype by more dramatic changes in heparan sulfate composition or content. However, complete ablation of sulfation (
) or inhibition of the polymerase that catalyzes formation of the heparan sulfate chains () result in early developmental defects, including failure to differentiate mesodermal precursors of the endothelium (; ; ), thus preventing further studies of the vasculature in adult animals.
mice based on normal birth weight, growth of the animals, reproductive capacity, and normal cutaneous wound healing (). Nevertheless, endothelial cells derived from the lungs of mutant animals showed alterations in growth factor binding, signaling, and sprouting behavior under conditions that promote proliferation in vitro (–
). Why, then, was angiogenesis selectively altered in the tumor environment but not under conditions of cutaneous wound repair? One possibility is that the tumor microenvironment places an especially high demand on the tumor vascular bed with respect to transfer of nutrients, removal of waste products, and gas exchange. If correct, then further stress, e.g., by hypoxia or induced tissue regeneration, might reveal underlying differences in angiogenic capacity in normal tissues. Alternatively, the composition of growth factors might differ in the wound environment, or other glycosaminoglycans, such as dermatan sulfate, which is abundant in wound tissue, might compensate for reduced heparan sulfate in the endothelium (; ). Further studies are needed to address this point.
As observed in other tumor models, the microvasculature in LLC tumors displayed a highly chaotic morphology compared with normal vasculature (; ). Altering heparan sulfate specifically in the endothelium resulted in less branched and tortuous tumor vessels and decreased overall density of the neovasculature (). Similar morphological changes have been observed in tumor microvasculature after in vivo treatment with anti-VEGF antibodies or VEGF receptor tyrosine kinase inhibitors (; ), suggesting that altering heparan sulfate in the endothelium may preferentially affect VEGF-mediated responses. Other observations that support this idea include the following: transfection of VEGF nullizygous tumor cells with VEGF rescues tumor growth and vascular density/branching, whereas altered tumor growth and a hypobranched/hypodense vasculature results from transfection with the non–heparin binding isoform, VEGF (); mouse embryos engineered to express only VEGF exhibit a marked decrease in capillary branching accompanied by increased vascular caliber (); altering Ndst1 expression reduces the formation of VEGF–VEGFR complexes in mutant tumor vasculature (); and reduction of expression of in heterozygous mutants resulted in reduced tumor growth and vascularization (). The modest dilatation of mutant tumor microvasculature that we observed could reflect incomplete inactivation of VEGF signaling, as the deficiency in Ndst1, although nearly fully penetrant, does not completely reduce sulfation of the heparan sulfate chains ().
The potentiation of growth factor action by heparan sulfate proteoglycans is thought to depend on formation of ternary complexes in which the heparan sulfate chain acts as a scaffold to approximate ligand and receptor, thus affecting signaling intensity and duration (). In the extracellular matrix, heparan sulfate proteoglycans also act as reservoir for growth factors (), providing stability and enhancing the capacity for gradient formation (). Many tumors secrete heparanase, which can degrade the extracellular matrix and liberate bound growth factors (; ), and overexpression of heparanase in tumor cells results in enhanced angiogenesis and tumor growth (). Because endothelial cells produce matrix proteoglycans, loss of Ndst1 activity could, in theory, result in reduced storage of growth factors in the surrounding matrix.
mice could reflect a combination of cell-autonomous growth factor signaling defects as well as reduced availability of growth factors in the matrix.
Heparan sulfate chains do not exist as free polysaccharides but, rather, occur covalently bound to proteoglycan core proteins. The major heparan sulfate proteoglycans in endothelial cells consist of the glycosylphosphatidylinositol-linked glypicans, membrane-spanning syndecans, and secreted proteoglycans (perlecan, agrin, and collagen XVIII). Some of these have roles in angiogenesis based on mutational studies. Mutants defective in syndecan-1 show an increase in vessel length during corneal angiogenesis, possibly because of differences in release of inflammatory mediators (), whereas overexpression of syndecan-1 causes abnormal angiogenesis during cutaneous wound repair (). In contrast, syndecan-4 deficiency impairs fetal vessels in the placental labyrinth () and angiogenesis during wound repair (). Deletion of exon 3 of perlecan core protein, which bears a major heparan sulfate containing domain, impairs wound angiogenesis as well as tumor angiogenesis (). Mice lacking collagen XVIII show delayed regression of blood vessels postnatally in the vitreous along the surface of the retina and abnormal outgrowth of retinal vessels () but enhanced intimal neovascularization in atherosclerotic aortas (). The variation in phenotype in mutants and tissues may reflect differences in the expression of individual proteoglycans in diverse vascular beds, other functional domains within the core proteins (; ), or changes in heparan sulfate in the supporting tissue in systemic mutants rather than cell-autonomous effects on endothelial cells. Additional studies are needed to analyze the effect of altering specific proteoglycans selectively in endothelial cells.
Our findings validate heparan sulfate and the enzymes involved in its biosynthesis as potential targets for chemotherapeutic intervention. Targeting heparan sulfate could prove advantageous over strategies that focus on individual proangiogenic growth factors, such as VEGF (e.g., bevacizumab; ), as heparan sulfate can modulate the activity of multiple angiogenic factors (e.g., hepatocyte growth factor, PDGF, heparin-binding EGF, angiopoietin, tumor necrosis factor-α, interleukin-8, and others; ; ), most of which bind to heparan sulfate. Previous studies have shown that tumor cell proliferation also depends on heparan sulfate (for review see ). Because heparan sulfate biosynthesis in endothelial cells and tumor cells utilizes a common set of enzymes, agents that target heparan sulfate would provide a two-prong approach for blocking both tumor cell division and formation of the supporting vasculature.
LLC LL/2 (CRL-1642; American Type Culture Collection) was grown in DME (Invitrogen) containing 1.5 g/l sodium bicarbonate and supplemented with 10% FBS (HyClone). B16BL6 murine melanoma cell line was a gift from I. Fidler (University of Texas MD Anderson Cancer Center, Houston, TX) and was grown in MEM (Invitrogen) and supplemented with 7% FBS. Murine lung microvascular endothelial cells were isolated from mice essentially as described previously (), except after straining, cells were mixed with 4.8 ml 17% Nycodenz (Sigma-Aldrich) in Gey's balanced salt solution, overlaid with 1 ml PBS (), and centrifuged at 1,400 , and the endothelial rich band was collected. Cells were then labeled with rat anti–mouse CD31 (0.2 μg/10 cells; Caltag), incubated with goat anti–rat antibody conjugated to magnetic beads, and passed over a magnetic column (Miltenyi Biotec) followed by washing, and purified endothelia were eluted from the column into gelatin-coated plates. Cells were maintained one or two passages in DME supplemented with 20% FBS, 100 μg/ml endothelial growth supplement (Sigma-Aldrich), 100 μg/ml heparin (Scientific Protein Laboratories), and nonessential amino acids (Invitrogen). Culture media were supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin. All cells were maintained under an atmosphere of 5% CO and 100% humidity at 37°C. Cell proliferation was examined using the Cell Titer-Blue viability assay (Promega) on primary endothelia. Cells were plated at 5,000 cells/well in 96-well plates, grown in complete medium, and assayed on successive days for fluorescence (excitation 488 nm/emission 575 nm; CytoFluor II fluorimeter) after addition of Cell Titer-Blue reagent according to the manufacturer's instructions. Fluorescence is proportional to the number of viable cells.
C57Bl/6 mice harboring a conditional mutation in exon 2 of (
) were crossed with transgenic mice expressing Cre recombinase under control of the (Tie2 receptor) promoter/enhancer (
; ; ).
males were mated with
females generating equal numbers of mutant and wild-type animals. Genotyping of mice was performed by PCR using genomic DNA from tail biopsies as described previously (). Mice heterozygous for VEGFR-2 (
) were purchased from The Jackson Laboratory, with genotyping performed using primers as described previously (). All animals were fully backcrossed on C57Bl/6 background (>10 generations). Mice were housed under barrier conditions in AAALAC-approved vivaria, following standards and procedures approved by the local Institutional Animal Care and Use Committee. They were weaned at 3 wk, maintained on a 12-h light–dark cycle, and fed water and standard rodent chow .
To assess the efficiency of inactivation, quantitative real-time PCR (Applied Biosystems) on genomic DNA was performed using primers flanking the downstream loxP recombination site of the floxed construct (; = 3 per group). Cycle threshold values from duplicate assays were used to determine the initial starting quantity of loxP DNA from mutant versus wild-type genomic DNA samples.
To examine endothelial mRNA, total RNA was isolated from mutant versus wild-type primary lung endothelia, reverse transcribed (Superscript III; Invitrogen), and amplified using gene-specific intron-spanning primers (primer sequences: forward, GGACATCTGGTCTAAG, and reverse, GATGCCTTTGTGATAG; forward, TGGTCCAAGGAGAAAACCTG, and reverse, GGTACGACCTCCGAGTCAAA; forward, CCACTGCCTTGTGTC, and reverse, GGAGTACGCTCGGTC; forward, CTAACTACTTCCACTC, and reverse, ATGTGCACTGCATACC). Cycle threshold values from triplicate assays were used to calculate fold expression compared with expression of β-actin in the same sample.
Tumor cells were harvested with Trypsin/EDTA (Invitrogen) and were inoculated subcutaneously at 4 × 10 cells per mouse in 50 μl DME into the left hindquarters of 10–12 wk-old mice. Tumor growth was measured using calipers over a 15-d period, and tumor volume was estimated using the following formula: vol = length × (width)/2. Mice were killed after 15 d (beyond which tumor ulceration became frequent), and their tumors were dissected, measured directly with calipers, and removed for cryo- and formalin fixation for further histologic analysis. Mouse bone marrow transplantation studies were performed as previously described (). In a pilot experiment, all nontransplanted irradiated mice died within 1 wk. Conversely, all mice that underwent marrow transplantation survived until termination of the tumor experiment without ill appearance.
In situ microvascular FGF-2 reactivity was assessed on frozen tissue sections by incubating specimens with biotinylated FGF-2 (), and bound factor was detected with HRP-conjugated streptavidin (1:500; Jackson ImmunoResearch Laboratories). For detection of VEGF–VEGFR2 complexes, acetone-fixed frozen sections were labeled with biotinylated GV39M antibody (1:50; East Coast Biologics) followed by treatment with Cy3-conjugated streptavidin (1:500; Jackson ImmunoResearch Laboratories). Laminin in the same sections was visualized by treatment with rabbit anti-laminin antibody (1:500; DakoCytomation) followed by FITC-labeled goat anti–rabbit antibody (1:50; Chemicon).
LLC LL/2 cells were resuspended in 10 ml complete medium at (2 × 10 cells/ml) and were rocked in a 50-ml silicon-coated flask on a gyratory shaker in culture. After 48–72 h, tumor spheroids of similar diameters (600–1,000 μm) were selected, washed in serum-free medium, and individually placed into dorsal skinfold chambers using a sterile plastic pipette as previously described (). Observations of tumor angiogenesis were made day 12 after tumor spheroid implantation in mutant and wild-type mice as described previously (). Unanesthetized mice were placed in a plexiglass tube containing a longitudinal slit from which the chamber projected for exposure to the microscope. The tube was immobilized on a plexiglass platform, which was placed on the stage of an intravital microscope (Biomed; Leitz) for viewing the tumor spheroid microcirculation. Images for all mice were recorded at room temperature through a silicone intensified tube camera (SIT68; DAGE-MTI) attached to the microscope (Nikon) using 4× or 10× water-immersion objectives. The camera was connected to a monitor (Panasonic) and an S-VHS video recorder (HC-6600; JVC). For each chamber, the tumor spheroid vasculature was identified using a 4× objective, and regions of highest tumor vascular density were recorded with the 10× objective. Still images of four to six tumor spheroid fields per chamber were recorded at 100× magnification and scanned for grid intersection using a 12 × 12 grid overlay that overlies the image, and branch points per field were measured using Photoshop CS2. Digitized images were also analyzed on Volocity 2.6.1 software (Improvision), which was used to obtain total linear (skeletal) length of vasculature per field, as well as mean vascular caliber per field.
Under general anesthesia, mice were wounded using a 3-mm-diameter full-thickness dorsal skin punch biopsy. Wounds were then photographed daily over 4 d after injury, and wound area was measured using ImageJ software. The wounds were biopsied to include the margin to normal surrounding skin and snap-frozen in Tissue-Tek OCT (Sakura), and median sections were taken to include the wound fibrin plug at center, flanked by the wound granulation tissue and surrounding normal dermis/epidermis. Acetone-fixed sections were incubated with biotinylated anti–mouse CD31 (1:100; BD Biosciences), followed by HRP-conjugated streptavidin (1:500; Jackson ImmunoResearch Laboratories). Wound microvascular density for any given mouse was then determined by calculating the mean number of CD31+ processes per 400× field using four fields surrounding the wound center.
Primary lung endothelial cells were harvested at the first or second passage using 5 mM EDTA in PBS. The cells were treated with heparin (1 mg/ml in PBS), washed with PBS, and incubated with biotinylated 0.6 μg/ml FGF-2 in buffer (0.5% BSA/2 mM EDTA in PBS) at 4°C for 1 h with gentle stirring on an orbital shaker (). Bound FGF-2 was detected by flow cytometry using streptavidin-phycoerythrin. A set of cells was exposed only to streptavidin-phycoerythrin as a negative control. To detect VEGF binding, cells (4 × 10/ml) were incubated for 1 h at 4°C with recombinant human biotinylated VEGF (1.1 μg/ml in PBS; R&D Systems). Bound VEGF was detected with avidin-FITC reagent (1:1,000; R&D Systems) and flow cytometry. As controls, some cells were incubated with an irrelevant biotinylated protein (soybean trypsin inhibitor) or Alexa Fluor 488–conjugated EGF (Invitrogen; each at 8 μg/ml for 45 min at 4°C) followed by flow cytometry.
Primary lung endothelia were labeled for 48 h with 37 μCi/ml H
SO (655 Ci/mmol; NEN Life Science Products) in reduced-sulfate medium (Ham's F12 containing 10% dialyzed, filter-sterilized FBS). Isolation of purified heparan [S]sulfate from cultured cells was performed essentially as described previously (). Binding of 0.2 μg FGF-2 and 1 μg VEGF was measured using a nitrocellulose filter binding assay using 5,000 cpm per assay (). Controls included competition with 100 μg/ml heparin or use of bovine serum albumin in place of growth factor.
[H]Heparan sulfate was prepared from primary endothelial cells by culturing them in growth medium containing [6-H]glucosamine and 1 mM reduced glucose. Radiolabeled chains were degraded with heparinases and analyzed by anion-exchange HPLC as described previously ().
Murine lung endothelial cells were isolated from 8–12-wk-old mutant mice or their wild-type littermates as described previously (). Purity was assessed by flow cytometry using CD31 mAb (FACSCalibur; BD Biosciences), uptake of DiI-acetylated-LDL (10 μg/ml; Biomedical Technologies), and reaction with polyclonal antibody to von Willebrand factor (DakoCytomation).
Matrigel (BD Biosciences) was added to each well of a 96-well plate and allowed to polymerize at 37°C for 1 h. Mutant or wild-type cells at first or second passage were harvested with trypsin/EDTA (BioWhittaker) and resuspended in DME/20% FBS. Cells (5,000/well) were then added to duplicate wells in 100 μl of growth medium in the presence or absence of recombinant FGF-2 (Invitrogen) or murine VEGF (Calbiochem). In a separate experiment, additional wells were seeded in the presence or absence of heparinase at 0.3 mU/ml (heparin lyase III; Seikagaku) and 10 ng/ml FGF-2. After 48 h, the degree of endothelial sprouting over the gel surface was measured by determining the net lengths of endothelial processes (in duplicate wells) viewed under phase-contrast light microscopy. The data is reported as fold stimulation relative to baseline growth without added growth factor. In some experiments, 100 μg/ml of unfractionated heparin (Scientific Protein Laboratories) was added to the wells during the sprouting period.
Primary lung endothelial cells were seeded at equal density on gelatin-coated 12-well culture dishes immediately after harvest from the animals and allowed to grow to near confluence under standard culture conditions. After incubation for 4 h in serum-free DME without supplements, cells were exposed to FGF-2 or VEGF (10 ng/ml) for 10–60 min. Samples were solubilized and mixed with an equal volume of 125 mM Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, 100 mM dithiothreitol, and 0.02% bromophenol blue. They were separated by electrophoresis in 4–15% gradient gels followed by electrotransfer to a nitrocellulose membrane, which was then probed using a polyclonal antibody (1:1,000; Cell Signaling Technology) that detects phosphorylated Thr/Tyr forms of p44/42 MAPKs (Erk1/2). Bound antibody was detected using peroxidase-conjugated anti–rabbit antibody (1:2,000; Bio-Rad Laboratories) followed by visualization using SuperSignal chemiluminescent substrate for HRP detection (Pierce Chemical Co.). To ensure equal protein loading, membranes were stripped and reprobed using polyclonal antibodies against total Erk1/2. In a separate experiment, a subset of wells was treated with 0.5 mU/ml heparinase during the periods of serum starvation and growth factor stimulation. For Akt signaling, membranes were probed using 1:1,000 anti-phospho (Ser)-Akt monoclonal antibody (Cell Signaling Technology). The bands were quantified by measurement of band intensity relative to background on a densitometer (GS800; Bio-Rad Laboratories).
Cre primary murine lung endothelia with Adenoviral-Cre (Ad-Cre; UCSD Gene Therapy Core) at 10 pfu/ml for 1.5 h in normal culture growth medium. Infection with Adenoviral-GFP (Ad-GFP) was used as a control. Cells were added to Matrigel-coated wells in chamber slides. Permeabilization and TUNEL labeling were performed directly on chamber slides according to the manufacturer's instructions (In Situ Cell Death Detection kit; Roche), and samples were stained with Vector Blue Alkaline Phosphatase Substrate kit (Vector Laboratories) and nuclear fast red stain.
Mean values of tumor volume, microvascular density on tumor sections or wound sections, microvascular skeletal length, caliber, grid intersection, and branching, and mean relative sprouting responses on extracellular matrix, were compared using one-tailed tests. P < 0.05 was considered significant. Comparison of wound contraction rates in mutant versus wild-type mice was performed using a two-factor repeated-measures analysis of variance, testing for group differences in the change over time (SPSS Software v. 13.0). |
A detailed molecular understanding of the mechanisms by which cells migrate is important not only to our view of normal physiological processes, such as embryonic development and wound repair, but also to our ability to intervene in the progression of inflammatory disease and cancer. The integrin family of heterodimeric matrix receptors plays a central role in normal and pathophysiological modes of cell migration by acting not only to physically couple cells to the ECM but also to function as signaling molecules that transmit information across the plasma membrane (). Of the numerous intracellular signaling events that are triggered by integrin engagement, perhaps the most pertinent to cell migration is their capacity to influence cytoskeletal dynamics via the activation of Rho subfamily GTPases (; ). Indeed, cells can use different migrational modes to move with varying degrees of speed and directionality depending on the nature of Rho GTPase signaling downstream of integrins. For instance, metastasizing tumor cells often move randomly and rapidly undergo amoeboid shape changes, and this depends on the ability of β1 integrins to activate Rho kinase (ROCK) via the small GTPase, RhoA (). Alternatively, during processes such as wound healing, fibroblasts migrate directionally and with high persistence (i.e., the tendency to continue traveling in the same direction without turning), and this can be determined by the degree of Rac signaling downstream of α5β1 integrin ().
To an extent, patterns of migratory behavior are dictated by characteristics that are intrinsic to particular cell types. However, both normal cells and those derived from tumors can switch between different modes of migration, and signaling pathways activated downstream of integrins can contribute to this. For instance, epithelial cells expressing αvβ3 integrin migrate persistently, but the same cells migrate randomly upon expression of the α5β1 heterodimer (). This is a consequence of the ability of α5β1 to activate ROCK, which in turn phosphorylates and inhibits the actin-severing protein cofilin.
Several integrins engage in endo–exocytic cycling, and many of the Rab GTPases and kinases that control their return to the plasma membrane are now becoming clear (; ). α5β1 integrin recycles to the plasma membrane from a perinuclear recycling compartment via a “long-loop” pathway requiring Rab11 and activity of the PKB/GSK-3β axis (). Conversely, αvβ3 integrin travels more rapidly back to the cell surface via a “short loop” that is controlled by Rab4 and requires association of protein kinase D1 (PKD1) with the integrin (). Receptors for growth factors and chemokines are also endocytosed and then recycled back to the cell surface, and it is now clear that this process influences the way they signal (). Indeed, many receptors remain competent to signal in endosomal compartments, and recycling pathways can resensitize receptors to prolong signaling outputs, as is the case for CXCRs () and the β-adrenergic receptor (). Furthermore, a recent study has suggested that recycling acts to constantly retarget internalized receptor tyrosine kinases to the leading edge, thus keeping downstream signaling localized during the directional migration of border cells ().
It has been proposed that receptor recycling pathways act to transport integrins forward during cell migration (). Indeed, the localization of αvβ3 integrin to focal complexes at the front of migrating cells is dependent on the short-loop pathway (; ), but how this contributes to migration is not yet clear. It is possible that anterograde vesicular transport could contribute directly to persistent migration by constantly retargeting integrins to the leading edge, thus reinforcing the cell's polarity axis. Alternatively, trafficking may influence migrational modes by altering integrin signaling. The precise mechanistic link between integrins and Rho signaling is as yet undefined, and endosomal recycling pathways provide an interesting means of reconciling the respective localizations of integrins and their Rho signaling counterparts.
To resolve these issues, we have developed a strategy to target short-loop αvβ3 recycling and have precisely determined its contribution to the speed and persistence of cell migration. Indeed, we find that short-loop recycling has a profound effect on the persistence of migration. This is not, however, because of its ability to transport αvβ3 forward during cell migration but, rather, because it can antagonize α5β1 recycling and the signaling of this integrin to cofilin. Thus, we have revealed that vesicular transport makes a major contribution to cell migration via its capacity to dictate the nature of downstream integrin signaling, which in turn influences the migrational mode of fibroblasts.
We previously found that PKD1 can associate specifically with αvβ3 integrin via a motif contained within the C-terminal 14 amino acids of the β3 cytodomain (). This association recruits PKD1 to αvβ3 at endosomes and drives the rapid return of the heterodimer to the plasma membrane in response to growth factor treatment. To further characterize this integrin–kinase interaction, we expressed His-tagged PKD1 in Cos-1 cells and purified the kinase to near homogeneity by Ni-affinity chromatography (). To ensure a preparation of maximally active kinase, Cos-1 cells were treated with phorbol myristate acetate (PMA) for 30 min before lysis in the presence of phosphatase inhibitors. Purified PKD1 bound directly and with high affinity (
[apparent] in the low nanomolar range) to GST-β3 integrin cytodomain (). There was no detectable association between purified active PKD1 and the cytoplasmic sequences of the αv, α5, or β1 integrin subunits, indicating that the interaction was specific for β3 integrin.
After activation with growth factors, cellular PKD1 is phosphorylated at several residues (). As we have previously shown that treatment with a growth factor such as PDGF is necessary for coimmunoprecipitation of αvβ3 and PKD1 (), we sought to determine whether phosphorylation was necessary for integrin binding. Indeed, treatment of PKD1 with alkaline phosphatase (which led to ∼80% dephosphorylation of the kinase [not depicted]) reduced the affinity of integrin–kinase association by approximately fivefold (). PKD1 is auto- (and possibly trans-) phosphorylated at Ser in its C terminus (; ; ), but no clear cellular role for this has been described. We therefore mutated Ser of PKD1 to alanine and determined the ability of this mutant kinase to bind to αvβ3 integrin. Indeed, purified PKD1 had strikingly reduced ability to bind to GST fusion proteins of the β3 integrin cytodomain (). Moreover, when expressed in fibroblasts, PKD1 did not coimmunoprecipitate with αvβ3 (), indicating that autophosphorylation of this residue is a prerequisite for integrin–kinase association. In agreement with a previous report (), we found that mutation of Ser had no influence on the PKD1 activity, as determined by the ability of purified PKD1 to phosphorylate one of its best-characterized substrates, the N-terminal portion of c-Jun (; ).
Suppression of cellular PKD1 levels by RNAi, expression of catalytically inactive PKD1s, and/or mutant β3 subunits that cannot bind to PKD1 oppose short-loop recycling of αvβ3 (). However, these strategies will be likely to compromise other aspects of PKD1 and integrin signaling, such as the recruitment of c-Src to αvβ3 () and the role of PKD1 in Golgi transport (). With this in mind, we determined the influence of PKD1 on integrin recycling via the short-loop pathway but also quantified other indices of integrin, PKD1, and endocytic function. Short-loop αvβ3 recycling was driven by the addition of growth factors such as PDGF and lysophosphatidic acid and by the addition of 10% serum (all of which lead to PKD1 activation) to serum-starved cells (). However, after expression of PKD1, these agents were unable to drive the delivery of αvβ3 to the plasma membrane, indicating that this PKD1 mutant acts in a dominant-negative fashion to oppose growth factor–driven short-loop integrin recycling (). Moreover, PKD1 did not inhibit the return of integrins to the plasma membrane via the long loop (), the recycling of internalized [I]Tfn (), or the endocytosis of integrins and the Tfn-R (not depicted), indicating that this mutant PKD1 selectively targets short-loop αvβ3 recycling.
To gain information as to how PKD1 exerts this dominant-negative effect on αvβ3 recycling, we overexpressed His-tagged PKDs and measured activation of the endogenous kinase using a reporter antibody recognizing activating phosphorylations within the kinase domain of PKD1 (phospho-Ser). Indeed, expression of His-PKD1 or His-PKD strongly suppressed phosphorylation of the endogenous kinase at Ser (Fig. S1, available at ), indicating that these overexpressed recombinant kinases can compete effectively with endogenous PKD1 for the upstream activating kinase (PKC) that phosphorylates these residues, thus providing a mechanistic rationale for the dominant-negative influence of PKD1 on integrin αvβ3 recycling and function.
To assess TGN to plasma membrane transport, we used vesicular stomatitis virus G protein (VSVG) from the temperature-sensitive (ts) 045 mutant of vesicular stomatitis virus, which is misfolded and retained in the ER at 40°C but moves out of the ER, though the Golgi complex and to the plasma membrane upon temperature shift to 32°C (). ts045 VSVG appeared at the plasma membrane over a time course of ∼2 h after shift to 32°C and, consistent with the previously established role of PKD1 in TGN to plasma membrane transport (), this was markedly reduced by expression of a short hairpin RNA (shRNA) targeting PKD1 (). However, expression of PKD1 did not suppress delivery of VSVG to the plasma membrane (), indicating that although this mutant kinase completely ablated growth factor–driven αvβ3 recycling (), it did not compromise PKD1's action at the TGN. Collectively, these data highlight the potential effectiveness of PKD1 as a molecular tool, not only to enable comparison of the respective roles played by PKD1 in Golgi transport and integrin recycling but, more particularly, to determine the contribution made by short-loop αvβ3 recycling to cell migration.
We () and others () previously determined that suppression of PKD1 leads to reduced cell migration and an impaired ability of migrating fibroblasts to establish their characteristic fan-like morphology. To determine the precise contribution of short-loop αvβ3 recycling to cell migration, we manipulated αvβ3-PKD1 association and Rab4-dependent recycling in fibroblasts, collected time-lapse videos of these cells migrating into a scratch wound, and followed individual cell movement using cell-tracking software. Expression of PKD1 compromised the ability of cells to migrate directionally into the wound () and, rather than migrating with the fan-like morphology characteristic of fibroblasts (, arrow at 300-min time point), PKD1-expressing cells appeared to migrate by extending thin and often pointed protrusions (, arrow at 250-min time point). We therefore proceeded with a more in-depth analysis involving the assembly of overlays of representative trajectories described by cells during the first 5 h of their migration into the wound () and the extraction of parameters such as the persistence and speed of migration from track plots (), persistence being defined as the ratio of the vectorial distance traveled to the total path length described by the cell. Untransfected fibroblasts and those expressing wild-type PKD1, Rab4, or control shRNA migrated largely perpendicular to the wound edge and maintained a high degree of persistence (). However, suppression of short-loop αvβ3 recycling, by PKD1, RNAi of PKD1, or dominant-negative Rab4, markedly reduced persistent migration such that the cells migrated randomly for up to 5 h after wounding. It is interesting to note that expression of PKD1 or Rab4 (both of which suppress short-loop recycling of αvβ3) reduced persistence without greatly affecting the migration speed, whereas RNAi of PKD1 (which affects both integrin recycling and TGN anterograde transport) reduced both the speed and persistence of migration (). Moreover, migrational persistence was unaltered by inhibition of long-loop recycling by dominant-negative Rab11, indicating that this key parameter of cell movement relies particularly on the short-loop pathway. To further investigate the requirement for short-loop recycling in migrational persistence, we used a strategy by which endogenous levels of mouse αvβ3 integrin were reduced by expression of a shRNA targeting the mouse sequence of β3 integrin (Fig. S2, available at ), followed by expression of either the wild-type human αvβ3 heterodimer or β3 integrins with cytodomain mutations that reduce binding to PKD1. Clearly, suppression of αvβ3 levels by shRNAi profoundly reduced migrational persistence without much affecting the speed of migration (), and persistence was completely restored by expression of human αvβ3 integrin or a β3 integrin mutant (β3; ) with an unaltered ability to recruit PKD1 (). In contrast, β3 integrin mutants (β3 and β3) that cannot bind PKD1 and are consequently unable to enter the short-loop pathway () do not restore persistent migration in β3 knockdown cells (). Moreover, a similar reduction in persistence was observed after the addition of a cyclic peptide (cyclo-RGDfNV) that competitively inhibits binding of ECM ligands to αv (but not β1) integrins (; ), indicating that αvβ3 needs not only be competent to recycle via the short loop but must also engage ligand to support persistent and directional fibroblast migration.
It is now generally accepted that there is considerable interplay between cytoskeletal events directing cell polarization and the vesicular transport machinery. Indeed, the way that receptors are targeted to the plasma membrane can influence the generation and maintenance of cell polarity and vice versa. We therefore determined whether blockade of short-loop αvβ3 recycling altered the ability of cells to polarize their microtubule organizing center (MTOC) in response to wounding. Anterior orientation of the MTOC was detectable shortly after wounding, and this reached a maximum (which was largely maintained) after 2 h (). The rate at which MTOC orientation was initiated and the extent to which it was maintained was unaffected by expression of either PKD1 or Rab4 (). This clearly indicated that integrin recycling plays no role in the ability of these cells to sense the wound and polarize their microtubular cytoskeleton accordingly. Moreover, as we continued to track cell movement, it became clear that cells with compromised short-loop recycling, after having migrated randomly for ∼5 h, began to migrate persistently into the wound (). Collectively, these data indicate that although short-loop αvβ3 recycling is not required for wound sensing or the eventual acquisition of a proper migratory phenotype, it is likely to alter signaling events that influence the balance between persistent versus random migration.
Whether a cell chooses persistent over random migration likely depends on the balance between αvβ3 and α5β1 integrin signaling. It is thought that αvβ3 promotes persistent and directional migration and that this requires appropriate levels of Rac signaling downstream of this integrin (). Conversely, α5β1 tends to promote random migration, and this is a consequence of its ability to activate the Rho–ROCK–cofilin pathway (). Indeed, the increased cellular phospho-Ser-cofilin levels that result from α5β1-driven activation of Rho (and expression of a Ser-phosphomimetic mutant of cofilin) strongly suppress persistence and promote random migration (). As it is possible that the nature of signaling downstream of integrins may be dictated by their trafficking, we investigated whether the influence of the short-loop αvβ3 recycling pathway on migrational persistence could be indirectly implemented through α5β1 recycling and signaling. Indeed, manipulations that compromise the short-loop recycling of αvβ3 (such as expression of PKD1, dominant-negative Rab4, or PKD1 binding–deficient β3 integrin mutants β3 and β3) acted to increase the rate at which α5β1 was returned to the plasma membrane by at least twofold (). Conversely, overexpression of wild-type αvβ3 or a “control” β3 integrin mutant (β3) that binds to PKD1 profoundly suppressed α5β1 recycling (). Furthermore, in experiments where cells were either spread onto fibronectin for 30 min or wounded with a pipette tip and then analyzed by Western blotting, phospho-Ser-cofilin levels were markedly promoted by inhibition of αvβ3 recycling, and the use of an α5β1 function-blocking antibody (mAb16) and a ROCK inhibitor (Y27632) indicated that this increase in phosphocofilin was dependent on both α5 integrin and ROCK (). Moreover, PKD1-driven increases in phosphocofilin were only detectable up to 5 h after wounding (i.e., during the period in which cells were migrating randomly); thereafter, levels of this index of ROCK signaling were indistinguishable from that of control cells (). Collectively, these observations show a clear reciprocal relationship between short-loop αvβ3 recycling and the trafficking of α5β1 and ability of this integrin to act via ROCK to promote cofilin phosphorylation. In addition, the time course of Rho signaling downstream of α5β1 inversely correlates with migrational persistence in a way that accounts for the resumption of this mode of migration at later times after monolayer wounding (compare and with ).
Given these relationships, we sought to directly determine whether the loss of persistence resulting from inhibition of the short-loop pathway was a consequence of increased α5β1 signaling. Indeed, addition of mAb16 or Y27632 restored persistent migration in cells expressing PKD1, shRNAs targeting PKD1, dominant-negative Rab4, or PKD1 binding–deficient β3 integrin mutants (). Moreover, persistent migration was partially restored by inhibition of α5β1 or ROCK signaling in cells treated with cyclo-RGDfNmeV to block the interaction of αvβ3 with its ECM ligands (). These data clearly show that the requirement for αvβ3 short-loop recycling (and its ligand engagement) in persistent migration is neither direct nor absolute but is mediated via the ability of this pathway (when active) to antagonize α5β1 integrin recycling and subsequent signaling to the ROCK cofilin pathway (). Thus, when the αvβ3 short loop is blocked, the resulting deregulation of α5β1 recycling and signaling promotes random migration in favor of persistence ().
PKD1 is thought to promote the fission of vesicles emanating from the TGN and thus enhance the transport of Golgi-derived cargo to the plasma membrane (). However, this has been controversial because of difficulties in demonstrating localization of PKD1 to the TGN () and the fact that evidence supporting a role for the kinase in Golgi to plasma membrane transport has come primarily from the use of kinase-dead mutants. Despite these caveats, the recent identification of the Golgi-localized phosphatidylinositol 4-kinase IIIβ as a physiological PKD substrate () and our observation that RNAi of PKD1 suppresses plasma membrane delivery of VSVG protein clearly support a role for this kinase in Golgi to plasma membrane transport. Furthermore, observations of TGN-derived vesicles being transported toward the lamellipodium as fibroblasts migrate (; ) demand that a contribution of PKD1-regulated Golgi transport to cell migration must be considered. Here, we show that the functional significance of autophosphorylation of PKD1 at Ser is to regulate direct interaction with αvβ3 and thus influence integrin recycling. However, as phosphorylation at this residue has no detectable effect on the transport of VSVG protein from the TGN to the plasma membrane, we have been able to use PKD1 as a molecular tool to address key questions concerning the relative contributions of PKD1-regulated Golgi transport and integrin recycling to cell migration. First, differential effects of PKD1 on αvβ3 recycling and VSVG transport indicate that this integrin is unlikely to return to the plasma membrane via the TGN (as is the case for certain recycling proteins). Second, suppression of PKD1 activity by expression of kinase-dead PKD1 or by RNAi influences both speed and directionality, whereas expression of PKD1 selectively targets migrational persistence. This indicates that PKD1 controls αvβ3 recycling to influence directionality, with PKD1-regulated Golgi traffic acting to additionally enhance the migration speed of fibroblasts.
Although the surface distribution of αvβ3 in migrating fibroblasts is polarized toward the cell front (), Rab4 is tightly localized to endosomes in the juxtanuclear region that face the direction of travel (unpublished data). Thus, the relevant matrix receptors and the endosomes that traffic them are distributed along the lamellipodial–perinuclear axis of the migrating cell. Moreover, this level of organization depends on flux of αvβ3 through the short loop, as expression of PKD1 or PKD1 binding–deficient β3 integrins dissipates the polarized distribution of surface αvβ3 (; unpublished data) and delocalizes Rab4 endosomes from the anterior perinuclear zone (unpublished data). Given these observations, it is tempting to suggest that the short loop directly reinforces persistent migration by transporting αvβ3 to and from the lamellipodium along the axis of polarity. However, inhibition of α5β1 signaling in cells with compromised αvβ3 short-loop recycling enables persistent migration despite a lack of proper polarization of αvβ3 and Rab4. Therefore, although the short loop may indeed transport αvβ3 toward the leading edge, this process is not an absolute requirement for persistent migration when the α5β1–ROCK–cofilin pathway is down-regulated. In addition to generating polarized surface distributions and restricting signaling spatially (), endocytosis/recycling can oppose receptor desensitization (), in part by acting to clear occupied receptors of ligand and returning them to the plasma membrane competent to bind fresh ligand. As our data indicate that αvβ3 needs to be both rapidly cycling and competent to engage ligand to promote persistent migration, it is probable that short-loop recycling acts to continuously resensitize αvβ3 to ligand occupation, thus maintaining sufficient αvβ3 downstream signaling to tonically inhibit α5β1 recycling.
Epithelial cells expressing αvβ3 (and not α5β1) migrate persistently, and the appropriate activation of Rac by this integrin is likely to be key to this process (). Conversely, if cells express α5β1 (and not αvβ3), they migrate randomly because of activation of the ROCK–cofilin pathway and the antagonistic effect this has on Rac-driven stabilization of the lamellipod (). Therefore, under situations where the expression profile of fibronectin-binding integrins is biased, one is able to predict a cell's migratory behavior. However, in fibroblasts and endothelial cells, α5β1 and αvβ3 expression is closely matched and, because of the relatively small size of the intracellular pool of these integrins (∼10 and 20% of the quantity of surface integrin for αvβ3 and α5β1, respectively) and their capacity to reach the plasma membrane via more than one route, experimental manipulations that target particular integrin recycling pathways (such as those used in the present study) do not greatly alter the amount of αvβ3 or α5β1 that is expressed at the cell surface (Fig. S3, available at ). There is a clear reciprocal relationship between the rates at which αvβ3 and α5β1 recycle; i.e., blockade of αvβ3 short-loop recycling doubles the rate at which α5β1 returns to the plasma membrane via the Rab11 pathway. The mechanistic connection underlying this relationship is not mediated by alterations in PKB/GSK-3β signaling (unpublished data), but the rapidity of α5β1 recycling is closely correlated with the intensity of cofilin signaling downstream of this integrin. Thus, the way in which an integrin is handled by the recycling pathway may dictate its ability to connect with and activate Rho-signaling pathways. Furthermore, our data suggest that the contribution of recycling to migrational persistence is more easily interpreted in terms of its influence on the signaling capacity of integrins rather than processes such as vectorial transport of matrix receptors to the leading edge and their subsequent incorporation into the adhesive and migratory machine. It is now becoming more apparent that the characteristics of signaling downstream of receptor tyrosine kinases and G protein–coupled receptors depend on how they are trafficked through the endosomal and recycling pathways (). In this regard, it will be interesting to investigate a potential role for the Rab11 pathway in resensitization and prolongation of α5β1 signaling and whether recycling endosomes constitute a platform for assembly of signalosomes that include guanine nucleotide exchange factors or GTPase-activating proteins for RhoA. In addition to Rho GTPase signaling, ligation of α5β1 integrin has been linked to activation of Calmodulin-dependent protein kinase II (CamKII) in myeloid cells (). Furthermore, as ligation of αvβ3 strongly suppresses the ability of α5β1 to communicate with CamKII, the possibility that this and other examples of integrin “cross-talk” involve alterations in the endo–exocytic behavior of α5β1 should be considered.
Using a strategy to selectively target the Rab4-dependent short-loop recycling of αvβ3 integrin, we demonstrate a clear connection between this pathway and a persistent mode of fibroblast migration. Short-loop recycling exerts its influence by counteracting the trafficking and signaling of another integrin, the α5β1 heterodimer, and there is no obligatory requirement for short-loop αvβ3 recycling when α5β1 signaling is compromised. These data show that the short loop does not form part of the machinery integral to persistent cell migration, but acts to dictate the nature of integrin downstream signaling, which in turn influences the cell's decision to migrate with persistence or to move randomly on 2D matrices. The ability of β1 integrins to signal to RhoA determines the mode of tumor cell invasiveness (), and a key challenge for the future will be to determine the influence that recycling pathways have on integrin signaling and the choice between elongated and amoeboid migration of tumor cells through 3D matrices.
αv, β3, α5, and β1 integrins and Rab4, Rab4, and Rab11 were in pcDNA3 are as described by and . The mouse sequences for PKD1 and PKD1 were tagged with a hexa-Histidine at the 5′ end (N terminus), cloned into pcDNA3, and verified by sequencing. The shRNA mU6pro vector targeting PKD1 and the validation of its efficacy is described by , and the shRNA sequences targeting mouse β3 integrin (5′-CAGCTCATTGTTGATGCTT-3′ and 5′-GTCAGCCTTTACCAGAATT-3′) were cloned into the mu6pro vector as described by . ts045-VSVG is as described previously () and was a gift from J. Lippincott-Schwartz (National Institutes of Health, Bethesda, MD). All plasmids were purified by CsCl banding before transfection into NIH3T3 fibroblasts by Fugene 6 or Amaxa Nucleofection. PCR-amplified DNA fragment corresponding to the indicated regions of the human sequence of β3 integrin were subcloned into the pGEX-4T-1 vector. GST fusion proteins were expressed in strain BL-21 and purified as described previously ().
Cos-1 cells transfected with His-PKD1 or His-PKD1 were treated with 1 μM PMA for 15 min to activate the kinase and then lysed in 200 mM NaCl, 75 mM Tris, 15 mM NaF, 1.5 mM NaVO, 7.5 mM EDTA, 7.5 mM EGTA, 1.5% Triton X-100, 0.75% Igepal CA-630, 50 μg/ml leupeptin, 50 μg/ml aprotinin, and aminoethyl benzene sulfonyl fluoride (AEBSF) and scraped from the dish with a rubber policeman. Lysates were passed three times through a 27-gauge needle and clarified by centrifugation at 10,000 for 10 min. The clarified lysates were loaded into a 1-ml His-TRAP affinity column (GE Healthcare), and the kinase was eluted with a linear gradient of imidazole. 1-ml fractions were collected, and the peak of purified His-PKD1 was identified by SDS-PAGE followed by staining with colloidal Coomassie. The kinase was dialysed overnight into kinase buffer (25 mM Hepes, pH 7.4, containing 25 mM MgCl, 0.5 mM NaVO, 0.5 mM EDTA, and 0.5 mM DTT), glycerol was added to 50% (vol/vol), and the kinase was stored at −20°C. Kinase assays to assess the catalytic activity of PKD1 were performed in kinase buffer in the presence of 100 μM ATP, 4.4 μCi γ-[P]ATP, and 3 μg c-Jun 1–89 GST fusion protein (a gift from M. Dickens, University of Leicester, Leicester, UK) per reaction. Reaction products were resolved is 12% SDS-polyacrylamide gels, which were dried and exposed to x-ray film to visualize bands.
GST-integrin cytodomain fusion proteins were bound at saturating concentrations to wells of microtitre plates (Immunol. 2; Dynatech Laboratories) in 0.05 M NaCO, pH 9.6, at 4°C, and the wells were blocked with PBS containing 0.1% (vol/vol) Tween-20 (PBS-T). Various amounts of purified His-PKD1 or His-PKD1 were added to the wells in PBS-T and incubated for 1 h at 15°C. After three washes with PBS-T, PKD1 was detected by serial incubations with polyclonal rabbit anti-PKCμ (sc-639; Santa Cruz Biotechnology, Inc.) and horseradish peroxidase–conjugated anti-rabbit IgG, followed by chromogenic reaction with ortho-phenylenediamine as described previously ().
Cells were grown to 90% confluence, serum-starved for 30 min, and treated with a combination of 10 ng/ml PDGF-BB and 0.6 mM primaquine for 12 min. After this, cells were washed twice in ice-cold PBS, lysed in 200 mM NaCl, 75 mM Tris, 15 mM NaF, 1.5 mM NaVO, 7.5 mM EDTA, 7.5 mM EGTA, 1.0% octyl β-thioglucopyranoside, 50 μg/ml leupeptin, 50 μg/ml aprotinin, and AEBSF and subjected to immunoprecipitation using magnetic beads coupled to a mouse anti-human β3 integrin monoclonal antibody (clone VI-PL2; BD Biosciences) as described previously (). Unbound proteins were removed by extensive washing in octyl β-thioglucopyranoside–containing buffer and specifically associated proteins resolved by SDS-PAGE (8% gels under reducing conditions for detection of PKD1; 6% gels under nonreducing conditions for β3 integrin) and analyzed by Western blotting as described previously ().
NIH3T3 mouse fibroblasts and Cos-1 cells were grown in DME with 10% (vol/vol) fetal calf serum and 100 U/ml penicillin, 100 μg/ml strepto mycin, and 0.25 μg/ml amphotericin B at 37°C with 10% CO. For integrin recycling assays, immunoprecipitations, and preparation of purified PKD1, cells were grown to 50% confluence, fed with fresh DME containing 10% (vol/vol) fetal calf serum, and transfected using Fugene 6 (Roche Diagnostics) according to the manufacturer's instructions. The ratio of Fugene 6 to DNA was maintained at 3 μl Fugene/1 μg DNA. For cell migration studies and measurement of phosphocofilin signaling, transfections were performed using the Nucleofector system (Amaxa). In brief, cells were grown to 80% confluence, removed by trypsinization, washed in PBS, and resuspended in Amaxa solution R with 5 μg DNA. After electroporation (in the Nucleofector; program T-20), the cells were replated in 6-well dishes.
Integrin recycling assays were performed as described previously (). I-transferrin recycling assays were performed essentially as described previously () with some modifications. In brief, serum-starved cells were incubated with I-labeled transferrin (0.1 μCi/well; NEX212 [NEN Life Science Products]) for 1 h at 4°C in PBS with 1% (wt/vol) BSA. The tracer was allowed to internalize for 15 min at 22°C (to label early endosomes) or 30 min at 37°C (to label the recycling compartment). Tracer remaining at the cell surface was removed by incubation with acid-PBS (corrected to pH 4.0 by the addition of HCl) at 4°C for 6 min, and the tracer was allowed to recycle at 37°C in serum-free DME supplemented with 1% BSA and 50 μM desferoxamine (D9533; Sigma-Aldrich). The quantity of I recycled into the medium is expressed as a percentage of the number of counts incorporated during the internalization period.
Confluent monolayers were wounded with a plastic pipette tip and placed on the stage of an inverted microscope (Axiovert S100; Carl Zeiss MicroImaging, Inc.) in an atmosphere of 5% CO at 37°C. Cells were observed using a 20× phase-contrast objective, and images were collected every 20 min using a digital camera (C4742-95; Hamamatsu). Videos were generated and cell tracks analyzed using Andor Bioimaging software. The selective αv integrin antagonist cyclic peptide, cyclo-Arg-Gly-Asp-D-Phe-N(Me)-Val (cRGDfNV), was as described by and was added to the monolayers shortly after wounding at a concentration of 1 μM.
Wounded monolayers were maintained at 37°C for various times and fixed in ice-cold methanol. Fixed cells were incubated with an anti–γ-tubulin monoclonal antibody (clone GTU-88; Sigma-Aldrich), followed by a Texas red–conjugated secondary antibody and counterstaining with DAPI to visualize nuclei. The percentage of cells with the MTOC positioned in quadrants facing the wound (front) or at the cell rear (back) or neither (middle) with respect to the position of the nucleus was determined by visual examination of images captured on an epifluorescence microscope (Axiophot; Carl Zeiss MicroImaging, Inc.).
Cells transfected using the Nucleofector were trypsinized, incubated in suspension for 45 min, and plated onto plastic surfaces coated with 10 μg/ml fibronectin. Where indicated, 2 μM Y27632 (Calbiochem) or 2 μg/ml mAb16 (a gift from K. Yamada, National Institutes of Health, Bethesda, MD) were included 15 min before and throughout the plating period. Alternatively, monolayers were extensively wounded (evenly spaced 500-μm wounds; wounded area was ∼30% of monolayer area) with a plastic pipette tip, and cells were allowed to migrate into the wound for various times. Cells were lysed and subjected to Western blotting followed by detection with an antibody recognizing phospho-Ser-cofilin (3311; Cell Signaling Technologies).
Fig. S1 shows that overexpression of recombinant His-PKD1s inhibits activation of endogenous PKD1. Fig. S2 shows use of shRNAi to suppress cellular levels of mouse αvβ3 integrin. Fig. S3 shows the influence of mutant Rab4 and PKD1 on the surface expression of αvβ3 and α5β1 integrins. Fig. S4 shows coimmunoprecipitation of endogenous PKD1 with endogenous mouse αvβ3 integrin. Online supplemental material is available at . |
Kinesin (; ) is a force-generating molecule that is probably ubiquitous and essential in organisms with microtubules. In neurons, it powers vesicular trafficking directed toward the synapse; in all cells, it is involved in the restructuring that occurs during cell division (). This exquisite molecular machine can convert up to 35% of the chemical energy released by ATP hydrolysis into mechanical energy (), as the dimeric form “walks” step-by-step along microtubule protofilaments (). Numerous measurements have characterized kinesin's biophysical properties (; ).
Atomic resolution structures of monomeric kinesin obtained by x-ray crystallography revealed a “switch I/switch II” nucleotide-sensing architecture, similar to that found in G proteins and myosins (; ). Moreover, a “relay helix” hypothesis emerged from the crystal structures and other data (). Based on the observation of two distinct conformations of the “switch II helix,” crystal structures of kinesin were designated as either “ATP-like” or “ADP-like.” In the hypothesis (presented in an adapted form below [see ]), the switch II helix of ADP-bound and nucleotide-free kinesin (the first two nucleotide states, respectively, of microtubule-bound kinesin) would be positioned away from the nucleotide catalytic site. This conformation of the switch II helix, it was proposed, would resemble that found in the class of kinesin crystal structures regarded as ADP-like. In ATP-bound kinesin, according to the hypothesis, the switch II helix moves toward the nucleotide active site, mimicking the switch II helix conformation found in a second class of kinesin crystal structures, regarded as ATP-like. This switch II helix movement was hypothesized to control kinesin's presumed force-delivering element, the neck linker, by obstructing the neck linker's docking onto the core of kinesin in the ADP-like position and permitting docking in the ATP-like position.
The first direct tests of the relay helix hypothesis have been incomplete and contradictory. A pair of newly published cryo-EM structures at a resolution of ∼1 nm of the KIF1A kinesin () are consistent with the relay helix theory for two of the three principal nucleotide states, finding the switch II helix ADP-like in ADP-bound, microtubule-bound kinesin and ATP-like in microtubule-bound kinesin with the nonhydrolyzable ATP analogue 5′-adenylyl-imidodiphosphate (AMPPNP). However, nucleotide-free kinesin was not included in the KIF1A study. Furthermore, a second very recent cryo-EM study of a different, unconventional kinesin (Kar3) is seemingly at odds with the relay helix theory (). In Kar3, the switch II helix was visualized apparently in the ADP-like position for both the ADP–kinesin–microtubule and AMPPNP–kinesin–microtubule states. Moreover, in the nucleotide-free state of Kar3, the switch II helix was only partially visible and was interpreted to have restructured as a loop. Thus, the generality of the relay helix theory is in doubt, and the role of the nucleotide-free state of kinesin in particular is highly unclear.
Another puzzle related to the relay helix theory concerns the communication between the nucleotide pocket and the switch II helix. Instead of showing clearly defined, nucleotide-dependent states for switch II, the crystal structures showed a range of positions for the switch II helix—apparently unconnected to the identity of the active site nucleotide. For example, it was possible to crystallize KIF1A with an ATP-like switch II helix conformation but with ADP in the active site (), in contrast to ATP-like switch II helix conformations found in crystal forms reported with either AMPPCP (5′-adenylyl-methylenediphosphate; ) or AMPPNP (). On the other hand, the more recent structures of KIF2C kinesin display an ADP-like switch II helix conformation in the presence of either ADP or AMPPNP (). Also, electron paramagnetic resonance and crystallography studies of the human kinesin heavy chain (KHC) confirmed that in solution kinesin freely exchanges between ATP- and ADP-like conformations in the presence of either ADP or AMPPNP (). Thus, although microtubule-bound kinesin is hypothesized to have the conformation of its switch II helix controlled by the active site nucleotide, crystallography experiments to date have only identified “uncontrolled” switch II helix movements.
We have previously speculated how microtubule binding may activate kinesin's switch II nucleotide sensing mechanism by structuring kinesin's loop “L11,” N-terminally adjacent to the switch II helix (). However, other subsequent reports have not addressed this aspect of kinesin's mechanism. Here, we use a new method of cryo-EM data analysis that, when combined with high-quality image data, can attain better than nanometer-resolution reconstructions of the kinesin–microtubule complex. Our resulting structure of nucleotide-free kinesin complexed to the microtubule exhibits at least three remarkable features. By directly visualizing the switch II helix, we find that the helix is in an ADP-like position, confirming one aspect of the relay helix theory for our KHC kinesin construct—in striking contrast to the nucleotide-free Kar3 structure with an apparently “melted” switch II helix. Furthermore, we show that the switch II helix is N-terminally extended by rearrangement of the loop L11 (which is unstructured in crystal structures of our construct). This structured extension, apparently stabilized by microtubule contacts, is a prime candidate for explaining the microtubule-induced activation of kinesin's switch II nucleotide sensing mechanism. Finally, we infer a possible role for switch I in disrupting kinesin's interactions with ADP upon binding to the microtubule.
To determine the structure of the asymmetric 13-protofilament microtubule, we applied single-particle reconstruction techniques (see Materials and methods) to our set of nucleotide-free kinesin-decorated microtubule images. The essential step of the single-particle analysis was to determine the orientation of a given microtubule's symmetry-disrupting “seam.” Once the seam orientations were identified in our imaged, kinesin-decorated microtubules, we produced a 15 Å–resolution reconstruction of the entire, asymmetric microtubule volume (see Materials and methods). The reconstruction, shown in , clearly reveals the microtubule seam, validating our reconstruction techniques. Seams in microtubules have been imaged before (; ), but 3D reconstructions have never been produced of the native 13-protofilament form, because of technical limitations in image processing. Here, we have applied a method capable of reconstructing the 13-protofilament microtubule form and have derived a cryo-EM reconstruction of this form, including the asymmetric seam.
To maximize the resolution of the nucleotide-free kinesin– microtubule complex, we averaged together the 13 unique protofilaments reconstructed in to produce a new motor– tubulin complex with the highest possible signal-to-noise ratio. After this step and subsequent image processing (see Materials and methods), we derived the kinesin–tubulin complex shown in –
. We note that in the averaging step, we assumed that all 13 copies of the motor–tubulin complex were identical—an assumption that could in principle be violated because our structure contained an asymmetric seam. However, inspection of our nonaveraged reconstruction () does not reveal obvious conformational differences between different sites (for example, differing motor occupancy or orientation), justifying our assumption at least to the first approximation.
At the resolution of the symmetry-averaged structure, ∼9-Å (see Materials and methods) helices in tubulin were distinct from one another, even the packed antiparallel helices H11 and H12 (). All helices in kinesin were resolved but were slightly less distinct, possibly because of imperfect occupancy (estimated to be ∼75%) on the microtubule lattice. However, the large β sheet at the core of kinesin appeared as a continuous twisted sheet of density, a loop (L8 in kinesin) appeared as a distinct arm connecting kinesin to the microtubule surface, and the switch II helix was clearly resolved at the microtubule binding interface.
To evaluate tubulin's secondary structure elements, the coordinates of bovine tubulin (Protein Data Bank [PDB] ID 1JFF; ) were fitted into the cryo-EM map using cross-correlation implemented in Fourier space (see Materials and methods). The agreement between positions of helices in our map and the fitted tubulin crystal structure was excellent ( and Video 1, available at ).
Two crystal structures of monomeric human kinesin, those of the K349 construct (used in our experiments) with and without a docked neck linker (PDB IDs 1MKJ [] and 1BG2 [], respectively), were fit into our map using cross-correlation (see the previous section). The positioning of the molecules was unambiguous, with alignments between the two fitted core motor domains differing by <2°, when the principal axes of the moments of inertia were compared (see Materials and methods). The fit of the crystal structure into our molecular envelope is excellent in most regions of the protein chain ( and Video 2, available at ). However, the docked neck linker present in 1MKJ was found to be out of density (). No compensating region of unoccupied density was found near the docked neck linker region. Thus, our density map strongly suggests that the neck linker is disordered in our experimental conditions. This result agrees with numerous other experiments using techniques including single molecule fluorescence (using bifunctional labels; ) and electron paramagnetic resonance ().
In all crystal structures observed to date, a disordered neck linker in kinesin is accompanied by a switch II helix in the ADP-like position; furthermore, the switch II helix in the ADP–kinesin–microtubule complex (with a disordered neck linker) was likewise in the ADP-like position (). It therefore seemed likely that the nucleotide-free kinesin–microtubule complex would have an ADP-like switch II helix conformation as well (). Our cryo-EM map allowed this prediction to be tested directly, as density corresponding to the switch II helix was clearly apparent.
We compared the position and orientation of the density corresponding to the switch II helix in our map with the switch II helix positions predicted by crystal structure fits, using ∼20 kinesin crystal structures from PDB. As shown in (a and b), our helical density falls within the switch II helix orientations classified as ADP-like from K349 and KIF1A crystal structures, as well as others (not depicted). We quantified this observation by fitting a long switch II helix from Kar3 () into our density using cross-correlation and comparing the principal axis of the moment of inertia of this fitted helix with the principal axes of the K349 and KIF1A helices. This analysis showed that the KIF1A ADP-like switch II helix was rotated ∼8° counterclockwise (in the plane of ) compared with our helical density, whereas the K349 ADP-like helix was rotated ∼6° clockwise. Thus, ADP-like switch II helices can vary over a considerable range of orientations relative to the core motor domain, and our switch II helix density lies squarely in this range.
On the other hand, fitted ATP-like kinesin structures displayed switch II helices distinct from our helix density, as shown in . In particular, the C termini of the crystal structure helices (, right) are well out of density and far from the apparent helical axis of the density. A moment of inertia comparison showed the ATP-like switch II helices to be rotated ∼12–15° clockwise in the plane of , relative to the helix density. This angular difference, substantially greater than the estimated error of our orientation determination (∼5°), demonstrates that the switch II helix orientation in our structure is measurably distinct from that found in the ATP-like crystal structures. Collectively, these results constitute direct evidence to support the “relay helix” prediction of kinesin's switch II helix ADP-like conformation in the nucleotide-free, microtubule-bound state.
An N-terminal extension of the switch II helix, relative to the crystal structures of our K349 construct, is evident in our cryo-EM density map in . The visible density extends the helix by several turns. Furthermore, the extension in our map is accompanied by an extra “sphere” of density, closely associated with the microtubule surface, in the vicinity of the N terminus of the helix. Intriguingly, these features correspond to the architecture of switch II seen in several crystal structures (; ; ), including that of Kar3 (mutation R598A; ), as shown in . The structural element at the N terminus of the switch II helix, loop L11, is often disordered in crystal structures but visible in that of Kar3 and projects through the sphere of density in our alignment.
The structural alignment in serves as a useful indicator of residues likely to be involved in the kinesin–microtubule interaction for kinesin's nucleotide-free state. This alignment indicates that the two space-filled residues shown in (corresponding to Asn 255 on the K349 switch II helix and Thr 241 on K349 loop L11), highly conserved in the kinesin family (), account for the two microtubule contact points of the switch II helix extension–L11 assembly. Both of these residues are involved in the microtubule-induced restructuring of kinesin seen in our cryo-EM structure. In the fitted ADP-like K349 crystal structure in , not only does the sidechain of Asn 255 face away from the microtubule surface (not depicted) but the backbone departs from the α-helical structure of subsequent residues 256–271 of the switch II helix. In contrast, the equivalent residue Asn 650 contained within the α-helical extension of Kar3 () has swiveled ∼180° to point its sidechain at microtubule helix H11′ and/or the subsequent loop on the microtubule surface. Residue Thr 241 is not present in the K349 crystal structure and is presumably disordered, whereas the equivalent residue Ser 636 in the Kar3 model faces helix H3′ on the microtubule surface. Thus, our modeling suggests that rearrangement of critical conserved microtubule-binding residues in kinesin supports the restructuring of switch II and L11 seen in our nucleotide-free kinesin–microtubule complex.
We compared the microtubule-bound orientation of our fitted K349 core domain with that of ADP- and AMPPNP-bound KIF1A reported by (PDB IDs 2HXF and 2HXH). A least-squares superposition and moment-of-inertia analysis showed only a 5° rotation of the core domain of our structure compared with ADP-bound KIF1A. On the other hand, AMPPNP-bound KIF1A was reoriented by 15° relative to our structure. Thus, our nucleotide-free K349 core domain orientation is much more similar to ADP-bound, microtubule-complexed KIF1A. Because kinesin's orientation on the microtubule is hypothesized to depend on whether the switch II helix is ATP- or ADP-like, the similarity between our structure and ADP- bound KIF1A is further confirmation that the switch II helix is ADP-like in both structures.
To compare the nucleotide pocket in our experimental map with that expected from crystal structures of K349–ADP, we generated a 9-Å–resolution synthetic map using the fitted PDB coordinates of ADP-like K349 (but with ADP omitted) and tubulin from . At this resolution, several structured loops are visible as lobes of density in both synthetic and experimental maps. Examples include kinesin's loops L5 and L8 ( and ), as well as the loop between tubulin's β strands B9 and B10 (not depicted). This observation suggests that it is feasible to detect, in our maps, conformational change of kinesin's nucleotide binding pocket, which contains two functionally important loops: the ADP-coordinating “P-loop” and the switch I nucleotide response element.
When viewed at an isocontour level of ∼2 σ, as shown in (a and b), the synthetic map displays a lobe of density marking the location of the P-loop. In the experimental density map, however, this lobe does not appear even at lower contour levels (). This difference, which was consistently observed in reconstructions using two independent half datasets (unpublished data) suggests a rearrangement of the P-loop in kinesin's nucleotide-free state relative to crystal structures.
Although no alternative density site for the P-loop is apparent at the 2 σ contour level in our map of nucleotide-free kinesin, at a lower threshold (1.2 σ) a “bridge” is revealed in our experimental map between the density for switch I and the P-loop site, as shown in . The bridge, which features a distinctive protuberance, was consistently observed in two independent half dataset reconstructions (Fig. S3, available at ). The appearance of this bridge is accompanied by a rotation and shortening of helix α3, N-terminal to the switch I loop, and a movement of the switch I density itself toward the nucleotide site as it forms the bridge ().
These observations suggest a model for ADP release by kinesin after microtubule binding (). In guanine nucleotide-binding proteins, whose mechanism of nucleotide release is somewhat better understood, a consistent feature of the nucleotide-free state is disruption of the P-loop—evidently contributing to nucleotide release (). This disruption appears to be caused by any of a variety of mechanisms, including direct interactions between the P-loop and residues of the guanine-nucleotide exchange factor (GEF), or an indirect pathway such as GEF-modulated conformational change of switch II that in turn disrupts the P-loop. As shown in , our data are consistent with a disruption of the P-loop but suggest that interactions by switch I may contribute to the disruption.
A probable trigger for the apparent restructuring of switch I in our structure is the microtubule-induced restructuring of L11 and the N-terminal extension of switch II as described (). As seen in , switch I in our structure approaches the extended switch II N terminus, possibly attracted by new contacts made available by the L11/switch II restructuring. Indeed, based on the Kar3 model of , absolutely conserved residue Glu 250 in the switch II helix extension would be positioned at the interface, where our density for switch I meets the helical extension (not depicted). Furthermore, in a KIF1A ADP-like crystal structure—with an extended switch II helix—the equivalent of K349's Glu 250 forms a salt bridge with the equivalent of K349's Arg 203 from switch I. Thus, given the proximity between the switch II helix and switch I in our structure, it is likely that this conserved interaction forms there as well.
This potential chain of communication, from switch II to switch I to the nucleotide pocket, is supported by experiments that showed that mutations in either switch I (equivalent to R203A in our K349 construct) or switch II (equivalent to N255K or E236A in K349) decoupled microtubule binding from kinesin's ADP affinity (). Our structure suggests a more detailed explanation for the decoupling mechanism, particularly in the case of the Asn mutation. If the N255K mutation destabilizes the helical extension by changing the residue's microtubule interactions, then in our mechanism, the extension would no longer be present to attract switch I toward the nucleotide site while kinesin is bound to the microtubule. Thus, microtubule binding would no longer eject ADP and would be uncoupled from ADP affinity.
Together with the ADP and AMPPNP kinesin–microtubule complexes of , our nucleotide-free structure completes a nanometer-resolution picture of the nucleotide binding and hydrolysis cycle for the plus end–directed kinesins. Notably, the position of the switch II helix was identified for all three of these nucleotide states and in each case was found to agree with the predictions of the relay helix theory.
Previously reported structures of Kar3 complexed to the microtubule () do not appear to follow the relay helix paradigm, as mentioned in the Introduction. Although it is hard to reconcile these results with ours and those of , we note that Kar3 moves toward the minus end of the microtubule, whereas our KHC construct and KIFA are a plus end–directed motors. As such, Kar3 possesses a different force-delivering element, the helical “neck” (; as opposed to conventional kinesin's neck linker). Thus, it is at least conceivable that minus end–directed kinesins have evolved a switch II mechanism distinct from the “relay helix.”
The mechanism that communicates the state of the nucleotide to the switch II helix is a crucial, unknown component of the proposed relay helix mechanism for kinesin. This mechanism presumably involves the formation of a hydrogen bond between switch II's absolutely conserved Gly 234 and ATP's γ-phosphate—a critical feature conserved as far as is known across proteins with functional switch II domains (). One possibility for this mechanism is our “connecter” hypothesis, depicted in (c and d).
We predict that this connecter, consisting of our observed N-terminal extension of the switch II helix as well as the C-terminal, microtubule-contacting segment of L11, forms upon microtubule binding and persists after ATP binds in kinesin's active site (). In this process, the microtubule contacts would act as a kind of glue, stabilizing the connecter so that ATP-induced movement of Gly234 and the adjacent, N-terminal segment of L11 would be transmitted through the relay helix to the neck linker. The apparent microtubule interactions of conserved residues Thr 241 and Asn 255 in our structure, sufficient to stabilize and restructure a portion of L11 relative to crystal structures, reinforce this idea. The fact that some crystal structures of kinesin show N-terminal extensions of the switch II helix relative to our construct show that such an extension is possible, if not necessarily stable, in the absence of microtubules. Furthermore, density consistent with the helical extension is evident in both ADP- and AMPPNP-bound forms of the KIF1A–microtubule complex ().
Though an N-terminal extension of the switch II helix is not seen in the AMPPNP–KIF1A crystal structure (), it is unlikely in any case that the conformations of L11 and the adjoining Gly 234 are the same in the ATP–kinesin–microtubule complex as in the crystal structure. This is because the AMPPNP–KIF1A crystal structure lacks a hydrogen bond between the nucleotide's γ-phosphate and KIF1A's Gly 251 (the equivalent of Gly 234 in K349). The switch II helix extension we propose in the ATP state of the kinesin–microtubule complex could help lead to the formation of such a hydrogen bond. Further work to extend the resolution of the kinesin–microtubule complex will be required to fully elaborate the role of L11 and Gly 234 in kinesin's ATP-sensing mechanism.
Our structure has allowed us to propose further details of how microtubule binding affects kinesin's interactions with bound nucleotide—in particular, how it may release bound nucleotide upon microtubule binding and how microtubule binding may permit the switch II helix to respond to the identity of kinesin's active-site nucleotide. Furthermore, our structure determination methods open up a route to high-resolution characterization of lateral interactions of the microtubule protofilaments—even those that occur at the microtubule seam. Future work will extend the resolution of both kinesin and the microtubule, allowing us to test the various aspects of the structural mechanisms of these proteins.
Human monomeric kinesin K349 cys-lite, derived from the conventional human KHC, was expressed from a plasmid (a gift from N. Nabor, University of California, San Francisco, San Francisco, CA) and prepared as described previously (). We chose the mutant form 220C, which has motility and hydrolysis properties close to wild-type kinesin (), for its high purification yield (∼100 mg per 2-liter culture). After elution from the final column (mono Q), ∼1-ml aliquots of ∼2 mg/ml were frozen in elution buffer + 20% sucrose. For use in EM experiments, protein was thawed on ice and dialyzed overnight against binding buffer (25 mM Pipes, pH 6.8, 25 mM NaCl, 1 mM EGTA, and 2 mM MgCl). The resulting protein solution was concentrated to 15–20 mg/ml using centricon concentrators (Millipore). After this treatment, our kinesin preparation remained stable with substantial microtubule binding activity after several months at +4°C, making it an especially useful construct for extended experimental study.
Glycerol-free tubulin was purchased from Cytoskeleton, Inc., and polymerized according to a modified protocol to minimize aggregated, nonpolymerized tubulin. 25 μl of frozen tubulin aliquots was thawed on ice and ultracentrifuged at maximum speed for 10 min (100,000 rpm; TLA 120.2 rotor [Beckman Coulter]). The supernatant was added to a chilled glass vial, 1 mM GTP was added, and the vial was placed in a 37° incubator for 15 min. At this point, 1.25 μl taxol (2 mM in DMSO), diluted in 11.25 μl polymerization buffer (80 mM Pipes, pH 6.8, 1 mM EGTA, and 2 mM MgCl), was slowly and carefully stirred into the polymerization vial. After 15–30 more minutes, 12.5 μl of the polymerized microtubules were mixed with ∼25 μl of above-prepared kinesin solution and 5 μl apyrase (10 mg/ml in deionized water). After 5 min, this mixture was layered onto a room-temperature glycerol cushion (50% glycerol + binding buffer + 200 μM taxol) and ultracentrifuged at 40,000 rpm for 20 min. The resulting kinesin–microtubule pellet was briefly washed 2× with binding buffer + 200 μM taxol and resuspended in 15 μl of the same buffer.
Approximately 350 images were collected on film using a microscope (JEM-4000; JEOL) operating at 400 kV and 60 kx magnification, with defocus values ranging from 0.8 to 2 μm. Developed films were scanned using a robotic system incorporating a scanner (CoolPix; Nikon) operating at 6.3 μm/pixel (). The final digitized images had a sampling of ∼1 Å per pixel.
13-protofilament microtubules were selected manually using the boxer program of the EMAN package (), whose interactive functions facilitated the division of each microtubule into short (750- × 750-pixel) overlapping segments. Subsequent single-particle image processing was performed using customized scripts written for the SPIDER package (), roughly following the methodology described by .
To generate an initial model for reference-based alignment, the atomic microtubule model of was “decorated” using the crystal structure of human monomeric kinesin in orientations reflecting the AMPPNP-bound kinesin–microtubule complex of . Reference projections of this volume (filtered to ∼1-nm resolution) were then calculated at 0.3° intervals around the microtubule axis, including up to ±25° out-of-plane tilt. Although we did not expect our final map to exactly resemble this initial atomic model, the differences between our final map and the initial reference model served as a control to ensure that our reconstruction methods were not simply reproducing the model with which we started.
Reference-based analysis of kinesin-decorated 13-protofilament microtubules (Figs. S1 and S2, available at ) was complicated by the presence of a seam that disrupts the quasi-helical symmetry (). The practical implication of the seam was that a complete 360° axial rotation of the decorated reference microtubule produced 13 fairly close matches against an experimentally imaged microtubule segment, of which only one gave the correct seam alignment. However, small differences in the magnitude of the image-reference cross-correlation indicated the true position of the seam (Fig. S2). Furthermore, seam determinations along consecutive microtubule segments typically agreed, amplifying our certainty that we had correctly identified the seam position. The low signal-to-noise ratio in our images caused some fraction (frequently 25% or more) of boxed segments to yield incorrect seam identifications that were one or more protofilaments away from the correct seam, identified by the majority of segments. However, this type of error decreased when increasingly accurate reference models were used in subsequent rounds of refinement.
After reference-based alignments were derived for the microtubule segments (Figs. S1 and S2), a 3D volume was obtained by weighted back projection. To increase the quality of our final map, the back-projection process was modified to directly integrate contrast transfer function (CTF) correction via a customized C program. The goal of the modification was to implement CTF correction with CTF-squared weighting of the Fourier components of the images, resembling the helical reconstruction methods presented, for example, by . Within the program, the Fourier transform of each microtubule segment was first multiplied by its experimentally determined CTF (which included astigmatism). The CTF was derived by straightening entire microtubules using previously determined reference alignment parameters; the entire microtubule was then Fourier transformed to find the position and shape of the layer line absences indicating CTF minima. Subsequently in the program (but before back projection), the Fourier transform of each image was multiplied by a general, exponential weighting function derived from all other image segment transforms overlapping in Fourier space () and divided by the sum of the squares of the CTFs for the overlapping Fourier space measurements. The general exponential weighting function used was identical to that used in SPIDER's bp 3d command. The resulting Fourier space object was inverse Fourier transformed into an image and back projected through the target volume; this process was repeated for all microtubule segments to produce a final volume.
Another feature of our volume reconstruction methods took advantage of the pseudosymmetry of the 13-protofilament microtubules used here. As mentioned above, the seam disrupted true helical symmetry in the microtubules. However, it was possible to average the 13 pseudosymmetry mates in Fourier space, during back projection, to improve the resolution of the basic kinesin–tubulin subunit along one protofilament. The most straightforward way to describe our symmetry-averaging method is to describe the real-space analogue. Rotating a microtubule volume by × 360°/13 and translating by /13 of the pseudohelical repeat distance of 12 nm (where is an integer) results in a new volume that superimposes on the original volume except for superimposing α tubulin on β tubulin in of the protofilaments. Applying this process for = 0–12 and superimposing the 13 resultant volumes will reveal one protofilament where kinesin and α and β tubulin superimpose for all 13 volumes. Thus, averaging the 13 volumes produced this way results in a strange volume that has one correctly averaged protofilament. To apply this idea in Fourier space before back projection, we created 13 copies of each segment image, corresponding to = 0–12. We then added × 360°/13 to the reference-determined axial orientation parameter of segment and translated the segment in the axial direction of the microtubule by /13 of the repeat distance of 12 nm. Thus, the input to our back-projection algorithm consisted not only of the original segment images but also 12 additional copies related by symmetry.
To compensate for amplitude attenuation at higher resolutions, the reconstruction was sharpened by applying a B-factor of −100, which was derived by comparing resolution-dependent amplitudes of our map to a synthetic map generated from PDB coordinates. The resolution of our reconstructions was estimated by two independent methods. First, the program RMEASURE () reported a resolution of 15.0 Å for the asymmetric reconstruction and 9.0 Å for the symmetry-averaged reconstruction (using the program's conservative output, with an estimated Fourier shell correlation [FSC] cutoff of 0.5). Second, the FSC () was compared between two reconstructions: one using odd-numbered image segments and the other using even-numbered image segments. The resulting curve descended below FSC = 0.5 at 15.1-Å resolution for the asymmetric reconstruction and at 9.1-Å resolution for the symmetry-averaged reconstruction.
X-ray crystal structures of tubulin and kinesin were fit into our map using the COLORES program from the SITUS package (). Principal axes of the moments of inertia of fitted crystal structures were computed using the program GEM ().
The coordinates of kinesin and tubulin, fitted into our density map, are available online from PDB (; ID 2P4N). The density map has been deposited in EMDB and is available online (; accession no. ).
Figs. S1 and S2 describe details of the single-particle analysis and 3D reconstruction procedures. Fig. S3 shows two independent half dataset reconstructions of kinesin's nucleotide pocket. Videos 1 and 2 are 3D reconstructions of tubulin and kinesin, respectively, from our density map, with superimposed ribbon diagrams of fitted PDB structures. Online supplemental material is available at . |
I wanted to become a writer for a very long time. My father was a classicist, and I was tempted to follow him into the humanities. But he influenced me very strongly out of his own frustration with the limited potential for discovery—he said science is so much more exciting. I started reading more and more popular science and original scientific research when I was in high school and I got gripped by it. But I got gripped by it more from an abstract, formal point of view than by tinkering around with a chemistry set in a basement.
I read the papers and I was just blown away by how beautiful they were. Rothman trained as a theoretical physicist as an undergraduate. In many ways, the papers are more typical of a physicist's approach than of a biologist's in that simplification, abstraction, and synthesis are the most important part whereas many biologists tend to get lost in bewildering experimental detail. I think if Jim Rothman had worked on a different topic I might still have joined him. For me, it was almost a matter of style over substance.
It was essentially a classic Rothman-style transport assay but applied to the neuronal synapse and using light as the readout rather than a chemical reaction product. The principle was inspired by earlier work—you separate an enzyme and substrate in space and then measure how transport brings them together.
To my knowledge this was the first paper where the idea of using genetically encoded probes to observe the function of neural circuits was proposed rather forcefully. The pHluorins became the dominant implementation of the technology because of issues of technical utility, but I think the conceptual foundation actually lies in the earlier synaptolucin study.
I'm very, very proud of that paper because, as with the synaptolucins, it formulated a new experimental strategy—again this combination of genetics and optics but used for a different purpose. Rather than passively observing cellular or neuronal activity, we actively controlled it. That was, I think, another important conceptual step.
It was not a deliberate choice. I wanted to do certain types of experiments, and I couldn't fill that experimental need unless I went out and tried to develop something new. For example, there was a clear sense that one needed a new way to look at networks of neurons.
Also, there is a beauty to these developmental efforts that is quite rewarding in its own right. It's really a pleasure to see something through from the conceptual stage, to trying to get it to work and then actually using it. Seeing the first images of a fruit fly smelling an odor as revealed by synaptopHluorin and looking back over the entire arc that began with the engineering of the GFP mutant, I have to say this was a very satisfying moment.
This biological engineering that my lab does—it's almost a little bit like poetry. You have to stay within a tight framework imposed by the biological reality, and there is a tension between innovation and what you actually can do.
Ultimately, what tends to prevail is simplicity. Although there is iterative improvement, it's always the earliest implementations that are conceptually the most important. It's more difficult to come up with a really new idea of how to do things, even if you don't get the ultimate implementation in the first try.
I had never in my life worked with fruit flies; I think I had never even seen one in a lab. So I became faculty and student at the same time and had to apprentice myself to some fly people. I still feel like a little bit of a parvenu among real Drosophilists, but we know enough now to get by.
The initial aim was to study a neural circuit with the synaptopHluorin system. But then very quickly, in the summer of 1999—it was one of those moments where I even remember the time and the date and the room I was in—I had the idea of using light not only to observe but also to control. That then quickly became the second focus of the lab.
I had the advantage of being a newcomer to neurobiology. I was not too weighed down by received wisdom, maybe not too weighed down by neuroscience knowledge in general. But I had worked in a leading cell biology lab. I had seen that to establish causality and dissect a complex mechanism it's essential to be able to control it. In neuroscience, I felt there was still way too much observation and not enough intervention. So I thought wouldn't it be wonderful if one could use these two ingredients that I had relied on with the synaptopHluorins, namely genetics and optics, and combine them again but for this opposite way of communicating with the experimental system.
Initially, we were completely alone. Now, of course, many people have begun to work with this and similar systems.
There is a whole range of approaches that becomes possible with the ability to control specific groups of neurons. This allows a connection to brain tissue that is noninvasive and physiological. You can get wiring diagrams of neuronal circuits. You can apply spatiotemporal patterns of input activity and measure what kinds of inputs a target cell or a group of target cells is looking for. This would be not just a mapping of anatomic connectivity but rather a way of deducing the input/output characteristics of a circuit. A still higher level of complexity would be to see what exact features of activity patterns are relevant for perception, action, cognition, memory, and so forth.
We initially were attracted to particular circuits for reasons of genetic accessibility. Getting the sensor and the actuator in the right place is not trivial. In the fly olfactory system I knew I could find promoters active in specific, functionally meaningful populations of neurons.
I want to understand multicellular information processing at a depth that currently exists only for single-cell phenomena. Essentially, we want to derive a degree of mechanistic understanding that is common for cell biologists but at a level of organization that includes multiple cell types and many more players.
Yes. This is where I feel most comfortable—at the interface between cellular and systems problems.
I am excited to take on a leadership role—developing neuroscience at Oxford—that extends beyond the confines of my lab. Leaving the increasingly translational focus of US medical schools behind will also be a relief. I think that, before you can translate, you have to have something to say. |
The presence of voltage-gated Na channels at the nodes of Ranvier ensures fast saltatory propagation of action potentials in myelinated nerves. The accumulation of these channels at nodes is tightly regulated by the overlaying myelinating Schwann cells (; ; ). In the peripheral nervous system (PNS), the nodal axolemma is contacted by an ordered array of microvilli that project radially from the outer collar of two adjacent myelinating Schwann cells. These Schwann cell microvilli are embedded within a poorly defined filamentous matrix (i.e., the gap substance) that was referred to as the “cement disc” by Ranvier (). The nodal gap substance consists of proteoglycans and nonsulfated mucopolysaccharides, which contribute to the ability of a wide variety of metallic cations to label the nodes of Ranvier (). Proteoglycans that are present at peripheral nodes include versican (; ), NG2 (), and syndecans (; ), as well as hyaluronic acid and its binding protein hyaluronectin, which are associated with proteoglycans in the ECM (; ). Several ECM and ECM-associated proteins are also enriched at PNS nodes, such as collagen α4(V) (), laminin α2β1γ1 and α5β1γ1 (), dystroglycan, and some members of the dystrophin–glycoprotein complex (; ). Schwann cell–specific ablation of dystroglycan (), and to a lesser extent of laminin γ1 (), causes disruption of microvillar organization and reduction in nodal Na channel clustering, suggesting that the microvilli play a direct role in node assembly. This notion is further supported by observations demonstrating that Schwann cell microvillar processes align with nascent nodes (; ).
At the nodal axolemma, Na channels associate with two cell adhesion molecules (CAMs), NrCAM and the 186-kD isoform of neurofascin (). Growing evidence suggests that during development, Na channels are recruited to clusters containing these axonodal CAMs that were first positioned by glial processes (; ; ; ; ; ; ). Neurofascin and NrCAM interact with gliomedin, which is concentrated at the Schwann cell microvilli (). During myelination, gliomedin accumulates at the edges of myelinating Schwann cells, where it is associated with early clusters of Na channels. In myelinating cultures, both the expression and correct localization of gliomedin are essential for node formation. Gliomedin is a type II transmembrane protein that is characterized by the presence of olfactomedin and collagen domains in its extracellular region, a domain organization shared by members of a specific subgroup of the olfactomedin proteins, termed colmedins (). In addition, gliomedin contains a putative α-helical, coiled-coil sequence at its juxtamembrane region, which serves as an oligomerization motif in collagenous transmembrane proteins (; ). The olfactomedin domain of gliomedin was shown to mediate its interaction with neurofascin and NrCAM (). The aggregation of this domain using a secondary antibody was sufficient to induce nodelike clusters along the axons of isolated dorsal root ganglion (DRG) neurons. These observations led us to propose that the focal presentation of gliomedin to the axon during myelination causes the initial clustering of the axonodal CAMs into higher-order oligomers, which facilitates the recruitment of ankyrin G and Na channels (). We report that gliomedin is cleaved from the cell surface by a furin protease, and then assembles into high–molecular weight multimers and incorporates into the ECM by binding to HSPGs. We propose that these unique features endow gliomedin its ability to cluster the axonodal CAMs, thereby facilitating node formation.
We have previously shown that the OLF domain of gliomedin mediates it interaction with axonal neurofascin and NrCAM (). To examine the existence of additional ligands for gliomedin in peripheral nerves, we used a soluble Fc-fusion protein containing its extracellular domain but lacking its OLF domain (COL-Fc) in binding experiments on mixed DRG neurons/Schwann cell cultures (). Whereas OLF-Fc bound to neurons as expected, COL-Fc binding was detected to a subpopulation of Schwann cells that were aligned with axons. Binding experiments using cultures of rat Schwann cells, revealed that both COL-Fc () and an Fc-fusion protein containing the entire extracellular domain of gliomedin (ECD-Fc; ) labeled the cells, as well as areas between adjacent cells, suggesting that they interact with Schwann cell ECM. In contrast, OLF-Fc did not bind to Schwann cells (). To determine whether COL-Fc interacts with Schwann cell ECM, we have grown Schwann cells in the presence of ascorbate, which induces the formation of ECM fibrils and basal lamina–like structures (). In ascorbate-treated Schwann cells, binding of COL-Fc was detected to the cell surface, as well as to ECM deposits located between the cells (). COL-Fc still bound these ECM deposits in cultures that were pretreated with ammonium hydroxide and Triton X-100, a procedure that completely removed the cells from the slide but left the ECM fibrils intact (). To determine whether the interaction of COL-Fc with the ECM was specific to Schwann cells, we examined whether it binds astrocytes, which are known to accumulate similar ECM structures in culture (). As depicted in (H and I), COL-Fc bound to the cell surface of astrocytes, but it did not bind to cell-free areas or to cultures pretreated with ammonium hydroxide and Triton X-100. These results demonstrate that the extracellular region of gliomedin contains distinct domains that mediate its interaction with both neurons and Schwann cells. They further suggest the existence of a novel glial ligand for gliomedin embedded within the Schwann cell ECM.
The domain organization of gliomedin is reminiscent of transmembrane collagens, all of which are type II transmembrane proteins that are cleaved from the cell membrane by furin-type endoprotease (). Hence, the ability of gliomedin to bind ECM deposits that are not cell surface–associated raises the question of whether it is secreted from Schwann cells. To examine this possibility, we immunoprecipitated gliomedin from cell lysates and media of cultured Schwann cells using an antibody to the cytoplasmic tail (Ab836), or an antibody directed to the olfactomedin domain (Ab320). As a control, we used human embryonic kidney (HEK)-293 cells that were transfected with a C-terminal, myc-tagged, full-length gliomedin cDNA. Immunoblots were then performed using an antibody to myc tag or an antibody that recognizes a short peptide sequence between the collagen and the olfactomedin domain (Ab720; recognition sites of the various antibodies is schematically depicted in ). Ab836 and Ab320 immunoprecipitated an 89-kD protein from the lysates of Schwann cells that correspond to the transmembrane form of gliomedin (, middle). In transfected HEK-293 cells, the transmembrane form of gliomedin appeared as a doublet consisting of the 89-kD and a weaker ∼92-kD band. In addition, Ab320 specifically precipitated a 91-kD protein from the medium of both cell types and a 45-kD protein from the transfected HEK-293 cells. The 45-kD protein was also detected in the medium of Schwann cells that were maintained in culture for longer periods of time (, right). Treatment of gliomedin immunocomplexes with N-glycosidase revealed that the core protein of the secreted form is smaller by ∼11 kD than the transmembrane protein (62 vs. 73 kD; ). This analysis demonstrates that gliomedin is secreted from Schwann cells as a major 91-kD and a minor 45-kD protein.
To directly determine the cleavage sites in gliomedin, we used an Fc-fusion protein in which the Fc region was fused to the C-terminal region of the full-length gliomedin. The rationale behind this approach was that because gliomedin is a type II transmembrane protein, cleavage of the protein should result in the release of an Fc-tagged protein to the medium, which could be purified using protein A agarose. Two distinct bands of ∼116 and 80 kD were purified from the medium of HEK-293 expressing this construct and were subjected to N-terminal amino acid sequencing (). The N-terminal sequence of the lower band was identified as DDTLV, which corresponds to position 278 of rat gliomedin (NP_852047), which is located just before the beginning of the olfactomedin domain. Despite several attempts, we were unable to obtain the N-terminal sequence of the larger band. Analyzing the amino acid sequence of gliomedin between the transmembrane and the collagen domains (position 38–138) for functional sites using the Eukaryotic Linear Motif server (; ), revealed the presence of a putative furin cleavage site (RNKR) at position 91–94. To determine whether this proteolytic site is important to the processing of gliomedin, we have constructed a mutant of the full-length gliomedin by replacing arginine at position 91 with glycine and arginine at position 94 with alanine (R91G94A). This mutant was transfected to HEK-293 cells, and the medium of these cells was analyzed 2 d later for the presence of gliomedin (). Mutating these arginines caused a marked reduction in the secretion of gliomedin to the medium, and was accompanied by an accumulation of the 92-kD transmembrane form in the cells. These results indicate that proteolytic cleavage of the mature 92-kD form by a furin-like enzyme mediates the shedding of the extracellular domain of gliomedin from the cell surface. In addition, and less frequently, another cleavage of the molecule at position 278 may occur, which separates the olfactomedin domain from the collagen triplex. Immunolabeling of teased rat sciatic nerves demonstrated a nodal staining of gliomedin using Ab320 and Ab720, but not when Ab836 was used (Ab836 does recognize gliomedin in Schwann cells; see ), demonstrating that only the extracellular domain of gliomedin was detected at the nodes (). Furthermore, the transmembrane form of gliomedin was not detected at nascent nodes during their development at sites that were labeled with the antibodies to the extracellular domain (unpublished data). Thus, we concluded that the main form of gliomedin found at the nodes of Ranvier is the cleaved form that contains the entire extracellular domain, including its collagen and olfactomedin domains.
The presence of a protease-resistant collagen domain in gliomedin indicates that it may assemble into homotrimers or higher-order multimers after cleavage from the cell surface. To test this possibility, we immunoprecipitated gliomedin from Schwann cells that were grown in the absence or presence of ascorbate, which stimulates the deposition of fibrillar collagen by Schwann cells (). Under reducing conditions, only a monomeric form of gliomedin with an apparent molecular weight of 89 kD protein was detected in both untreated and ascorbic acid–treated cells (). However, in the presence of a chemical cross-linker (BS), only high-molecular weight multimers of gliomedin were detected in ascorbic acid–treated cells (). These multimers were detected using an antibody to the extracellular, but not to the cytoplasmic tail, of gliomedin (, right). Collectively, these results demonstrate that ascorbate treatment of Schwann cells induced the secretion, cleavage, and oligomerization of gliomedin.
To determine which region in gliomedin mediates its self-association, we examined the ability of purified Fc-fusion proteins containing its entire extracellular domain (ECD-Fc), olfactomedin domain (OLF-Fc), ECD lacking the olfactomedin domain (COL-Fc), ECD lacking the coiled-coil domain (ECDdCC-Fc), and ECD lacking its N-terminal linker region (ECDdNTR) to precipitate a full-length myc-tagged gliomedin from transfected HEK-293 cells (). Similar amounts of the Fc-fusion proteins were used, as detected by immunoblotting with an antibody to human Fc (unpublished data). As an additional control, we used HEK-293 cells expressing a myc-tagged PSD93. Whereas ECD-Fc and COL-Fc specifically pulled-down gliomedin, ECDdCC-Fc was less efficient (estimated 50% reduction), and only a weak signal was obtained using ECDdNTR-Fc. OLF-Fc was unable to recognize and precipitate gliomedin (). To further explore the functional significance of these observations, we examined whether self-association of gliomedin is required for its interaction with neurofascin and NrCAM. To this end, COS7 cells expressing neurofascin (NF186) were incubated with the different Fc-fusion proteins as either dimers (i.e., caused by the presence of the Fc) or as multimers by preclustering them with a secondary antibody to human Fc (). It was found that, in contrast to ECD-Fc and ECDdCC-Fc, which bound to NF186-expressing cells without preclustering, only multimeric forms of ECDdNTR-Fc and OLF-Fc were able to recognize neurofascin. Similar results were obtained using DRG neurons (unpublished data). Thus, although the olfactomedin domain of gliomedin mediates its direct interaction with the axonodal CAMs (), multimerization of the extracellular domain of gliomedin, which is conferred by other sequences present at the N-terminal region of the molecule, is essential for this interaction to take place.
The observation that a secreted extracellular domain of gliomedin can bind to Schwann cells, has prompted us to examine whether the endogenous gliomedin is incorporated into the Schwann cell ECM. Thus, we immunolabeled cultured rat Schwann cells with antibodies to gliomedin, together with antibodies to the known components of Schwann cell ECM, laminin (), and the α4(V) collagen chain (). In Schwann cells grown in the absence of ascorbic acid, gliomedin was mostly present on the cell surface, whereas α4(V) immunoreactivity was detected on the cell surface and between the cells. Growing the cells in the presence of ascorbic acid for 48 h resulted in a dramatic incorporation of gliomedin into fibrillar ECM deposits, where it was found to be colocalized with α4(V), mainly in the larger fibrils (). Similar results were obtained using antibodies to perlecan and the α1(V) chain of collagen (unpublished data). Previous studies have shown that after ascorbic acid treatment, α4(V) incorporates into detergent-insoluble ECM material (). To examine whether gliomedin was incorporated into similar structures, we immunoprecipitated gliomedin from ascorbic acid–treated Schwann cells that were extracted with a modified RIPA buffer, or with a RIPA buffer containing 1% SDS (). Cells grown in the absence of ascorbic acid were used as a control. Western blot analysis of these immunoprecipitants showed that the inclusion of SDS significantly increased the amount of gliomedin detected in the lysates of ascorbic acid–treated cells, but had no effect on the immunorecovery of gliomedin from untreated cells. A large amount of gliomedin was detected in the insoluble material obtained after the extraction of ascorbic acid–treated cells, with the modified RIPA buffer lacking SDS (, bottom), whereas the addition of SDS to the lysis buffer significantly reduced the amount of gliomedin in the insoluble pellet.
To ascertain that gliomedin was incorporated into Schwann cell ECM, cultures of Schwann cells grown in the presence of ascorbic acid were washed with ammonium hydroxide and Triton X-100 to remove the cells, and then immunolabeled using antibodies to gliomedin and laminin (). In nontreated cultures, gliomedin immunoreactivity was detected on the surface of the cells, as well as in fibrillar structures that were also labeled for laminin (top). After cell removal, most of the gliomedin immunoreactivity was associated with laminin-positive fibrils, further supporting the conclusion that it is incorporated into the ECM network. Notably, the extracellular domain of neurofascin (NF155-Fc; NF155 and NF186 are used interchangeably, as they both bind gliomedin) bound to ECM deposits that remained after Schwann cell removal, demonstrating that this axonodal CAM still recognizes the ECM-incorporated multimers of gliomedin (). The binding of Fc-fusion proteins containing the extracellular domain of Necl1/SynCAM3, which is a CAM that interacts with Schwann cells, was abolished after the removal of the cells (), demonstrating that the ammonium hydroxide and Triton X-100 wash used did not create nonspecific Fc-binding sites on the slides. Finally, similar experiments done using rat astrocytes revealed that although these cells secrete large quantities of gliomedin to their culture medium (not depicted), NF155-Fc binding was only detected on the cell surface (), indicating that gliomedin was not incorporated into the ECM produced by these cells. This conclusion is further supported by the observation that the extracellular domain of gliomedin did not bind astrocyte ECM, which suggests that Schwann cells produce a specific ECM ligand for gliomedin (). Collectively, these results demonstrate that upon ascorbic acid treatment, gliomedin is secreted from the cells and forms high-molecular weight multimers that are entrapped in the Schwann cell ECM.
At the perinodal space, gliomedin is concentrated on the Schwann cell microvilli (), which are embedded in the proteoglycan-rich perinodal matrix. To examine whether the association of gliomedin with Schwann cells ECM is mediated by heparan sulfate proteoglycans (HSPGs), we used NF155-Fc as a specific affinity reagent to detect gliomedin in ascorbate-treated Schwann cells that were incubated with heparin. As depicted in , in the absence of heparin, gliomedin was present on both the cell surface and in ECM deposits between adjacent cells. In cultures incubated for 30 min with heparin, gliomedin was released from the ECM and was only present on the cell surface (). The removal of gliomedin from Schwann cell ECM by heparin was even clearer using cultures that were prewashed with ammonium hydroxide and Triton X-100 to remove the cells (). In contrast to gliomedin, heparin had no effect on the distribution of α4(V) () or other Schwann cell ECM components, such as laminin (unpublished data). Immunolabeling with an antibody to the cytoplasmic domain of gliomedin indicated that, in heparin-treated Schwann cells, NF155-Fc bound to the transmembrane form of gliomedin present on the cell surface ().
The efficient removal of secreted gliomedin by heparin indicates that its association with Schwann cells ECM requires HSPGs. To establish whether gliomedin interacts with HSPGs, we tested the effect of heparin on the binding of its extracellular domain (ECD-Fc) to Schwann cells. As depicted in , heparin completely inhibited the binding of ECD-Fc, suggesting that gliomedin binds to glycosaminoglycan side chains of HSPGs expressed by Schwann cells. To examine whether gliomedin binds directly to heparin, we incubated the Fc-fusion proteins containing the different domains of gliomedin with heparin-Sepharose and detected the bound material by immunoblotting with an antibody to human Fc (). As control, we used protein A beads, which precipitated all of the Fc-fusions tested. The extracellular domain of gliomedin (ECD), as well as an ECD lacking the olfactomedin domain (COL), the coiled-coil domain (ECDdCC), or its N-terminal linker region (ECDdNTR), bound to the heparin affinity matrix. In contrast, neither the olfactomedin domain alone (OLF) nor an Fc-fusion containing the extracellular region of the gliomedin-related myocilin protein were able to bind heparin, indicating that the heparin-binding sequence in gliomedin is located outside its olfactomedin domain. Collectively, our results demonstrate that the incorporation of gliomedin to Schwann cell ECM is mediated by its binding to HSPGs.
xref
#text
Dissociated rat DRG cultures were grown in Neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen) and 50 ng/ml NGF (Neurobasal medium; Alomone Labs) for 5 d before being used in binding experiments. Purified DRG neurons were established by treating dissociated mixed cultures with two cycles (2 d each) of NB medium containing 10 μM uridine/10 μM 5′-Fluoro 2′-deoxyuridine (Sigma-Aldrich) to eliminate fibroblasts and Schwann cells. Schwann cells isolated from P4 rat sciatic nerve were grown in Schwann cell proliferation medium (DME, 3% FBS, 10% NDFβ conditioned medium, 2 μM forskolin, 2 mM glutamine, and 1 mM pyruvate) on primaria culture dishes (Falcon). To stimulate ECM assembly, Schwann cells were treated with 50 μg/ml L-ascorbic acid (Sigma-Aldrich) for 48 h. To remove cells from the culture dish, cultures were treated with PBS + 0.5% Triton X-100 and 20 mM ammonium hydroxide for 10 s and subsequently washed with PBS.
Gliomedin Fc fusions COL-Fc (residues 49–294), OLF-Fc (residues 288–543), ECD-Fc (residues 49–543), and NF155-Fc were previously described (). Fc fusions ECDdNTR-Fc and ECDdCC-Fc (residues 139–543 and 62–543, respectively) were generated by cloning the corresponding cDNA to pSX-Fc. Myocilin-Fc was generated by cloning full-length myocilin amplified by PCR from rat Schwann cell cDNA into pCX-Fc. Generation of the mutant proteins gliomedin R91G and R94A was done using the 5′ primers CCCATGAGTGCAGCCGGCAATAAGCGAGC (for R91G, NaeI site was inserted), and GCAGCGCGCAATAAGGCTAGCCACGGCGGCGAG (for R94A, NheI site was inserted) and their complementary 3′ primers. PCR was performed using Pfu DNA Polymerase (Promega) on pRKpF10 () as a template, followed by a DpnI digestion.
Polyclonal and monoclonal antibodies to gliomedin (Ab720 and mAb94) were previously described (). Polyclonal antibody 320 was generated by immunizing rabbits with the aforementioned purified recombinant protein OLF-Fc. Polyclonal antibody 836 was generated against a synthetic peptide corresponding to residues 1–16 that comprise the cytoplasmic tail of gliomedin. Polyclonal antibodies to α4(V) and to α1(V) were obtained from David Carey. Rabbit anti–laminin antibody was purchased from Sigma-Aldrich.
Cells were grown on poly-L-lysine–coated (Sigma-Aldrich) coverslips. For binding experiments, conditioned media containing various Fc-fusions were either mixed with a Cy3-conjugated anti–human Fc antibody for 30 min before the binding procedure or were not mixed (“nonclustered”). After a 30-min incubation of the cells with the conditioned media at RT, cells were washed and fixed with 4% PFA for 5 min at RT. In the case of nonclustered binding, coverslips were incubated with a Cy3-conjugated anti–human Fc antibody after fixation for 30 min at RT. For antibody labeling, cells were fixed in 4% PFA, washed with PBS, and incubated in blocking solution (PBS, 10% normal goat serum, 0.1% Triton X-100, 1% glycine) for 30 min. Primary antibodies diluted in blocking solution were added for 1 h at RT, followed by washing with PBS and incubation with secondary antibodies diluted in blocking solution for 40 min. Coverslips were then washed, mounted in elvanol, and analyzed on a microscope (Eclipse E1000; Nikon; objectives 20×/0.5 NA, 40×/1.3 NA, and 60×/1.4 NA) equipped with a camera (ORCA-ER; Hamamatsu). Fluorescence images were acquired using Openlab software (Improvision) and figures were mounted using Photoshop software (Adobe). In the case of laminin and collagen V staining, live cultures were incubated with primary antibodies for 45 min, followed by washing and fixing. Secondary antibodies were added after a 30-min blocking as described for nonclustered cells. To assess the effect of heparin on the gliomedin labeling, Fc fusion binding or antibody labeling were performed after a 30-min treatment with 10 μg/ml heparin in PBS supplemented with Ca and Mg. Teased sciatic nerves were prepared and immunolabeled as previously described ().
For immunoprecipitation, Schwann cells or transfected cell conditioned media and cell extracts (lysed in 50 mM Hepes, pH 7.2, 150 mM NaCl, 1 mM EGTA, 10% glycerol, 1% Triton X-100, 2 mM PMSF, and 10 μM aprotinin and leupeptin) were immunoprecipitated by polyclonal anti-gliomedin antibodies cross-linked to protein A–Sepharose beads using Dimethyl Pimelimidate (Sigma-Aldrich). Samples were analyzed by electrophoresis on 10% polyacrylamide SDS gels. For cross-linking experiments, Schwann cells were treated with BS (Pierce Chemical Co.) according to manufacturer's protocol before cell lysis. Cell extracts were subjected to the aforementioned immunoprecipitation. To asses the solubility of gliomedin after ascorbate treatment, immunoprecipitation was performed in an SDS-free RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% NP-40, 0.25% Na-deoxycholate, 1 mM EDTA, 2 mM PMSF, and protease inhibitors) or in a RIPA buffer containing 0.1% SDS. For pull-down experiments, protein A beads bound with various gliomedin Fc-fusion proteins were mixed with cell extracts of HEK-293T cells transfected with either gliomedin-myc or PSD93-myc as control, for 4 h at 4°C. Beads were subsequently washed and analyzed by electrophoresis and immunoblot using anti-myc antibodies. For heparin pull-down experiments, conditioned media of various gliomedin Fc-fusion proteins and myocilin-Fc were mixed with either protein A–Sepharose or heparin–Sepharose beads for 1 h at RT. Beads were subsequently washed and resolved by electrophoresis and immunoblot using anti–human antibodies. N-terminal sequencing of a secreted Fc-tagged gliomedin was determined using Procise 491 protein sequencer (Applied Biosystems). |
The skin is composed of an epithelial (epidermis and hair follicle [HF]) and a mesenchymal compartment (dermis, subcutis, and dermal papilla [DP]) joined and maintained together by a basement membrane (BM). The interfollicular epidermis contains multiple layers of keratinocytes at different stages of differentiation, from a basal layer of undifferentiated, proliferating keratinocytes attached to the BM, to terminally differentiated, cornified cells (). The HF is an epidermal appendage, which arises as an epithelial cone from the fetal epidermis after a series of epithelial–mesenchymal cues. The mature HF epithelium consists of a central hair shaft (HS), surrounded by an inner and an outer root sheath (IRS and ORS, respectively). HS and IRS differentiation from the hair matrix (HM) is induced by mesenchymal cues from the connective tissue sheath and the DP. The mature HF has the ability to involute and regenerate, with cyclically alternating periods of rapid growth (anagen), apoptosis-driven regression (catagen), and relative quiescence (telogen). During each growth period, the progeny (transient amplifying [TA] cells) of epithelial stem cells located in the bulge region of the ORS extends into the mesenchymal compartment and generates a new HM. Here, epithelial cells change migration direction and terminally differentiate into IRS or HS ().
Basal keratinocytes express several integrins, including α2β1, α3β1, α9β1, αvβ5, and α6β4 integrins (). The α6β4 integrin is the core component of hemidesmosomes anchoring keratin filaments to the BM, whereas α3β1 and α9β1 integrins link the actin cytoskeleton to the BM. The α2β1 is found around the entire basal keratinocytes, where it is thought to mediate cell–cell interactions. ORS cells express α2β1, α3β1, and α6β4 integrins at different levels according to the region of the HF (). In vitro studies with keratinocytes and genetic manipulations in mice revealed that β1 integrins regulate adhesion and differentiation of epidermal cells and play an essential role for hair germ invagination, ORS cell migration, and sustained HM proliferation during HF morphogenesis (; ; ; ). An important and still largely unanswered question is how integrins mediate these functions in skin and HFs. Because integrin cytoplasmic domains lack actin binding sites and enzymatic activity, signaling is implemented through accessory molecules such as talin, α-actinin, and integrin-linked kinase (ILK; ). ILK is composed of N-terminal ankyrin repeats, a pleckstrin homology–like domain and a putative, C-terminal kinase domain (; , ). ILK was given its name based on the enzymatic activity of its kinase domain (; ; ), which was shown to phosphorylate several target proteins, including protein kinase B (PKB)/Akt and glycogen synthase kinase (GSK) 3β. The significance of the ILK activity, however, is controversial as in vitro and in vivo results in flies, worms, and mice point toward an adaptor rather than an enzymatic function of ILK (; ; ; ; ; ). A recent report proposed that the ILK activity is biologically relevant for transformed epithelial cells but not normal cells (). Whether the controversy may indeed be ascribed to the different biological systems used in the past to investigate ILK function awaits further studies. Another important function of ILK is its ability to link integrins to the actin cytoskeleton and to modulate actin reorganization (; ; ; ). Almost all proteins that bind ILK bind and/or regulate actin dynamics. They include PINCH1 and PINCH2, which bind actin modulators and connect ILK to growth factor receptors, the parvin family of F-actin binding proteins, and paxillin, which recruits actin binding and regulatory proteins, including vinculin, talin, α-actinin, and FAK (for reviews see ; ). HF development and cycling is crucially dependent on the inactivation of GSK-3β in HM cells (; ). Active, nonphosphorylated GSK-3β can phosphorylate β-catenin bound to a protein complex, collectively called the β-catenin degradation complex. Phosphorylation of GSK-3β inactivates the kinase and leads to stabilization and translocation of β-catenin to the nucleus, where it associates with the Lef1/Tcf family of DNA binding proteins to activate the transcription of target genes, such as cyclin D1, c-myc, homeobox containing transcription factors, Lef1, and hair-specific keratins (; for review see ). ILK can modulate the stability of β-catenin either through phosphorylating GSK-3β (; ) or through inhibiting the β-catenin degradation complex () and could therefore play a central role for HF morphogenesis.
To test the function of ILK during epidermis and HF development, we deleted the ILK gene in keratinocytes. We found that loss of ILK compromises epidermal keratinocyte adhesion and disrupts HF formation, leading to progressive hair loss. The HF defect was not due to an abnormal β-catenin stability, HM differentiation, or stem cell maintenance. Instead, the accumulation of proliferating ORS cells points to an impaired HF downward growth in vivo.
To delete the ILK gene in keratinocytes, floxed ILK mice were intercrossed with animals carrying the keratin 5 (K5)–Cre transgene (ILK-K5 mice). Littermates carrying heterozygous floxed ILK gene and the K5-Cre transgene served as controls (ILK Co). K5-mediated Cre expression deleted the ILK gene in back skin at around embryonic day 15, decreased ILK levels in newborn skin, and led to the loss of the ILK protein thereafter (Fig. S1 A, available at ). Western blot analysis in back skin epidermis of 6-d-, 2-wk-, 4-wk-, and 10-wk-old mice confirmed the sustained absence of ILK (). Immunostained sections of 2-wk-old control mice revealed ILK in basal epidermal keratinocytes, ORS, HM, DP, and the arrector pili muscle (). ILK was absent from epidermis and HF epithelium of ILK-K5 skin but still present in DP ().
ILK-K5 animals were indistinguishable from control littermates at birth. At 1–2 wk, when control animals developed their hair coat, ILK-K5 animals had scattered hair with partial alopecia. This appearance endured until around 4 wk of age and was followed by progressive hair loss, leading to persistent alopecia by 6–8 wk (). A reticular pigmentation pattern developed on the back skin of 8-wk-old ILK-K5 mice (), whereas hair coat and hair cycle–dependent skin color changes occurred normally in control mice.
The epidermis of ILK-K5 mice was morphologically normal at birth and postnatal day (P) 2 but became progressively hyperplastic (at P7–9, four to five cell layers, and at P28, six to seven cell layers; ). Although basal keratinocytes were polarized and tightly attached to the BM in control skin, they appeared flattened in the mutant epidermis and were often detached along the dermal–epidermal junction (DEJ; , asterisks). The detachment became more severe with age (at P7, 5–10% of total epidermal length; at P14, 30–50%; and at P70, up to 70%) but did not result macroscopically in visible skin blisters.
The most striking phenotype was a severe impairment of HF development in ILK-K5 mice characterized by a progressive growth retardation, which was first visible at around P2 (). By P14, control mice had completed HF morphogenesis, with all hair bulbs residing deep in the subcutis. In contrast, ILK-K5 HFs diverged into two subpopulations. Approximately 33% of the mutant HFs reached the final stages of HF morphogenesis but were shortened and profoundly distorted. They displayed substantial hyperplasia of the ORS with up to six cell layers and condensed DPs (, ▴). Approximately 66% of the mutant HFs were arrested in their development. They failed to reach down deeper than the reticular dermis and showed defective morphogenesis with distorted or absent HS formation and misshapen HM and DP (, ▪). A plausible explanation for the varying HF populations is the combination of an asynchronous HF morphogenesis () and the perinatal loss of ILK protein expression: fully developed HFs (, ▴) lost ILK late in morphogenesis, whereas arrested HFs (▪) lost ILK early in morphogenesis. At P28, none of the ILK-K5 HFs was able to initiate anagen characterized by HF downgrowth into the subcutis (). By 10 wk of age, the ILK-K5 HFs were resorbed () and melanin condensates within the dermis gave rise to a reticular skin pigmentation ().
Cell detachment in ILK-K5 skin points to a compromised integrin function that could be caused by altered expression, activity, localization, or weaker linkage to the actin cytoskeleton. Integrin function was tested with adhesion assays using fibronectin (FN), collagen I (Col I), collagen IV (Col IV), and laminin 332 (LM332) as substrates. Although interaction with poly--lysine was similar between ILK-K5 and control keratinocytes, adhesion to the ECM substrates was significantly diminished in ILK-K5 keratinocytes ().
Integrin expression determined by FACS revealed strong β1 integrin expression and a comparable Mn-triggered activation of β1 integrins on freshly isolated control and ILK-K5 keratinocytes. However, a subpopulation of ILK-K5 cells expressed lower levels of β1 integrin (Fig. S1 B). The expression levels of the α6, β4, and αv integrin subunits were not changed on ILK-K5 keratinocytes, whereas the α3 and α2 integrin chains were slightly up-regulated (Fig. S1 B). In situ immunostaining revealed differences in integrin localization at the cellular level. In control skin, β1 integrin was expressed around the entire surface of basal keratinocytes () and β4 and α6 integrins along the DEJ ( and ). In ILK-K5 skin, the β1 integrin subunit was present on basal keratinocytes but also on many suprabasal cells (), which maintained K14 expression (, middle). The localization of α6 and β4 integrins on ILK-K5 basal keratinocytes was comparable to control skin, with the exception of a few areas lacking detectable α6 and β4 integrin and some suprabasal cells showing a strong staining for α6 and β4 integrin (). The latter cell population likely expressed similar levels of α6β4 integrins as basal keratinocytes, as FACS analysis of freshly isolated keratinocytes did not distinguish two populations of α6β4-expressing keratinocytes (Fig. S1 B).
The decreased keratinocyte adhesion was associated with severe BM defects. Although control skin showed a linear staining of LM332 along the DEJ and around HFs, ILK-K5 skin displayed irregular deposits of LM332 at the DEJ with areas of massive LM332 (, asterisks) diffusion into the dermis and dotlike deposits adjacent to integrin-positive suprabasal keratinocytes (, right, arrowhead). EM of a control skin revealed a regular BM structure at the DEJ, whereas mutant skin showed an abnormal BM with discontinuities in the lamina densa between hemidesmosomes (). The number of hemidesmosomes was normal except in areas with detached epidermis, where the number was reduced. Collectively, these data demonstrate that loss of ILK weakens integrin-mediated adhesion of basal keratinocytes to the BM and abrogates BM integrity.
ILK-deficient epidermis was hyperplastic (). Ki67 immunostaining revealed that P4 epidermis from control as well as ILK-K5 skin contained comparable numbers of proliferating cells almost exclusively in the basal layer. At P7, however, ILK-K5 skin contained normal numbers of proliferating cells in the basal layers and, in addition, a significant number of proliferating cells in the suprabasal layers (Fig. S2, A and B, available at ). The ectopic keratinocyte proliferation was also observed in Ki67-immunostained skin (). It occurred in areas with aberrant and normal BM and was associated with β1 and β4 integrin expression ().
The defective keratinocyte adhesion could trigger a chronic wound healing response with infiltrating inflammatory cells, which in turn may induce the ectopic proliferation of suprabasal keratinocytes in vivo. To test this, we searched skin sections from control and ILK-K5 mice for the presence of granulocytes and macrophages. As expected, granulocyte and macrophage infiltrates were absent from P7 as well as P14 control skin (Fig. S3, A and B, available at ). ILK-K5 skin also lacked granulocyte and macrophage infiltration at P7 (Fig. S3, A and B), when abundant proliferation of suprabasal keratinocyte was already evident (Fig. S2 B). At P14, however, macrophages accumulated around ILK-K5 HFs and granulocytes beneath the epidermis (Fig. S3, A and B).
The presence of proliferating, integrin-positive keratinocytes in suprabasal layers points to an aberrant differentiation and/or mislocalization of undifferentiated ILK-K5 keratinocyte. To investigate differentiation, we analyzed the expression of epidermal keratins. K14 was expressed in basal cells and weakly extended into the first suprabasal layer of control epidermis (). In ILK-K5 skin, K14 was expressed suprabasally in up to five cell layers (). Normal suprabasal cells switched off K14 and K5 expression and instead expressed K10 (). In ILK-K5 epidermis, K10 was absent from basal cells but strongly expressed in the four to five suprabasal cell layers. In addition, there were often patches of cells lacking K10 but expressing integrins (, asterisks) and high levels of K14 ( and ). Furthermore, although loricrin was confined to the stratum granulosum and appeared as a thin linear signal in control epidermis, in ILK-K5 epidermis, loricrin was found in two to three cell layers, which contained large and round keratinocytes with prominent nuclei (). These data suggest that loss of ILK sustains proliferation and expression of basal layer markers in suprabasal cell layers and delays keratinocyte differentiation.
ILK-deficient keratinocytes have a flattened shape (), suggesting that their polarity was impaired. To investigate keratinocyte polarity in vivo, we compared F-actin and the distribution of cell–cell adhesion molecules between control and ILK-K5 epidermis. In control epidermis, F-actin distributed to the apical and lateral plasma membranes of basal keratinocytes, whereas in ILK-K5 epidermis, the F-actin was also present at the basal plasma membrane zone facing the BM, where it frequently colocalized with nidogen (). Similar F-actin defects were also seen in mutant HFs ().
In normal skin, E-cadherin and its junctional adaptor protein β-catenin were found at the lateral and apical plasma membrane of basal keratinocytes ( and not depicted). In ILK-K5 skin, the E-cadherin and β-catenin staining was normally distributed in areas where the epidermis was attached to the dermis. In areas where the epidermis was detached from the BM, both E-cadherin and β-catenin were redistributed to the basal side of basal keratinocytes ( and not depicted). In epidermal lysates, E-cadherin and β-catenin protein levels were indistinguishable between control and ILK-K5 samples (). The expression and localization of desmosomal components such as plakoglobin and desmoplakin ( and not depicted), as well as the ultrastructure of desmosomes (Fig. S3 C), were unaffected in all areas of the ILK-K5 epidermis. We conclude that ILK controls cell polarity by maintaining the integrity of the actin cytoskeleton and BM and not by regulating E-cadherin expression or the formation of cell–cell junctions.
A possible role of ILK for hair epithelium differentiation stems from the observation that ILK controls β-catenin–Lef1–mediated gene transcription either by phosphorylating and inactivating GSK-3β () or by stabilizing β-catenin (). To test whether GSK-3β and the downstream β-catenin–Lef1 complex were affected by the loss of ILK, we performed a series of different experiments. Immunoblotting of lysates from freshly isolated keratinocytes revealed that the total levels of GSK-3β and the extent of phosphorylation of Ser9 did not differ between control and ILK-K5 samples (Fig. S2 D). Immunostaining revealed that Lef1 and nuclear β-catenin were present in the precortical HM and HS cortex of control as well as fully developed ILK-K5 HFs (, ▴). Moreover, both proteins could clearly be detected in the Ki67-positive HM cells of prematurely growth-arrested ILK-K5 HFs (, ▪). To determine the activity of the nuclear β-catenin–Lef1 transcription factor complex, ILK-K5 mice were intercrossed with reporter mice, in which β-galactosidase expression is controlled by nuclear β-catenin–Lef1 (). The expression of β-galactosidase was clearly visible in the HS of control and fully developed ILK-K5 HFs (, ▴) and in cells of growth-retarded ILK-K5 HFs (, ▪). Normal activity of the β-catenin–Lef1 complex was further confirmed by determining the β-catenin–Lef1–dependent expression of IRS-specific keratins. The IRS keratin K6irs1 () and K6irs2-4 (not depicted) were normally expressed in ILK-K5 HFs. Similarly, the expression of ORS keratins and several HS-specific markers (e.g., hHa1) was also normal in both populations of ILK-K5 HF (even though the localization of the K6irs1-positive cells in the shortened mutant HF was abnormal; ). Altogether, these findings demonstrate that ILK regulates neither the phosphorylation of GSK-3β and the stability and activity of β-catenin in HFs nor the differentiation of HM into the IRS or HS.
Loss of β1 integrin expression leads to reduced proliferation of epidermal keratinocytes and HF matrix cells (; ). To assess whether altered proliferation of the ILK-K5 HM accounts for the abnormal hair development, we performed BrdU incorporation assays and determined Ki67 expression. At P7, both fully developed ILK-K5 HFs (, ▴) as well as growth-retarded HFs (, ▪) showed an elevated number of proliferating cells in the ORS. To quantify the number of proliferating cells, we counted their numbers on fully developed ILK-K5 HFs (, ▴), thereby ensuring comparison of identical HF developmental stages. Counting of proliferating cells in P7 ILK-K5 HFs revealed that the increased number of proliferating ORS cells was associated with a slight but not significantly lower amount of proliferating cells in the HM (). At P14, however, the number of proliferating cells significantly diminished in the HM and further increased in the ORS (), suggesting that ILK-deficient, rapidly proliferating TA cells are capable of proliferating but accumulate in the ORS. Moreover, neither TUNEL assays nor immunostaining for activated caspase-3 revealed an elevation in apoptotic cell numbers, indicating that cell survival was unaffected in the ILK-K5 HFs (unpublished data).
The ORS cells originate from the CD34-positive stem cell population that is located in the hair bulge (). To determine whether ILK loss led to the elimination of CD34-positive cells, we immunostained P24 skin sections. Both control and ILK-K5 HFs contained CD34-positive cells in their hair bulges (). The formation of secondary hair germs is driven by the proliferation of hair bulge-derived TA cells triggered by the inductive activity of the DP. At P24, normal HFs are at the onset of anagen, and Ki67+ TA cells appeared adjacent to the DP (, left). Ki67 staining of ILK-K5 skin revealed the presence of two types of HFs: ∼65% contained proliferating cells, suggesting that ILK-K5 HFs were principally capable of entering early stages of anagen (, middle). The remaining ILK-K5 HFs lacked proliferating cells (, right), likely because they were detached from the DP (, right) or connected to a malformed DP () and, hence, did not receive the inductive signals. Collectively, these data suggest that ILK-K5 HFs contain CD34-positive stem cells that give rise to TA cells, which require ILK to migrate down to the HM or to trigger the downward growth of hair germs.
We first performed transwell migration assays and observed that migration of primary ILK-K5 keratinocytes on LM332, as well as their invasion through laminin-rich matrigel, was significantly impaired (). Next, we scratched monolayers of primary keratinocytes and observed the closure of the scratch over 12 h using time-lapse video microscopy. After scratching, control keratinocytes displayed directional migration and invaded the denuded area () with a mean wound closure speed of 42.3 μm/h, leading to the closure of the scratch within 12 h (). In contrast, ILK-K5 keratinocytes often stopped and migrated back- and sideward (), with a reduced wound closure speed of 9.7 μm/h (). Furthermore, single-cell tracking at the migration front revealed a migration velocity of 0.7 ± 012 μm/min by ILK-K5 keratinocytes versus 0.9 ± 016 μm/min by control cells (P < 0.01).
To more closely evaluate the migration defect, we performed time-lapse microscopy of single keratinocytes. Control keratinocytes formed broad, usually single and stable leading edge lamella with a mean persistence of 985 ± 339 s that allowed single cells to directionally migrate (; and Video 1, available at ). In sharp contrast, ILK-K5 lamellae were instable and collapsed within 618 ± 332 s (). Furthermore, the mutant cells constantly extended new lamellae toward different directions simultaneously, which gave rise to frequent changes of the migration direction and consequently prohibited directional movement ( and Video 2).
To precisely characterize lamellipodia behavior, we monitored and quantified the plasma membrane extension rates of migrating cells using kymography () over a period of 20 min. The lamellipodia of ILK-K5 keratinocytes persisted for a significantly shorter time () and protruded more frequently than those of control keratinocytes (). Collectively, these data indicate that ILK is important for the stability and dynamics of the lamellae/lamellipodia and hence for directional migration of keratinocytes.
The reduced adhesion of ILK-K5 keratinocytes to the ECM ( and ) can diminish the fixation of plasma membrane protrusions to the ECM, impair cytoskeletal reorganizations, and compromise integrin-triggered signaling, which in turn can cause the abnormal formation of leading-edge lamellipodia and impaired directional migration.
To test whether ILK is critical for the formation of integrin adhesion sites and integrin signaling, we isolated control and ILK-K5 keratinocytes. Both cultured cell types had comparable integrin profiles, β1 integrin activity, and α6β4-containing migration track patterns at the rear of the cell (Fig. S4, A and B, available at ). The size of ILK-K5 cells was smaller, reaching a threefold smaller spreading area 40 h after plating on a mixture of Col I and FN (Fig. S4 C). Talin staining of adherent cells revealed that ILK-K5 keratinocytes formed fewer focal complexes (FCs) in the leading-edge lamellipodia (). Additional immunostaining for paxillin and FAK showed that only 30% of the cells contained mature FAs () whose number per cell and size were significantly reduced (). The number of FAs in relation to the cell contact area, however, was not altered between ILK Co and ILK-K5 keratinocytes. In line with the severe spreading defect, ILK-K5 keratinocytes contained fewer stress fibers than control cells ().
ILK can associate with several FA components, which in turn can modulate the activity of adaptor and signaling proteins, including FAK and Rac1 (). Therefore, we tested whether their function is affected in ILK-K5 cells. Although total FAK levels were normal in ILK-K5 keratinocytes, the auto-activated form of FAK (pY397-FAK), as well as other tyrosine residues, such as Y861, were reduced ().
To test whether Rac-1 can be activated upon cell adhesion, we determined the levels of GTP-loaded Rac1 before and after cell seeding on LM322. Both ILK-K5 and control keratinocytes activated Rac1 to a similar extent (), indicating that the absence of ILK does not impair Rac1 activation in keratinocytes. Moreover, growth factor–induced activation of Rac1 became similarly increased in control and ILK-K5 keratinocytes (unpublished data).
t
h
e
p
r
e
s
e
n
t
p
a
p
e
r
,
w
e
r
e
p
o
r
t
t
h
a
t
a
k
e
r
a
t
i
n
o
c
y
t
e
-
r
e
s
t
r
i
c
t
e
d
d
e
l
e
t
i
o
n
o
f
t
h
e
I
L
K
g
e
n
e
i
n
m
i
c
e
l
e
a
d
s
t
o
a
b
n
o
r
m
a
l
H
F
m
o
r
p
h
o
g
e
n
e
s
i
s
a
n
d
e
p
i
d
e
r
m
a
l
d
e
f
e
c
t
s
w
i
t
h
b
l
i
s
t
e
r
s
,
e
c
t
o
p
i
c
k
e
r
a
t
i
n
o
c
y
t
e
p
r
o
l
i
f
e
r
a
t
i
o
n
i
n
s
u
p
r
a
b
a
s
a
l
c
e
l
l
l
a
y
e
r
s
,
a
n
d
a
b
n
o
r
m
a
l
k
e
r
a
t
i
n
o
c
y
t
e
d
i
f
f
e
r
e
n
t
i
a
t
i
o
n
.
M
u
t
a
n
t
H
F
s
p
r
o
d
u
c
e
d
p
r
o
l
i
f
e
r
a
t
i
n
g
p
r
o
g
e
n
i
t
o
r
c
e
l
l
s
,
w
h
i
c
h
a
c
c
u
m
u
l
a
t
e
d
i
n
t
h
e
O
R
S
a
n
d
f
a
i
l
e
d
t
o
r
e
p
l
e
n
i
s
h
t
h
e
H
M
.
I
n
v
i
t
r
o
e
x
p
e
r
i
m
e
n
t
s
r
e
v
e
a
l
e
d
t
h
a
t
I
L
K
-
d
e
f
i
c
i
e
n
t
k
e
r
a
t
i
n
o
c
y
t
e
s
w
e
r
e
u
n
a
b
l
e
t
o
f
i
r
m
l
y
s
t
a
b
i
l
i
z
e
l
a
m
e
l
l
i
p
o
d
i
a
,
l
e
a
d
i
n
g
t
o
i
m
p
a
i
r
e
d
d
i
r
e
c
t
i
o
n
a
l
m
i
g
r
a
t
i
o
n
a
n
d
p
r
o
v
i
d
i
n
g
a
p
o
t
e
n
t
i
a
l
e
x
p
l
a
n
a
t
i
o
n
f
o
r
t
h
e
a
c
c
u
m
u
l
a
t
i
o
n
o
f
p
r
o
g
e
n
i
t
o
r
c
e
l
l
s
i
n
t
h
e
O
R
S
.
To obtain mice with a keratinocyte-restricted deletion of the ILK gene, transgenic mice expressing under the control of the keratin-5 promoter () were crossed with floxed ILK mice (; ). Offspring were genotyped as described previously (). BatGal transgenic mice carry Lef1/Tcf binding sites in front of a minimal promoter and the gene () and were intercrossed with the ILK mutant mice.
Primary keratinocytes were cultured in keratinocyte growth medium containing 8% FCS and low Ca (45 μM) on cell culture dishes coated with a mixture of Col I (Cohesion) and FN (Invitrogen) to subconfluence as described previously (). Protein lysates from keratinocytes or epidermis were separated by SDS gel electrophoresis, blotted, and incubated with the indicated antibodies.
For GTPase pull-down assays, keratinocytes were cultivated to 70% confluence. Cells were then serum starved overnight and detached by Trypsin/EDTA treatment (Invitrogen). Detached cells were resuspended in serum-free keratinocyte growth medium and kept for 30 min in suspension. For adhesion-induced GTPase activation, cells were plated on a LM332-rich matrix produced by Rac-11P/SD squamous cell carcinoma cells for 30 min (). Cells were washed twice with PBS and then lysed in lysis buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1% Nonidet P-40, 10% glycerol, 2 mM MgCl, 1 mM NaF, and 1 mM NaVO; all from Sigma-Aldrich) supplemented with protease inhibitor cocktail tablets (Complete Mini, EDTA-free; Roche) and containing biotinylated PAK-CRIB peptide (a gift from J. Collard, Netherlands Cancer Institute, Amsterdam, Netherlands). Lysates were centrifuged at 20,000 for 10 min at 4°C, and the supernatant was subsequently incubated for 45 min at 4°C. Next, lysates were incubated with streptavidin-conjugated agarose beads (GE Healthcare) for 30 min at 4°C. Beads were washed three times with lysis buffer, resuspended in 2× SDS sample buffer, and boiled for 5 min at 95°C. The supernatant was subjected to SDS gel electrophoresis, Western blotting, and immunodetection by the indicated antibodies. The following antibodies were used for Western blot analysis: mouse mAb against ILK (clone 3; BD Biosciences); rat mAb against α-tubulin (); rabbit pAb against PKB/Akt and phospho-PKB/Akt (Ser473; Cell Signaling Technology); mouse mAb against GSK-3β (BD Biosciences); rabbit pAb against phospho–GSK-3β (Ser9; Biosource International); mouse mAb against cyclin D1/2 (Upstate Biotechnology); rabbit pAb against cyclin A (Santa Cruz Biotechnology, Inc.); rabbit pAb against p42/44 MAPK (Cell Signaling Technology); mouse mAb against phospho-p42/44 MAPK Thr202/204 (New England Biolabs, Inc.); rat mAb against E-cadherin (Zymed Laboratories); rabbit pAb against β-catenin (Sigma-Aldrich); rabbit pAbs against FAK (Upstate Biotechnology) and pFAK (Tyr397and Tyr861; Biosource International); mouse mAb against Rac1 (BD Biosciences); and goat anti–rat HRP, goat anti–mouse HRP, and goat anti–rabbit HRP (Bio-Rad Laboratories).
Skin samples were fixed in 4% PFA in PBS, pH 7.2, overnight, dehydrated in a graded alcohol series, and embedded in paraffin (Paraplast X-tra; Sigma-Aldrich) or frozen unfixed in OCT (Thermo Shandon). Immunohistochemistry of skin sections was performed as described previously (). For cellular immunostainings, keratinocytes were seeded on chamber slides (Nunc) coated with 5 μg/ml of purified LM332 or 30 μg/ml Col I and 10 μg/ml FN and allowed to spread for 40 h. Cells were washed in PBS, fixed in 4% PFA, and incubated with the indicated antibodies. To determine BrdU incorporation, mice were injected with BrdU (100 μg/g body weight) 2 h before killing. Assessment of proliferation of cultured keratinocytes was performed with the Cell Proliferation ELISA according to the manufacturer's protocol (Roche). The following antibodies were used for immunohistology: rabbit pAb against ILK (Cell Signaling Technology); FITC-conjugated mAb against integrin α6 (BD Biosciences); rat mAb against β1 integrin (Chemicon); rat mAb against β4 integrin (BD Biosciences); rabbit pAb against laminin-5 (); rabbit pAbs against keratins 6, 10, and 14 and loricrin (Covance); rat mAb against E-cadherin; rabbit pAb against β-catenin; rabbit pAb β-catenin (); rat mAb against nidogen (Chemicon); rabbit pAb against desmoplakin (Research Diagnostics); rabbit pAb against plakoglobin (Santa Cruz Biotechnology, Inc.); rabbit pAb against Lef1 (obtained from R. Grosschedl, Max Planck Institute of Immunobiology, Freiburg, Germany); rat Ki67 (Dianova); guinea pig pAbs against HF keratins (K6hf, K6irs1, K6irs2, K6irs3, K6irs4, hHa4, hHa5, hHb2, hHb5, CK5, and CK14; made by L. Langbein, German Cancer Research Center, Heidelberg, Germany); rat mAb against CD34 (clone RAM34; eBioscience); FITC-conjugated mouse mAb and POD-conjugated mAb against BrdU (Roche); rabbit pAb against cleaved caspase-3 (Asp175; Cell Signaling Technology); mouse mAb against paxillin (BD Biosciences); rabbit pAbs against FAK (Upstate Biotechnology) and phospho-FAK (Tyr397 and Tyr861; Biosource International); mouse mAb against Talin (Sigma-Aldrich); phalloidin Alexa488 (Invitrogen); goat anti–mouse Cy3, goat anti–rat Cy3, goat anti–rabbit FITC, and donkey anti–rabbit Cy3 (Jackson ImmunoResearch Laboratories); goat anti–rabbit Alexa488 (Sigma-Aldrich); and goat anti–rat Alexa488 (Invitrogen). Images were collected at room temperature by confocal microscopy (DMIRE2; Leica) using the Leica Confocal Software (version 2.5 Build 1227) with 63× NA 1.4 or 100× NA 1.4 oil objectives or by bright field microscopy (Axioskop; Carl Zeiss MicroImaging, Inc.) with 10× NA 0.3, 20× NA 0.5, or 40× NA 0.75 objectives, a camera (DC500; Leica), and IM50 software.
Flow cytometry was performed as described by . Antibodies used for FACS analysis are as follows: FITC-conjugated hamster mAb against integrin β1; rat mAb against integrin β1 9EG7; FITC-conjugated rat mAb against integrin α6; biotinylated rat mAb against integrin αV; rat mAb against integrin β4; FITC-conjugated hamster mAb against integrin α2; biotinylated rat mAb against integrin α5 (all obtained from BD Biosciences); mouse mAb against integrin α3 (BD Biosciences); Streptavidin-Cy5 (BD Biosciences); mouse mAb anti–rat FITC (BD Biosciences); and goat anti–mouse FITC (Jackson ImmunoResearch Laboratories).
Adhesion of epidermal keratinocytes to ECM proteins (poly--lysine [Sigma-Aldrich], Col I, Col IV [a gift from R. Timpl, Max Planck Institute of Biochemistry, Martinsried, Germany], FN, and LM332) was measured as described previously (). Transwell migration and matrigel invasion assays of primary keratinocytes were performed as described by .
Monolayers were treated with 4 μg/ml Mitomycin C (Sigma-Aldrich) for 4 h before scratching with a 200-μl plastic micropipette to obtain wound widths of 500–600 μm. Live-cell recordings were performed immediately after wounding for 12 h at 37°C and 5% CO using a microscope (Axiovert; Carl Zeiss MicroImaging, Inc.) equipped with 10× NA 0.3, 20× NA 0.4, 40× NA 0.6, and 100× NA 1.3 objectives, motorized scanning table (Märzhäuser) and a stage incubator (EMBL Precision Engineering). Images were captured every 10 min with a cooled charge-coupled device camera (MicroMAX; Roper Scientific) using the MetaMorph software (Universal Imaging Corp.) for microscope control and data acquisition. Wound closure was quantified by measuring the distance between both leading edges moving toward the wound in 20 randomly chosen regions. At least four independent scratch-wound experiments were used for calculations. Migration velocity was determined by calculating the slope of a linear regression line. Single-cell tracking of cells within the leading edge was performed using MetaMorph software, choosing 15 cells each in at least three independent experiments.
Cells were seeded on Col I/FN-coated dishes (MatTek Corporation) and allowed to spread for the indicated time. Four images were taken by the live-cell recording unit for each time point, and cell area was assessed using MetaMorph software.
Lamellipodia dynamics and lamella stability was analyzed using kymography (). We monitored at least 10 migrating cells over a period of 20 min with a frame rate of 4 s using the live-cell imaging unit (100× NA 1.3 objective). Subsequently, eight areas of interest across the cell lamella with a 1-pixel width were defined. The 1-pixel-wide images were pasted side-by-side to generate a composite image of membrane dynamic at a single point along the cell lamella. As described by , slopes of these lines were used to calculate the velocities, and projections of these lines along the x axis (time) were used to calculate the persistence of protrusions.
Transmission EM was performed as described previously ().
Statistical evaluation was performed with SPSS software (SPSS, Inc). Statistical significance between data groups was determined by Whitney U test and subdivided into three groups (*, P < 0.05; **, P < 0.01; ***, P < 0.001).
Fig. S1 shows the ILK expression on newborn and 2-d-old skin sections and the integrin-expression pattern on freshly isolated control and ILK-K5 keratinocytes. Fig. S2 shows the numbers of proliferating cells in basal and suprabasal layers of control and ILK-K5 epidermis, in vitro proliferation of primary control and ILK-K5 keratinocytes, and the phosphorylation levels of GSK-3β and PKB/Akt. Fig. S3 shows immunostaining for Mac1 and Gr1 on skin sections of 7-d- and 2-wk-old mice and transmission EM of desmosomal contacts in the epidermis. Fig. S4 shows the integrin-expression pattern and immunostaining for integrins on cultured primary control and ILK-K5 keratinocytes and spreading kinetics of freshly isolated keratinocyte. Video 1 shows time-lapse video microscopy of control keratinocytes. Video 2 shows time-lapse video microscopy of ILK-K5 keratinocytes. Online supplemental material is available at . |
italic
xref
#text
AMPK contains three protein subunits, α, β, and γ, which form a heterotrimer. The α subunit (AMPKα) encodes a highly conserved serine/threonine kinase, and the other subunits are regulatory. From a forward genetic screen for mutants affecting larval neuronal dendrite development (), we identified several lethal mutations in . The ethylmethanesulfonate mutants, and , contain a single amino acid change (S211L, completely conserved) and a premature stop codon (Q295 STOP), respectively, whereas has a 16-bp deletion creating a stop codon (Y141 STOP; ). All mutants, whether homozygous or in trans with a deletion covering the locus, displayed a completely penetrant and nearly identical phenotype, with greatly enlarged plasma membrane domains in dendrites, but not in axonal compartments (; unpublished data). In addition, and could be rescued to viability with either a chromosomal duplication carrying a wild-type gene, a wild-type AMPKα transgene, or a transgene that is tagged with the red fluorescent protein mCherry (; see Materials and methods). The requirement for is cell autonomous because transgene expression within only neurons rescues the phenotype (). Therefore, these mutations represent the first knockouts of the single AMPKα catalytic subunit in the genome and allow the genetic analysis of AMPK function in vivo.
Although mutants display a strong phenotype in larval neuronal dendrites, no phenotype was observed in early larval mutants (unpublished data), probably because of the large maternal contribution of this protein. To explore the relationship between AMPKα and LKB1 function without the confounding issues caused by the differing maternal contributions of each protein, we chose to examine follicle cells of the ovary. The follicle cells that surround the oocyte have a typical epithelial architecture with a highly polarized actin cytoskeleton in which the apical surface is marked by dense actin bundles in the apical microvilli, the lateral cortex is covered by a thin actin mesh, and the basal side contains a prominent network of parallel actin stress fibers. This polarized organization of actin typifies many epithelia, including the main mammalian tissue culture model for polarized epithelial cells, MDCK cells (). We did not observe any actin phenotypes in mutant follicle cells using standard detection procedures (). Because AMPK is maximally activated under low cellular energy levels, we also tested the influence of energy stress by strongly reducing the availability of sugar in the culture medium. Under these conditions, mutant cells display a strong actin phenotype (). The density of basal stress fibers is strongly reduced, whereas the amount of apical F-actin increases. This phenotype is highly penetrant under these starvation conditions (98%; = 49) and is also observed with the two other alleles of .
Because this phenotype reflects a disruption of the apical–basal polarity of the actin cytoskeleton, we examined other polarity markers within these cells. mutant clones induced in adult flies fed with high-sugar diets did not show any polarity phenotypes, which is consistent with the absence of an actin phenotype under these conditions (). Under energetic starvation conditions, however, mutant cells show a fully penetrant loss of polarity. Apical markers, such as atypical PKC (aPKC) and Crumbs (Crb) lose their cortical localization completely and appear to be down-regulated, as do the lateral markers Discs large (Dlg) and Coracle (Cora; ). In contrast, Dystroglycan (Dg), which is normally enriched at the basal cortex, extends into the lateral domain, and occasionally even reaches the apical membrane (). This suggests that the phenotype represents an expansion of the basal domain at the expense of the lateral and apical domains.
Although most aspects of apical–basal polarity are completely disrupted in mutant clones under energetic stress, E-cadherin (ECad) is usually still enriched at the adherens junctions, suggesting that the altered polarity is not a secondary consequence of a loss of intercellular adhesion. The subapical localization of Bazooka (Baz) with cadherin is also maintained in most cases (). This indicates that Baz is not in a complex with aPKC in columnar follicle cells, but is instead associated with the adherens junctions, as has recently been described in the embryo and in neuroepithelial cells of the Zebrafish neural tube (; ).
A considerable proportion of mutant clones show a more severe phenotype, in which the cells round up and lose their epithelial organization to form multiple layers of cells (). In these cases, Baz is now also absent from the cell cortex. Finally, larger mutant clones, particularly at the anterior or the posterior of the egg chamber, show a complete loss of epithelial organization and overproliferate to form small, tumorlike growths ().
As one proposed function for AMPK is to sense and maintain cellular ATP levels, the polarity phenotype observed under starvation conditions could be caused by low cellular ATP concentrations. To test this hypothesis, we examined cells that were mutant for (). Tend encodes a mitochondrial cytochrome oxidase subunit; therefore, mutants have reduced intracellular ATP concentrations to levels sufficient to maintain cell survival and growth, but not cell division (). This cell cycle block is believed to require AMPK activation. In agreement with a role for Tend in cell cycle progression, we did not observe clones bigger than four to six cells under energetic starvation conditions (). In contrast to mutant cells, however, mutant cells showed no polarity defects, ruling out the possibility that the phenotype is a secondary effect of low ATP levels. We also tested the effect of specific nutrient starvation by feeding flies only glucose, but these conditions did not induce any polarity phenotypes in mutant cells (). Thus, AMPKα is specifically required to maintain epithelial polarity and growth control under conditions of energetic stress.
Because our results indicate that plays a role in epithelial polarity, we assessed whether the localization of the protein itself is polarized. We also examined LKB1 localization, as it is a potential regulator of AMPK. Transgenic wild-type fusion proteins for both AMPKα and LKB1 rescue lethal null mutants to viability, and should therefore mimic the localizations of the endogenous proteins. LKB1-GFP is mainly found at the apical and lateral cortex of the follicle cells, and is absent from the basal domain (). This basal exclusion is surprising, as cortical localization of LKB1 requires its membrane targeting by prenylation of a conserved CAAX motif (). This suggests that the lipid composition of the basal domain is different from the rest of the plasma membrane and/or that LKB1 posttranslational modifications are asymmetrically controlled. In contrast, mCherry-AMPKα does not show any enrichment or asymmetric localization at the plasma membrane, and it is found distributed throughout the cytoplasm, but absent from the nucleus (). The localization of LKB1 suggests that AMPK could be activated specifically at the apical and lateral cortices of the cells. To test this hypothesis, we used an antibody against the LKB1 phosphorylation site of AMPK (phospho-T184). The immunostaining is reduced to background levels in both and mutant clones. This confirms the specificity of the antibody and indicates that LKB1 is the principle AMPK kinase in these cells (). In wild-type cells, PhosphoT184-AMPK is found diffusely in the cytoplasm (). The effect of AMPK on apical–basal polarity is therefore not related to a polarized distribution of the kinase or its localized activation by LKB1.
Because LKB1 activates AMPK, we wondered if similar phenotypes could be observed in mutant cells. clones can lead to severe polarity defects in follicle cells in normally fed flies (). However, these defects are observed only in large clones that are induced in the stem cells that give rise to the follicular epithelium, whereas small mutant clones, which are induced after the formation of the epithelium, have no effect on follicle cell polarity or the organization of the actin cytoskeleton ( = 24; ). This suggests that LKB1 is required for the establishment of epithelial polarity in well-fed flies, but not for its maintenance, as is the case for PAR-1 (). In contrast, under conditions of glucose starvation, small clones that were induced after the formation of the follicular epithelium show a fully penetrant polarity phenotype (100%; = 21). Under these conditions, we observed a loss of the polarized localization of Dlg, aPKC, Crb, and Cora (). However, Baz distribution is usually not affected by loss of function (unpublished data). Dg extends laterally and occasionally localizes to the apical domain (). The actin cytoskeleton is also disturbed, with more F-actin apically and a decreased density of stress fibers on the basal side. Finally, large clones lose their epithelial organization completely and overproliferate to form small neoplasms (). Thus, mutant cells exhibit identical phenotypes to mutant cells under low-energy conditions.
Because and mutant clones lead to very similar polarity defects and LKB1 phosphorylates AMPKα, we wondered if a constitutively active form of AMPKα could rescue the phenotype. Therefore, we generated transgenic lines carrying a construct, in which Threonine184 is replaced by an aspartate, which should mimic the activating phosphorylation of this site by LKB1 (). The expression of the transgene in mutant clones fully rescues their starvation-dependent polarity and overproliferation phenotypes ( = 37), whereas the Gal4 driver alone has no effect (). Furthermore, AMPKα-T184D–expressing mutant clones also have a normal actin cytoskeleton (100%; = 13; ). Thus, the phosphomimetic version of AMPKα completely rescues the mutant phenotype under conditions of energetic stress.
The recovery of null mutations in has allowed the first in vivo analysis of AMPK function in a multicellular organism, which has revealed an unexpected role for the kinase in the maintenance of epithelial polarity, but only under conditions of energetic stress. This implies that at least one of the pathways that normally maintain cell polarity cannot function when cellular energy levels are too low, and that AMPK activation compensates for this defect.
A surprising feature of the polarity phenotype is that it has opposite effects on the actin cytoskeleton and the cortical polarity cues. In mildly affected clones, basal actin is strongly reduced, with a corresponding increase in the amount of apical actin. In contrast, mutant clones show an expansion of the basal markers into the lateral and apical regions, as well as a loss of lateral and apical markers. Thus, the effects on actin may be independent of other polarity defects, suggesting that AMPK acts though different pathways to regulate actin and cortical polarity in opposite ways.
It is unclear how AMPK regulates the actin cytoskeleton, but it is possible that it acts on only one side of the cell and that the reciprocal changes on the other are caused by a change in the concentration of free G-actin or an actin nucleator, as has been shown for mutants during cellularization (). For example, loss of AMPK could increase actin polymerization apically, thereby depleting the pool of free actin that can polymerize basally. Alternatively, mutants may prevent the formation of basal actin stress fibers, and thus increase the concentration of free actin, which enhances apical actin polymerization.
The cortical polarity defects of mutant clones also suggest a reciprocal relationship between the basal and apical/lateral membrane domains because the basal domain, marked by Dg, is dramatically expanded, whereas the determinants for the lateral domain (Dlg) and the apical domain (aPKC and Crb) disappear from the cortex. This suggests that there is some form of mutual antagonism between the basal and lateral domains that maintains a sharp boundary between them, as has been described for apical and lateral domains through the inhibitory phosphorylation of Baz (PAR-3) by lateral PAR-1, and of PAR-1 by apical aPKC (; ). If this model is correct, AMPK could be required to restrict the extent of the basal domain, with the expansion of this domain in mutants leading to the exclusion of lateral and apical markers. Indeed, the overexpression of Dg has been found to cause a similar loss of apical and lateral markers to that seen in clones (). Alternatively, AMPK could be necessary to maintain the localization of the apical and lateral determinants, which in turn prevent the basal domain from extending into these regions.
Mutations in not only disrupt the polarity of the follicle cell epithelium, but also cause the cells to overproliferate, giving rise to a tumorous phenotype. One possible explanation for this phenotype is that it is caused by the mislocalization and down-regulation of Dlg. Dlg is a member of a class of tumor suppressors in that also includes Lgl and Scribble, and follicle cell clones mutant for any of these genes overproliferate to form invasive tumors that are similar to those formed by and clones under low-energy conditions (; ; ). Furthermore, the tumor suppressor function of these proteins is probably conserved in humans because Scribble restricts proliferation by repressing the G1/S transition, and is a target of the papilloma virus E6 oncoprotein (; ). This may account for the observation that AMPK is required to trigger the G1/S checkpoint under conditions of energetic stress (). However, it has also been shown in mammals that AMPK activates TSC2 to repress the insulin–TOR pathway, and thus it functions as a tumor suppressor that inhibits cell growth and division (, ). Loss of this repression might provide an alternative explanation for the overgrowth of mutant clones.
Although the molecular pathways involved remain to be elucidated, our results demonstrate that mutant cells lose their polarity under low-energy conditions and overproliferate to give rise to tumorlike growths. The activation of AMPK depends on its phosphorylation by LKB1, and loss of LKB1 produces an identical tumorous phenotype. Thus, the novel functions of AMPK reported in this work may provide a basis for the tumor suppressor function of LKB1.
An ethylmethanesulfonate mutagenesis screen on the X chromosome was performed as previously described (). Early second instar larvae were visually screened for dendritic defects using fluorescent microscopy. The mutants, lethal at late second instar stages, were mapped to ∼150 kb on the X chromosome using a molecularly defined deficiency (), an undefined deficiency (), and a duplication of the Y chromosome (). Predicted coding regions for genes in the region were sequenced using PCR amplicons made from mutant genomic DNA, and one gene (; ; NM_057965) was discovered that had mutations in all three alleles.
The wild-type transgene was cloned into the vector () as an EcoRI–BglII fragment of an EST, corresponding to an transcript (). The mCherry-AMPKα fusion protein was made using a mCherry construct (provided by R. Tsien, University of California, San Diego, San Diego, CA) at the N terminus fused in-frame to into the vector. The transgene rescues viability and fertility when expressed by in either or mutants. The phosphomimetic activated form of AMPKα (AMPKα T184D) was made by PCR-based, site-directed mutagenesis converting base C549 to G549. The transgenes were introduced into a stock by element–mediated transformation.
alleles were recombined with for mitotic recombination. Other mutant stocks used were and and were expressed in follicle cells using the driver. Flip-out experiments were performed by crossing and to and heat-shocking pupae. For rescue experiments, two independent stocks were established and crossed together: ; and , ,
Adult flies were placed in vials containing “normal” food media (5% glucose, 5% yeast extract, 3.5% wheat flour, and agar 0.8%), energetic starvation medium (1% yeast extract, 3.5% wheat flour, and agar 0.8%), or specific nutrient-starvation medium (5% glucose and agar 0.8%). Clones were induced by heat-shocking adult females at 37°C for 2 h on two consecutive days. Females were dissected 2 d after the last heat shock.
Immunofluorescence on ovaries was performed using standard procedures. Primary antibodies were used as follows: rat anti-ECad (1:1,000; ); mouse anti-Crb (cq4; 1:50; Developmental Studies Hybridoma Bank); Guinea pig anti-Cora (1:2,000; ) rabbit anti-aPKC (1:500; Sigma-Aldrich); rat anti-Baz (1:500; ), mouse anti-Dlg (1:50; Developmental Studies Hybridoma Bank); rabbit anti-Dg (1:1,000; ); and rabbit anti–phospoT385-AMPK (1:100; Cell Signaling Technology). Actin staining was performed with rhodamine-conjugated phalloidin (Invitrogen). Second instar larvae were dissected in 4% paraformaldehyde, as previously described (). Secondary antibodies coupled with Cy5 (anti–rabbit and anti–guinea pig) or Texas red (anti–mouse and anti–rabbit; Jackson Immuno-Research Laboratories; 1:500) were used. Images of follicle cells were collected on a confocal microscope (Radiance 2000; Bio-Rad Laboratories) with a 40×/1.3 NA objective (Plan Fluor; Nikon) using LaserSharp software. Live images of dendrite morphology were acquired using a confocal microscope (LSM 510; Carl Zeiss MicroImaging, Inc.) by using the 488-nm argon line to excite GFP. Larvae were covered in a glycerol solution at 22°C and gently covered with a coverslip (22 × 50 mm; Fisher Scientific) to restrict movement, but not cause bursting of the body wall. Images were taken using a Pan-Neofluar 40×/1.3 NA oil immersion lens with a 2-μm optical slice and LSM Imaging software (Carl Zeiss MicroImaging, Inc.). Images were resized and cropped with Photoshop (Adobe), and imported into Illustrator (Adobe) for labels and arrangement. |
The vascular endothelial growth factor (VEGF) and PDGF family members are closely related. In three PDGF/VEGF-like factors (PVFs) have been identified that function through a single receptor to mediate guidance of cell migration (; ). A PVF identified in was shown to bind the human VEGF receptors (VEGFRs) VEGFR1 and VEGFR2, which mediated angiogenesis (). Sequence analysis of VEGF, PDGF, and PVF members predicts that VEGF and PDGF evolved from a common ancestor (). Both VEGF and PDGF belong to the cystine-knot superfamily of signaling molecules, which is characterized by having a cystine-knot structure formed by eight cysteine residues ().
The most abundant and active member of the VEGF family is VEGF-A (; ), which undergoes alternative splicing to produce several different isoforms. The predominant human isoforms are VEGF-A and -A, which lacks a heparin-binding domain (). Three VEGFR tyrosine kinases (RTKs; VEGFR1-3) that form homodimers on ligand binding have been identified (; ). VEGF-A binds to VEGFR1 and VEGFR2, but not VEGFR3, but most signal transduction is mediated by VEGFR2 (; ). All three VEGFRs are structurally related to the PDGF class III RTK subfamily, which are all characterized by seven extracellular immunoglobulin-like domains with an intracellular tyrosine kinase domain interrupted by a noncatalytic region (). These and other structural similarities between VEGFRs and PDGF receptors (PDGFRs) suggest a close evolutionary relationship ().
The PDGF family consists of four different PDGF chains (A–D), which assemble into functional homodimers or a PDGF-AB heterodimer, and two PDGFR tyrosine kinases (α and β), which form a homodimer or heterodimer on ligand binding (; ). PDGF-AA binds only PDGFRα, whereas PDGF-BB binds both homodimer and heterodimer PDGFRs. The less abundant PDGF-CC and -DD bind to PDGFRα and PDGFRβ homodimers, respectively, with both binding to the PDGFRαβ heterodimer. PDGF-C and -D have a novel N-terminal CUB domain and are structurally more similar to the VEGF family than the PDGFs (; ).
VEGF-A and PDGF-BB are both critical factors in promoting the recruitment and proliferation of vascular cells (; ). Adult bone marrow–derived mesenchymal stem cells (MSCs), which can differentiate to vascular cells (; ; ), may be recruited during angiogenesis and to sites of vascular injury (; ). Although PDGF isoforms induce human MSC migration (), less is known of VEGF-A–mediated effects, with several studies reporting no VEGFR expression in MSCs (; ). In this investigation, we examined the role of VEGF-A in regulating MSC migration and proliferation.
We report that VEGF-A can directly signal through PDGFR, which represents a novel VEGF-A/PDGFR signaling mechanism. This study provides important new insights into how VEGF signaling regulates MSC recruitment and proliferation during tissue regeneration and disease.
e
c
r
u
i
t
m
e
n
t
a
n
d
p
r
o
l
i
f
e
r
a
t
i
o
n
b
y
v
a
s
c
u
l
a
r
g
r
o
w
t
h
f
a
c
t
o
r
s
a
r
e
c
r
i
t
i
c
a
l
e
v
e
n
t
s
d
u
r
i
n
g
b
l
o
o
d
v
e
s
s
e
l
g
r
o
w
t
h
,
r
e
p
a
i
r
,
a
n
d
d
i
s
e
a
s
e
.
I
n
t
h
i
s
s
t
u
d
y
,
w
e
e
x
a
m
i
n
e
d
t
h
e
c
h
e
m
o
t
a
c
t
i
c
a
n
d
m
i
t
o
g
e
n
i
c
e
f
f
e
c
t
s
o
f
V
E
G
F
-
A
o
n
M
S
C
s
.
In this study, we demonstrate that VEGF-A regulates MSC migration and proliferation, despite the fact that RT-PCR or flow cytometry analysis provided no evidence for VEGFR1-3 expression. Furthermore, using a human phospho-RTK array, which is more sensitive than immunoprecipitation, VEGF-A resulted in no detectable VEGFR1-3 tyrosine phosphorylation. However, VEGF-A–induced PDGFRα and PDGFRβ tyrosine phosphorylation was clearly confirmed, highlighting that VEGF-A exerts its effect on MSCs by the stimulation of PDGFRs.
Using complementary approaches, we provide evidence of a novel VEGF-A/PDGFR signaling mechanism, showing that VEGF-A can signal using both PDGFRs. Heparin-binding domains are important modulators of VEGF subtype binding and biological activity, VEGF-A binds heparin, but VEGF-A does not (; ). Because we demonstrated that both VEGF-A isoforms stimulated MSC and HDF migration, heparin binding is unlikely to be an important determinant. Pretreatment of MSCs with a PDGF RTK inhibitor significantly reduced VEGF-A–stimulated MSC migration. Neutralizing either cell surface PDGFRα or PDGFRβ using a specific blocking antibody also resulted in significant inhibition of VEGF-A–induced migration. Furthermore, blocking either PDGFRα or PDGFRβ expression using siRNA oligonucleotides significantly attenuated VEGF-A–induced migration, with RTK array analysis confirming decreased tyrosine phosphorylation of the respective PDGFRs. Thus, both PDGFRs are essential for VEGF-A–induced migration, suggesting that both PDGFR homodimers (-αα and -ββ) and/or a heterodimer (-αβ) mediate VEGF-A/PDGFR signaling.
PDGFRα neutralization by antibody blocking or siRNA knockdown resulted in a greater decrease in VEGF-A–induced migration than corresponding PDGFRβ inhibition, which may reflect VEGF-A binding affinity to PDGFRα. We have previously shown that the MSCs used in this study express abundant PDGFRα, have a high ratio of PDGFRα to PDGFRβ, and, importantly, virtually every cell coexpressed both receptors (). The two PDGFRs have different PDGF binding affinities; PDGFRβ has a higher affinity for PDGF-B or -D, whereas PDGFRα has a higher affinity for PDGF-A or -C (Betsholtz et al., 2004). PDGF-C and -D are structurally more similar to VEGF-A than to PDGF-A or -B (), and both bind to a PDGFRαβ heterodimer (). MSCs exposed to PDGF-BB resulted in PDGFRα and PDGFRβ tyrosine phosphorylation levels being 2.2- and 6.0-fold higher, respectively, than corresponding VEGF-A–stimulated receptors. In comparison, VEGF-A induced similar levels of PDGFRα and PDGFRβ tyrosine phosphorylation, which may reflect a preference for PDGFRαβ stimulation. Thus, the data suggest that heterodimeric PDGFRαβ, at least in part, mediates VEGF-A/PDGFR signaling. The biological functions of PDGF-activated heterodimeric PDGFRαβ are not defined ().
Interestingly, phospho-RTK array analysis revealed that in addition to VEGF-A–induced PDGFRα and PDGFRβ tyrosine phosphorylation, VEGF-A stimulated EGFR, EphA7, and Axl tyrosine phosphorylation. PDGF-BB also stimulated EGFR phosphorylation, as well as FGFR3, but not EphA7 or Axl receptors, indicating that EphA7- and Axl-induced tyrosine phosphorylation were VEGF-A specific. Because siRNA knockdown of either PDGFRα or PDGFRβ had little impact on EphA7 or Axl tyrosine phosphorylation levels, the mechanism of VEGF-A–induced, ligand-independent dimerization and activation of EphA7 and Axl receptors remains to be determined.
We demonstrated that either VEGF-A or -A isoforms were able to induce MSC and HDF migration, and that both cell types expressed NP-1 and -2 transmembrane glycoproteins. Although VEGF-A binds to NP-1 and -2, VEGF-A binds to neither (). NPs are not known to signal independently after VEGF binding, but are proposed to act as coreceptors and facilitate binding of certain VEGF subtypes to VEGFRs (). Thus, although we cannot discount a role for NPs, in the absence of VEGFRs, to facilitate VEGF-A binding to PDGFRs, NPs are unlikely to be involved in mediating VEGF-A–induced chemotactic or mitogenic effects.
Along with finding that VEGF-A was able to induce MSC migration, we also demonstrated that a low concentration of VEGF-A at the cell surface can inhibit both PDGF-AA– and -BB–mediated chemotaxis, indicating that VEGFA competes with PDGF ligands for PDGFR occupancy. Because both MSCs and HDFs were shown to express abundant VEGF-A transcript, it is tempting to speculate that autocrine expression of VEGF-A may act to regulate PDGF-induced chemotaxis in these cell types.
The demonstration that, in the absence of VEGFRs, VEGF can use PDGFR-mediated signaling in both MSCs and HDFs, suggests the intriguing possibility that under certain circumstances, VEGF may have an impact on a wider range of target cells than previously recognized.
VEGF-A is a crucial factor in promoting the recruitment and proliferation of vascular cells during both physiological and pathological angiogenesis and neovascularization (; ). The local oxygen concentration controls the expression of VEGF, which is mediated, at least in part, by the transcription factor, hypoxia-inducible factor 1 (). Therefore, in pathological hypoxic microenvironments, such as tumorigenesis (), disease progression is often associated with increased VEGF-A and vascular remodeling. MSCs are actively recruited during tumor neovascularization (; ) and engraft into established tumor lesions (), forming the basis for novel therapeutic approaches. Thus, VEGF-A/PDGFR signaling, especially during tissue hypoxia, is likely to be an important determinant in the recruitment and proliferation of MSCs and other PDGFR-positive cells.
Human MSCs from normal bone marrow of 20- and 26-yr-old females and 18-, 22-, and 24-yr-old males (Cambrex), HUVECs from 35- and 29-yr-old females, and a pooled batch of HUVECs and HDFs from 23- and 32-yr-old males (Cascade Biologics) were maintained as previously described (). MSCs were analyzed at passage 4, whereas HUVECs and HDFs were analyzed at passage 6. All were grown in serum-free medium for 24 h before analysis. The MSCs used in this study express a wide range of smooth muscle cell markers, including the smooth muscle cell–selective marker smoothelin-B (), and can also differentiate into osteoblast, chondrocyte, and adipocyte lineages (). They are positive for CD29, CD44, CD105, and CD166, but negative for hematopoietic cell markers CD14, CD34, and CD45. We have also demonstrated that they are negative for the specific pericyte marker 3G5 (; ).
All growth factors, PDGF-AA (221-AA), PDGF-BB (220-BB), TGF-β (240-B), VEGF-A (293-VE), and VEGF-A (298-VS), were obtained from R&D Systems. Three different batches of VEGF-A were used during this study, all containing BSA carrier protein (50 μg BSA/1 μg cytokine). We excluded the possibility that the VEGF-A may contain contaminant PDGF-BB (which binds to both PDGFRs). Immunoblot analysis (using anti–PDGF-B) readily detected 1 ng PDGF-B, but 100 ng VEGF-A produced no PDGF-B immunoreactivity, indicating that any potential PDGF-BB contamination must be <1 ng (Fig. S2). However, the minimum concentration of PDGF-BB that induced a detectable PDGFRα or PDGFRβ tyrosine phosphorylation response was ≥5 ng and 2 ng, respectively (Fig. S2). Thus, any potential contamination of <1 ng PDGF-BB (in 100 ng VEGF-A) would not induce a detectable PDGFR tyrosine phosphorylation response.
In addition, VEGF-A from two different suppliers (Invitrogen and Autogen Bioclear) was also tested, and both showed similar biological effects to the VEGF-A obtained from R&D Systems.
Anti–human PDGFRα (MAB322) and PDGFRβ (AF385) antibodies were used to specifically neutralize PDGFRs, whereas anti–human VEGFR1 (AF321) and VEGFR2 (MAB3572) antibodies were used to specifically neutralize VEGFR1 and VEGFR2 (R&D Systems). PDGFR tyrosine kinase was inhibited using PDGFR tyrosine kinase inhibitor III (50 nM PDGFRα IC; PDGFRβ IC, 80 nM with IC ≥ 30 μM for EGFR, FGFR, Src, PKA, and PKC; ; Calbiochem). VEGFR2 tyrosine kinase was inhibited using VEGFR2 inhibitor V (IC < 2 nM with IC > 50 μM for VEGFR1, EGFR, FGFR1 and PDGFRβ; ; Calbiochem).
Semiquantitative RT-PCR was performed as previously described (). Each primer pair was designed using the same parameters, resulting in similar Tm values (58.8–60.0) and product lengths as shown. VEGFR-1 (99-bp), forward (5′-GCGACGTGTGGTCTTACG-3′) and reverse (5′-GGCGACTGCAAAAGTCCT-3′); VEGFR-2 (81-bp), forward (5′-CATCCAGTGGGCTGATGA-3′) and reverse (5′-TGCCACTTCCAAAAGCAA-3′); VEGFR-3 (87-bp), forward (5′-GATGCGGGACCGTATCTG-3′) and reverse (5′-ATCCTCGGAGCCTTCCAC-3′); VEGF-A (98-bp), forward (5′-CACCCATGGCAGAAGGAG-3′) and reverse (5′-CACCAGGGTCTCGATTGG-3′); NP-1 (77-bp), forward (5′-GCAGTGGCTCCTGGAAGA-3′) and reverse (5′-AGTCGCCTGCATCCTGTC-3′); NP-2 (83-bp), forward (5′-ATTCGGGATGGGGACAGT-3′) and reverse (5′-CCCGAGGAGATGATGGTG-3′); and GAPDH (71-bp), forward (5′-AAGGGCATCCTGGGCTAC-3′) and reverse (5′-GTGGAGGAGTGGGTGTCG-3′). An additional pair of primers for all three VEGFRs (VEGFR1-3) that was designed to different sequence regions were also used (primer sequences not shown).
For single-color flow cytometry, MSCs, HUVECs, or HDFs (4×10 cells/ml) were incubated with either PE-conjugated anti–human VEGFR1-PE (FAB321P), VEGFR2-PE (FAB357P), or VEGFR3-PE (FAB3492P) antibodies, or control anti–IgG-PE antibody (IC002P; R&D Systems). VEGFR1 (MAB4711), VEGFR2 (MAB3572), VEGFR3 (MAB3491) antibodies, or control anti-IgG (MAB002) antibody (R&D Systems) were also used after secondary labeling with a FITC secondary antibody (Dako Cytomation). HDFs were also incubated with either anti–human PDGFRα-PE (sc-21789PE) or PDGFRβ-PE (sc-19995PE) antibodies (Santa Cruz Biotechnology). For each sample, 100,000 cells were counted using a FACscan cytometer (Becton Dickinson) at a flow rate of <200 events/s.
Cell migration was determined using a modified Boyden chamber assay. Cell culture filter inserts of 8 μm pore size, 6.5 mm diam (Becton Dickinson), were coated on the underside with 10 μg/ml fibronectin in PBS, overnight at 4°C. MSCs (1×10) were added to the upper chamber with 10 or 20 ng/ml growth factors in the lower chamber and cells allowed to migrate to the membrane underside for 5 h at 37°C in a humidified atmosphere of 5% CO in air. In some experiments, cells were preincubated with receptor neutralization antibodies or kinase inhibitors (10 μg/ml anti-VEGFR1 or anti-VEGFR2 neutralization antibodies, 100 nM VEGFR2 tyrosine kinase inhibitor [VEGFR2-TK], 2 μM PDGFR tyrosine kinase inhibitor [PDGFR-TK], and 10 μg/ml anti-PDGFRα or -PDGFRβ neutralization antibodies) for 30 min at 37°C before growth factor exposure. After migration, cells on the upper membrane surface were removed and migratory cells on the membrane underside were fixed using 5% (wt/vol) glutaraldehyde and stained using 0.1% (wt/vol) crystal violet solution. Filter inserts were inverted and the number of migratory cells on the membrane underside (cells/field using a 10× NA 0.3 Olympus UPlanF1 objective lens) was determined, at room temperature, by visualizing the crystal violet–stained cells directly on insert undersides by phase-contrast microscopy, without use of fluorochromes (BX51; Olympus). Images were captured using a computerized imaging system (MetaMorph imaging v 5.0; Molecular Devices) and CoolSNAP (Photometrics) camera system.
MSCs (2,000 cells/well) in growth medium were seeded in 96-well plates and incubated with 10 ng/ml growth factors at 37°C in a humidified atmosphere of 5% CO in air. Experiments were also conducted using MSCs pretreated with 10 μg/ml receptor neutralization antibodies for 30 min before growth factor addition. At the end of each time point, a CyQuant cell proliferation assay kit (Invitrogen) was used to detect MSC proliferation. Cells were treated in situ according to the manufacturer's protocol. To generate a standard curve, a serial dilution of MSCs (2,000–20,000) were also aliquoted into separate wells and treated the same as sample cells. Plates were read using a scanning multiwell fluorometer at a wavelength of 480 nm, and cell numbers were calculated using the standard curve for each plate.
A human Phospho-RTK Array kit (R&D Systems), which has a greater sensitivity than immunoprecipitation analysis, was used to simultaneously detect the relative tyrosine phosphorylation levels of 42 different RTKs in untreated or growth factor–treated MSC lysates. Each array contains duplicate validated control and capture antibodies for specific RTKs. MSCs cultured for 24 h in serum-free medium were stimulated with 20 ng/ml growth factors for 10 min at 37°C in a humidified atmosphere of 5% CO in air, and then immediately placed on ice, washed twice with chilled PBS, and isolated using chilled lysis buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% NP-40, 2.5 mM EDTA, 1 mM sodium orthovanadate, 10% glycerol, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). Total protein concentration was quantitated using a BCA assay kit (Pierce Chemical Co.). RTK array analysis was performed according to the manufacturer's protocol. In brief, array membranes were blocked, incubated with 500 μg MSC lysate overnight at 4°C, washed, and incubated with anti–phosphotyrosine-HRP for 2 h at room temperature, washed again, and developed with ECL Western blotting detection reagent (GE Healthcare), and RTK spots were visualized using Kodak XAR film. Average pixel density of duplicate spots were determined by Gene Tools v3 software (Syngene), with values normalized against corner duplicate phosphotyrosine-positive control spots, which were assigned a value of 100.
MSCs (5×10 cells), together with 3-μg siRNAs, were transfected by electroporation using a human Nucleofector kit (Amaxa) and cultured overnight in growth medium. Validated siRNAs, which were functionally tested to provide ≥70% target gene knockdown, were used for PDGFRα and PDGFRβ knockdown and a scrambled siRNA control (QIAGEN).
Cells were isolated using ice-cold lysis buffer and 100 μg lysates precleared using 10% (wt/vol) protein A–Sepharose (GE Healthcare), and then incubated with monoclonal anti–human PDGFRα (MAB1264) or PDGFRβ (MAB1263; R&D Systems) overnight at 4°C. Immune complexes were isolated by incubation with 10% (wt/vol) protein A–Sepharose for 2 h. Immunoblot analysis was performed as previously described (), using a monoclonal anti–human antibody for phosphorylated tyrosine (PY99; sc-7020; Santa Cruz Biotechnology). Human phospho-PDGFRα, phospho-PDGFRβ and soluble PDGF-BB levels were all detected by ELISA kits, performed according to the manufacturer's protocol (R&D Systems).
After stimulation of MSCs in serum-free conditions with growth factor, 1 mM 3, 3′-Dithiobis[sulfosuccinimidyl propionate] (DTSSP; Pierce Chemical Co.) was directly added to the medium and incubated for 30 min at room temperature, and the cross-linking reaction was quenched using 20 mM Tris, pH 7.5, for 15 min at room temperature. DTSSP is a membrane-impermeable thiol-cleavable reagent that is used for cross-linking molecules at the cell surface. PDGFRs were immunoprecipitated from cell lysates using anti–human PDGFRα (MAB1264) or PDGFRβ (MAB1263; R&D Systems). Proteins conjugated to PDGFR–DTSSP complexes were dissociated by adding 5% β-mercaptoethanol and boiling for 5 min. Growth factors associated with PDGFRs were resolved by SDS-PAGE and detected by immunoblot analysis, as previously described (), using the following corresponding monoclonal anti–human antibodies: VEGF-A (MAB293), PDGF-B (MAB220), or TGF-β (MAB240; R&D Systems).
In all quantitation experiments, results are expressed as the mean ± the SD. Statistical differences between sets of data were determined by using a paired test on SigmaPlot 8.0 software, with P < 0.05 considered significant.
Fig. S1 shows that inhibition of PDGFRα or PDGFRβ attenuated VEGF-A–induced MSC proliferation. Fig. S2 shows that VEGF-A contained no detectable PDGF-BB contamination. Fig. S3 shows that VEGF-A did not change soluble PDGF-BB levels. |
Actin filament dynamics are essential for various cell activities, including cell migration, morphological change, and polarity formation. These events are regulated by a variety of actin-binding proteins, which cooperatively act in the assembly/disassembly and reorganization of actin filaments in cells (; ). Cofilin/actin-depolymerizing factor (ADF) family proteins, ubiquitously expressed in eukaryotes, are key regulators of actin filament dynamics (; ; ; ). In vitro studies have demonstrated that cofilin stimulates actin filament disassembly by accelerating the off rate of actin monomers from the pointed ends of actin filaments (depolymerization) and by severing actin filaments (; ; ; ). Depletion or inactivation of cofilin in or mammalian cells results in aberrant F-actin accumulation, implicating cofilin in actin filament disassembly in the cell (; ; ; ; ; ).
Conversely, cofilin is required for actin filament assembly in the cell, as seen in the case of stimulus-induced lamellipodium formation (; ; ). The observations that cofilin preferentially binds to the ADP-bound actin in filaments and enhances actin filament disassembly from the pointed ends in the rear of the lamellipodium have led to the treadmilling model, where cofilin contributes to actin filament assembly by replenishing actin monomers for polymerization (; ; ). An alternative model has recently proposed that cofilin is involved in stimulus-induced actin filament assembly by severing actin filaments to create free barbed ends that are used as nucleation sites for actin polymerization (; ; ). This model is based on the observation that in MTLn3 mammary adenocarcinoma cells cofilin inactivation inhibited EGF-induced barbed end formation and lamellipodium extension in the cell periphery, without changing the G/F-actin ratio in the cell (; ). However, in other types of cells, conflicting results have suggested that the G/F-actin ratio decreases after cofilin inactivation (; ). Thus, it remains unclear whether cofilin contributes to stimulus-induced actin filament assembly in the cells by supplying actin monomers through its depolymerizing/severing activity, creating free barbed ends through its severing activity, or both of these two processes. To define the extent to which cofilin plays a role in these two possible processes in stimulus-induced actin filament assembly, it is essential to determine the G/F-actin ratio quantitatively in both cofilin-active and -inactive cells.
In this study, we have assessed the actin monomer pool in the cytoplasm of living cells by measuring the fluorescence decay of Dronpa (Dp)-labeled actin photoactivated in a small region of the cytoplasm. Dp is a GFP-like protein whose fluorescence can be reversibly switched off and on by photobleaching and photoactivation, respectively (). Together with F-actin sedimentation assays, we provide evidence that cofilin is involved in the generation of more than half of the actin monomers in the cytoplasm. Using cofilin mutants, we also show that the severing activity, rather than the depolymerization activity, of cofilin is predominantly involved in maintaining the actin monomer pool in the cell. We also demonstrate that actin monomers in the cytoplasm are incorporated into the tip of the lamellipodium at the rate dependent on the actin monomer pool size in the cytoplasm. Furthermore, in cofilin-inactivated or -depleted cells, in which >80% of total cofilin is converted to the inactive phosphocofilin by LIM-kinase overexpression or total cofilin expression is decreased ∼80% by RNA interference, stimulus-induced actin monomer incorporation at the cell periphery is attenuated, but the incorporation of microinjected actin monomers is not. Our results suggest that cofilin contributes to stimulus-induced actin filament assembly in the cell periphery by supplying an abundant pool of actin monomers to the cytoplasm.
To assess actin monomer pool size in the cytoplasm of living cells, we expressed Dp-labeled actin in COS7 cells and measured the fluorescence decay of Dp-actin in the cytoplasm. The level of Dp-actin expressed in COS7 cells was <1%, compared with endogenous actin, and the expression had no apparent effect on F-actin assembly (Fig. S1, available at ). After the fluorescence of the whole cell was photobleached to the background level, Dp or Dp-actin was locally photoactivated in a small square region of the cytoplasm, and fluorescence images were acquired every 0.4 s. Both Dp ( and Video 1) and Dp-actin ( and Video 2 A) rapidly diffused from the photoactivated region throughout the cytoplasm. Quantitative analyses of the time-dependent changes in fluorescence intensity in the photoactivated region () and the fluorescence decay at 0.8 s after photoactivation () indicate that the rate of fluorescence decay of Dp-actin is slightly slower than that of Dp.
To determine whether the rate of fluorescence decay of Dp-actin is related to the actin monomer pool in the cytoplasm, we examined the effects of actin-modulating drugs on the diffusion of Dp-actin. When COS7 cells were pretreated with jasplakinolide (Jasp), a drug that induces actin polymerization, Dp-actin in the photoactivated region was almost immobile ( and Video 2 B), and the rate of fluorescence decay was correspondingly reduced (). In contrast, treatment of the cell with latrunculin A (LatA), a drug that induces actin filament disassembly, slightly accelerated the fluorescence decay rate of Dp-actin, compared with untreated cells (). These results suggest that the rate of Dp-actin fluorescence decay reflects the relative levels of G/F-actin in the cytoplasm, with fast decay rates corresponding to a high G-actin population. Although the rate of Dp-actin fluorescence decay theoretically also depends on the turnover rate of actin filaments, the half-life of actin filaments in the cell ranges from tens of seconds (in the leading edge of migrating keratocytes) to several minutes (in stress fibers; ; , ; ). Therefore, the rates of fluorescence decay of Dp-actin measured in this study, at least in the initial phase after photoactivation (less than a few seconds), appear to reflect primarily the G/F-actin ratio in the cytoplasm. Thus, the level of G-actin relative to total actin in the cytoplasm of living cells can be estimated by measuring the rate of fluorescence decay of Dp-actin immediately after photoactivation.
To examine the role of cofilin in regulating the actin monomer pool in the cytoplasm, we analyzed the effect of overexpression of LIM-kinase 1 (LIMK1), which inactivates cofilin by phosphorylation of Ser-3 (), on the fluorescence decay of Dp-actin in COS7 cells. By expression of wild-type (WT) LIMK1, >80% of total cofilin was converted to phosphocofilin in COS7 cells (Fig. S2, available at ). Expression of LIMK1 (WT) clearly reduced the mobility of photoactivated Dp-actin in the cytoplasm and the rate of fluorescence decay in the photoactivated region (; and Video 2 C). In contrast, expression of kinase-dead LIMK1(D460A) had no apparent effect on fluorescence decay (; and Video 2 D). These results suggest that LIMK1-mediated cofilin inactivation markedly reduced the G-actin pool in the cytoplasm. The fluorescence decay of Dp-actin in LIMK1-expressing cells was reduced by 59%, compared with control cells expressing Dp-actin alone (), which indicates that cofilin contributes to the production of more than half of the G-actin pool in control cells. When a nonphosphorylatable, constitutively active cofilin(S3A) mutant was coexpressed with LIMK1, it partially rescued the mobility of Dp-actin and increased the fluorescence decay rate, compared with cells expressing LIMK1 alone (; and Video 3 A). The partial rescue is probably due to the low actin-disassembling activity of cofilin(S3A) mutant (Fig. S3 B). In contrast, coexpression of cofilin(WT) had no apparent effect (; and Video 3 B), most likely because of its phosphorylation and inactivation by LIMK1. Similarly, the phosphorylation-mimic cofilin(S3E) mutant had no effect (). Furthermore, coexpression of the cofilin phosphatase Slingshot-1, which neutralizes LIMK1 activity (), almost completely blocked the inhibitory effect of LIMK1 on fluorescence decay (; and Video 3 C). These observations suggest that the LIMK1-induced decrease in Dp-actin fluorescence decay is primarily caused by cofilin phosphorylation and inactivation and that the cytoplasmic G-actin pool depends largely on cofilin activity.
We also examined the effect of knock down of cofilin/ADF expression using siRNA in MCF-7 human breast carcinoma cells. Similar to LIMK1 overexpression in COS7 cells, double knock down of cofilin and ADF remarkably reduced the mobility of Dp-actin and the rate of fluorescence decay of Dp-actin in the cytoplasm of MCF-7 cells ( and Video 4, available at ), which further supports the theory that cofilin plays a critical role in maintaining the G-actin pool in the cytoplasm.
Cofilin stimulates depolymerization (dissociation of actin monomers from the pointed end) and severing of actin filaments. To elucidate which of these two activities plays a dominant role in increasing the G-actin pool in the cytoplasm, we expressed LIMK1 with (S3A,Y82F)- or (S3A,S94D)-cofilin mutants in COS7 cells and examined the effects of these mutants on the LIMK1-mediated inhibition of Dp-actin diffusion. Similar to the reported characteristics of Y82F- and S94D-cofilin mutants (, ), the in vitro analyses showed that cofilin(S3A,Y82F), but not cofilin(S3A,S94D), has severing activity (Fig. S3, C and D), whereas both mutants retain the ability to promote dilution-induced F-actin disassembly (which reflects the combined depolymerization and severing activities; Fig. S3 B). Cofilin(S3A,Y82F), which retains the severing activity, significantly recovered the LIMK1-mediated inhibition of the fluorescence decay of Dp-actin, to the extent similar to cofilin(S3A) (; and Video 3 D). In contrast, cofilin(S3A,S94D), which exhibits no severing activity, did not block LIMK1 inhibition of fluorescence decay (; and Video 3 E). These results suggest that the severing activity of cofilin plays a dominant role in increasing the actin monomer pool in the cytoplasm.
We also analyzed the effects of LIMK1(WT or D460A) expression on the G/F-actin ratio in cells using F-actin sedimentation assays. COS7 cells were cotransfected with plasmids for Myc-actin and LIMK1 at a molar ratio of 1:5, to ensure the coexpression of LIMK1 in almost all Myc-actin–expressing cells. Cell lysates were centrifuged to separate G- and F-actin and analyzed by anti-Myc immunoblotting. Expression of LIMK1(WT), but not LIMK1(D460A), markedly reduced the ratio of G- to F-actin, compared with mock-transfected cells (). G-actin accounted for 41 and 18% of total actin (G- plus F-actin) in mock-transfected and LIMK1-expressing cells, respectively, thus indicating that 56% of the G-actin in control cells was shifted to F-actin by LIMK1 expression (). These data are consistent with results from the fluorescence decay of Dp-actin (). The LIMK1-induced reduction in actin monomer content was substantially blocked by coexpression of cofilin(S3A) or cofilin(S3A,Y82F), but not by cofilin(WT), cofilin(S3E), or cofilin(S3A,S94D) (). Together with data from the Dp-actin fluorescence decay assays, these results strongly suggest that cofilin contributes to the production of more than half of the G-actin in the cell and that the severing activity of cofilin plays a dominant role in this process.
We next examined whether the G-actin population in the cytoplasm is correlated to actin filament assembly in the lamellipodium. We analyzed the fluorescence decay of Dp-actin in the cytoplasm and its incorporation into the lamellipodium in the same cell. We used N1E-115 cells because they stably produced lamellipodia by expression of active Rac(V12). After cotransfection of the cells with Dp-actin and Rac(V12) and cell-wide photobleaching, Dp-actin was photoactivated in a rectangular region in the cytoplasm (). Dp-actin photoactivated in the cytoplasm was continuously incorporated into the tip of the lamellipodium and then flowed retrogradely toward the cell body ( and Video 5 A, available at ). Expression of LIMK1 slowed the fluorescence decay of the photoactivated Dp-actin in the cytoplasm and repressed its incorporation into the lamellipodium ( and Video 5 B). shows the time courses of Dp-actin fluorescence decay in the photoactivated region. shows the rate of Dp-actin incorporation into the lamellipodium, measured as the length of Dp-actin fluorescence from the tip of the lamellipodium to the cell center per unit time after photoactivation. These analyses revealed that LIMK1 expression suppresses both the fluorescence decay of Dp-actin in the cytoplasm and its incorporation into the lamellipodium. The inhibitory effects of LIMK1 were substantially blocked by coexpression of cofilin(S3A) ( and Video 5 C), but not by cofilin(WT) or cofilin(S3E) (). The rate of Dp-actin incorporation into the lamellipodium is linearly correlated with the fluorescence decay in the cytoplasm in each cell (), which indicates that actin monomer assembly at the tip of the lamellipodium is highly dependent on the cytoplasmic actin monomer pool size. Because cofilin elevates the cytoplasmic G-actin levels, cofilin likely enhances actin filament assembly at the tip of the lamellipodium by increasing the G-actin level in the cytoplasm.
When Dp-actin in the front region of the lamellipodium was photoactivated, it flowed retrogradely and then rapidly diffused throughout the cytoplasm ( and Video 6, available at ). The fluorescence intensity in the front region declined continuously until it reached a plateau at 15%, whereas the intensity in the rear of the lamellipodium initially increased, reached a maximum level at 20 s after photoactivation, and declined to a level similar to that in the front region (). These results suggest that most of the actin monomers that disassembled from the rear of the lamellipodium diffused into the cytoplasm.
Previous studies have shown that EGF induced actin filament assembly in the periphery of MTLn3 carcinoma cells (, ). As this EGF-induced assembly was suppressed by cofilin inactivation without changing the G/F-actin ratio in the cell, it was suggested that cofilin promotes actin filament assembly by creating new barbed ends through its severing activity (; ). We have shown here, however, that LIMK1-mediated inactivation or knock down of cofilin/ADF led to a large decrease in the actin monomer population in the cytoplasm. We also showed that the rate of actin monomer incorporation into the cell periphery depends on cofilin activity, as well as the cytoplasmic actin monomer population. We therefore hypothesized that cofilin is involved in stimulus-induced actin filament assembly in the cell periphery by supplying actin monomers in the cytoplasm and that cofilin inactivation inhibits actin filament assembly by decreasing the cytoplasmic actin monomer concentration to levels at which actin assembly is not feasible. To test this hypothesis, we analyzed the effects of LIMK1 overexpression on EGF-induced actin filament assembly by measuring the incorporation of Alexa Fluor 546–labeled actin (Alexa546-actin) monomers into actin filaments in the cell periphery by two distinct experimental procedures. First, Alexa546-actin was added to the cell after EGF stimulation using a cell permeabilization approach; second, Alexa546-actin was microinjected into the cell at the time of EGF stimulation.
We first analyzed changes in the number of barbed ends in COS7 cells before and after EGF stimulation, by applying Alexa546-actin monomers and saponin (cell-permeabilizing agent) to the outside of the cells, according to previously reported procedures (, ). COS7 cells expressing super-enhanced CFP (SECFP)-tagged LIMK1(WT) or LIMK1(D460A) were left unstimulated or stimulated with EGF for 1–3 min and then incubated with cell-permeabilization buffer containing 0.025% saponin and 0.45 μM Alexa546-actin monomers. At a concentration of 0.45 μM, Alexa546-actin monomers were incorporated only into barbed ends (). In response to EGF stimulation, Alexa546-actin incorporation at the cell periphery markedly increased in cells not expressing LIMK1 (, arrowheads) and in cells expressing LIMK1(D460A) (, arrows), indicating that EGF stimulation increased the number of barbed ends at the cell periphery of cofilin-active cells. In contrast, in cells expressing LIMK1(WT), EGF-induced Alexa546-actin incorporation at the cell periphery was not observed (, arrows). Quantitative analysis revealed that Alexa546-actin incorporated into the cell margin increased 2.1–2.6-fold after EGF stimulation in cells expressing LIMK1(D460A), but no increase was observed in cells expressing LIMK1(WT) (). Thus, the number of actin filament barbed ends at the cell periphery was considerably increased after EGF stimulation of cofilin-active cells, but not of cofilin-inactive cells, which suggests that cofilin is required for EGF-induced barbed end formation at the cell periphery.
Previous studies have suggested that cell stimulation activates the Arp2/3 complex, which enhances de novo nucleation and arborization of actin filaments to generate dendritic actin filament structures (; ; ). Arp2/3 complex–mediated dendritic actin structure formation exponentially increases the number of barbed ends, and this process requires actin monomers for polymerization. To examine whether cofilin contributes to EGF-induced barbed end formation directly, by severing actin filaments, or indirectly, by supplying actin monomers for Arp2/3-mediated dendritic actin structure formation, we analyzed the effect of microinjection of Alexa546-actin monomers into cells expressing LIMK1(WT) or LIMK1(D460A). If cofilin primarily contributes to EGF-induced barbed end formation by severing actin filaments, injected Alexa546-actin will be incorporated into the periphery of control cells, but not into the periphery of cells expressing LIMK1(WT), after stimulation with EGF. Alternatively, if cofilin contributes to barbed end formation indirectly by supplying actin monomers, microinjected Alexa546-actin monomers will be incorporated into the periphery of both control and LIMK1(WT)-expressing cells in response to EGF stimulation. As shown in (top) and in Video 7 A (available at ), Alexa546-actin monomers injected into cells expressing LIMK1(WT) were initially diffuse in the cytoplasm and then incorporated into the distal cell margin 1–3 min after EGF stimulation. The YFP-actin distribution remained largely unchanged after EGF stimulation, probably because most YFP-actin was already assembled into F-actin as a result of cofilin inactivation (, middle; and Video 7 A). Alexa546- actin injected into control cells expressing LIMK1(D460A) was similarly incorporated into the cell periphery after EGF stimulation (, top; and Video 8 A). In this experiment, YFP-actin was also incorporated into the cell periphery after EGF stimulation, because YFP-actin monomers were abundant in the cell (, middle). Quantitative analysis of Alexa546-actin fluorescence intensity in the cell margin revealed that Alexa546- actin incorporation into the cell margin significantly increased with time after EGF stimulation in both LIMK1(WT)- and LIMK1(D460A)-expressing cells (, right). The time course of Alexa546-actin incorporation in LIMK1(WT)-expressing cells was comparable to that of LIMK1(D460A)- expressing cells.
To further examine the role of cofilin in stimulus-induced actin assembly, we performed Alexa546-actin microinjection studies on neuregulin (NRG)-stimulated MCF-7 cells, in which cofilin/ADF were knocked down by siRNA. Alexa546-actin monomers injected were efficiently incorporated into the cell periphery of both cofilin/ADF and control siRNA MCF-7 cells in response to NRG stimulation (; and Videos 9 A and 10 A, available at ). Similar to the case of LIMK1 overexpression, YFP-actin was incorporated into the cell periphery in control siRNA cells, but it was almost unchanged in cofilin/ADF siRNA cells. In both LIMK1(WT)-expressing COS7 cells and cofilin/ADF siRNA MCF-7 cells, injected Alexa546-actin monomers were also incorporated into the cell periphery even in the absence of cell stimulation, although the level of incorporation was substantially lower than in stimulated cells (, graphs; Fig. S4, A and C; and Videos 7 and 9). This behavior probably reflects the fact that actin monomers are depleted in cofilin-inactive cells, and therefore, the ratio of the injected Alexa546-actin monomers is relatively high in the actin monomer pool available for polymerization. Thus, we conclude that actin monomers injected into the cytoplasm were effectively incorporated into actin filaments in the cell periphery in response to cell stimulation in both cofilin-active and -inactive cells. These results suggest that cofilin is required for stimulus-induced barbed end formation and actin filament assembly in the cell periphery primarily by supplying actin monomers.
#text
Jasp, LatA, and Alexa Fluor 546 were purchased from Invitrogen. EGF and NRG were purchased from Sigma-Aldrich and R&D Systems, respectively. Actin was purified from rabbit skeletal muscle. G-actin was labeled with Alexa Fluor 546 C maleimide, according to the manufacturer's protocols (Invitrogen).
Plasmids coding for Dp and SECFP were provided by A. Miyawaki (Riken, Wako, Japan). Expression plasmids for GFP (pEGFP-C1) and DsRed-monomer (mDsRed-C1) were purchased from CLONTECH Laboratories, Inc. Plasmids for YFP-actin and CFP-SSH-1L were constructed as described previously (; ). Expression plasmids for Dp-actin, LIMK1-SECFP, LIMK1-mDsRed, and cofilin-mDsRed were constructed by subcloning PCR-amplified Dp, SECFP, mDsRed, β-actin, LIMK1, or cofilin cDNA into the pEGFP-C1 vector. The plasmid for Myc-actin was constructed by inserting β-actin cDNA containing a Myc epitope tag into the pEGFP-C1 vector. Human cofilin siRNA plasmid (target sequence GGAGGATCTGGTGTTTATC), human ADF siRNA plasmid (target sequence GCAAATGGACCAGAAGATC), and control siRNA plasmid (target sequence TCTTCCCCCAAGAAAGATA, corresponding to the mutated human SSH-1L oligo) were constructed as described previously ().
Cells were cultured in DME supplemented with 10% (COS7 and MCF-7 cells) or 15% fetal calf serum (N1E-115 cells). Cells were transfected with expression plasmids using Lipofectamine 2000 (Invitrogen). Cells were used for various assays after being cultured for 18–24 h (COS7 and N1E-115 cells) or 4 d (MCF-7 cells) after transfection.
Photoactivation of Dp or Dp-actin and fluorescence imaging were performed using a laser-scanning confocal imaging system (LSM 510; Carl Zeiss MicroImaging, Inc.) equipped with a PL Apo 63× oil-immersion objective lens (NA 1.4) at 37°C in DME (without phenol red) supplemented with 10% fetal calf serum. The pinhole was set to 5.2 Airy units to obtain a greater depth of field. After photobleaching the whole cell by intense irradiation with a 488-nm laser, Dp or Dp-actin was photoactivated in a 5.8-μm region (400 pixels, with no obvious F-actin accumulation) in the cytoplasm by intense irradiation with a 458-nm laser for 1 s. Immediately after photoactivation, fluorescence images were automatically acquired every 0.4 s for 7.2 s by weak irradiation with the 488-nm laser. In pharmacological experiments, cells were pretreated with 1 μM Jasp for 20 min or 1 μM LatA for 2 min on the microscopic stage before photobleaching. For N1E-115 cells, Dp-actin was photoactivated for 1 s in a 14.25- × 2.85-μm rectangular region (4,500 pixels) in the cytoplasm, and fluorescence images were acquired every 2 s for 20 s to measure the fluorescence decay of Dp-actin in the photoactivated region. The whole cell was again photobleached, and Dp-actin was again photoactivated in the same region for 1.8 s, and fluorescence images were acquired every 5 s for 40 s at the maximum gain level of the photomultiplier to measure Dp-actin incorporation into the lamellipodium. The fluorescence images were analyzed by ImageJ (). The rate of Dp-actin incorporation was measured as the distance Dp-actin moved from the tip of the lamellipodium toward the cytoplasm during 20 or 40 s after photoactivation. The time course of fluorescence decay in the photoactivated region was calculated as the mean fluorescence intensity in the region versus time. The mean fluorescence intensity in the same region just before photoactivation was subtracted as the background. The fluorescence decay includes a contribution of photobleaching over the time of imaging after photoactivation. The contribution of photobleaching was reduced by minimizing the excitation laser power. A photobleaching correction was performed by capturing images without photobleaching and photoactivation under the same laser-scanning conditions. The time course of photobleaching was well approximated by a single exponential decay function using Excel (Microsoft), and the fluorescence decay data were corrected for photobleaching by the exponential decay function.
COS7 cells were transfected with plasmids for Myc-actin and LIMK1-SECFP at a molar ratio of 1:5 or with plasmids for Myc-actin, LIMK1-SECFP, and cofilin-mDsRed in a molar ratio of 1:5:10. After 24 h, cells were lysed in lysis buffer (50 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM MgCl, 0.2 mM CaCl, 1 mM dithiothreitol, 0.2 mM ATP, 1% NP-40, and 2 μM phalloidin). Cell lysates were centrifuged at 100,000 for 30 min at 4°C. Equal amounts of pellet and supernatant were subjected to SDS-PAGE and analyzed by immunoblotting with an anti-Myc antibody (9E10; Roche).
COS7 cells expressing LIMK1(WT or D460A)-SECFP were stimulated with 100 ng/ml EGF, permeabilized for 1 min with 0.025% saponin and 0.45 μM Alexa546-actin monomers, and fixed with formaldehyde, according to the method described previously (, ). The images were obtained using a fluorescence microscope (DMIRBE; Leica) equipped with a PL APO 63× oil-immersion objective lens (NA 1.3; Leica) and a cooled charge-coupled device camera (CoolSNAP HQ; Roper Scientific) driven by Q550FW Imaging Software (Leica). Incorporation of Alexa546-actin into the cell periphery was quantified by measuring the mean fluorescence intensity in a region 2 μm from the cell edge, using a customized macro in ImageJ ().
COS7 cells cotransfected with plasmids for YFP-actin and LIMK-SECFP at a molar ratio of 1:1 were cultured for 24 h and placed in DME (without phenol red) with 10% fetal bovine serum for 3 h before microinjection. MCF-7 cells cotransfected with plasmids for YFP-actin, cofilin siRNA, and ADF siRNA at a molar ratio of 1:2:2, or with plasmids for YFP-actin and control siRNA at a molar ratio of 1:4, were cultured for 4 d and placed in DME (without phenol red) without serum for 3 h before microinjection. 24 μM of Alexa546-actin monomers were injected into cells using a microinjection system (FemtoJet; Eppendorf) equipped with Femtotips 2. Immediately after the injection, cells were treated with 100 ng/ml EGF or 50 ng/ml NRG, and fluorescence images were acquired every 30 s for 9.5 min at 30°C. The images were obtained using the aforementioned fluorescence microscope equipped with a PL FLUOTAR 40× dry objective lens (NA 0.75; Leica). Incorporation of Alexa546-actin into the cell periphery was quantified by measuring the mean fluorescence intensity in a region 2 μm from the cell edge using ImageJ.
Fig. S1 shows the expression level and localization of Dp-actin in COS7 cells. Fig. S2 shows the level of P-cofilin in LIMK1-expressing COS7 cells. Fig. S3 shows the biochemical characterization of cofilin mutants. Fig. S4 shows the fluorescence images of Alexa546-actin microinjected into COS7 cells expressing LIMK1(WT or D460A) or MCF-7 cells transfected with cofilin/ADF siRNA or control siRNA, without cell stimulation. Videos 1 and 2 show the time-lapse fluorescence of Dp or Dp-actin in COS7 cells. Video 3 shows the time-lapse fluorescence of Dp-actin in COS7 cells coexpressing LIMK1(WT) with cofilin mutants or SSH-1. Video 4 shows the time-lapse fluorescence of Dp-actin in the MCF-7 cells transfected with cofilin/ADF or control siRNA. Videos 5 and 6 show the time-lapse fluorescence of Dp-actin photoactivated in the cytoplasm or in the lamellipodium in RacV12-expressing N1E-115 cells. Videos 7 and 8 show the time-lapse fluorescence of Alexa546-actin and YFP-actin in COS7 cells expressing LIMK1(WT) (Video 7) or LIMK1(D460A) (Video 8) with or without EGF stimulation. Videos 9 and 10 show the time-lapse fluorescence of Alexa546-actin and YFP-actin in MCF-7 cells transfected with cofilin/ADF siRNA (Video 9) or control siRNA (Video 10) with or without NRG stimulation. Online supplemental material is available at . |
One well-characterized inhibitor of axonal growth is myelin-associated glycoprotein (MAG), an Ig superfamily member expressed by myelinating glia (). In the central nervous system (CNS), MAG is localized to periaxonal oligodendroglial membranes of myelin sheaths and participates in axon–glia interactions (). Germline ablation of is not sufficient to promote axonal regeneration in vivo (), but acute inactivation of MAG in retina-optic nerve cultures leads to enhanced optic nerve regeneration (). In addition to its role as a regulator of neuronal growth, MAG has also been shown to regulate the long-term stability and integrity of axon-myelin associations in the optic nerve () and other fiber systems (). The mechanisms that enable MAG to exert its pleiotropic effects are not well understood and are only now starting to be defined.
MAG is a member of the Siglec family of sialic acid–binding Ig-lectins with an ectodomain comprised of five Ig-like repeats (). MAG binds to the neuronal cell surface and inhibits growth in a sialic acid–dependent neuraminidase (VCN)–sensitive manner (; ). Select gangliosides, including GD1a and GT1b, support MAG binding in a sialic acid–dependent manner, and postnatal cerebellar granule neurons (CGNs) isolated from mice lacking complex gangliosides are substantially less inhibited by MAG, indicating that gangliosides play an important role in MAG inhibitory neuronal responses (; ). A soluble fusion protein of MAG comprised of the first three Ig repeats, MAG(1–3)-Fc, binds to neurons in a sialic acid–dependent manner but is not sufficient to bring about inhibition (). This suggests that sialic acid–independent sites located in Ig repeats 4 or 5 of the MAG ectodomain are important for neurite outgrowth inhibition.
More recently, MAG has been found to interact with members of the Nogo receptor family, including neuronal Nogo66 receptor (NgR)-1 and NgR2 (; ; ). NgR1 has been proposed to function as the ligand-binding component of a tripartite NgR1–p75–Lingo-1 receptor complex that signals MAG inhibition (; ; ). Upon MAG binding to the neuronal cell surface, p75 undergoes α- and γ-secretase–dependent proteolytic cleavage, and processing of p75 is important for RhoA activation and subsequent inhibition of neurite outgrowth (). Similar to p75, the structurally related protein TROY associates with NgR1 and Lingo- 1. In the mature CNS, p75 expression is restricted, and TROY has been proposed to serve as a functional substitute in neurons that lack p75 (; ). In spite of the growing number of cell surface receptor components implicated in MAG inhibition, their role and relative contribution to growth inhibition in different cell types has not yet been examined. In this study, we provide evidence that MAG uses distinct and cell type–specific mechanisms to signal growth inhibition in different neuronal cell types, a finding that may have important implications for the development of strategies aimed at promoting neural repair after CNS injury.
Neurite outgrowth of postnatal retinal ganglion cells (RGCs), a population of myelinated CNS neurons, is strongly inhibited by MAG. On CHO-MAG feeder cells, Thy-1–immunopanned RGCs from postnatal day (P) 7–10 rat retina are strongly inhibited (neurite length = 10.2 ± 0.6 μm) compared with control CHO cocultures (neurite length = 26.0 ± 1.6 μm). To examine whether sialoglycans are important for MAG-mediated inhibition of RGCs, cultures were treated with increasing concentrations of VCN to remove cell surface terminal sialic acids (). Interestingly, neurite length on CHO-MAG cells is not significantly enhanced in the presence of VCN either at 2.5 (11.6 ± 0.7 μm; P = 0.427) or 5.0 mU/ml (11.5 ± 1.3 μm; P = 0.569) of enzyme when compared with CHO-MAG control cultures not treated with VCN (10.2 ± 0.6 μm). Increasing the VCN dose to 7.5 mU/ml inhibits the growth of RGCs and results in significantly reduced neurite length on control CHO cells (20.9 ± 1.9 μm; P = 0.012). Together, our results suggest that sialoglycan-independent mechanisms are sufficient for MAG inhibition of RGCs.
As an additional cell type to examine whether the loss of terminal sialic acids is important for MAG inhibition, we used rat P7–8 CGNs. Neurite outgrowth of CGNs cultured on CHO-MAG cells (19.0 ± 0.8 μm) is strongly inhibited compared with CGNs grown on CHO cells (38.6 ± 1.3 μm). Consistent with a previous study (), we found that in the presence of 7.5 mU/ml VCN, neurite length on CHO-MAG is significantly increased (25.0 ± 1.3 μm) compared with untreated CGNs on CHO-MAG (19.0 ± 0.8 μm; P < 0.001). Neurite outgrowth on CHO cells is not significantly altered in the presence (38.9 ± 1.6 μm) or absence of VCN (38.6 ± 1.3 μm; P = 0.838). Increasing the VCN dose to 15 mU/ml did not lead to any further promotion of the neurite growth of CGNs on CHO-MAG (; and not depicted). To assess VCN activity, we took advantage of a previous observation that NgR1 and NgR2 show a VCN-dependent drop in molecular weight (). Because of the limited yield of immunopanned RGCs, hippocampal neurons were treated for 4 h with 2.5 mU/ml VCN. Immunoblotting of lysed cells showed an ∼3-kD shift in the molecular mass of NgR1, indicating that terminal sialic acid moieties were removed successfully (). In addition, we show that VCN treatment of feeder cells does not change neuronal growth behavior (Fig. S1, A and B; available at ). Together, our results provide evidence for a neuronal cell type–specific requirement of terminal sialic acids for MAG inhibition.
To begin to address whether previously identified protein components of the MAG receptor complex are necessary for neurite outgrowth inhibition, RGCs from wild-type and -deficient mice were assayed for MAG responsiveness (). On CHO cells, the neurite length of wild-type (33.8 ± 1.8 μm) and RGCs (33.1 ± 1.7 μm) is indistinguishable. Furthermore, on CHO-MAG cells, the neurite length of wild-type (15.0 ± 0.9 μm) and RGCs (15.2 ± 1.0 μm) is very similar (). Statistical analysis revealed no significant difference in neurite length between the two genotypes (P = 0.929).
In a parallel approach, we examined MAG inhibition of wild-type and dorsal root ganglion (DRG) neurons. On CHO control cells, the neurite length of wild-type (79.9 ± 2.5 μm) and (78.6 ± 2.8 μm) DRG neurons is comparable. Consistent with previous studies (; ), we found that postnatal DRG neurons from mutants (64.8 ± 1.8 μm) are disinhibited on MAG compared with wild-type controls (53.4 ± 1.7 μm; ). The disinhibition of neurons is statistically significant (P < 0.001) but still incomplete, and fiber length is decreased compared with CHO control cultures. Previously, it was reported that CGNs of and wild-type mice show strong and comparable inhibition on crude CNS myelin (). To expand on this study, we examined the importance of in the MAG-mediated inhibition of CGNs. On CHO-MAG cells, wild-type (14.4 ± 0.7 μm) and -deficient (12.0 ± 0.4 μm) CGNs are both strongly inhibited (Fig. S2, A and B; available at ). On CHO cells, the neurite length of wild-type (28.3 ± 1.0 μm) and (30.7 ± 0.8 μm) CGNs is comparable. Importantly, in the presence of the Rho kinase inhibitor Y27632 or dibutyryl-cAMP, the MAG inhibition of -deficient CGNs is completely released (Fig. S2 B). Together, our findings suggest that MAG inhibits the growth of RGCs and CGNs through activation of the RhoA pathway in a -independent manner.
Although is strongly expressed in postnatal CGNs, there is little, if any, present in postnatal and adult RGCs either in steady state () or after optic nerve injury (; ). In neurons that lack p75, the TNF receptor family member TROY has been proposed to serve as a functional substitute (; ). Because TROY is expressed in RGCs (), we asked whether deficient RGCs are less responsive to MAG inhibition. As shown in (E and F), neurite outgrowth of RGCs isolated from P8 mutant (10.4 ± 0.8 μm) and wild-type (10.8 ± 0.8 μm) pups is strongly inhibited by MAG.
(34.4 ± 2.4 μm) and wild-type (32.2 ± 1.8 μm) RGCs plated on CHO cells (P = 0.305).
mice and found no significant release of growth inhibition (P = 0.689; Fig. S2, C and D). Together, these findings suggest that TROY does not serve as a functional substitute for p75 in RGCs and is not necessary to bring about MAG inhibition in RGCs or CGNs.
Next, we assessed the role of in the MAG-mediated inhibition of RGCs (). When plated on CHO cells, neurite outgrowth of wild-type RGCs (29.5 ± 1.8 μm) and
RGCs (24.7 ± 1.4 μm) is robust, but
RGC neurite length is decreased by 16.3% compared with wild-type littermate controls.
RGC neurite outgrowth on control CHO cells (P = 0.006), the percent outgrowth inhibition of wild-type (57%) and
RGCs (60%) on CHO-MAG cells is very similar ().
RGCs (9.8 ± 0.7 μm) is decreased compared with wild-type RGCs (12.0 ± 0.8 μm; ), suggesting that the reduced growth of
RGCs is substrate independent.
RGCs do not display enhanced fiber growth in the presence of MAG.
To expand on this observation, we considered the possibility that MAG uses multiple and perhaps independent pathways to signal growth inhibition. Consistent with this idea, we found that the combined loss of terminal sialic acids and in RGCs leads to a significant increase in neurite length on CHO-MAG cells compared with untreated controls (). On CHO-MAG cells, the neurite length of -null RGCs increases significantly from 9.8 ± 0.7 to 14.1 ± 1.2 μm after VCN treatment (P = 0.036). MAG inhibition of wild-type RGCs treated with VCN is not significantly attenuated (P = 0.108). Importantly, the release of MAG inhibition of -deficient neurons treated with VCN is significant but not complete. This suggests the existence of additional mechanisms for MAG inhibition in RGCs. Importantly, in the presence of the Rho kinase inhibitor Y27632 or dibutyryl-cAMP, MAG inhibition of RGCs is fully released (Fig. S3, available at ). Together, our data imply that there are multiple and independent pathways for MAG to bring about neurite outgrowth inhibition in RGCs.
In this study, we used a mouse genetic approach combined with the enzymatic removal of terminal sialic acid moieties to dissect the molecular composition of the neuronal receptor complex that signals MAG inhibition. The principal findings of our study are as follows: loss of terminal sialic acids in CGNs but not RGCs is sufficient to attenuate MAG inhibition; in DRG neurons but not in RGCs or CGNs, is important for MAG-elicited inhibition; -deficient RGCs and CGNs are strongly inhibited by MAG; and loss of in RGCs is not sufficient to attenuate MAG inhibition, but the combined loss of and terminal sialic acids in RGCs attenuates MAG inhibition. Based on these observations, we conclude that MAG uses distinct and cell type–specific mechanisms to signal neurite outgrowth inhibition.
The neuronal distribution of receptor components implicated in MAG inhibitory signaling is distinct and only partially overlapping (,; ; ; ; ; ). When coupled with previous findings that MAG inhibits a broad spectrum of postnatal neurons (), this implies the existence of distinct and perhaps cell type–specific mechanisms for MAG inhibition. Initially, our studies focused on RGCs, a population of heavily myelinated neurons. Remarkably, neither the loss of sialoglycans nor individual components of the NgR1 receptor complex alone is sufficient to overcome MAG inhibition in RGCs. Interestingly, when combined with VCN treatment, the loss of in RGCs leads to a significant disinhibition on MAG substrate (P = 0.036).
RGCs is still incomplete when compared with CHO control cultures. We conclude that neither the selective loss of nor the enzymatic cleavage of sialoglycans alone is sufficient to overcome MAG inhibition of RGCs. When combined, however, enzymatic removal of terminal sialic acids unmasks a contribution of in MAG inhibition. This suggests that NgR1 and sialoglycans function independently and in redundant signaling pathways to bring about MAG inhibition.
Select gangliosides and the neuronal glycoproteins NgR1 and NgR2 have been shown to support MAG binding in a sialic acid–dependent manner. Functional studies revealed that CGNs treated with VCN or CGNs isolated from mice deficient for complex gangliosides are more resistant to MAG inhibition (; ). Previously, it has been found that the first three Ig repeats of MAG bind to neurons in a sialic acid–dependent manner but are not sufficient to inhibit growth ().
In spite of the broad distribution of MAG-binding gangliosides, we find that enzymatic removal of terminal sialic acids is not sufficient to attenuate the MAG inhibition of RGCs. This is in marked contrast to CGNs, a neuronal cell type in which the enzymatic removal of terminal sialic acids substantially attenuates MAG inhibition (). The differences in MAG inhibition observed between VCN-treated CGNs and RGCs is unexpected and suggests that MAG uses cell type–specific mechanisms to bring about inhibition. When compared with RGCs, CGNs are a population of largely unmyelinated neurons that express low levels of NgR1 and lack NgR2 protein expression (; ). This suggests that Nogo receptors play a minor role, if any, in the MAG-mediated inhibition of CGNs. In RGCs, however, NgR1 and NgR2 are more abundantly expressed, providing a potential explanation for why the selective loss of terminal sialic acids in RGCs is not sufficient to attenuate MAG inhibition. In this model, NgR1 and NgR2 would have to participate in MAG inhibition in the absence of terminal sialic acids. Consistent with this idea, NgR1 and NgR2 both harbor multiple MAG docking sites, including sialic acid–dependent as well as sialic acid–independent sites (). We propose that in the presence of VCN, sialic acid–independent mechanisms of NgR1 and NgR2 are sufficient to bring about MAG inhibition. Furthermore, the incomplete release of MAG inhibition in VCN-treated -null RGCs may be a reflection of NgR2-dependent mechanisms.
RGCs are still inhibited by MAG, NgR2 does not support the binding of p75 or TROY (unpublished data), suggesting novel, as yet unidentified signal transduction mechanisms for NgR2. Future studies are needed to determine whether the loss of MAG-binding gangliosides combined with the functional ablation of NgR1 and NgR2 is sufficient to fully overcome MAG inhibition in RGCs and other neuronal cell types.
The transmembrane-spanning proteins Lingo-1, p75, and TROY have been identified as the signal transducing components in the NgR1 receptor complex. Upon MAG binding to the neuronal cell surface, p75 undergoes regulated intramembrane proteolysis, and release of the intracellular domain of p75 is important for RhoA activation and inhibition of neurite outgrowth (). Our functional studies with -deficient neurons revealed that p75 is important in DRG neurons but not in RGCs or CGNs for MAG inhibition ( and S2). This suggests a more limited and cell type–specific requirement for p75. Commensurate with this observation, an independent study () found that -deficient DRG neurons are more resistant to CNS myelin inhibition than wild-type littermate controls. Under similar experimental conditions, no difference in inhibition between CGNs of wild-type and -deficient mice was observed (). In the present study, we expand on these findings by showing that in DRG neurons but not in CGNs participates in MAG inhibition. It is of interest that CGNs are disinhibited on Nogo66 but not on MAG substrate (), This suggests that Nogo66 and MAG use distinct receptor systems to bring about growth inhibition. Because CGNs do not express the MAG-specific receptor NgR2, the mechanism used by MAG to inhibit the growth of CGNs is unknown.
RGCs express low levels of p75, and, after injury to the optic nerve, p75 is not up-regulated in RGCs (, ; ). Consistent with this, RGCs from mice are strongly inhibited by MAG. TROY is expressed in RGCs and has been proposed to serve as a functional substitute in neurons that lack (; ). However, similar to wild-type controls, RGCs are strongly inhibited by MAG. This suggests that in RGCs, TROY does not serve as a functional substitute for p75, a cell type that normally lacks p75 expression. Consistent with our functional studies, it was recently reported that TROY has a restricted neuronal expression in the adult mouse CNS ().
Commensurate with its pleiotropic effects, a growing number of MAG receptors has been identified. However, their relative contribution to MAG-elicited outgrowth inhibition has not yet been examined systematically. In this study, we provide evidence for cell type–specific mechanisms for MAG-elicited growth inhibition, an observation that may have important implications for the development of strategies aimed at overcoming the growth inhibitory barrier of adult CNS myelin. We propose that depending on the CNS fiber tracts injured, cell type–specific strategies may be needed to promote axonal growth and regeneration. It has not yet been examined whether other myelin inhibitors, including Nogo-A and OMgp, use cell type–specific mechanisms to inhibit neuronal growth. The lack of robust regenerative growth in spinal cord–injured mice with germline ablation of individual receptor components (; ; ; ) or myelin inhibitory proteins (; ; ; ) suggests that multiple and at least partially redundant mechanisms exist to bring about CNS myelin inhibition. Our studies also provide evidence that within a specific neuronal cell type, multiple and partially redundant pathways contribute to MAG inhibition. Based on these findings, the existence of one major cell surface receptor system that serves as a convergence point for multiple myelin inhibitors appears unlikely. Additional studies are needed to determine whether the combined loss of multiple receptors or the ablation of multiple growth inhibitors leads to a more robust neuronal growth response in vitro and after CNS injury in vivo.
Mice mutant for (provided by M. Tessier-Lavigne, Genentech, South San Francisco, CA; ), (provided by H. Federoff, Aab Institute for Biomedical Research, Rochester, NY; ), and () have been described previously. To assay MAG-mediated growth inhibition, primary neurons were cultured on confluent monolayers of CHO cells that either express recombinant long MAG on their surface or on CHO cells lacking long MAG (). P7–8 rat and mouse CGNs were purified in a discontinuous Percoll gradient (). P14 mouse DRG neurons were dissected and incubated in 0.05% trypsin and 0.1% collagenase, triturated, and cultured in SATO medium for 18–20 h (). RGCs were isolated by anti-Thy1 immunopanning (). In brief, P7–10 rat or mouse retina was dissected in DME (Cellgro). Retinae were digested in 2 ml DME and 30 μl papain solution (10 U/ml; MP Biomedicals) for 30 min at 37°C. The papain was removed by three washes in DME/10% FBS. The cell suspension was then plated onto 60-mm nontissue culture-coated Petri dishes that were precoated with goat anti–mouse IgG (Jackson Immuno-Research Laboratories) and anti-Thy1.1 (Serotec, Inc.) for rat or goat anti–mouse IgM (Chemicon) and with anti-Thy1.2 (Sigma-Aldrich) for mouse RGCs. After a 1-h incubation at room temperature, unbound cells were removed by several rinses in PBS. Bound cells were lifted using 0.125% trypsin/EDTA and were washed three times in DME/10% FBS. Cells were resuspended in SATO medium, plated on CHO feeder layers, and cultured for 20–24 h before fixation (). Rho kinase inhibitor (Y27632; Calbiochem), dibutyryl-cAMP (Sigma-Aldrich), and VCN (Calbiochem) were used at the concentrations indicated and directly added to the culture medium at the time of plating. VCN preferentially cleaves α2-3–, α2-6–, and, to a lesser extent, α2-8–linked terminal sialic acids. The loss of sialic acids in VCN-treated neurons results in a molecular weight shift of NgR1 that was assessed by immunoblotting as described previously ().
Quantification of neurite outgrowth was performed as described previously (). In brief, cultures were fixed in 4% PFA and blocked in 1% horse serum and 0.1% Triton X-100 in PBS, and neurons were stained with anti–class III β-tubulin antibody (TuJ1; Promega). Images were taken using an inverted microscope (IX71; Olympus) attached to a digital camera (DP70; Olympus), and neurite length was quantified using UTHSCSA ImageTool for Windows. For quantification of neurite outgrowth, processes equal or longer to approximately one cell body diameter were measured. For all experiments, the mean and SEM of neurite length was determined from multiple independent experiments. Several independent experiments were performed for quantification of the neurite length of primary neurons isolated from the following mice: mice (DRGs, = 3 [+/+] and = 3 (−/−); CGNs, = 5 [+/+] and = 5 [−/−]; and RGCs, = 6 [+/+] and = 8 [−/−]), mice (RGCs, = 3 [+/+] and = 3 [−/−]; and CGNs, = 2 [+/+] and = 3 [−/−]), and mice (RGCs, = 5 [+/+] and = 4 [−/−]). For quantification of the neurite length of mouse RGCs treated with 5 mU VCN, several independent experiments were performed: = 5 () and = 4 (). RGCs from wild-type (+/+) and heterozygous (+/−) mice behave similarly in neurite outgrowth assays and were included as controls in some experiments. All data were analyzed using one-way analysis of variance followed by Dunn's or Holm-Sidak post-hoc comparisons. All statistics were performed using SigmaStat 3.0 for Windows (Systat Software).
Fig. S1 shows that the loss of cell surface sialic acid moieties on feeder cells does not alter neuronal growth behavior. Fig. S2 shows that and are not necessary for MAG-mediated inhibition of CGNs. Fig. S3 shows that MAG inhibition of RGCs is released in the presence of Rho kinase inhibitor or dibutyryl-cAMP. Online supplemental material is available at . |
Mitochondrial shapes range from interconnected networks to punctiform organelles in different cell types. Recent experiments have demonstrated that the mitochondrial morphology is an important determinant of mitochondrial function (). In fact, the mitochondrial reticulum can be rapidly remodeled by dynamic fission and fusion events in response to the physiological requirements of a cell (). For example, extensive mitochondrial fragmentation is observed during apoptotic cell death and in hyperglycemic conditions (; ).
Mitochondrial fission and fusion rely on the function of multiple proteins that mediate the remodeling of the outer and inner mitochondrial membrane (). In mammalian cells, fission requires dynamin-related protein 1 (DRP1; ; ) along with hFis1 (; ; ), endophilin B1/Bif-1 (; ), MTP18 (), GDAP1 (), DAP3 (), and potentially other unknown proteins. Similar to other dynamins, DRP1 and its yeast orthologue Dnm1p contain an N-terminal GTPase domain, a coiled-coil middle domain involved in protein self-assembly, and a C-terminal GTPase-activating domain (; ; ; ). In vitro, DRP1 forms oligomeric ringlike structures and tubulates liposomes in a nucleotide-dependent manner, suggesting that it functions directly in membrane constriction and/or scission (). In vivo, DRP1 forms cytosolic tetramers and is recruited to defined loci along the mitochondrial surface, which often mark sites of mitochondrial fission (; ; ; ). The function of mitochondrial DRP1 puncta that are not associated with active constriction sites is unclear.
The recruitment of DRP1 from the cytosol to the mitochondrial surface is likely mediated by membrane-associated receptors. Three proteins—Fis1p, Mdv1p, and Caf4p—have been described to form a complex with Dnm1p (; ; ; ; ; ; ; ). In mammalian cells, only the outer membrane protein hFis1 has been proposed to fulfill this function, and evidence suggests that the nature of this complex is rather transient (; ). However, given the uniform distribution of hFis1p on the mitochondrial surface and the fact that its removal does not alter the recruitment of DRP1 to mitochondria (; ), it is likely that other factors contribute to DRP1 assembly and/or function. For instance, sumoylation and ubiquitination were recently reported as potential regulators of the DRP1 activity (; ; ). Components of the sumoylation machinery have also been reported to modulate the oligomerization and function of dynamin 1 (), suggesting that this posttranslational modification may control the activity of several members of the dynamin family of GTPases.
Recent evidence indicates that DRP1 participates in mitochondrial fission during apoptotic cell death (; ; ; ; ). Interference with DRP1 function during apoptosis results in a block of mitochondrial fission, leading to long interconnected organelles, as well as a delay in cell death. Mitochondrial fission is an early apoptotic event, occurring within the same time frame as activation of the proapoptotic Bcl-2 family member Bax and permeabilization of the mitochondrial outer membrane that leads to the release of multiple inner membrane space proteins and loss of the mitochondrial membrane potential (ΔΨ; for reviews see ; ). It is presently unclear how a fission-mediating GTPase such as DRP1 is integrated into this series of events. Down-regulation of DRP1 expression by RNAi or overexpression of a dominant-negative mutant DRP1 K38A delays but does not block Bax recruitment and activation on the mitochondrial membranes and inhibits cytochrome release from mitochondria (; ; ; ). This implies that both Bax and DRP1 function upstream of cytochrome release in the apoptotic cascade of events, with DRP1 potentially participating in efficient Bax activation. Importantly, the overexpression of DRP1 K38E blocks remodeling of the mitochondrial cristae (), an event that allows for the translocation of cytochrome from the intracristal stores to the intermembrane space, from where it is released (; ). This is consistent with recent data indicating that interference with DRP1 function does not affect the release of SMAC (second mitochondria-derived activator of caspases)/Diablo, which resides in the intermembrane space, but strongly inhibits the release of cytochrome that is mostly confined to cristae (; ). Whether DRP1-mediated fission is coupled to inner membrane remodeling or whether DRP1 has a more direct function in cristae remodeling remains to be determined.
Although it has been established that the membrane remodeling machinery participates in apoptosis, two important questions remain. First, it is unclear whether the process of fragmentation or alterations in membrane curvature (including cristae remodeling) are requisite for apoptosis. Second, it remains unclear how the fission proteins are activated by the apoptotic machinery. It has been observed that mitochondria-associated Bax coalesces at DRP1-containing fission sites and mitochondrial tips during cell death, implying that they may function together in membrane remodeling (). On the other hand, colocalization between Bax and Mfn2, a GTPase involved in mitochondrial fusion, may account for a block in fusion observed during apoptosis (). The link between the fusion machinery and Bax was recently highlighted by the discovery that Bax is required for the regulation of Mfn2 activity and lateral assembly into foci along the mitochondrial tubules (). Also, upon apoptotic stimulus, endophilin B1/Bif-1, a membrane-shaping protein, has been reported to interact with Bax and to control the recruitment and activation of both Bax and Bak, another Bcl-2–like homologue (; ). It is conceivable that channel-forming properties of Bax are actually regulated by changes in membrane shape induced by endophilin B1/Bif-1, although this hypothesis needs to be explored further.
Because the fission/fusion proteins also function at steady state, their properties are likely to be altered during cell death, allowing them to participate in membrane remodeling that is specific for the mitochondrial apoptotic pathway. In fact, during apoptosis, DRP1 has been shown to accumulate on mitochondrial tubules (; ; ; ), suggesting that apoptotic signals control its recruitment and/or dissociation from the mitochondrial membrane. However, the molecular details of this event are unknown. In this study, we use FRAP of the mitochondrial YFP-DRP1 puncta as well as biochemical recruitment experiments to gain insight into the cycling dynamics of DRP1 in healthy and apoptotic cells. The data show that the increased association of DRP1 with the mitochondrial membranes that has been observed during apoptosis is not caused by an increase in the recruitment of DRP1. Instead, the membrane-associated pool of DRP1 becomes irreversibly locked on the membrane during cell death in an hFis1-independent but Bax/Bak-dependent manner. The shift from rapid cycling early in apoptosis to stable mitochondrial association observed after fragmentation is accompanied by the stable sumoylation of DRP1, an event that is also dependent on the presence of Bax/Bak. This is the first study describing specific Bax/Bak-dependent biochemical transitions in DRP1 properties that link the induction of cell death to the initiation of changes in mitochondrial membrane dynamics.
In an effort to uncover the mechanisms of DRP1 association with mitochondrial membranes, we examined the dynamic properties of this event using FRAP (). GFP-Dnm1p has been shown to rescue the null mutant in (; ; ), and fluorescently tagged DRP1 has been used extensively in worm and mammalian studies to examine the properties of this GTPase (; ; ; ; ; ). Therefore, we constructed the YFP-DRP1 fusion protein and confirmed that it is recruited into punctate structures on the mitochondria similar to the endogenous protein (). The association of YFP-DRP1 with membranes is dynamic, as indicated by the efficient kinetics of FRAP (half-time = 50 s; ). The cytosol rather than mitochondrial membranes constitutes the major supplier of unbleached fluorophores because the recovery is observed even when entire organelles are photobleached (unpublished data). The recovery of YFP-DRP1 fluorescence to the mitochondria is limited to ∼80%, indicating that a portion of the membrane- associated protein cannot be replaced by unbleached molecules (). This immobile DRP1 population may represent a scaffold that is stably associated with the mitochondrial membrane. Consistent with this, we occasionally notice fluorescence recovery to the same puncta, suggesting that at least a portion of DRP1 is recruited to predetermined sites (). The small size of the YFP-DRP1 puncta combined with the high mobility of the organelles complicates quantitative analysis of the frequency of recovery to the same puncta, although it is clear that many of these sites are conserved during the bleaching and recovery process.
Biochemical fractionation and immunofluorescence experiments indicate that DRP1 accumulates on mitochondrial membranes during apoptosis (; ; ; ; ). However, it has not been demonstrated whether this accumulation is caused by increased recruitment of DRP1 from the cytosol or stabilization of existing membrane-associated DRP1 complexes. To explore this question, we performed in vitro recruitment experiments using mitochondria-enriched membrane fractions purified from untreated or staurosporin (STS)-treated suspension HeLa (sHeLa) cells (). Purified mitochondria retained cytochrome () and maintained ΔΨ, as revealed by the accumulation of a potential-sensitive dye, MitoFluor red (). As expected, mitochondria isolated from apoptotic cells contained higher levels of DRP1 on the membrane (; ; ; ; ). Experiments were performed in the presence of nontreated cytosol or cytosol purified from apoptotic cells to ensure that accessory factors potentially necessary for DRP1 recruitment were present in the reaction mix. Purified GST-DRP1 was recruited to the mitochondrial membranes in a temperature-sensitive manner, confirming the specificity of the reaction (). Surprisingly, in vitro, the capacity of apoptotic mitochondria to recruit GST-DRP1 was comparable with that of the nontreated mitochondria (). Similar results were obtained using cytosol from sHeLa cells expressing YFP-DRP1 as a source of the recombinant protein and mitochondrial fractions purified from cells that have been treated with STS for longer periods of time (). Thus, accumulation of DRP1 on the mitochondrial membrane during apoptosis is not caused by an increase in the intrinsic capacity of mitochondria to recruit DRP1 from cytosol.
Next, we assessed the cycling properties of YFP-DRP1 during apoptosis in vivo using FRAP. As reported previously, we noticed an accumulation of DRP1 on the mitochondrial membranes during apoptosis (; ), resulting in puncta that appear larger and brighter (compare with ). Strikingly, we observed a complete block in the fluorescence recovery of YFP- DRP1 in HeLa cells treated with STS or anti-Fas–activating antibodies (). In fact, the mobile fraction of membrane-associated YFP-DRP1 decreased abruptly from 80 ± 15% in nontreated cells to 12 ± 5% in STS- and anti-Fas–treated cells (compare with ). Thus, in apoptotic cells, YFP-DRP1 puncta appear very stable and are not subjected to rapid turnover, as observed in the steady state. Based on the biochemical recruitment experiments () and on the FRAP data, we conclude that DRP1 recruited during cell death becomes irreversibly locked on the membrane, eventually leading to the saturation of membrane-associated loci and an overall block in cycling.
To position the block in DRP1 cycling in the series of apoptotic events, we plotted the maximal values of fluorescence recovery from 42 cells as a function of time after STS treatment (). The data revealed that the block in cycling is a relatively synchronous event within the total population of cells. Photobleaching of YFP-DRP1 showed that 85% of cells treated from 0 to 70 min with STS displayed a maximal fluorescence recovery to mitochondrial puncta. There was much more variability between 70 and 110 min of STS treatment, with some cells recovering to 95% but others showing a sharp decrease of fluorescence recovery to only 10% of the prebleach levels. After 120 min of STS treatment, the recovery was reduced to <35% of prebleach levels of YFP-DRP1. In addition, examination of the videos from these 42 photobleaching experiments revealed that YFP-DRP1 fluorescence recovered to mitochondria undergoing fission (, B and B′; and Video 1, available at ). Thus, actively cycling DRP1 participates in fission during apoptosis-induced fragmentation. Conversely, fluorescence recovery of YFP-DRP1 was blocked in cells displaying fragmented and immotile mitochondria (, C and C′; and Video 2). The mitochondrial ΔΨ was maintained in these cells, as indicated by the mitochondrial accumulation of MitoFluor red (′). Altogether, these data indicate that the block in YFP-DRP1 cycling occurs after mitochondrial fragmentation but before the loss of ΔΨ.
To position the block in DRP1 recycling relative to cytochrome release, we fixed HeLa cells at 0, 30, 60, 90, 120, 150, and 180 min after STS treatment and immunostained with anti–cytochrome antibodies. The percentage of cells displaying cytochrome release from the mitochondria was scored, and the data were directly compared with the time course of maximal fluorescence recovery of YFP-DRP1 (). The release of cytochrome was observed only in 10% of cells at the time when DRP1 becomes trapped on the mitochondrial membrane (80–100 min after STS treatment). This indicates that the transition in DRP1 cycling occurs before outer membrane permeabilization, supporting the data obtained from live cells demonstrating that the transition precedes the loss of ΔΨ. Finally, the transition did not require caspase activity because the general caspase inhibitor z-Val-Ala-Asp(OMe)-fluoromethyl ketone (zVAD-fmk; or the caspase 3 inhibitor V) did not affect the transition in DRP1 cycling, only inhibiting events downstream of outer mitochondrial membrane permeabilization ( and ).
The aforementioned data indicate that YFP-DRP is actively cycling during apoptotic fragmentation. To test whether fragmentation is mechanistically required for changes in DRP1 association with the membrane, we blocked mitochondrial fission by silencing hFis1 using a siRNA approach. The knockdown was confirmed by Western blotting and indirect immunofluorescence (). We then transfected hFis1 siRNA-treated cells with YFP-DRP1 and performed photobleaching studies on cells either untreated ( = 10) or treated with STS for >150 min ( = 10). In untreated hFis1-silenced cells, YFP-DRP1 fluorescence recovered within the same time frame as in wild-type (WT) cells (compare with ), which is consistent with previous evidence that hFis1 is not required for the recruitment of DRP1 at the steady state (; ). Upon treatment with STS, we noticed a block in YFP-DRP1 recovery comparable with WT cells (compare with ). Interestingly, given that hFis1 inhibits mitochondrial fragmentation, the transition in YFP-DRP1 cycling was observed upon mitochondria that remained highly tubulated (7/10 photobleached cells; , STS overlay). The remaining three cells displayed an arrest in YFP-DRP1 cycling on partially fragmented mitochondria. This delayed fragmentation is likely caused by residual hFis1 present in RNAi-treated cells, as the silencing efficiency was ∼70–80%. Together, these data indicate that fragmentation events that precede the trapping of DRP1 on the membrane are not required for the transition to occur and that hFis1 is not involved in the mechanisms that stabilize the mitochondrial association of DRP1.
Because hFis1, a known component of the mitochondrial fission machinery, was not required for the stable association of DRP1 with mitochondria during cell death, we examined the potential role of the apoptotic machinery in this transition. Bax translocation from the cytosol to the mitochondrial membrane is an early apoptotic event that correlates with mitochondrial fragmentation and cytochrome release (; ; ). To determine when the cycling of DRP1 is altered relative to Bax translocation, we cotransfected CFP-Bax and YFP-DRP1 and monitored DRP1 cycling by FRAP before, during, and after STS-stimulated Bax translocation (). As soon as Bax puncta became detectable on the mitochondria, we observed colocalization between CFP-Bax and YFP-DRP1 as reported previously (, B′ and C′; ). YFP-DRP1 recovery at that time was comparable with that of untreated cells (). However, as Bax puncta on the mitochondria became more prominent, we noticed a sharp drop in the fluorescence recovery of YFP-DRP1 (). We grouped the results from multiple bleaching experiments into two categories: the first one examined the recycling of YFP-DRP1 in STS-treated cells displaying a cytosolic distribution of CFP-Bax ( = 9), and the second one regrouped apoptotic cells showing the mitochondrial recruitment of CFP-Bax ( = 17). As indicated in , the recovery of YFP-DRP1 in apoptotic cells with cytosolic CFP-Bax was comparable with untreated cells. In contrast, the block in YFP-DRP1 recycling was clearly observed in STS-treated cells in which CFP-Bax was mitochondrial (). Thus, the block in cycling of YFP-DRP1 correlates with Bax translocation to the mitochondria.
To determine whether the proapoptotic proteins Bax and Bak are necessary for the transition in DRP1 cycling, we monitored the fluorescence recovery of YFP-DRP1 in WT or Bax/Bak double knockout (DKO) baby mouse kidney (BMK) cells (). As observed by FRAP, the mobility of YFP-DRP1 in nontreated BMK DKO cells was comparable with that in WT cells, suggesting that Bax and Bak are not necessary for normal DRP1 cycling (unpublished data). To assess YFP-DRP1 cycling properties during apoptosis, STS treatment periods longer than 3 h were chosen to ensure that most cells were past the transition point mapped in . As expected, the fluorescence recovery of YFP-DRP1 was drastically reduced in apoptotic BMK WT cells, similar to HeLa cells (, left; and quantification in B). However, the kinetics of fluorescence recovery in BMK DKO cells were comparable with those in nonapoptotic cells, indicating that Bax/Bak are required for the block in DRP1 cycling (observed recovery in STS-treated DKO cells is ∼80%, as in control cells [], with a half-time of 55 vs. 50 s in control cells; ).
By subcellular fractionation, the association of DRP1 with mitochondrial membranes was similar in nontreated WT and DKO cells, confirming that at steady state, Bax and Bak do not modulate the membrane-binding properties of DRP1 (). Conversely, the increased membrane association of DRP1 characteristic of apoptotic WT cells was not detected in DKO cells, indicating that Bax/Bak are required for the stabilization of DRP1 on the mitochondrial membrane ().
The apoptotic stimulus not only induced a stable membrane- associated form of DRP1 but also promoted the appearance of a high molecular weight DRP1-reactive band in cytosolic and mitochondrial fractions isolated from BMK cells (). Strikingly, the emergence of this high molecular weight species was absolutely dependent on the presence of Bax/Bak, as it was absent from STS-treated DKO cells (). We previously reported that DRP1 is transiently conjugated to small ubiquitin-like modifier-1 (SUMO-1) at steady state in COS-7 cells, resulting in the presence of a substoichiometric 150-kD band (). In the present study, we also detected DRP1 conjugates at steady state in BMK cells (, cytosol; long exposure), and the absence of Bax/Bak did not affect the overall pattern of these substoichiometric conjugates. To determine whether the apoptotic-induced Bax/Bak-dependent band was indeed the result of an increase in sumoylation, we immunoprecipitated DRP1 from BMK cell extracts treated or untreated with STS. The DRP1-reactive 150-kD band detected in the immunoprecipitates was indeed labeled by anti–SUMO-1 antibodies, confirming its identity as a DRP1–SUMO-1 conjugate ().
To confirm the involvement of sumoylation during apoptosis-induced mitochondrial fragmentation, we determined the localization of YFP–SUMO-1 in live HeLa cells treated with STS. We have previously observed YFP–SUMO-1 association with mitochondria under steady-state conditions (). However, given the lability and substoichiometric amounts of the conjugate at steady state, this is not easily detectable. Strikingly, upon the stimulation of cell death, YFP–SUMO-1 was readily observed to accumulate at sites of mitochondrial constriction and at mitochondrial tips (). This accumulation was observed during active apoptotic fragmentation as well as after fragmentation (Fig. S2, available at ). In addition, we observed partial colocalization between mitochondria-associated endogenous DRP1 and YFP–SUMO-1 during apoptotic death (). The lack of YFP–SUMO-1 from some DRP1-positive puncta likely reflects the transitory and substoichiometric nature of this posttranslational modification (as demonstrated in biochemical fractionation experiments; ). Interestingly, the absence of DRP1 from some YFP–SUMO-1–positive puncta indicates the existence of additional mitochondrial SUMO-1 substrates as previously reported ().
Finally, we wanted to examine whether the stable sumoylation of DRP1 was specific for apoptosis or whether this conjugation event was common to other treatments that induce nonapoptotic mitochondrial fragmentation or remodeling. As expected, treatment of BMK cells with STS induced the DRP1 conjugate, whereas incubation with carbonyl cyanide m-chlorophenylhydrazone or oligomycin (; ) did not trigger the appearance of high molecular weight DRP1 bands (Fig. S3, available at ). Together, these data demonstrate that the sumoylation of DRP1 is specifically stimulated during apoptosis in a manner dependent on Bax/Bak.
In mammalian cells, the assembly dynamics of the mitochondrial fission complex are unknown. In this study, we document the cycling properties of DRP1 at steady state and during apoptotic death. Using FRAP, we demonstrate that the cytosolic DRP1 protein binds reversibly to membranes in healthy cells. Its cycling kinetics are comparable with those of human dynamin 2 at the centrosome (half-time = 60 s; ). Interestingly, the yeast orthologue of DRP1, Dnm1p, was shown to recycle in a slow and inefficient manner, which may reflect important differences in the mechanisms of recruitment and retention of these two proteins (). Association of DRP1 with mitochondrial membranes appears more stable than that of many other proteins recruited to intracellular membranes. For example, examination of multiple components of the exocyst machinery in yeast revealed two separate classes of proteins: those with time constants ranging from 12 to 23 s (Sec4p, Sec5p, Sec6p, Sec8p, Sec10p, Sec15p, and Exo84p) and those with a time constant of ∼60 s (Sec3p and Exo70p; ). Coated vesicle components GGA1, clathrin, COPI, and Arf1 have been reported to recycle with half-times of recovery of ∼10–35 s (; ; ). Therefore, the cycling kinetics of DRP1 is slower compared with other membrane- associated protein complexes. Interestingly, we also observed that the fluorescence recovery of DRP1 often occurred at preexisting sites, indicating that DRP1 or its mitochondrial receptors generate stable membrane domains.
Several lines of evidence indicate that DRP1 plays an important role in the efficient progression of apoptosis (for review see ). However, mechanisms that are responsible for regulation of the mitochondrial fission machinery during cell death are unknown. In this study, we define two phases of DRP1 cycling during apoptosis. The first phase is characterized by active DRP1 cycling and is independent of Bax/Bak, whereas the second phase is characterized by a stable membrane association of DRP1 that is Bax/Bak dependent. The combined biochemical and FRAP data suggest that the mitochondrial accumulation of DRP1 observed during the second phase results from stabilization of the membrane-bound protein rather than the stimulation of DRP1 recruitment (; ; for review see ). This change in DRP1 properties occurs after Bax translocation to the membrane and depends on the presence of Bax/Bak, revealing that DRP1 not only functions downstream of Bax but that DRP1 cycling is regulated by Bax.
There are several mechanisms through which Bax/Bak could affect the membrane binding of DRP1. After the induction of apoptosis, Bax-positive puncta appear at fission sites, where they colocalize with DRP1 just before the block in cycling (). Therefore, Bax recruitment and oligomerization at DRP1 foci may directly or indirectly induce a shift in DRP1 cycling dynamics and promote its stable association with membranes. Interestingly, a BAR (Bin/amphiphysin/Rvs) domain–containing fission protein, endophilin B1/Bif-1, is known to bind to Bax and was recently shown to be required for Bax activation and membrane recruitment during apoptosis (). As endophilin B1/Bif-1 function is required upstream of Bax, it would also be requisite for the transition in DRP1 cycling. It is possible that a stabilized microdomain induced by endophilin B1/Bif-1 bound to activated Bax oligomers may create an environment in which DRP1 is stably recruited. In turn, the constricting ability of DRP1 may also contribute to the formation of curvature within the outer membrane that is required for the efficient insertion of Bax (, ; ). Together, the recruitment of these proteins would create a stable microdomain that strengthens itself with a positive feedback mechanism. Evidence that the dominant-negative DRP1 delays (but does not block) Bax activation is consistent with this idea (). In the absence of Bax, these microdomains would instead be dynamic and transient, with DRP1 rapidly cycling on and off the membrane. In addition, Bax may induce changes in the calcium flux from the ER, which has been shown to be a required event for DRP1 recruitment and cristae remodeling (; ). The molecular details of these pathways remain to be explored, but our data show a substantial Bax/Bak-dependent transition in the nature of DRP1 puncta on the mitochondria during cell death.
It has been suggested that hFis1 functions as a receptor/adaptor for DRP1 in mitochondrial fission (; ). However, the FRAP data reported here demonstrate that hFis1 is not required for DRP1 cycling on and off membrane at steady state, which is consistent with previous observations in mammalian cells (). This is distinct from yeast, in which Fis1p appears to function both upstream and downstream of Dnm1p recruitment (; ; ; ). Because the silencing of hFis1 results in a clear inhibition of mitochondrial fragmentation, it is more likely that hFis1 functions downstream of DRP1 recruitment in the fission process. This downstream function of hFis1 is consistent with the second requirement for yeast Fis1p in which, together with Mdv1p, they facilitate the oligomeric assembly of Dnm1p that is required to drive fission (; ; ). Importantly, the silencing of hFis1 does not preclude the appearance of the transition in YFP-DRP1 cycling during apoptosis (). This is explained by the fact that Bax and caspase 3 activation, although slightly reduced, is still detectable in hFis1 knockdown cells (Fig. S1, available at ; and not depicted; ).
It is important to consider the function of the Bax/Bak-dependent stable membrane-associated form of DRP1. Our data indicate that the block in YFP-DRP1 cycling occurs after fragmentation and does not require mitochondrial fission because it is observed on tubulated mitochondria in the hFis1 siRNA-treated apoptotic cells (). Thus, it is likely that the function of YFP-DRP1 that is stably associated with membranes is unrelated to its role in mitochondrial fragmentation. Another important piece of data indicates that the block in cycling occurs before cytochrome release and loss of ΔΨ. It has been shown that Bax oligomerization is required for permeabilization of the outer membrane (; ) and that the remodeling of the inner membrane is required to facilitate the complete release of cytochrome from intracristae stores (; ; ; ). Our previous study revealed that DRP1 plays a role in apoptosis because it is requisite not only for mitochondrial fission but also for cristae remodeling (). Importantly, this DRP1-dependent remodeling requires a functional permeability transition pore (), which is consistent with the appearance of a stable DRP1 form before the loss of ΔΨ (). This suggests that the apoptosis-specific role for DRP1 may not reside in its ability to divide mitochondria but in the remodeling of both the inner and outer membranes that precedes the permeabilization of the outer membrane. Thus, we propose that during apoptosis, an actively cycling DRP1 mediates mitochondrial fission, whereas the Bax/Bak-dependent stable membrane-associated DRP1 is involved in later apoptotic events, such as membrane remodeling that leads to the complete release of cytochrome and eventual loss of ΔΨ.
Finally, we report a correlation between the stable association of DRP1 with the mitochondrial membranes and stimulation of the sumoylation of DRP1 during cell death. Importantly, both events are linked in their requirement for Bax/Bak, indicating that sumoylation may stabilize the association of DRP1 with mitochondrial membranes.
Posttranslational modification by SUMO commonly regulates the assembly and disassembly of protein complexes, protein localization, stability, and function (). Sumoylation is known to be highly substoichiometric because it often generates intermediates that result in new protein interactions or conformational states that persist even after SUMO removal (). Recent studies implicate SUMO as a regulator of nuclear localization and/or functions of proteins that are involved in apoptosis (; ; ; ; ). In the present study, we report DRP1 as the first and only known mitochondrial SUMO target whose modification is up-regulated during apoptotic cell death. The detection of SUMO-positive dots associated with mitochondrial fission sites reinforces the notion that sumoylation plays a role in mitochondrial remodeling during apoptosis. Of note, the sumoylation of mitochondrial targets appears to be a selective rather than global event because overall mitochondria-associated sumoylation is not increased during apoptosis (unpublished data), indicating that specific proteins are being targeted, such as DRP1. Although we have documented the stimulation of DRP1 sumoylation, we have not provided evidence that this modification is functionally linked to DRP1 cycling during apoptosis. Our efforts to investigate this question have been complicated by the requirement to selectively block the second phase of DRP1 cycling and Bax/Bak-dependent sumoylation of DRP1. Further analysis of the role of sumoylation in DRP1 function during apoptosis will require a thorough characterization of its sumoylation/desumoylation cycle and of how it regulates the protein properties. Future work will focus on the dissection of specific Bax/Bak-dependent changes in DRP1 conjugation and cycling dynamics and their function in apoptotic membrane remodeling.
Antibodies were obtained from the following providers: α-DRP1 (BD Biosciences), α-Hsp60 (Sigma-Aldrich), α–SUMO-1 (Zymed Laboratories), α-Bax (Upstate Biotechnology), α–cytochrome (BD Biosciences), mouse IgG (Sigma-Aldrich), α-hFis1 (Biovision), and α-Tom20 rabbit serum (gift from G. Shore, McGill University, Quebec, Canada).
YFP-DRP1, YFP–SUMO-1, and Oct-DsRed were described previously (). pECFP-C3–human Bax was constructed using the pAdLoxGAL4-pEGFP-C3-hBax construct (gift from R. Slack, University of Ottawa, Ontario, Canada).
HeLa and BMK WT and DKO cells (gift from E. White, Rutgers University, New Brunswick, NJ) were grown in DME and RPMI 1640 medium (Invitrogen), respectively. sHeLa cells were maintained in F12 medium (Invitrogen) with 0.1 mM MEM nonessential amino acids (Invitrogen). For subcellular fractionation, sHeLa cells were grown in flasks for 5–6 d in 2 liters of MEM for suspension culture (Sigma-Aldrich) with 10 mM Hepes, pH 7.4, and 0.1 mM MEM nonessential amino acids. All media used were supplemented with 10% FBS and antibiotics.
Cells grown on glass coverslips were transfected with LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions. 16 h later, cells were treated with 1 μM STS (LC Laboratories) and 10 μM zVAD-fmk (MP Biomedicals) or with 1 μg/ml α-Fas–activating antibodies (Upstate Biotechnology), 40 μM caspase 3 inhibitor V (Calbiochem), and 10 μg/ml cycloheximide (Sigma-Aldrich). For live microscopy, cells were mounted in media with 20 mM Hepes, pH 7.4, and 50 nM MitoFluor red (Invitrogen) and were imaged at 37°C. For immunofluorescence, cells were fixed with 4% PFA/PBS for 15 min, permeabilized with 0.2% Triton X-100/PBS for 4 min, and blocked with 5% BSA, 5% FBS, and 0.02% Triton X-100 in PBS for 30 min at RT. Cells were labeled with primary antibodies and with goat α–rabbit Rhodamine X red and goat α–mouse AlexaFluor647 secondary antibodies (Invitrogen). Cells were visualized with a 100× NA 1.4 oil immersion objective (Olympus) at 1 airy U on a laser-scanning confocal microscope (IX80; Olympus) operated by FV1000 software version 1.4a (Olympus). To directly compare recovery curves from different treatments, data were normalized to correct for variations in background fluorescence (F) and loss of fluorescence during the bleach according to the formula F(t) = 100 × (F(t) − F)(F − F)/(F(t) − F)(F − F), where F(t) = region of interest intensity, F(t) = total cell intensity at any given time point (t), F = initial intensity of the region of interest, and F = initial intensity of the entire cell (). The resulting normalized data were then averaged for different cells, and the associated SD was calculated. The percent mobile fraction was calculated according to the formula M = 100 × (F − F)/(F − F), where F, F, and F are the normalized fluorescence intensities at the asymptote, immediately after the bleach, and before the bleach, respectively (). Maximal fluorescence recovery at t = ∝ and half-time of recovery were calculated using nonlinear regression (curve fit) in Prism 4.03 software (GraphPad).
Transfection of HeLa cells was performed using siGENOME SMARTpool transfection reagent (Dharmacon) according to the manufacturer's protocol. In brief, cells were seeded at 80% confluency in 10-cm dishes containing coverslips and transfected with either equal amounts of four siRNA oligonucleotides (hFis1; antisense sequences 5′-PUAACAGACCGCACAGCUCCUU-3′, 5′-PUUAGAUAGUACUGCAUGCCUU-3′, 5′-PUGAUGAAUGAUCUUUGAGCUU-3′, and 5′-PUGGCUGUUAAGCGUUUCUUUU-3′) or nontargeting sequence siRNA (glyceraldehyde-3-phosphate dehydrogenase) oligonucleotides. Cells were exposed to the transfection mixture for 16 h in complete OPTI-MEM medium + 10% FBS, at which time the transfection medium was replaced. 48 h after transfection, the cells were retransfected in the same way. 96 h after the initial transfection, cells were collected and analyzed for hFis1 expression. YFP-DRP1 was transfected 16 h before analysis by indirect immunofluorescence or by FRAP.
Cells treated with DMSO or 1.0–1.5 μM STS were collected by trypsinization or centrifugation and lysed using a ball-bearing cell breaker in buffer A (220 mM mannitol, 68 mM sucrose, 80 mM KCl, 0.5 mM Mg[CHCOO], 10 mM Hepes, pH 7.4, 2 mg/ml BSA, and protease inhibitor cocktail [Roche Diagnostics]). Lysates were centrifuged at 3,000 rpm for 10 min at 4°C. The supernatant was spun at 10,000 rpm for 15 min at 4°C, yielding a mitochondria-enriched pellet and a light membrane fraction. Light membranes were spun at 70,000 rpm for 40 min at 4°C in a TLA 100.4 rotor (Beckman Coulter) to yield cytosol. Equal protein amounts from each fraction were analyzed by SDS-PAGE and Western blotting.
The assay was performed in a 50-μl volume with 50 μg of mitochondria and 50 μg of cytosol from sHeLa cells in the presence of 50–100 nM GST or GST-DRP1 in 110 mM mannitol, 70 mM sucrose, 80 mM KCl, 0.25 mM EGTA, 3 mM Mg(CHCOO), 15 mM Hepes, pH 7.4, 2 mM KHPO, 1 mM ATP(K), 0.08 mM ADP, 5 mM Na succinate, 1 mM DTT, and 1 mg/ml BSA. Soluble components were centrifuged before the assay at 80,000 rpm for 30 min at 4°C in a TLA 100.4 rotor (Beckman Coulter). Reactions were incubated for 1 h on ice or at 37°C with mild shaking and were spun through 1 ml of 250 mM sucrose and 10 mM Hepes, pH 7.4, at 13,000 rpm for 20 min at 4°C. Membrane pellets were analyzed by SDS-PAGE. To prepare virally expressed GFP and YFP-DRP1, sHeLa cells were infected with adenoviruses at 22 plaque-forming units/cell (gifts from R. Slack). 48 h later, cells were treated with DMSO or 1 μM STS for 4 h and lysed by sonication in buffer A lacking BSA. The lysate was precleared for 5 min at 13,000 rpm, and the resulting supernatant was spun at 80,000 rpm for 40 min in a TLA 100.4 rotor. 10 μg of cytosol was used per recruitment reaction.
GST and GST-DRP1 were produced in an BL21 strain. In brief, fusion proteins were induced with 0.5 mM IPTG at 22°C for 20 h. Bacteria were lysed in a French press in buffer B (20 mM Hepes, pH 7.4, 100 mM NaCl, 2 mM MgCl, 1 mM DTT, and protease inhibitors). The lysate was supplemented with 1% Triton X-100 and spun at 35,000 rpm for 30 min at 4°C in a Ti55.2 rotor (Beckman Coulter). The supernatant was incubated overnight with glutathione–Sepharose beads (GE Healthcare). Beads were washed in buffer B supplemented with 0.1% Triton X-100, and the fusion proteins were eluted from the beads according to a standard protocol.
BMK cells were treated with 1 μM STS or 0.01% DMSO for 4 h and scraped in 10 mM Hepes, pH 7.4, 50 mM NaCl, 2 mM Mg(CHCOO), 20 mM -ethylmaleimide, and protease inhibitor cocktail. Cells were solubilized with 1% Triton X-100 and spun at 80,000 rpm in the TLA 100.4 rotor (Beckman Coulter). Precleared supernatant was incubated overnight with protein G–Sepharose beads (GE Healthcare) coupled to α-DRP1 antibody or total mouse IgG. Specifically bound proteins were resolved by SDS-PAGE and analyzed by Western blotting.
Supplemental videos and images within the figures were acquired on a laser-scanning confocal microscope (IX80 FV1000; Olympus). HeLa cells were transfected overnight with YFP-DRP1 or YFP–SUMO-1 and treated for 1 μM STS and 10 μM zVAD-fmk, respectively. For imaging, cells were mounted in a chamber in DME media supplemented with 20 mM Hepes, pH 7.4, and 50 nM MitoFluor red (Invitrogen) at 37°C. Fig. S1 shows that apoptosis proceeds in cells silenced for the expression of hFis1. Fig. S2 shows that YFP–SUMO-1 is associated with apoptotic mitochondria during and after fragmentation. Fig. S3 shows that the stable SUMO-1 conjugate of DRP1 is specific to apoptotic fragmentation. Videos 1 and 2 show HeLa cells that were transfected with YFP-DRP1 and Oct-DsRed and treated with 1 μM STS. Videos 3 and 4 show HeLa cells expressing YFP–SUMO-1–positive puncta that were imaged in time-lapse microscopy to follow the mitochondria labeled with MitoFluor red. Online supplemental material is available at . |
Low oxygen tension is not only a pathophysiological component of many human disorders, including cancer, heart attack, and stroke, but it is also critically important in normal fetal development and cell differentiation (; ). The transcription factor hypoxia-inducible factor-1 (Hif-1) has emerged as the central regulator of hypoxic gene expression (; ; ; ; ). Hif-1 is a heterodimer consisting of two subunits, Hif-1α and Hif-1β, both of which are basic helix-loop-helix/Per-Arnt-Sim domain proteins (). Transcriptional activation by Hif-1 occurs upon its binding to the hypoxia response element (HRE) within its target genes. Whereas Hif-1β protein is constitutively expressed, Hif-1α protein is subject to rapid degradation by oxygen-dependent proteolysis (; ; ; ; ). Under hypoxic conditions, Hif-1α protein is stabilized, initiating a multistep pathway of activation that includes nuclear translocation, dimerization with its partner Hif-1β, recruitment of transcriptional coactivators, and binding to the HREs of Hif-1 target genes ().
Endochondral bone formation is a two-stage mechanism; chondrocytes first shape a template, the “cartilage anlage,” in which osteoblasts then differentiate to form bone (; ; ; ).
The chondrocytic fetal growth plate is virtually avascular, but it requires blood vessel invasion to be substituted by bone (; ). We previously showed that the fetal growth plate has an out–in gradient of oxygenation. More importantly, by using a Cre-lox strategy with a Col2a1 promoter-driven Cre (Col2a1-Cre) and a floxed Hif-1α allele, we provided evidence that Hif-1α is essential for cell growth and survival of growth plate chondrocytes in vivo, as chondrocytes lacking functional Hif-1α undergo massive cell death in the center of the growth plate (; ). However, this genetic model did not allow us to address the role of Hif-1α in early chondrogenesis, as deletion of Hif-1α occurred in cells that were already committed to become chondrocytes.
An essential and specific function of differentiated chondrocytes is matrix synthesis. We have also recently reported that hypoxia and Hif-1α support cartilaginous matrix formation (, ). Thus, we speculated that hypoxia and Hif-1α may be permissive factors in chondrocyte differentiation.
The goal of this study was to investigate the roles of Hif-1α in the formation of mesenchyme condensations, in the commitment of mesenchymal cells toward chondrocytes, and in early stages of chondrocyte differentiation.
To evaluate the role of Hif-1α in limb bud mesenchyme, we first ascertained the presence of hypoxia in precartilaginous condensations by injecting the hypoxia marker EF5 into pregnant female mice at embryonic day (E) 12, a stage at which precartilaginous condensations are well formed and chondrocytes are just starting to differentiate (). EF5 bound mesenchymal condensations that give origin to both the axial () and the appendicular skeleton (), whereas, with the exception of the skin, no significant binding was detected in the surrounding soft tissues. These data demonstrate that mesenchymal condensations that give origin to the endochondral skeleton are hypoxic during development.
Consistent with the EF5 detection, whole mount in situ hybridization showed detectable Hif-1α mRNA in the limb bud mesenchyme and in axial condensations as early as E10.5 (unpublished data). This expression persisted in limb bud mesenchymal condensations and in axial condensations at E11.5 and E12.5 (unpublished data). A similar pattern of Hif-1α protein was detected by whole mount immunohistochemistry at E10.5 () and by immunohistochemistry at E12 (Fig. S1, A and B, available at ). Notably, Hif-1α protein was also highly expressed in the apical ectoderm of the limb bud at E13.5 (). No staining was seen in embryos treated with secondary antibody alone (unpublished data).
We then evaluated the transcriptional activity of Hif-1α in limb bud mesenchyme by generating two types of hypoxia-inducible reporter mice. Five HREs were placed in front of either a retroviral E1b or murine hsp68 promoter fragment, driving the LacZ reporter gene to create the 5XHRE-E1b/LacZ and 5XHRE-hsp/LacZ transgenes, respectively (). The 5XHRE-E1b fragment has previously been used to successfully direct a hypoxia-specific induction of luciferase in cell culture (). The hsp68 promoter fragment is a well-characterized promoter capable of being activated in distinct patterns by defined heterologous enhancer elements in transgenic mice (). Neither the hsp68 nor E16 fragment alone is activated by hypoxia.
To confirm that the 5XHRE enhancer fragment used in our transgenic models was also hypoxia inducible in vivo, E8.5 5XHRE-E1b/LacZ embryos were cultured under normoxia or anoxic conditions for 24 h, and then analyzed for LacZ expression by X-gal staining (). In contrast to embryos cultured at 21% oxygen, a high level of LacZ expression was observed in embryos exposed to 0% oxygen for 24 h. This finding demonstrates that the 5XHRE fragment used in our experiments is induced by hypoxia to drive reporter expression in vivo. Notably, the embryos kept in normoxic conditions did not show any detectable signal. Although it is conceivable that E8.5 embryos are mildly hypoxic in utero, the short half-life of β-galactosidase () combined with the absence of stimulation of the reporter in the presence of oxygen likely explains this result. In addition, it is important to note that both the 5XHRE-E1b/LacZ and 5XHRE-hsp/LacZ are reporter constructs for both Hif-1α and Hif-2α; therefore, stimulation of these constructs in hypoxic conditions might also result in part from Hif-2α activity.
For either transgenic line, numerous founders were generated. For the purpose of our study, we have analyzed one 5XHRE-hsp/LacZ founder line in detail. Consistent with the EF5 findings and the expression of Hif-1α mRNA and protein, whole mount β-galactosidase staining at E10.5 showed a high level of LacZ expression in the limb bud mesenchyme (). This expression was even more evident at E12.5, and at this age it overlapped with regions of mesenchyme condensations (). Similar to the expression of Hif-1α mRNA and protein, β-galactosidase staining was also detectable in axial condensations ().
To dissect the role of Hif-1α in early chondrocyte differentiation, we conditionally inactivated Hif-1α in limb bud mesenchyme, using a Prx1 promoter-driven Cre transgenic mouse (), and a mouse homozygote for a floxed Hif-1α allele (Hif-1α; ). The Prx1-Cre transgenic line expresses Cre recombinase prevalently in limb bud mesenchyme starting from E9.5 (). Both Hif-1α and Prx-1 Cre mice were indistinguishable from wild-type animals (unpublished data). Newborn Hif-1α; Prx-1 Cre (CKO) mice, were viable, but had a characteristic shortening of both forelimbs and hindlimbs (, arrows; the length of newborn tibia is 3.4 ± 0.3 mm in control vs. 1.5 ± 0.2 mm in mutant; newborn ulna 3.9 ± 0.2 mm in control vs. 1.2 ± 0.5 mm in mutant). Real-time PCR analysis of genomic DNA extracted from newborn control and CKO paws, after removal of the skin, showed that efficiency of deletion of the Hif-1α gene was 75 ± 2.5% at this age. The result is particularly substantial, especially in light of the tissue heterogeneity of the specimens. In addition, at E14.5, accumulation of Hif-1α protein was severely decreased in CKO forelimb autopod when compared with control (). Given the nature of the immunohistochemistry, it is difficult to establish whether the remaining signal in the CKO autopod is background, or if it reflects some residual Hif-1α protein.
Despite the report of Prx1-Cre expression in the skull and, to some degree, in the axial skeleton, but consistent with a robust expression of Cre in limb bud mesenchyme (), no obvious abnormalities could be observed elsewhere than in the limbs of mutant mice. Mutant limbs were severely shorter and misshapen in comparison to controls (). This finding suggests that in the absence of Hif-1α, the process of endochondral bone development was severely impaired. Histological analysis of the limb proximal bones, i.e., of the stylopod and zeugopod, at birth confirmed their extreme shortening, their severe deformities, and the massive central cell death phenotype concomitant to aberrant proliferation of viable chondrocytes (Fig. S2, C–F, available at ; and not depicted) that we previously reported in growth plates of mice in which Hif-1α had been conditionally inactivated in chondrocytes using a Col2a1 promoter-driven Cre (). The central cell death phenotype was already massive at E13.5 in the stylopod and zeugopod of CKO mice (Fig. S2 B; and Fig. S4, F and H).
To address the role of Hif-1α in early chondrogenesis, we then carefully analyzed the phenotype of CKO mice during early limb development (). The online version of this article contains supplemental material. Surprisingly, no obvious histological evidence of spatially localized loss of cell viability could be histologically observed in the distal portion of the mutant limbs, i.e., in the autopod (see below). This result, i.e., lack of the central cell death phenotype in the CKO autopod, was clearly different from what we had observed in the CKO stylopod and zeugopod. We decided to take advantage of this finding, and in our subsequent analysis, we focused exclusively on the autopod. Normally, undifferentiated mesenchymal cells in the limb bud start to condense around E11.5, and cells differentiate into chondrocytes soon after (). The precartilaginous condensations of E12.5 forelimbs appeared similar in CKO and control (), as further confirmed by both PNA staining () and Sox9 mRNA expression (). Thus, our data suggest that Hif-1α is not required for the formation of precartilaginous condensations. Consistent with the undifferentiated state of the cells at this stage, Col2a1 mRNA was expressed at very low levels in both CKO and control limbs ().
At E13.5, however, the CKO autopod presented a remarkable delay in cartilage formation compared with controls (, and not depicted). Cells in the cartilaginous elements of control limbs showed typical chondrocyte morphology, whereas cells in CKO condensations resembled undifferentiated mesenchymal cells, with no evidence of hyaline matrix in between (). Alcian blue staining confirmed the paucity of proteoglycan accumulation in the mutant autopod in comparison to the control element (). Lastly, Col2a1 mRNA expression at this stage was slightly lower in mutant autopods than in controls ().
At E14.5, histology and in situ hybridization analysis confirmed a severe delay in chondrocyte differentiation in the forelimb autopod (, and not depicted). Col2a1 mRNA expression was significantly lower in CKO compared with control, and, paradoxically, more intense in the distal, rather than in the proximal, portion of the mutant digital ray, whereas expression of Sox9, L-Sox5, and Sox6 mRNAs was similar in both mutants and controls (, and not depicted). Similar results were obtained in the E14.5 hindlimb autopod (unpublished data).
The impairment in early chondrogenesis observed in CKO autopod could affect later steps of chondrocyte maturation, including hypertrophic differentiation. Consistent with this hypothesis, the autopod of E14.5 CKO forelimbs showed fewer regions of Indian hedgehog (Ihh) and Col10a1-expressing chondrocytes (). In addition, the pattern of expression of Col10a1 mRNA was clearly abnormal in the mutant versus control, as Col10a1 mRNA was detected in the distal, but not in the proximal portion of the mutant digital ray (). We do not have a good explanation for this abnormal distribution at the moment. Similar findings were also observed in the autopod of E14.5 hindlimbs (unpublished data). The massive early loss of cell viability precluded a meaningful analysis of the effect of the loss of Hif-1α in early chondrogenesis in stylopod and zeugopod. However, we previously described that cell death is restricted to the core of the cartilaginous elements (), and we took advantage of this to assess the degree of hypertrophic differentiation (late chondrogenesis) of chondrocytes located away from the core. In situ hybridizations analysis of superficial sections obtained from E14.5 zeugopod and stylopod confirmed a marked delay of hypertrophic differentiation in the mutant specimens (Fig. S2, K–P).
Histological analysis at birth of the autopod of both forelimbs and hindlimbs showed a severe reduction of hypertrophic chondrocytes and bony trabeculae in the metacarpals, metatarsals, and phalangeal elements of the mutant mice when compared with control (; and Fig. S3, C and D, available at ). Consistent with these data, at P9, bony trabeculae were present in control, but not in CKO, tarsal bones (Fig. S3, K and L; see the absence of pink staining in L), and no mineralized secondary ossification center was detectable in the radius and ulna of CKO animals (). In addition, talus and calcaneus in CKO mice showed extensive cartilaginous remnants that were absent in control bones (Fig. S3, M and N, blue staining).
In situ hybridization analysis of chondrogenic markers in the autopod of newborn hindlimbs further proved that removal of Hif-1α in limb bud mesenchyme retards hypertrophy (Fig. S4, available at ). Expression of Ihh, Col10a1, and OP were completely absent from the region corresponding to the future phalangeal elements (Fig. S4, E–J), indicating that no hypertrophic chondrocytes were present in these areas, as also shown by histological analysis (Fig. S3, C and D). Furthermore, the distance between either the two Ihh or Col10a1 expression domains, as well as the extension of OP expression were significantly reduced in the prospective metacarpals (Fig. S4, E–H and I–J, respectively), suggesting that the overall replacement of cartilage by bone was significantly delayed. Conversely, we observed an increase in Sox9 expression in CKO bones compared with controls (Fig. S4, C and D). Because during endochondral bone development, Sox9 mRNA expression decreases over time, this observation further demonstrates that the CKO bones were, overall, younger than controls.
Immunohistochemistry analysis revealed that accumulation of Hif-1α protein is particularly abundant in the prospective joints of E13.5 and E14.5 forelimb autopods, both in the digital rays and in the wrist (; and Fig. S1, C and D). Consistent with this finding, in situ hybridization showed detectable VEGF mRNA expression, which is a classical target of Hif-1α, in the same regions, at a similar stage (). In addition, analysis of EF5 staining demonstrated that the developing joints at E13.5 are highly hypoxic (). Notably, careful histological analysis of the autopod at E13.5 revealed a remarkable thickening of an avascular perichondrium around the area of the future joints () that persisted at E14.5 ( and Fig. S3 A). Moreover, at E15.5, after the joint space had formed, articular chondrocytes showed a significantly higher degree of hypoxia than the rest of the cartilaginous element ().
Consistent with this model, the digital ray was not yet segmented into metacarpal and phalange elements in CKO forelimbs at E13.5 and E14.5, whereas this segmentation had already occurred in control limbs (; and , respectively). Distal joints, however, eventually developed at the right location in newborn CKO forelimbs (). Similar results were also observed in the digital ray of the hindlimbs (Fig. S3, A–D). No segmentation of the digital ray into metatarsal and phalange elements was yet evident at birth in CKO hindlimbs (Fig. S3 D), whereas a definition of metatarsals and phalanges had already occurred at E14.5 in control elements (Fig. S3 A). Also in this case, metatarsals and phalanges were eventually all properly segmented in 9-d-old (postnatal day [P] 9) CKO animals (unpublished data). Of note, thickening of the perichondrium was not affected in E13.5 and E14.5 mutant autopods (; ; and Fig. S3 B), further supporting the hypothesis that this event precedes specification of the joints and can, thus, have a critical role in joint development. We also looked at joints in stylopod and zeugopod of mutant mice and observed that, in contrast to control, the interzone that normally forms between the scapula and the humerus was still absent in E13.5 mutant limbs (Fig. S3, E and F), suggesting that the role of Hif-1α in joint development is not limited to the autopod. At other sites, such as the elbow, the analysis was not conclusive, as a consequence of both the massive cell death and the severe deformities of the mutant elements (Fig. S3, G and H).
The aforementioned joint phenotype was even more severe in the ankle and wrist, which are extremely hypoxic during development ( and not depicted). Consistent with the previously described impairment of early chondrogenesis, E14.5 CKO wrist had no defined cartilaginous elements, which was different from control (unpublished data). In newborn animals, the CKO wrists contained dislocated and extremely misshapen elements (). The malformations persisted postnatally. At P9, despite an identical number of bones in mutant and control, many bones were still misshapen, and some only partially segmented (, c and 3) and/or dislocated in the wrist of CKO animals (). Similar, but more severe, abnormalities were observed in CKO ankles (Fig. S3, I–N). Notably, the ankle of CKO mice was a single skeletal element at birth, whereas individual elements were present in control ankles (Fig. S3, I and J). At P9, all the expected bones were present in CKO ankle, but some were still only partially segmented (Fig. S3, K and N; bones 2 and 3 are fused, and c is partially fused with 4/5).
Collectively, our data indicate that lack of Hif-1α severely affects joint development, but, at this stage, they do not allow us to distinguish whether delay of joint specification or rather of joint cavitation was the cause of the phenotype. To address this issue, we looked at the expression of GDF5, a marker of joint specification that is detected even before interzone regions can be recognized histologically, and that represents one of the earliest known markers for joint formation (; ). In E12.5 forelimbs, GDF5 mRNA appeared to be similarly expressed in the interdigital tissues of both CKO and control limbs (unpublished data). At E13.5, expression of GDF5 mRNA was already detectable in the control early prospective joints, but not in the CKO autopods (), indicating a delay in joint specification. At E14.5, control autopods presented strong, sharp stripes of GDF5 expression in the regions of prospective joints (), whereas in CKO, GDF5 expression was weaker and diffuse (), confirming the requirement of Hif-1α for joint development. In addition, consistent with a severe delay of segmentation, cells occupying the future joint regions were still present in the autopod of newborn CKO hindlimbs and, although they did not have detectable levels of Col2a1 mRNA, they did express Sox9, and also, weakly and with a rather diffuse pattern, GDF5 mRNA (Fig. S4, B, D, and L, respectively).
To define the functional consequences of hypoxia in cartilage at the molecular level, and to get some insights into the molecular mechanism by which Hif-1α regulates early chondrogenesis and joint formation, metatarsals were isolated from E15.5 wild-type embryos and, after a few days in culture, exposed to 21 or 1% O for 8 h.
At 1% O, Hif-1α protein is stabilized and transcriptionally active (; ; ). Approximately 90% of the genes examined did not show any significant difference in response to hypoxia, as genes displaying less than −1.5- to +1.5-fold change were considered within normal range (). The genes modulated by hypoxia in chondrocytes belonged to a variety of biological categories, including transcription, cell cycle, apoptosis, adhesion, and angiogenesis (). For a complete list of the hypoxia-regulated genes, see Table S1 (available at ).
In addition to classical hypoxia-regulated genes, numerous novel targets of hypoxia, some of which had been previously involved in endochondral bone development, were identified (). Interestingly, expression of the master transcription factors of chondrogenesis, Sox9, L-Sox5, and Sox6, which play a particularly important role in early chondrogenesis (; ; , ), was virtually identical in hypoxic and normoxic specimens. The same results were obtained using primary chondrocytes cultured in hypoxic conditions for 8 h (unpublished data). These data confirm recently published findings showing that the Sox family of transcription factors is not a target of hypoxia ().
GDF5, Wnt14, and Noggin are main regulators of joint development (; ; ; ; ). The abnormal GDF5 expression pattern observed in mutant mice raises the possibility that lack of Hif-1α may delay joint segmentation by interfering with GDF5 expression. We thus investigated whether the expression of these factors could be directly regulated by hypoxia. Microarray experiments showed that exposure to 1% hypoxia for 8 h did not significantly induce GDF5 mRNA expression in metatarsal explants (), suggesting that this factor is not a direct, transcriptional target of Hif-1α. In the same microarray assay, we also searched for levels of expression of Wnt14 and Noggin. As for GDF5, we did not observe any significant difference in Wnt14 and Noggin mRNA expression in hypoxic versus normoxic conditions (). The same results were obtained with primary chondrocytes cultured in hypoxic conditions for 8 h (unpublished data).
In the same assays, Col2a1, aggrecan, and hyaluronan synthases mRNAs were also not differentially regulated by hypoxia (unpublished data). Notably, however, expression of prolyl-4-hydroxylase α (I) (P4haI) mRNA was increased by approximately threefold after 8 h of exposure to hypoxia in both metatarsal explants () and primary chondrocytes (not depicted). P4haI is a very well-documented gene of the hypoxia signature that is directly regulated by Hif-1α (), and it plays an essential role in matrix accumulation by controlling posttranslational modifications of the α-I chain of collagen molecules.
In contrast to the stylopod and zeugopod, and consistent with our histological findings, no detectable TUNEL-positive cells were found in the cartilaginous elements of either control or CKO autopod, whereas positive cells were observed, as expected, in the interdigital space of both (, and not depicted). These results indicate that the abnormal chondrogenesis in the autopod of CKO limbs cannot be attributed to cell death.
Hif-1α controls angiogenesis, at least in part, by regulating expression of VEGF (). In the growth plate, VEGF is important for both chondrocyte survival and blood vessel invasion (; ; ; ). Thus, we looked at VEGF mRNA expression by in situ hybridization in E14.5 forelimbs. Consistent with previous studies (), we observed that at this stage, VEGF mRNA was not detectable in chondrocytes forming the digital ray of control specimens, despite accumulation of Hif-1α protein (). This finding argues against a role of VEGF downstream of Hif-1α in early chondrogenesis.
To investigate whether removal of Hif-1α in limb bud mesenchyme would alter angiogenesis with yet unknown mechanisms, we then analyzed the expression of VEGF-receptor2 (VEGFR2), which is a marker of endothelial cells (). At both E12.5 and E14.5, VEGFR2 mRNA expression was comparable in the soft tissue surrounding the precartilaginous and cartilaginous elements, respectively, of control and CKO (). Notably, at E14.5, this marker was also detectable in the bone collar of control elements (), whereas CKO forelimbs lacked the bone collar structures at this age because of their delay in hypertrophic differentiation ( and not depicted).
#text
Pregnant females were injected (i.p.) at the appropriate stage with 10 mM EF5 at 1% of body weight; staining was performed using a Cy3-conjugated antibody, as previously described ().
Embryos were processed and immunohistochemistry was performed as previously described (). Embryos were incubated with a primary antibody against Hif-1α (R&D Systems) diluted 1:100.
Wholemount in situ hybridization using DIG-labeled RNA probes was performed essentially as previously described (). The hybridization was performed at 65°C overnight, and the signal was detected using the colorimetric BM-Purple substrate (Roche).
The 5XHRE-E1b/LacZ transgene was constructed by blunt-ending a 4.26-kb NcoI–PstI nuclear localization signal/LacZ–containing fragment from pIRESLacZ (a gift from A. Nagy, Samuel Lunenfeld Research Institute, Toronto, Canada) into the HindIII–XbaI sites (luciferase removed) of p5XHRE-Luciferase (). The 5XHRE-hsp/LacZ transgene was constructed by blunt-ending a 0.3-kb KpnI–HindIII 5XHRE- containing fragment from p5XHRE-Luciferase into the HindIII site of pHspLacZpA (compliments of J. Rossant, Samuel Lunenfeld Research Institute, Toronto, Canada). Transgenic founder mice were identified by Southern blot analysis of AccI–EcoRI–digested tail DNA probed with a 1.1-kb LacZ-containing fragment from SphI-digested mb-STOP-LacZ (). Genotyping was performed on genomic DNA by PCR. Yolk sac or tail DNA was amplified for 40 cycles (1 min at 94°C, 1 min at 55°C, and 1.5 min at 72°C) on a thermal cycler. The LacZ primers 5′-GGTGATTTTGGCGATACGC-3′ and 5′-TGCAGGAGCTCGTTATCGC-3′ produce a 191-bp product.
E8.5 embryos were dissected individually into cold sterile L15 media, staged for somite number, and transferred to 1 ml DME media containing 20% fetal calf serum and 1× nonessential amino acids. Embryos were cultured either in normoxic (21% O or 5% CO at 37°C) or in anoxic (hypoxia chamber; 0% O or 5% CO at 37°C) conditions for 24 h. Embryos were then processed for X-Gal staining.
Whole mount X-Gal staining of freshly dissected mouse embryos was performed as previously described (). Cultured embryos were stained with X-Gal solution at 37°C overnight.
Male hemizygous Prx-1 Cre transgenic mice in the Swiss-Webster background () were bred with animals that were homozygous for a floxed Hif-1α allele (Hif) in the FVB/N background (). Males heterozygous for the floxed Hif-1α and homozygous for the Prx-1 transgene were crossed with female mice homozygous for the floxed Hif-1α allele to generate Hif-1α; Prx-1 Cre mutant mice, which were genotyped as previously described (; ).
Alizarin red S staining was performed as previously described (). For light microscopy, tissues from E12.5, E13.5, and E14.5 mice (delivered by caesarean section), as well as newborn and p9 mice, were fixed in 10% formalin/PBS, pH 7.4, and stored in fixative at 4°C. Paraffin blocks, sections, and H&E staining were realized by standard histological procedures. In situ hybridizations were performed using complementary S-labeled riboprobes, as previously described (). For TUNEL assay, paraffin sections from hindlimbs of newborn mice were permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate. TUNEL assay was performed using an In Situ Cell Death Detection kit (Roche) according to the manufacturer's conditions. For PNA staining, sections were dewaxed, rehydrated, and pretreated for 20 min with 0.5% HO in methanol, and washed in PBS-T (PBS + 0.05% Tween-20). Slides were then incubated with biotinylated PNA (Vector Laboratories) at 100 μg/ml in PBS-T + 0.1 mM CaCl for 45 min at RT, and then washed extensively with PBS-T. Bound PNA was detected using the TSA- Biotin system (Perkin Elmer) and DAB substrate (Vector Laboratories) according to the manufacturer's instructions.
Frozen sections of E12 and paraffin sections of E13.5 and E14.5 mouse autopods were heated at 80°C in 0.1 M citrate buffer, pH 6.0, for 2 h. After quenching of the endogenous peroxidase by incubation in 3% HO/PBS for 10 min at room temperature, blocking was performed with a specific solution provided by a TSA kit (NEN). Sections were incubated with a primary antibody against Hif-1α (R&D Systems) at a dilution of 1:500 at 4°C overnight. After incubation with the appropriate biotinylated secondary antibody, detection of the binding was carried out using the Streptavidin-HRP system provided by the TSA kit, following the manufacturer's instructions.
All images were acquired with a microscope (Eclipse E800; Nikon) using Plan Apo 4×/0.2 NA, Plan Apo 10×/0.45 NA, Plan Apo 20×/0.75 NA, and Plan Apo 40×/0.95 NA lenses (Nikon), at ambient temperature, with air imaging medium. Cy3 fluorochrome was used for the EF5 detection. Photographs were taken using a SPOT camera (model 1.30) and SPOT software version 3.5.5 for Macintosh (both from Diagnostic Instruments). Images were assembled and legends were added using PowerPoint software (Microsoft). Figures were transferred into Photoshop software (Adobe) and brightness and contrast were modified by applying brightness/contrast adjustments to the whole image, with the strict intent of not obscuring, eliminating, or misrepresenting any information present in the original, including background.
Chondrocytes were isolated from newborn wild-type mice as previously described (). Chondrocytes were plated at a density of 4 × 10 cells per well of 6-well plates and grown in monolayer cultures in high-glucose DME (Invitrogen) supplemented with 10% FBS (HyClone) and 1% penicillin/streptomycin. At days 5, 10, and 20 after plating, respectively, cells were exposed to 21% (normoxic) or 1% (hypoxic) O for 8 h; then total RNA was isolated, as previously described ().
Metatarsal explants were obtained from E15.5 mouse hindlimbs and cultured as previously described (). After 3 d in culture, metatarsals were exposed to exposed to 21% (normoxic) or 1% (hypoxic) O for 8 h, and total RNA was isolated as previously described (). For these experiments, biological triplicates were used.
Deletion of Hif-1α was confirmed by real-time quantitative PCR analysis of genomic DNA extracted from five independent controls and mutant forelimb paws at birth after removal of the skin. Real-time PCR was performed as previously described (). Sequences of primers are available upon request. The Vhlh gene was amplified as an internal control. Cycle threshold (Ct) values were measured and calculated by the sequence detector software. Relative amounts of mRNA were normalized to Vhlh and calculated with the software program Excel (Microsoft). Relative genomic DNA contents were calculated as = 2, in which ΔΔCt = ΔE − ΔC, ΔE = Ct − Ct, and ΔC = Ct − Ct. A fourfold difference in efficiency of amplification was calculated between mutants and controls, indicating that the efficiency of deletion in the mutant samples was ∼75%.
Fig. S1 shows Hif-1α protein expression in early mesenchymal condensations and in prospective joints. Fig. S2 describes the phenotype of the stylopod and zeugopod of CKO animals characterized by early cell death, disorganized growth plate chondrocytes, and delayed hypertrophy. Fig. S3 presents the abnormal joint development in CKO hindlimb autopods/ankles and in CKO stylopods (shoulder) and zeugopods (elbow). Fig. S4 describes the abnormal distribution of chondrogenic and joint-specific markers in CKO hindlimb paws. Table S1 lists the genes regulated in metatarsal explants exposed to hypoxia for 8 h. The online version of this article is available at . |
Ubiquitylation and proteolysis of cell cycle regulators is essential for control of the cell cycle. In particular, the anaphase-promoting complex/cyclosome (APC/C) ubiquitin ligase regulates mitosis by directing the degradation of key substrates at different times during mitosis. The APC/C is regulated at several levels to ensure that its activity is restricted to mitosis and G1 phase and that it recognizes specific substrates at specific times. This is achieved through its phosphorylation in mitosis, through its interaction with coactivators (including Cdc20 and Cdh1), and by inhibitors such as regulator of cyclin A 1 (Rca1)/early mitotic inhibitor 1 (Emi1; ; ; ).
Interaction between the APC/C and its two coactivators is regulated during the cell cycle (for review see ). Cdc20 can only interact with the phosphorylated form of APC/C that is specific to mitosis and is itself degraded after metaphase, thereby restricting its activity to the first part of mitosis. In contrast, Cdh1 can interact with both the interphase and mitotic forms of the APC/C, and its binding depends on its phosphorylation status. In budding yeast, Cdh1 is phosphorylated by B-type cyclin–Cdk complexes from S phase to anaphase, and this prevents it from binding to the APC/C. The Cdc14 phosphatase subsequently dephosphorylates Cdh1 at mitotic exit (; ), and the Cdh1-bound form of the APC/C (APC/C) aids mitotic exit and regulates G1 phase. Cdh1 is phosphorylated in metazoans too, and this also prevents its binding to APC/C (; ; ). However, how Cdh1 phosphorylation is regulated during the cell cycle is less clear.
In metazoans, APC/C is negatively regulated in interphase in by Rca1 and in mammalian and cells by the orthologous protein Emi1. The protein Rca1 was originally identified as a regulator of cyclin A whose mutation induces an arrest in G2 phase of cell cycle 16, which is similar to that observed in cyclin A mutants (). Rca1 mutant cells block in G2 phase and prematurely degrade mitotic cyclins in a Cdh1-dependent manner (). Similarly, in extracts, Emi1 immunodepletion prevents mitotic entry, and cyclins are prematurely degraded ().
Emi1 was originally identified in a yeast two-hybrid screen for F-box proteins using Skp1 as bait (). Emi1 also contains a zinc-binding domain that is important for APC/C inhibition. Emi1 levels are regulated during the cell cycle: in mammalian cells, its transcription is induced at the G1/S transition under the control of E2F, and this has been reported to be required to stabilize cyclin A and allow cells to begin S phase (). Overexpressing Rca1 in G1-phase cells has also recently been shown to promote S phase in embryos in an F-box–dependent manner (). Emi1 is subsequently degraded at the entry to mitosis (; ) by the SCF ubiquitin ligase (; ). SCF recognizes Emi1 only after it has been phosphorylated by Plk1 (; ), which creates the phosphodegron DSGxxS. Recently, the Evi5 protein has been reported to stabilize Emi1 in S and G2 phases by blocking its phosphorylation by Plk1 ().
Emi1 has been proposed to have a wider role than Rca1 by acting on APC/C as well as APC/C and, thus, to play a key role in controlling activation of the APC/C in mitosis. Emi1 has been reported to inhibit APC/C when cells begin mitosis such that its degradation is essential for APC/C activation and the degradation of mitotic regulators (such as cyclins A and B and securin; ; ). Excess Emi1 inhibits cyclin B ubiquitylation when added to an in vitro ubiquitylation assay, and excess Emi1 added to egg extracts prevents cyclins A and B, securin, and geminin degradation (). Furthermore, overexpressing Emi1 in vertebrate cells induces a delay in mitosis (; ), which has been interpreted as evidence that it inhibits Cdc20. In contrast, genetic interactions in are not consistent with a role for Rca1 in the inhibition of Cdc20 ().
In this study, we have investigated the role and regulation of Emi1 in mammalian cells in vivo. We find that Emi1 is normally degraded in prophase but that a nondegradable version of Emi1 has no effect on the timing or extent of cyclin A destruction nor on the destruction of the majority of cyclin B1 and securin. Thus, we show that Emi1 does not regulate activation of the APC/C at mitosis. Rather, we find that the major requirement for Emi1 is to allow the accumulation of mitotic cyclins and geminin in G2 phase in order to allow cells to enter mitosis and to prevent rereplication.
We set out to test the role of Emi1 in regulating the APC/C at mitosis. We used a commercial antibody to characterize Emi1 in human cells. We confirmed that the antibody is specific for Emi1 because it recognized a band of the correct size (∼50 kD) on an immunoblot of whole cell lysate ( and Fig. S1 A, lane 1; available at ) that decreased after treating cells with siRNA oligonucleotides specific for Emi1 (Fig. S1 A, lanes 4–5). Furthermore, it recognized higher molecular mass bands of the correct size after transfecting plasmids encoding HA-tagged or fluorescent protein-tagged forms of Emi1 into human tissue culture cells (Fig. S1 A, lanes 2–3). A minor faster migrating band was also reproducibly detected whose significance is unknown but seemed to be related to Emi1, as it also disappeared after siRNA treatment (Fig. S1 A). Using this antibody, Emi1 protein levels were analyzed in HeLa cells synchronized by a thymidine/aphidicolin block and release regime (see Materials and methods); samples were collected during S and G2 phases and mitosis. The cell cycle phase was monitored by flow cytometry, which showed that most of the cells passed through mitosis between 12 and 15 h after release from the aphidicolin block. Emi1 levels decreased 11 h after release (), when most of the cells were in late G2 phase or mitosis, and before cyclin B1 levels decreased at 13 h, which is consistent with a previous study (). This was ∼2 h after we could detect an increase in Plk1 activity as judged by phosphorylation on S133 of cyclin B1, a known Plk1 phosphorylation site (; ).
The degradation of Emi1 had been reported to be essential for activation of the APC/C at mitotic entry and, thus, for the degradation of early mitotic regulators, including cyclin A. Therefore, it was important to characterize the exact timing of Emi1 degradation. With this aim, we generated a chimeric protein in which the Venus fluorescent protein () was fused to the N terminus of Emi1. A plasmid encoding Venus-Emi1 was injected into HeLa cells in G2 phase, and the fluorescence levels were measured as a read-out for Emi1 degradation. Venus-Emi1 started to be degraded in prophase usually before or occasionally at nuclear envelope breakdown (NEBD; ), as determined by differential interference contrast (DIC) images (). The earliest time we observed Emi1 proteolysis to start was 20 min before NEBD. We also analyzed Emi1 degradation in the nontransformed hTert–retinal pigment epithelium (RPE) cell line and observed very similar timings (Fig. S1 B). In vitro, Emi1 has to be phosphorylated by Plk1 before it can be degraded (; ), and the timing that we observe for Emi1 degradation is consistent with the time when Plk1 is activated ().
A β-TrCP recognition site (DSGxxS) had been identified in Emi1 (; ), and mutating either Asp or Gly in this motif prevented the recognition of other substrates by β-TrCP (). Therefore, we generated two different Emi1 mutants (D144A and G146V) in the β-TrCP recognition site and tested their stability. Both mutants were stable in HeLa () and hTert-RPE (Fig. S1 C) cells progressing through G2 phase and mitosis, confirming that the β-TrCP site was required for the degradation of Emi1 in prophase. We also generated a mutant (S145A/S149A) to ablate phosphorylation by Plk1 and found that this also stabilized Emi1 in mitosis (), which is consistent with previous studies (; ; ). In most of the experiments described in this and the next section, each of the D144A and G126V degradation mutants and the S145A/S149A phosphorylation mutants were used to rule out any effects specific to only one mutant. We could also demonstrate that the S145A/S149A mutant retained its biological activity by using it to rescue cells depleted of endogenous Emi1 by siRNA (unpublished data).
The timing of Venus-Emi1 degradation with respect to NEBD was variable and, in some cases, started 20 min before NEBD; in comparison, the earliest mitotic APC/C substrates (e.g., cyclin A) are degraded at or after NEBD (; ). When we compared the timing of Emi1 and cyclin A degradation in the same cell using Venus-Emi1 and cyclin A–CFP, we found that these were also variable and that Emi1 began to be degraded anywhere between 6 and 45 min (mean of 26 min; = 8), before cyclin A. This made it unlikely that Emi degradation was the trigger that directly activated APC/C at mitosis, although it could still be a prerequisite to activate APC/C.
To determine whether Emi1 had to be degraded to activate APC/C, we injected wild-type and nondegradable Emi1 constructs into cells and assayed the effect on the degradation of the three most well-characterized substrates of APC/C: cyclin A, cyclin B1, and securin. We assayed both the timing of degradation in vivo and the level of endogenous proteins remaining in cells arrested in mitosis after the injection of Emi1.
When we injected RNA encoding nondegradable Emi1 at 1 μg/μl, this produced levels of Emi1 similar to those found in previously published transient transfection experiments, and we found that as reported, cells delayed in mitosis (; ; ). However, in these experiments, Emi1 was expressed at 10–30-fold more than its endogenous level (see Materials and methods; ). Thus, the delay in mitosis might not be apparent if we expressed Emi1 closer to its normal levels, which would be consistent with cells lacking β-TrCP (and thus unable to degrade Emi1) that only exhibit a slight delay (∼30 min) in exit from mitosis after release from a nocodazole block (). Therefore, we reduced the amount of RNA encoding nondegradable Emi1 between 2- and 10-fold (0.1 and 0.5 μg/μl) and found that 51% of cells ( = 45) expressing lower amounts of Emi1 were able to perform anaphase successfully and reenter interphase. This indicated that the cells must have activated the APC/C and degraded most or all of their securin and mitotic cyclins.
The delay in mitosis seen with high levels of nondegradable Emi1 had been attributed to inhibition of the APC/C and, in particular, protecting cyclin A from destruction. To determine whether the timing of cyclin A degradation was affected by high levels of nondegradable Emi1, we analyzed cyclin A degradation in living cells after coinjection with high levels of any of the three nondegradable Emi1 mutants. We found that cyclin A–CFP degradation was unaffected by the presence of Emi1 (). To rule out any possible artifact caused by the presence of the tag on Emi1, we also analyzed cyclin A degradation in cells coinjected with untagged Emi1 D144A expressed from an internal ribosome entry site (IRES) vector coexpressing GFP and obtained similar results (Fig. S2 A, available at ). Similarly, the timing of cyclin B1 and securin degradation, which starts at metaphase (; ), was unaffected when nondegradable Emi1 mutants were coinjected with cyclin B1–CFP () or securin-Cerulean (). Similar results were obtained when the wild-type version of Venus-Emi was coinjected with cyclins A and B1 or securin (unpublished data). Thus, APC/C is activated upon entry to mitosis regardless of the presence of even supraphysiological levels of Emi1. Although the timing of securin destruction was unaffected, we did find that its extent was more sensitive to the levels of Emi1 than either cyclin A or B1 destruction.
We also analyzed the levels of the endogenous cyclins and securin in cells arrested in mitosis by high levels of nondegradable Emi1. Each of the three nondegradable Emi1 mutants were injected separately into G2-phase HeLa cells, and cells were followed by time-lapse DIC microscopy. After the cells were delayed for >2 h in mitosis, they were fixed and processed for immunofluorescence. Although immunofluorescence is not a truly quantitative approach, the relative levels of cyclins A and B1 and securin in noninjected control cells at different stages of mitosis were consistent with the known timings of their degradation. In noninjected cells, cyclin A levels were maximal in G2 phase and prophase, decreased in prometaphase and metaphase, and were almost undetectable in anaphase and telophase cells (), which is consistent with cyclin A degradation starting in prometaphase (; ). The immunofluorescence signal of endogenous cyclin A in cells expressing high levels of a nondegradable Emi1 that had delayed in a metaphase-like state was similar to the intensity of noninjected anaphase/telophase cells (), indicating that, like the cyclin A–CFP marker, endogenous cyclin A was degraded completely as normal. The three nondegradable Emi1 mutants produced identical results.
Cyclin B1 degradation normally starts when the spindle checkpoint is satisfied at metaphase (). Consistent with this, we observed maximum cyclin B1 levels in prometaphase control noninjected cells, whereas the signal was almost undetectable in anaphase and telophase cells (). When we analyzed the cells delayed by high levels of Emi1, we found that cyclin B1 levels were lower than in control prometaphase or metaphase cells but not as low as in control anaphase and telophase cells (). This showed that the bulk of cyclin B1 had been degraded but that a fraction of cyclin B1 remained, accounting for the cells delaying in mitosis. To rule out any artifact caused by the presence of the tag, we quantified the levels of either endogenous cyclins A (Fig. S2 B) or B1 (Fig. S2 C) in cells that had delayed in mitosis after injecting untagged Emi1 D144A expressed from an IRES vector. For both cyclins, results were identical to those obtained with Venus-D144A injection ().
Staining Emi1-arrested cells for endogenous securin showed that some securin also remained (unpublished data), which was consistent with the cells delaying in metaphase, and we confirmed that the sister chromatids were still attached by staining the cells with anti–Aurora B and autoimmune CREST antibodies to visualize the kinetochores and centromeres ().
To assay the effect of Emi1 on mitosis in nontransformed cells, we used hTert-RPE cells. Cells were stimulated to reenter in the cell cycle after serum starvation and were injected with nondegradable Emi1 when they had reached G2 phase. With high levels of Emi1, these cells also delayed in mitosis. We fixed the cells, measured the levels of endogenous cyclins A and B1 by immunofluorescence, and compared these with the levels in control noninjected cells. We found that as in HeLa cells, endogenous cyclins A and B1 had been degraded (Fig. S2, D and E). Thus, in both normal and transformed cells, Emi1 did not have to be degraded for all of cyclin A and the bulk of cyclin B1 to be destroyed. Although a fraction of cyclin B1 remained in cells expressing supraphysiological levels of Emi1, cyclin B1 destruction began at its normal time. In all, these results are clearly inconsistent with the model that Emi1 degradation activates APC/C and with the proposal that Emi1 is required to stabilize cyclin A in early mitosis.
Because Emi1 did not appear to control activation of the mitotic APC/C, we investigated its role in the cell cycle by depleting it by siRNA. Two different siRNA duplexes specific for Emi1 (siEmi1_1 and siEmi1_2) were effective in reducing Emi1 protein levels, although to different extents (see ). To analyze the effects of reducing Emi1 levels during the cell cycle, synchronized cells were treated with siRNA, and cells were analyzed by flow cytometry. Emi1 siRNA–treated cells arrested at the G1/S transition in an aphidicolin block and progressed through S and G2 phase after release from aphidicolin in a similar manner to that of control siRNA–treated cells (). Labeling cells with short pulses of BrdU to reveal the pattern of replication foci in S phase showed that Emi1-depleted cells also resembled control cells in having many evenly distributed foci in early S phase and fewer foci that tend to be closer to the nuclear envelope in mid- to late S phase (; ). These changes likely reflect the firing of early, mid-, and late origins of replication. However, Emi1-depleted cells differed from controls in that the foci persisted after cells had reached a 4N DNA content as judged by flow cytometry.
Around 15 h after release from the aphidicolin block, most of the control cells had progressed through mitosis, and the bulk of the cells had entered the following G1 phase (2N DNA content) after 20 h (). However, at these time points, cells treated with siRNA targeting Emi1 accumulated with a 4N DNA content. At 40 h after release from the G1/S block, only a minor fraction of these cells had entered the next cell cycle (); most of the cells had increased in ploidy compared with control cells ( and see ).
To determine whether depleting Emi1 caused a block in G2 phase or mitosis, siRNA-treated cells were followed by time-lapse microscopy 11 h after release from an aphidicolin block. shows the cumulative total of cells that entered mitosis as judged by NEBD. Only a minor fraction of cells transfected with siRNA against Emi1 entered mitosis (), indicating that the cells accumulating in the 4N DNA peak observed by flow cytometry were delayed in G2 phase and not in mitosis. Similar results were obtained with an independent siRNA oligonucleotide specific for Emi1 (Fig. S3, available at ).
In cells and egg extracts, mitotic entry was blocked when Rca1 or Emi was mutated or depleted (; ), and this correlated with a failure to accumulate mitotic cyclins. To test whether this was also the case in mammalian cells, we collected samples at the same time points as for flow cytometry analysis and analyzed the levels of Emi1 and cyclins A and B1 by immunoblotting. In cells treated with the control siRNA oligonucleotides, Emi1 levels decreased 12–15 h after release from a G1/S block, when most of the cells were entering mitosis. As expected, cyclin A levels decreased with a similar timing, whereas cyclin B1 levels decreased later (). In stark contrast, in Emi1-depleted cells, cyclin A and B1 levels were almost undetectable (). Treating Emi1-depleted cells with a proteasome inhibitor caused cyclin A to accumulate, showing that Emi1 was indeed required to stabilize it (). Thus, the impairment in mitotic entry in Emi1-depleted cells correlated with a failure to accumulate cyclins A and B1.
In budding and fission yeasts and in cells, the mitotic cyclins have an important role in preventing rereplication, which might provide an explanation for the increase in ploidy in Emi1-depleted cells after 40 h (). However, in human cells, geminin has been implicated as the primary block to rereplication (; ). To study this further, we depleted Emi1 by siRNA in asynchronous HeLa and hTert-RPE cells and harvested the cells 24, 48, and 72 h after transfection. We used two different siRNA oligonucleotides to deplete Emi1, one of which (Emi1_2) was less effective than the other (Emi1_1), particularly at later time points (). Flow cytometry analysis showed that in contrast to cells treated with the control oligonucleotides, which continued to proliferate normally, most of the cells treated with the Emi1 siRNA oligonucleotides accumulated with a 4N DNA peak 24 h after transfection (), as we had observed in synchronized cells (). At later time points (48 and 72 h after transfection), a substantial number of cells became polyploid (), and this correlated with the appearance of enlarged nuclei (). Nuclei became even larger 72 h after transfection, indicating that the cells were actively replicating their DNA. We confirmed that the increase in polyploidy and enlarged nuclear size were caused by extra rounds of DNA replication by labeling cells with BrdU and measuring DNA synthesis by flow cytometry (Fig. S4 A, available at ).
The rereplication phenotype was more pronounced with oligonucleotide siEmi1_1 than with siEmi1_2, which is consistent with the more efficient knockdown observed by immunoblotting (). Cells treated with the siEmi1_2 siRNA oligonucleotides showed a partial phenotype both by flow cytometry and the number of enlarged nuclei visible after Hoechst staining. In particular, the rereplication phenotype seemed to be reduced 72 h after transfection, most likely because the proliferating nontransfected cells diluted out the rereplicating cells that were still visible as large nuclei ().
To prove the specificity of this phenotype, we cotransfected siRNA oligonucleotides with an expression plasmid encoding Emi1. The siEmi1_1 siRNA oligonucleotide is directed against the STOP codon of Emi1 and its 3′ untranslated region; the Venus-Emi1 construct does not contain the 3′ untranslated region and is thus resistant to siRNA. Control and Emi1 siRNA duplexes were cotransfected with either a plasmid expressing Venus-Emi1 or an empty vector expressing Venus. Because the size of the nuclei correlates with the degree of ploidy (), we measured the area of the fluorescent nuclei when the Emi1 or control oligonucleotides were cotransfected with either pVenus or pVenus-Emi1. The nuclei of cells cotransfected with pVenus and the Emi1 siRNA oligonucleotides had clearly increased in size ( and S4 E), which is similar to those in which the siRNA oligonucleotides alone were transfected (). In contrast, cotransfecting pVenus-Emi1 with the Emi1 siRNA oligonucleotides prevented the increase in nuclear size ( and S4 E). Cotransfecting pVenus-Emi1 did not have a major effect on the size of nuclei in cells transfected with control siRNA oligonucleotides ( and S4 E).
Previous studies had shown that both cyclins A and E are capable of stimulating DNA replication; therefore, we predicted that cells lacking cyclin A through the depletion of Emi1 might be able to rereplicate their DNA because they contained cyclin E. Therefore, we assayed siRNA-treated cells for the levels of cyclins A, B1, and E. We found that cells depleted of Emi1 had very low levels of cyclins A and B1 but higher than normal levels of cyclin E (), which is consistent with the idea that cyclin E–Cdk activity was driving the rounds of rereplication. To assay whether cyclin E–Cdk2 was responsible for driving DNA synthesis, we treated Emi1-depleted cells with roscovitine to inhibit Cdk2 and found that this blocked rereplication (), and the cells remained with a 4N DNA content.
As mentioned in the previous paragraph, another APC/C substrate, geminin, is required to prevent rereplication in several human cell lines. Geminin binds, inactivates, and apparently stabilizes Cdt1, which loads the minichromosome maintenance (MCM) protein complex onto DNA origins. We assayed the levels of geminin and Cdt1 on immunoblots and found that geminin and Cdt1 were substantially reduced in Emi1-depleted cells compared with control cells, whereas Cdc6 did not appear to change substantially (). Thus, Emi1 is required to stabilize the mitotic cyclins, geminin, and Cdt1 in interphase cells.
In , Rca1 has been shown to be epistatic to fizzy related, the homologue of Cdh1. To determine whether APC/C was responsible for degrading cyclins A and B1 in cells depleted of Emi1, we cotransfected cells with siRNA duplexes targeting Emi1 and Cdh1 and assayed the effect on cell division and cyclin levels. Cells were transfected with siRNA duplexes against Emi1 together with either control oligonucleotides or in combination with siRNA oligonucleotides targeting Cdh1 or an equal amount of control duplex, and protein levels were assayed by immunoblotting (). This revealed that codepleting Cdh1 antagonized the effect of depleting Emi1 such that cyclin A could then accumulate (), as did cyclin B1 and geminin (). In agreement with this, in cells depleted of both Emi1 and Cdh1, nuclear size was very similar to control transfected cells and much smaller than Emi1-depleted cells (), indicating that codepleting Cdh1 stopped rereplication in cells lacking Emi1. Furthermore, the cell number of Emi1-depleted cells was much lower than control cells, as one would expect from rereplicating cells that stop dividing, whereas there were a similar number of control and Emi1/Cdh1-codepleted cells at 72 h after transfection, indicating that these cells were proliferating. In contrast, codepleting Cdc20 had no effect on the rereplication phenotype (unpublished data), which was to be expected because we found that Cdc20 was unstable in Emi1-depleted cells (unpublished data).
Flow cytometry analysis confirmed these conclusions: in contrast to cells depleted of Emi1 that arrested with a 4N DNA content, which increased overtime as the DNA was rereplicated, cells depleted of both Emi1 and Cdh1 entered and then exited mitosis (Fig. S5, available at ) with a similar time course to control cells. Assaying cell division by trapping mitotic cells with taxol and measuring the cumulative mitotic index confirmed that the bulk of the Emi1-depleted cells did not enter mitosis, whereas cells depleted of both Emi1 and Cdh1 accumulated in mitosis (). Therefore, in the absence of Cdh1, Emi1 does not appear to be required for cell division.
xref
#text
HeLa cells were cultured in Advanced DME (Invitrogen). Cells were synchronized at the G1/S transition by a thymidine/aphidicolin block. The day after seeding, cells were blocked with 2.5 mM thymidine (Sigma-Aldrich) for 24 h, released for 12 h, and then blocked again with 5 μg/ml aphidicolin (Sigma-Aldrich) for 24 h. Cells were then released into fresh DME. hTert-RPE cells were cultured in 10% FBS + DME-F12 medium (Sigma-Aldrich). For synchronization, cells were serum starved for 24 h and re-fed with 25% FBS-containing medium. Cell cycle analyses were performed by flow cytometry using a cytometry system (FACSCalibur; Becton Dickinson).
Chimeric Emi1 proteins were produced by PCR from IMAGE clone ID3960052 using Advantage polymerase (CLONTECH Laboratories, Inc.). The fragment was cloned into the pT7-Venus-C1 vector. Emi1 mutants were generated by QuikChange Site-Directed Mutagenesis (Stratagene) using the following oligonucleotides: CCAGTAGACTTTATGAAGCCAGTGGCTATTCCTCA (forward) and TGAGGAATAGCCACTGGCTTCATAAAGTCTACTGG (reverse) for the D144A mutant; GACTTTATGAAGACAGTGTCTATTCCTCATTTTCTC (forward) and GAGAAAATGAGGAATAGACACTGTCTTCATAAAGTC (reverse) for the G146V mutant; and CCAGTAGACTTTATGAAGACGCTGGCTATTCCGCATTTTCTCTACAAAGTGG (forward) and CCACTTTGTAGAGAAAATGCGGAATAGCCAGCGTCTTCATAAAGTCTACTGG (reverse) for the S145A/S149A mutant. IRES constructs were generating by pasting the Emi1 D144A and Emi1 G146V sequences into pIRES2-EGFP (CLONTECH Laboratories, Inc.). All clones were confirmed by automated sequencing. Cyclins A2 and B1 and securin constructs have been described previously (; ).
siRNA duplexes against Emi1 corresponding to the sequence GATTGTGATCTCTTATTAA (siEmi1_1, which starts at 1,337 bp and overlaps the STOP codon) and ACTTGCTGCCAGTTCTTCA (siEmi1_2, which starts at 569 bp) and against Cdh1 () were synthesized by Dharmacon. Nontargeting siRNA against luciferase (Dharmacon) or glyceraldehyde-3-phosphate dehydrogenase siRNA (Ambion) were used as negative control siRNAs. Oligonucleotides were transfected at a final concentration of 50 nM according to the manufacturer's instructions using Oligofectamine (Invitrogen). Asynchronously growing cells were transfected, and samples were taken 24, 48, and 72 h after transfection. 28 μM roscovitine (Sigma-Aldrich) was added 24 h after transfection, and DNA content was analyzed 24 h later. 50 μM MG132 (Calbiochem) was added 24 h after transfection for 6 h.
Cells were lysed directly in SDS loading buffer. Samples were then syringed, boiled, and run on NuPAGE 4–12% Bis-Tris gels except where otherwise indicated. Proteins were transferred to a polyvinylidene difluoride membrane. Membrane saturation and all of the following incubation steps were performed in 5% low fat milk in PBS–Tween 20 (0.2%). Anti-Emi1 antibody (Zymed Laboratories) was used at 1:100. Anti–cyclin A (BF683; Cell Signaling), cyclin B1 (GNS-1; BD Biosciences), and cyclin E (HE12; Abcam) antibodies were used at 1:1,000. Antiphospho-S133 cyclin B1 () was used at 1:500. Anti-Cdh1 antibody (a gift from T. Hunt and J. Gannon, Cancer Research UK, Cambridge, UK) was used at 1:50. Anti–geminin antibody (a gift from R. Laskey, Hutchison/Medical Research Council Research Centre, Cambridge, UK) was used at 1:500. Anti-Cdt1 and anti-cdc6 antibodies (gifts from K. Helin, Biotech Research and Innovation Centre, Copenhagen, Denmark) were used at 1:100 and 1:2,000, respectively. Anti-actin (AC-40; Sigma Aldrich) and anti-Hsp70 (H-5147; Sigma-Aldrich) antibodies were used at 1:1,000 and 1:5,000, respectively. HRP-conjugated secondary antibodies (DakoCytomation) were used at 1:5,000. The antibodies were detected using ECL-plus (GE Healthcare) or were analyzed with the Odyssey Infrared Imaging System (LI-COR) for quantitative immunoblotting. Only the output levels were adjusted to assemble the panels in Photoshop.
Fig. S1 shows the specificity of the Emi1 antibody and Venus Emi1 degradation in hTERT-RPE cells. Fig. S2 shows mitotic cyclin degradation in hTERT-RPE cells injected with Venus Emi1 and in HeLa cells after the injection of untagged Emi1. Fig. S3 shows the requirement for Emi1 for mitotic entry using an independent Emi1 siRNA oligonucleotide. Fig. S4 shows that hTERT-RPE cells have the same response as HeLa cells to depleting Emi1 by siRNA treatment and demonstrates the specificity of the phenotype by rescue with a plasmid expressing Emi1. Fig. S5 shows the rescue of mitotic entry in Emi1-depleted cells by codepleting Cdh1 as assayed by flow cytometry. Online supplemental material is available at . |
DNA methylation is generally thought to silence gene expression and reduce transcriptional noise by compacting chromatin structure. However, how this is brought about has not been systematically investigated. The impact of DNA methylation on histone modifications is well established in plants () but not in mammalian cells. In human cells, ablation of the DNA methyltransferase DNMT1 leads to a partial reduction of DNA methylation, predominantly at repetitive sequences. This is reported to result in a depletion of di- (H3K9me2) and trimethylation (H3K9me3) at H3K9 and a concomitant increase in H3K9 acetylation (H3K9ac; ). Indirect reduction of DNA methylation at pericentromeric heterochromatin by loss of the chromatin remodeling protein Lsh also results in increased levels of histone acetylation ().
and
embryonic stem (ES) cells (
/
ES cells), which have a 50% reduction in levels of DNA methylation (). Because of the remaining levels of DNA methylation in all of the aforementioned studies, it has not been clear what the absolute relationship between DNA methylation and histone modifications is in mammalian cells.
There has been recent interest in links between linker histones and DNA methylation. Reduction of linker histone levels in vivo can give rise to altered DNA methylation at specific genomic sites (). On the other hand, there is conflicting evidence about whether DNA methylation influences linker histone binding in chromatin. Unmethylated CpG islands appear to be depleted of H1 (), and H1-containing nucleosomes contain 80% of the 5′-methylcytosine (), suggesting that linker histones may prefer to bind to methylated DNA. This is confirmed by some in vitro binding experiments (; ). However, other studies suggest that DNA methylation does not affect H1 binding to nucleosomes (; ; ).
Experimentally or genetically induced alterations in levels of DNA methylation have implicated this epigenetic modification in mammalian higher order chromatin condensation and nuclear organization, especially at sites of constitutive heterochromatin that have the highest concentration of DNA methylation (; ; ; ; ). Chromocenter clustering during differentiation has been recently attributed to increasing levels of DNA methylation and methyl-CpG–binding proteins ().
; ). Chromatin structure in these cells was analyzed by biochemical, biophysical, and cytological assays and compared with that from wild-type (WT) cells. The absence of DNA methylation altered chromatin structure at the level of the nucleosome and at the level of nuclear organization. There is a genome-wide decrease in H3K9me2, an increase in histone acetylation, and an increased clustering of chromocenters in mouse ES cells that are devoid of DNA methylation. However, contrary to expectations, micrococcal nuclease (Mnase) digestion and sucrose gradient sedimentation analyses indicate that the compaction of the chromatin in general and of heterochromatin in particular is not affected. Instead, there is a surprising decrease in the mobility of linker histones in the absence of DNA methylation. These studies highlight the complex interplay between DNA methylation and chromatin structure and the need to assess the effects of epigenetic modifications at multiple levels of chromatin and nuclear organization.
cells show a partial loss of CpG methylation (), but, after prolonged passage in culture, virtually no (0.6%) CpG methylation remains (; ). This probably reflects the failure of Dnmt1 to efficiently maintain methylation (). Southern blotting shows that methylation is similarly lost at euchromatic (B2 repeat) and heterochromatic (minor and major satellite) parts of the genome ().
ES cells () and
cells transfected with a transgene (
; ) show near normal levels of DNA methylation.
In the mouse nucleus, pericentric heterochromatin comprising of major satellite repeats tends to cluster into chromocenters. Recently, it has been suggested that increasing levels of DNA methylation contribute to progressive chromocenter clustering during differentiation and that this is mediated through methyl-CpG–binding proteins ().
cells compared with WT. Using 3D FISH with a probe for major satellite, we analyzed the number of individual chromocenters visible in the nuclei of mutant ES cells and in their parental equivalents, J1 cells ().
ES cells (median = 12) was significantly less than in J1 WT cells (median = 19; P = 0.0000 in Mann-Whitney U analysis; = 85; taken from three independent experiments; ).
cells compared with . This was also confirmed using an independent WT ES cell line, CGR8. The median number of chromocenters in these cells is not significantly different (P = 0.9) from that in the J1 cells and is significantly (P = 0.000) larger than the chromocenter number in
cells.
These data indicate that the loss of DNA methylation leads to increased clustering of pericentric heterochromatin into a few large chromocenters in ES cells.
cells increased the number of chromocenters (median = 16; ).
cells is not simply caused by the subsequent loss of methyl-CpG–binding proteins because the number of chromocenters in
cells (median = 16.0) is significantly larger (P = 0.0000) than in
cells ().
To investigate whether the altered nuclear organization of heterochromatin in DNA methylation–deficient ES cells is caused by a change in underlying secondary chromatin structures, we analyzed the Mnase sensitivity of chromatin from and mutant cells. All cell lines showed identical digestion kinetics of bulk chromatin (). Major satellite has a less accessible chromatin structure (i.e., is digested more slowly;
= 11 min) than bulk chromatin or chromatin at minor satellite (
= 7 min), which is independent of DNA methylation (). Nucleosome repeat length of bulk chromatin and major and minor satellites was also identical between cell lines. As previously shown, dinucleosomes at major satellite are refractory to Mnase digestion compared with those at minor satellite (; ), but this is also unaffected by the absence of DNA methylation (unpublished data).
A 16-kb region of silent repetitive chromatin adjacent to the chicken β-globin locus, which is resistant to Mnase digestion (), has been shown to sediment through sucrose with a frictional coefficient consistent with a rodlike shape of approximately the dimensions expected of a compact 30-nm chromatin fiber (). Similarly, and consistent with the aforementioned Mnase sensitivity, the rate of sedimentation of mouse and human satellite DNAs in sucrose gradients suggests that they are packaged into 30-nm chromatin fibers that are more compact in shape than those from the bulk genome (; ).
cells have identical sedimentation rates ().
ES cells still sediment more rapidly than equivalently sized bulk chromatin fragments from the same cells (), indicating that these satellite regions remain more compact than bulk chromatin even in the complete absence of DNA methylation.
The surprising absence of any effect of DNA methylation on secondary chromatin structure suggests that the main influence might be on primary chromatin (nucleosome) structure.
cells, levels of H3K9me2 reduced concomitantly (). Adding back Dnmt3b (
restored H3K9me2 to WT levels.
cells.
cells was paralleled by a progressive increase in the levels of H3K9ac. Acetylation levels at H4K5 and H4K16 are also elevated ().
cells, levels of H3K9ac are not rescued in
cells () or in
cells rescued with Dnmt3a (not depicted). In human cells lacking Dnmt1, H3K9ac levels did return to those of WT when Dnmt1 was restored (). However, in that case, the effects of DNA methylation and, therefore, presumably histone modification were concentrated in repetitive sequences.
ES cells, the increased H3K9ac levels are refractory to rescue by the reintroduction of Dnmt3a or b and that this may reflect a problem in retargeting some histone deacetylases.
cells are similar to WT, this indicates that histone acetylation does not affect secondary chromatin fiber structure or the nuclear organization of heterochromatin in these assays.
No decrease in global levels of H3K9me3 was detected in mutant cells by Western blotting. However, histone methylation marks are differentially distributed among tandem and interspersed repeats in the mouse genome ().
cells at minor and major satellite repeats and at the interspersed B2 repeat. Levels of H3K9me2 were almost undetectable at minor satellite even in , but, at the major satellite, there was a large (63%) reduction in H3K9me2 and an increase in H3K9ac ().
cells ().
Major satellite, minor satellite, and B2 comprise 3%, 0.45%, and 2.39% of the mouse genome, respectively (; ). Normalizing our ChIP data for repeat abundance showed that the concentration (per kilobase) of H3K9me2 is higher at major satellite than at minor satellite or B2 (). In contrast, the concentration of H3K9me3 is higher at minor satellite, although its total abundance is less. Loss of DNA methylation redistributes H3K9me3 so that its concentration at major and minor satellites is similar.
The binding of linker histones to nucleosomes in order to form the chromatosome is a fundamental aspect of dynamic chromatin structure. The precise sites of linker histone binding to the nucleosome remain in dispute, but there is evidence, mainly from in vitro analyses, arguing both for (; ) and against (; ; ) a role of DNA methylation in linker histone binding. To investigate this in vivo, we used FRAP to analyze the mobility of linker histones H1 (subtype H1.4) and H5, which were tagged with GFP at their N termini, in WT and
ES cells. We chose this orientation to minimize any interference of GFP with the high-affinity C-terminal chromatin-binding domain of H1 (). We were able to select for somatic cells stably transfected with these constructs but not undifferentiated ES cell lines. However, in transient transfections of ES cells, viable expressing cells were still visible 72 h after transfection, the tagged linker histones localized correctly in the nucleus, and Western blotting with GFP and H1 antibodies confirms that the levels of GFP–linker histone were low compared with endogenous H1 (). Fractionation of Mnase-digested chromatin from these cells on a sucrose step gradient showed that the GFP fluorescence cosediments with chromatin (). To confirm that GFP–linker histone incorporates into nucleosomes, soluble polynucleosomes, which were released from GFP- or GFP-H5–transfected COS cells with Mnase, were fractionated on a nucleoprotein gel, and GFP fluorescence was analyzed by scanning (). We conclude that GFP–linker histones are correctly incorporated into nucleosomal chromatin in transiently transfected ES cells but that expression of the exogenous linker histone is incompatible with the long-term propagation of undifferentiated ES cells.
Linker histone mobility has been shown to be influenced by phosphorylation ().
cells were examined on an acid-urea gel that is able to discriminate proteins based on their phosphorylation state.
interphase cells (). This was confirmed by Western blotting of the cells with antibodies that detect phosphorylated or hyperphosphorylated H1 ().
Linker histone mobility was then analyzed in ES cells 24 h after transfection by following the fluorescence recovery every 7 s (for a total of 340 s) after photobleaching (). The recovery kinetics of H1 in ES cells (
= 50 s) is consistent with previous analyses in differentiated cells (; ).
for H5 was longer (70 s), which is consistent with the higher affinity of the arginine-rich H5 linker histone for the chromatin fiber ().
cells (), with
values increasing to 70 s for H1 and 100 s for H5 ().
for both H1 and H5 were considerably shorter (25–28 s) in
cells than in ().
cells is caused by the lack of DNA methylation and not by the elevated levels of histone acetylation because the latter persists in
cells ().
ES cells was indistinguishable from ( and ).
cells is likely caused by the lack of DNA methylation itself and not by the subsequent loss of this methyl-CpG–binding protein.
Linker histones retard Mnase digestion of the core particle, pausing digestion at 178 nucleotides before protection of the 168-bp chromatosome, which comprises the nucleosome core particle together with the DNA that is protected from digestion by linker histone binding to the dyad ().
cells is caused by its altered binding to the nucleosome core particle, the rate of Mnase trimming of nucleosomes might change. However, the chromatosome was trimmed at the same rate in WT,
, and other mutant cells ().
cells, its residence time at the nucleosome dyad of the chromatosome measured in vitro was not altered.
Using mutant mouse ES cells, we show that the absence of DNA methylation leads to altered nuclear organization with an increase in the level of chromocenter clustering (). This is reminiscent of the increased associations between juxtacentromeric heterochromatin that are seen in the nuclei of individuals with immunodeficiency centromeric instability facial anomalies (ICF) syndrome, which is caused by the mutation of DNMT3b and the consequent hypomethylation of satellite sequences ().
cells is in contrast to the suggested role of DNA methylation and methyl-CpG–binding proteins in promoting chromocenter clustering during the terminal differentiation of mouse cells () and might indicate differences between cell types.
cells suggests that there is some underlying changes in chromatin structure when DNA methylation is absent. Indeed, we have identified such changes at two levels of primary chromatin structure: histone modifications and linker histone binding.
ES cells lacking DNA methylation have globally elevated levels of acetylation at H3K9, H4K5, and H4K16 (). Increased acetylation of H4K5 has been previously reported in these cells ().
cells by both Western blotting and ChIP and a redistribution of histone methylation within satellite-containing heterochromatin compartments ().
ES cells ().
cells had substantial residual DNA methylation. Both Dnmts and methyl-CpG–binding proteins can physically associate with histone deacetylases (; ) and histone methyltransferases (,; ; ), so the altered histone modifications in cells that completely lack CpG methylation may reflect the loss of these interactions.
It has been suggested that histone modifications can directly affect secondary chromatin structures by, for example, altering nucleosome–DNA or nucleosome–nucleosome interactions and by neutralizing charge in the histone N-terminal tails (; ; ). However, neither Mnase digestion kinetics nor sucrose gradient sedimentation () revealed any evidence for a general decompaction of secondary chromatin structure in the absence of DNA methylation. Satellite sequences remain in a more compact structure than the bulk genome using these assays regardless of the state of DNA methylation.
Most unexpectedly, the loss of DNA methylation leads to the altered binding of linker histones.
cells compared with ( and ).
cells ().
cells, in which DNA methylation levels are restored but H3K9ac levels remain elevated compared with WT,
times are dramatically shortened (). This is consistent with the increased mobility of linker histones found in trichostatin A–treated cells (). The recovery kinetics that we found for H1 in WT and mutant ES cells are similar to those previously reported in differentiated somatic cells (; ). We do not detect the fast recovery kinetics that have recently been reported for H1 in ES cells (). This might be the result of differences in the linker subtypes or ES cells used or of differential behavior between C- and N-terminally tagged H1 ().
cells (), we suggest that the loss of DNA methylation does not alter the binding of linker histones to the dyad of the nucleosome but that it is binding to another site (perhaps linker DNA) that is enhanced (; ). This is consistent with the two-site binding model proposed by . The recent identification of a link between H1 depletion and altered DNA methylation at specific genomic regions () suggests that a better understanding of the mechanistic and regulatory interactions between these two key chromatin modulators is required.
Mouse WT J1, −,
(S/S allele) (gift from E.
(gift from A. Bird, Wellcome Trust Centre for Cell Biology, University of Edinburgh, Edinburgh, UK; ) ES cells were cultured under standard conditions in the presence of leukemia inhibitory factor. Cells were transfected using LipofectAMINE 2000 (Invitrogen).
Nearest neighbor analysis was performed as described previously (). Genomic DNAs were digested with methylation-sensitive isoschisomer pairs (HpaII–MspI) or a methylation-sensitive enzyme (HpyCH4IV) and fractionated on a 0.7% agarose gel in Tris–phosphate buffer supplemented with ethidium bromide (EtBr). Gels were Southern blotted onto Hybond N (GE Healthcare) in 20× SSC and were probed for minor and major satellites and the B2 repeat ().
ES cells were cultured on gelatine-coated microscope slides for 4 h before fixation in 4% PFA for 10 min and processed for 3D FISH as previously described (). Major satellite was detected by hybridization with digoxigenin-labeled pSAT (). After washing and detection, slides were counterstained with 50 ng/ml DAPI and mounted in Vectashield. Slides were examined with an epifluorescence microscope (Axioskop; Carl Zeiss MicroImaging, Inc.) equipped with a 100× NA 1.3 lens and a CCD camera (Micromax; Princeton Instruments). A Pifoc piezo-driven objective focusing device was used to capture images at 0.25-μm intervals through the z axis. Images were captured and analyzed using custom IPlab (BD Biosciences) scripts. The significance of differences between cell lines was assessed using a Mann-Whitney U nonparametric test.
Nuclei were prepared as previously described () but with a reduced concentration (0.05%) of NP-40 in nuclei buffer B. For Mnase sensitivity digests, the nuclei concentration was adjusted to 4 A260 in nuclei buffer R. 20 or 50 U/ml Mnase (Worthington) was added, and aliquots were removed into stop buffer (2% SDS, 200 μg/ml proteinase K, amd 10 mM EDTA) at various time intervals. Purified DNAs were fractionated on a 1% agarose gel in Tris-borate buffer in the presence of EtBr. To distinguish nucleosomal (146 bp) and chromatosome-protected (166 bp) DNA after MNase digestion, the nuclei concentration was adjusted to 4 A260 in nuclei buffer R and were digested with 180 or 360 U/ml Mnase. Purified DNAs were fractionated on a 4% (1% regular agarose and 3% 3:1 NuSieve agarose [Flowgen]) gel in Tris-borate buffer in the presence of EtBr. Gels were scanned on a phosphorimager (FLA5100; Fuji) equipped with a 532-nm laser and 575-nm bandpass filter and were transferred to Hybond N by Southern blotting if required. Soluble chromatin was prepared and fractionated on sucrose gradients as described previously (; ). Gel and blot images were analyzed using the Aida software package version 3.52 (Raytek).
Linker histones were isolated by extracting whole cell lysates with 5% perchloric acid and subsequently by precipitation with acetone. The linker histones were analyzed on either a 15% (80:1 acrylamide/bis-acrylamide) acid-urea gel () or a 18% (200:1 acrylamide/bis-acrylamide) SDS-PAGE gel ().
ChIP was performed as described previously () with antibodies recognizing the aforementioned H3K9me2, H3K9me3, and H3K9ac. Immunoprecipitated chromatin was dot blotted onto Hybond N+ (GE Healthcare). Membranes were probed with minor and major satellite repeats and the interspersed B2 repeat, and the blots were analyzed on a phosphorimager (FLA5100; Fuji).
Chicken linker histone H5 cDNA () was subcloned from pBR-H5 using a blunted NotI–XmnI fragment into the SmaI site of pUC18 to give pUCH5. The coding sequence was PCR amplified with primers 5′-GAAGATCTTCCGGAATGACGGAGAGCCTGGTC-3′ and 5′-GGCCGCTCGAGTTACTTCAGCTCACTTCTTCTTGGGCGATTT-3′. Human histone H1.4 (a gift from D. Doenecke, German Cancer Research Center, Heidelberg, Germany) genomic clone (GenBank/EMBL/DDBJ accession no. ) was PCR amplified using primers 5′-GAAGATCTTCCGGAATGTCCGAGACTGCGCCT-3′ and 5′-GGCCGCTCGAGTTACTTCAGCCTACTTTTTCTTGGCTGCCGC-3′. Both PCR fragments were digested with BglII–XhoI and cloned into a BglII–SmaI site of pDsRed1-C1 (CLONTECH Laboratories, Inc.). The linker histone–containing portion was removed using BspEI–BamHI and was cloned into EGFP-C1 at BspEI–BamHI. Constructs were checked by sequencing, by transient transfection, and by Western blotting for GFP.
To examine the nuclear distribution of the GFP fusion proteins, transfected cells were fixed in 4% PFA and visualized by fluorescence microscopy. To investigate the association of GFP–linker histone to chromatin, 3-kb nucleosomal fragments from transiently transfected cells were generated by Mnase digestion and fractionated on a 10/50% sucrose step gradient at 48 K for 105 min in a rotor (SW55; Beckman Coulter). Fractions were collected by upward displacement with continuous monitoring of the chromatin at 254 nm (). GFP fluorescence was measured in each fraction on a fluorometer (Envision; PerkinElmer) at 510 nm. Proteins were precipitated from the gradient fractions and analyzed by SDS-PAGE to confirm that the GFP–linker histone fusion protein was present. Binding of GFP-H5 to individual nucleosomes was investigated by isolating chromatin from COS7 cells transfected with GFP-H5. Soluble chromatin was dialysed overnight against TEP80 and were further digested with Mnase to prepare short oligonucleosomes. The chromatin fragments were fractionated on a 5% polyacrylamide gel in Tris-borate buffer at 4°C. The gel was analyzed for GFP fluorescence using a scanner (FLA2000; Fuji) equipped with a 479-nm laser and 520-nm bandpass filter. The EtBr-stained gel was scanned using a 479-nm laser and 580-nm bandpass filter.
For FRAP, a 3-μm-diameter region of interest (ROI) of the nucleus in the midfocal plane was bleached with 10–15 iterations at 100% power with an argon laser at 6.1 mA. The pinhole size for the confocal was set at 1 Airy U. The time series software option was used to specify the appropriate time delay between rounds of 3D image stack capture. Images were captured with a 100× objective at 7-s intervals for a total of 340 s using 8% of laser power. Each image was processed by an interactive script (IPLAB version 3.6; Scanalytics) to correct for nuclear rotation and cell movement. Loss of fluorescence attributed to the imaging process alone was assessed from the sum of pixel intensities in the cell. The fluorescence intensity for each ROI over time was then normalized to this (). |
Tripin is a mammalian protein of unknown function that was reported to contain a domain that is conserved amongst Shugoshin (Sgo) family members (). Sgo1 is a family of evolutionarily conserved proteins that was first identified in yeast (; ; ; ) and (MEI-S332; ; ) as mutants that failed to maintain centromeric cohesion during meiosis I. Although Sgo1/MEI-S332 are essential for maintaining centromere cohesion during meiosis I in yeast and flies, they are not essential for mitotic chromosome segregation in both species. Sgo2 is a paralogue of Sgo1 fission yeast, and studies in have shown that it acts both in meiosis and mitosis (; ). During meiosis I, Sgo2 specifies monopolar attachments of paired chromatids, as opposed to a role in centromere cohesion (; ). Recent studies in fission yeast showed that Sgo2 is important for bipolar attachments of chromosomes in mitosis, and it specifies the centromeric localization of the chromosome passenger proteins Bir1/survivin, Pic1/INCENP, and Ark1/Aurora B kinase (; ). As Aurora B kinase is a critical component of the error correction machinery at kinetochores that monitors defective attachments (; ; ), its loss from centromeres in Sgo2 mutants explains the defects in establishing stable bipolar attachments.
Comparison of the primary sequences of Sgo1 and Sgo2 amongst different species of fungi revealed a common coiled-coil domain near their N termini and a conserved basic region of ∼30 residues near their C termini (; ). Identification of mutations within the conserved elements in MEI-S322 () established that it was related to Sgo1 both in structure and in function. Although vertebrate proteins with these conserved elements were also identified, the first vertebrate Sgo1 was identified in a biochemical screen for microtubule-binding proteins in egg extracts (). Consistent with the microtubule-binding activity, both the frog and human Sgo1 were found to be essential for establishing kinetochore–microtubule attachments (; ; ). These proteins were also essential for maintaining chromatid cohesion during mitosis, as cells depleted of Sgo1 were delayed in mitosis with unattached kinetochores and separated chromatids (; ; ). The vertebrate Sgo1 was concentrated near the outer kinetochore, which is where one would expect to find a microtubule-binding protein (). Others reported that hSgo1 was concentrated at the inner centromere (; ), which is where one would expect to find a protein that is responsible for centromeric cohesion.
The annotation of Tripin as Shugoshin 2 (Sgo2) in the database was based on the presence of the conserved C-terminal basic region that is shared amongst Sgo family members. Both the mouse and human proteins lack the conserved coiled-coil domain that is present within the N terminus of Sgo1. Additionally, Tripin is significantly larger than Sgo1 (∼1,200 vs. 480 residues) and yeast Sgo2 (647 residues). A recent study linking hSgo1 to PP2A phosphatase also reported that Tripin/hSgo2 is localized at the inner centromere, where it is important for centromere cohesion (). Mechanistically, hSgo2 was proposed to maintain cohesion in a manner that is similar to that of hSgo1. Namely, hSgo1 and hSgo2 recruit PP2A to the centromere, where they can neutralize Plk1's ability to phosphorylate and release cohesin complexes (; ; ; ; ).
Our studies show that hSgo2 is, indeed, a component of the inner centromere and that it exhibits a dynamic localization pattern where it is concentrated in between sister kinetochores during prometaphase, but extends toward the kinetochore by metaphase. We show that hSgo2 is released from the inner centromere shortly after the onset of anaphase and does not reappear there until late G2/prophase. This pattern is similar to that reported for the localization of Sgo2 during meiosis II in mouse spermatocytes ().
Functionally, cells quantitatively depleted of hSgo2 exhibited kinetochore attachment defects that transiently delayed cells at metaphase. When the cells entered anaphase, they invariably contained lagging chromosomes, which suggested that the defective attachments were never corrected. We ascribe the attachment defects to the delocalization of the microtubule depolymerase MCAK. Although we have no evidence to indicate that hSgo2 is essential for centromere cohesion, we confirmed that hSgo2 is, indeed, associated with PP2A (). We speculate that this subpopulation of PP2A may regulate the targeting or activity of MCAK at the inner centromere and kinetochore.
Antibodies to Tripin/hSgo2 were generated to characterize its localization and expression patterns in HeLa cells. Consistent with its predicted size of 1,265 aa, the antibodies identified an ∼150-kD protein in HeLa lysates that is depleted by Tripin/hSgo2 siRNA (Fig. S1, A and B, available at ). Costaining with hBUB1 antibodies showed that hSgo2 is concentrated at the inner centromere (). hSgo2 is diffusely distributed in the nucleus during interphase (unpublished data). In cells whose chromosomes have begun to condense (mid to late G2), hSgo2 accumulated at foci that were coincident with hBUB1. At a later stage of G2, when the nascent kinetochore pairs have resolved, pairs of hSgo2 foci that were positioned internal to hBUB1 were clearly evident. After nuclear envelope breakdown, hSgo2 staining appeared as a single focus that was positioned in between the sister kinetochores as defined by hBUB1 staining. At metaphase, hSgo2 was distributed across the width of the centromere and extended toward, and sometimes overlapped, the kinetochores. In early anaphase cells, hSgo2 and hBUB1 remain colocalized, but kinetochores exhibiting only hBUB1 staining in the same cells were also observed. Thus, the release of hSgo2 from kinetochores does not appear to be regulated solely by mitotic timing. By late anaphase, neither hSgo2 nor hBUB1 were detected at kinetochores. The dynamic properties of hSgo2 were confirmed in real time by tracking cells expressing GFP/hSgo2 (unpublished data).
We monitored the stability of hSgo2 as a function of mitotic exit by probing lysates prepared from cells that were released from a nocodazole block (). Between 60 and 90 min after release, the majority of cells entered anaphase, as cyclin B levels abruptly declined. hSgo2, along with CENP-E, also began to decline at this time, although their kinetics of degradation appeared to lag behind cyclin B. Loss of hSgo2 (as with cyclin B and CENP-E) was dependent on the proteosome as it was stabilized when cells were treated with a proteosome inhibitor.
The localization pattern of hSgo2 is reminiscent of MCAK, which is a microtubule depolymerase (; ) that is concentrated at the inner centromere, but is redistributed toward the kinetochores in response to microtubule attachments (). Indeed, hSgo2 localization was found to be coincident with MCAK (), as recently reported in mouse spermatocytes (). In late prophase, hSgo2 and MCAK were colocalized at a single focus in between kinetochore pairs that were stained with ACA (). In prometaphase, we found examples where hSgo2 and MCAK staining were skewed toward one of the sister kinetochores (). For MCAK, this was shown to reflect its redistribution toward the leading kinetochore of a congressing chromosome ().
The relocalization of hSgo2 from the inner centromere to the kinetochore at metaphase led us to test whether its localization pattern was sensitive to microtubule attachments or kinetochore tension. HeLa cells were treated with a dose of nocodazole that suppressed microtubule dynamics, and thus reduced kinetochore tension without affecting attachment. The reduction in the interkinetochore distance of bipolar attached kinetochores in the drug-treated samples relative to controls (2.2 vs. 1.3 μm, respectively) confirmed that nocodazole reduced kinetochore tension. Despite the reduction in tension, hSgo2 was still able to redistribute from a single dot, seen at unattached kinetochores (, inset 1), to a bar that stretched between the bipolar attached kinetochore (, insets 2 and 3). However, when tension was reduced, the peaks of hSgo2 staining did not overlap with hBUB1 to the same extent as seen in kinetochores that are under maximal tension (, insets 1 and 2). Thus, the extent of hSgo2's redistribution from the centromere to the kinetochore is sensitive to tension, as previously suggested ().
Next, we used FRAP to compare the turnover rates of hSgo2 at kinetochores of different microtubule-binding status. Kinetochores expressing GFP/hSgo2 were photobleached, and the rate of recovery of the GFP signal was monitored (Fig. S2, available at ).
during G2 phase, when hSgo2 is first recruited to the nascent kinetochore, is 5.25 ± 2.1 s.
increased to 10.17 ± 5.93 and 9.17 ± 3.39 s in prometaphase and metaphase, respectively. We then compared the turnover rates in mitotic cells that were exposed to vinblastine or taxol.
was 13.22 ± 4.89 s, as opposed to 8.87 ± 1.67 s kinetochores in taxol-treated cells. Regardless of the microtubule-binding status or phase of the cell cycle, all of the kinetochores examined ( = 38) were able to recover >94% of the prebleached level of GFP/hSgo2. Thus, the slightly faster turnover rate of hSgo2 at metaphase kinetochores may be affected by increased microtubule attachments. This explanation cannot account for the more rapid turnover rates in G2, which may be governed by determinants that specify kinetochore assembly.
Consistent with recent findings, we found that hSgo2 exhibits a dynamic localization pattern (). hSgo2 accumulates at an early stage of kinetochore assembly when hBUB1 is first detected at the nascent kinetochore. At this time, kinetochores have not resolved into discrete pairs, as hBUB1 appears as single spot instead of pairs of foci. Once kinetochore pairs were resolved, hSgo2 was concentrated at the inner centromere. hSgo2 localization was not always centered between kinetochores because it tended to skew toward the leading kinetochore, as was reported for MCAK. Once stable bipolar attachments are made, hSgo2 was found to span the width of the centromere and partially overlap with the kinetochore, as was shown for MCAK. Consistent with the recent finding that the localization of Sgo2 in mouse was sensitive to tension (), we found that the extent of the overlap between hSgo2 and kinetochores (based on hBUB1 colocalization) was reduced when tension was reduced by nocodazole treatment. Thus, some aspects of the dynamic localization pattern of hSgo2 within the centromere–kinetochore complex appear to be sensitive to tension. However, we cannot rule out the role of microtubule attachments in the redistribution of hSgo2 from the inner centromere toward the kinetochore.
Functionally, we established that hSgo2 is essential for recruiting MCAK to the inner centromere. Cells depleted of hSgo2 exhibited a quantitative depletion of MCAK from centromeres by >95%. Although we detected some hSgo2 in immunoprecipitates obtained with MCAK antibodies, hSgo2 immunoprecipitates did not contain MCAK. As neither gel-filtration analysis of HeLa lysates nor yeast two-hybrid assays indicated that MCAK associated with hSgo2, MCAK is unlikely to be recruited to the centromere in a stable complex with hSgo2. Our studies also showed that hSgo2 localization is dependent on Aurora B. As Aurora B has been shown to specify the recruitment of MCAK to the centromere (; ), the combined data suggests the following linear assembly pathway: Aurora B → hSgo2 → MCAK. The relationship may be more complex, given how Aurora B is thought to specify MCAK localization. MCAK has been shown to be a substrate of Aurora B kinase in vitro and in vivo (; ), but it is not entirely clear whether recruitment of MCAK depends on these phosphorylation sites. This is based on the finding that mutating all five of the MCAK phosphorylation sites to phosphodefective and phosphomimic mutants did not prevent their assembly to the centromere (). Instead, the distribution of MCAK between the inner centromere and the kinetochore seemed to be affected by its phosphorylations status (; ). This suggests that the role of Aurora B in recruiting MCAK to the inner centromere may differ from its role in regulating the dynamic distribution of MCAK within the centromere and kinetochore. Recruitment of MCAK may depend on other proteins, such as hSgo2, that are also targets of Aurora B. One role for the PP2A that is associated with hSgo2 might be to locally regulate the spatial distribution of MCAK within the centromere and kinetochore. Finally, we showed that hBub1 is also required by hSgo2 and MCAK to localize to centromeres. As hBUB1 and Aurora B do not depend on each other for their localization to kinetochores, the assembly of hSgo2 and MCAK appears to depend on two parallel pathways. The significance of this is unclear, but the use of multiple pathways appears to be a common feature used for kinetochore assembly ().
The importance of hSgo2 in recruiting MCAK to centromeres also provides a mechanistic explanation for the kinetochore attachment defects seen in cells depleted of hSgo2. Time-lapse studies of chromosome dynamics in cells depleted of hSgo2 showed a delay in congression to the spindle equator. Whereas virtually all chromosomes eventually achieved alignment, kinetochores with merotelic and syntelic attachments were identified. These defective attachments accumulated because no MCAK was present to sever them. This interpretation is supported by the fact that similar defects were observed when MCAK was directly depleted from cells. Failure to resolve these defective attachments in the hSgo2-depleted cells before anaphase onset explains the high incidence of lagging chromosomes once cells enter anaphase.
Given that the microtubule depolymerase activity of MCAK has been shown to be negatively regulated by phosphorylations mediated by Aurora B kinase (; ), there may be another role for the PP2A that is associated with hSgo2. PP2A may provide a way to locally control MCAK activity so that only defective microtubule attachments are severed, while productive attachments are preserved. In this scenario, PP2A associated with hSgo2 may dephosphorylate and activate MCAK depolymerase activity. This model implies that hSgo2/PP2A, MCAK, and Aurora B activities must be highly regulated so that they can spatially restrict their actions to just a single defective attachment.
Our interpretation of the PP2A–hSgo2 interaction differs from that proposed for how the PP2A–hSgo1 interaction maintains centromeric cohesion (; ; ). As with hSgo1, hSgo2 is thought to target PP2A to the inner centromere, where it can neutralize the phosphorylation of cohesin subunit Scc3 introduced by Polo kinase 1 (). This is supported by their finding that depletion of hSgo2 resulted in a high incidence (15-fold increase) of prematurely separated chromatids. However, we found that cells depleted of ∼95% of hSgo2 only exhibited a small increase in chromatid separation, which was also seen in cells depleted of MCAK. We believe that the loss of hSgo2 and MCAK from the inner centromere affects the higher-order organization of this region in a way that indirectly weakens centromeric cohesion. This is partially confirmed at the EM level, which showed that depletion of hSgo2 and Aurora B compromised the organization of the inner kinetochore and the subjacent chromatin. Given that Aurora B lies upstream of hSgo2, we would expect that its removal should lead to a dramatic increase in the frequency of separated chromatids if hSgo2 is, indeed, essential for cohesion. On the contrary, inhibition of Aurora B kinase has been reported to not affect chromatid cohesion (; ). At present, we cannot provide a satisfactory explanation for the discrepancy in the functional roles of hSgo2 presented by the two studies. The microtubule attachment defects we identified in cells depleted of hSgo2 is fully consistent with its role in recruiting MCAK. It is formally possible that we failed to see cohesion defects because depletion of ∼95% of hSgo2 was insufficient to manifest the phenotype. It is noteworthy that a recent study showed that Sgo2 in fission yeast is not essential for cohesion (; ). Instead, both studies showed that Sgo2 facilitated chromosome biorientation, most likely via its role in recruiting the Aurora kinase complex to the centromere. Although our results differ in respect to the fact that hSgo2 was not important for recruiting chromosome passenger complexes to the centromere, its role in kinetochore attachments is consistent with those reported for the fission yeast Sgo2.
hSgo2 was PCR amplified from a cDNA library (Marathon-ready cDNA; CLONTECH Laboratories, Inc.) and confirmed by sequence analysis. The full-length cDNA or fragments were cloned into pENTR (Gateway) to facilitate transfer into mammalian and bacterial expression vectors by in vitro recombination reactions. The cDNA encoding N-terminal 469 aa of hSgo2 was inserted into the bacterial expression vector pDEST17 (Gateway) and recombinant protein was purified by Ni-beads under denaturing condition. Purified protein was used to immunize animals and coupled to Affigel-10 (BioRad Laboratories). The affinity column was used to purify antibodies from rabbit and rat sera.
SMARTpool and single siRNAs targeting hSgo1 () and hSgo2 (siRNA 1, 2, 3, and 4 sense sequences were as follows: UCAAAGACAUUACCUGAUAUU, GAACACAUUUCUUCGCCUAUU, UCGGAAGUGUUAUUUCUUAUU, and GAGAAACGCCCAGUCUAUUUU) were obtained from Dharmacon. siRNAs were diluted in serum-free OptiMEM and HiPerfect (QIAGEN) as per the manufacturer's instructions and added to cells so that the final concentration of siRNA was 20 nM. 24–36 h after transfection, cells were fixed and stained or lysed in SDS sample buffer.
Cells were fixed for 7 min in freshly prepared 3.5% paraformaldehyde/PBS, pH 6.9, extracted in KB (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 0.1% BSA) plus 0.2% Triton X-100 for 4 min at room temperature, and then rinsed in KB. In some cases, cells were preextracted for 2 min before fixing. Primary and secondary antibodies were diluted in KB and added to coverslips for 30 min at 37°C in a humidified chamber. Antibodies to tubulin (Sigma-Aldrich), Aurora B (BD Biosciences), and survivin (Novus Biologicals) were obtained commercially. Human ACA, INCENP, and MCAK antibodies were gifts from J.B. Rattner (University of Alberta, Calgary, Canada), W. Earnshaw (Edinburgh University, Edinburgh, UK), and L. Wordeman (University of Washington, Seattle, WA), respectively. Antibodies to hBUB1, hBUBR1, hBUB3, and Mad1 were obtained from our laboratory (; ; ). Antibodies were used at a final concentration of 0.5–1.0 μg/ml. Secondary antibodies conjugated to Alexa Fluor 488, 555, and 647 (Invitrogen) were used at 1 μg/ml. Images were visualized with a 100×/1.4 NA objective on a microscope (Eclipse TE2000S; Nikon) and 0.5-μm image stacks were captured with a charge-coupled device camera (Roper Scientific). Images are presented as maximum projections and quantitated as previously described (). Deconvolution was conducted with AutoQuant (Media Cybernetics).
For time-lapse studies, HeLa (GFP/H2B) were plated onto glass-bottomed 35-mm dishes (MakTek) in Hepes-buffered, phenol red–free medium, transfected with siRNA, and imaged with an UltraView spinning disc confocal microscope. Images were captured every 3–5 min overnight at 37°C. For FRAP experiments, HeLa cells were transfected with Lipofectamine 2000 (Invitrogen). GFP-labeled kinetochores were imaged with a 63× objective on a multiphoton laser scanning microscope (NLO510; Carl Zeiss MicroImaging, Inc.) that is equipped with a heated stage and objective heater. FRAP was performed essentially as previously described (). LSM software (Carl Zeiss MicroImaging, Inc.) was used to measure integrated fluorescence intensities of kinetochores. The intensities are normalized as the percentage of recovery. The normalized data was fit to the nonlinear regression curve in Prism (Graftpad Software).
Chromosome spreads were prepared as previously described (). In brief, mitotic cells were removed by shake-off, pelleted, hypotonically swollen in 75 mM KCl at 37°C for 20 min. Cells were pelleted, fixed with methanol/glacial acetic acid (3:1) for 5–10 min, dropped onto clean glass slides, and allowed to air dry. Slides were rehydrated in an 80°C steam bath for several seconds, dried on a 70°C heatblock, and stained with DAPI.
For EM, HeLa cells transfected with siRNAs were fixed in 3% glutaraldehyde and 0.2% tannic acid in 200 mM Na cacodylate buffer for 1 h at room temperature. Postfixation was in 2% OsO for 20 min. The cells were dehydrated in ethanol, and then infiltrated with Polybed 812 resin (Polysciences). Polymerization was performed at 60°C for 24 h. Silver-gray sections were cut with an ultramicrotome (Leica) equipped with a Diamond knife, and sections were stained with uranyl acetate and lead citrate and examined in an electron microscope (H-7000; Hitachi).
Fig. S1 shows that the specificity of hSgo2 antibody and the efficiency of hSgo2 siRNAs depletion by Western blot and immunofluorescence. Fig. S2 shows the turnover rates of GFP/hSgo2 at kinetochores at different cell cycle phases as determined by FRAP. Fig. S3 shows that cells depleted of hSgo2 by siRNA exhibit anaphase bridges and delocalization of MCAK from the centromere. In contrast, depletion of Sgo1 does not affect the centromeric localization of MCAK. Fig. S4 shows the effects of hSgo2 depletion on the phosphorylation state of MCAK, and cells depleted of MCAK exhibit mitotic defects similar to depletion of hSgo2. Coimmunoprecipitation experiments reveal a weak interaction between hSgo2 and MCAK and a clear association with endogenous and transfected hSgo2 with PP2A-C. Fig. S5 shows that hSgo2 is neither essential for centromere cohesion nor centromeric localization of chromosomal passenger proteins. Online supplemental material is available at . |
Lafora disease (LD; OMIM #254780) is an autosomal recessive neurodegenerative disorder resulting in severe epilepsy and death (; ). It is one of five major progressive myoclonus epilepsies. LD presents itself as a single seizure in the second decade of the patient's life (); this single event is followed by progressive central nervous system degeneration, culminating in death within 10 yr of the first seizure (Van Heycop ). A hallmark of LD is the accumulation of polyglucosan inclusion bodies called Lafora bodies (LBs; ; ) that are located in the cytoplasm of cells in most organs (; ; ). LB accumulation coincides with increased neuronal nonapoptotic cell death and a number of seizures in LD patients. Thus, it is hypothesized that LBs are responsible for these symptoms and, ultimately, for the death of the patient ().
Recessive mutations in (epilepsy of progressive myoclonus type 2 gene B)/ (), which encodes the E3 ubiquitin ligase malin (; ), are responsible for ∼40% of LD cases (). Of the LD cases not attributed to mutations in , ∼48% result from recessive mutations in (epilepsy of progressive myoclonus type 2 gene A; ; ; ). encodes laforin, which contains a carbohydrate-binding module (CBM) family 20 (CBM20; ) domain followed by the canonical dual specificity phosphatase (DSP) active site motif HCXXGXXRS/T (CxR) (; ; ; ). The CBM of laforin binds complex carbohydrates in vivo and in vitro (), and the DSP motif can hydrolyze phosphotyrosine and phosphoserine/threonine substrates in vitro (; ). However, no group has detected endogenous laforin localization in tissue culture cells or in wild-type tissues likely as a result of low levels of accumulation (; our unpublished data).
Of the 128 human phosphatases (; ), only laforin possesses a CBM. CBM domains are predominantly found in glucosylhydrolases and glucotransferases of bacterial, fungal, or plant origin (; ; ). The vast majority of enzymes containing a CBM use the domain to bind a specific type of carbohydrate and then enzymatically act on the sugar (). Accordingly, we recently showed that laforin liberates phosphate from the complex carbohydrate amylopectin, whereas other phosphatases lack this activity ().
disrupted the locus in a mouse model. Although this model faithfully recapitulated the disease, it yielded no molecular explanation for LD. Similarly, generated a transgenic mouse overexpressing inactivated laforin, and this mouse model also developed LD. Despite the availability of these two LD mouse models, the molecular etiology of LD remained unexplained. These limitations demonstrate the need to develop alternative model systems to elucidate the biology of LD. Although a molecular mechanism to explain LD has remained elusive, data cumulatively place laforin in the context of being intimately, if not directly, involved in regulating glycogen metabolism. Therefore, we focused on this indisputable aspect of LD for clues to its molecular etiology.
One protist that accumulates floridean starch (also called amylopectin granules) in its cytoplasm is (; ; ; for review see ). is an obligate intracellular parasite that can infect nearly any nucleated cell from a warm-blooded animal. Like most members of Apicomplexa, has a complex life cycle: in its intermediate hosts, it exists as a rapidly dividing tachyzoite or an encysted bradyzoite, depending on the host immune response. The bradyzoite forms floridean starch in its cytoplasm that is used as an energy source (for review see ). Recent studies characterized the biochemical composition of floridean starch (; ). We noted that the biochemical composition of floridean starch was remarkably similar to that of LBs described nearly 40 yr ago (, ; ). Although has been reported to be present only in vertebrates (, ), the similarity between floridean starch and LBs led us to explore the partially completed genome for a laforin orthologue.
The sequence of the genome, like the genome of many protists, was not accessible via GenBank when we initiated this study. Therefore, we searched the database (ToxoDB; ) for a laforin orthologue. We used the criteria that a laforin orthologue must contain both an amino-terminal CBM and a carboxy-terminal DSP domain (). DSP domains are readily recognized by the protein families database (pfam; ) and the National Center for Biotechnology Information's (NCBI) conserved domain database (CDD; ). However, CBMs are very degenerate at the primary amino acid level, and neither database consistently recognizes any of the 45 CBM families. Because CDD and pfam do not reliably recognize CBMs, we devised a multitiered search strategy to identify laforin orthologues (). First, we performed BLASTp () searches using the DSP motif HCXXGXXR as an index sequence and identified 20 proteins containing this motif. Because laforin contains an amino-terminal CBM and CBMs contain 80–100 amino acids, we eliminated two of these proteins because their HCXXGXXR motif was within the first 80 amino acids. We next performed a secondary BLAST using the NCBI nonredundant (nr) database with each of the remaining 18 proteins minus their DSP domain. If the protein contained a CBM, the BLAST identified other CBM-containing proteins. Using this strategy, we identified one protein, which we refer to as laforin (Tg-laforin), that met the aforementioned criteria. Tg-laforin and laforin (Hs-laforin) are 37% identical (). Importantly, Tg-laforin contains all of the residues important for carbohydrate binding as well as the signature residues of a DSP ().
With the discovery of a putative laforin orthologue in , we extended our search methods to identify additional orthologues using a variety of genome databases (Table S1, available at ). Using this strategy, we identified laforin orthologues in the four classes of vertebrates with sequenced genomes (mammals, aves, amphibians, and osteichthyes; ). In addition, we identified putative laforin orthologues in four additional protists: , , , and (). Although Hs-laforin contains 331 amino acids, the putative protist orthologues varied in predicted size from 323 to 727 amino acids. However, each putative orthologue contained the signature amino acids of a CBM20 and DSP; that is, four invariant aromatic amino acids (Hs-laforin F5, W32, W60, and W99) as well as DXCXGXR, respectively (). Despite exhaustive efforts (we searched ∼170 eukaryotic genomes and ∼670 bacterial and archaeal genomes), we did not identify any other putative laforin orthologues. Thus, laforin is absent in all traditional nonvertebrate model organisms (e.g., yeast, fly, and worms). Laforin orthologues exist in all classes of vertebrates in which sequence data is available and in the five protists that we identified ().
laforin (Cm-laforin) shares the least identity with Hs-laforin (). As such, we reasoned that if it exhibited similar in vitro characteristics as Hs-laforin, the other putative orthologues were likely to as well. To test whether the identified protist proteins had similar biochemical characteristics as Hs-laforin and were thus laforin orthologues, we cloned the putative orthologue from (Cm-laforin) and purified recombinant protein from bacteria (Fig. S1 A, available at ). Characteristic of all DSPs, Hs-laforin exhibits phosphatase activity against the artificial substrate para-nitrophenylphosphate (p-NPP; ; ). Cm-laforin also used p-NPP as an artificial substrate with similar kinetics as Hs-laforin () and displayed a similar specific activity (). In addition to activity against p-NPP, we recently showed that recombinant Hs-laforin releases phosphate from amylopectin () and that this activity is unique to laforin (). Additionally, we fused the CBM of laforin to the DSP VH1 related (VHR) and demonstrated that although this fusion protein was an active phosphatase, it did not liberate phosphate from amylopectin (). shows that like Hs-laforin, Cm-laforin displays a robust ability to release phosphate from amylopectin, whereas VHR does not hydrolyze phosphate from amylopectin. As predicted, the catalytically inactive Cm-laforin–C/S mutant displayed no activity against either substrate (). Additionally, Tg-laforin also displayed activity against both p-NPP and amylopectin (unpublished data).
Hs-laforin is the only phosphatase in the human genome that contains a CBM and, as such, is predicted to be the only phosphatase that binds carbohydrates. Cm-laforin and Tg-laforin bound amylopectin to the same extent as Hs-laforin ( and not depicted). Conversely, VHR did not bind amylopectin (). previously demonstrated that conserved tryptophan and lysine residues () that participate in binding to the sugar are necessary for Hs-laforin to bind amylopectin (). Accordingly, mutation of these corresponding residues in Cm-laforin also abolished its ability to bind amylopectin (). These mutations also considerably reduced the ability of Cm-laforin to release phosphate from amylopectin (Fig. S2 A, available at ) while only minimally affecting its p-NPP activity (Fig. S2 B). These data suggest that Cm-laforin must be positioned correctly via the CBM in order for the DSP domain to dephosphorylate amylopectin or that the CBM binding to the carbohydrate is needed to activate the DSP.
While laforin from all three species binds α-glucans in vitro, this result may not reflect the biological localization of laforin. Moreover, the localization of Hs-laforin has never been determined in wild-type cells or tissues (; our unpublished data). Because we identified multiple new systems to study laforin, we investigated laforin's localization in . A cell contains a chloroplast, mitochondrion, and nucleus and, when grown in continuous light, accumulates vast storages of floridean starch (, schematic; ). We fixed cells and probed them with an affinity-purified α–Cm-laforin antibody. We found that endogenous Cm-laforin localized in punctate accumulations throughout the cytoplasm of cells (). To further define the localization of Cm-laforin, we performed immunogold electron microscopy staining. Ultra-thin sections of cells were probed with the affinity-purified α–Cm-laforin antibody and a 10-nm gold particle–conjugated goat α–rabbit secondary antibody. Positive staining was observed surrounding the floridean starch granules (, arrowheads). No Cm-laforin was observed within the granules because before sectioning, no protein would have access to this region. In addition, no background staining was observed with the secondary antibody alone (Fig. S3, available at ). Thus, as we hypothesized, endogenous laforin binds the outer region of insoluble carbohydrates.
The key to the evolutionary lineage of laforin lies in the origin of the aforementioned five protists. The chromalveolate hypothesis postulates () that a distinct sequence of events led to the evolution of the kingdom Plantae and to subsequent progeny, including the five aforementioned protists. As illustrated in , a mitochondriate protist engulfed a cyanobacterium (; ) and eventually gave rise to the kingdom Plantae (). Once Plantae was established, a second endosymbiosis involving red algae () gave rise to the chromalveolates (; ). These engulfments were accompanied with the coevolution of “various manifestations of mitochondria” () and various forms of carbohydrate storage (). These combined evolutionary events resulted in organisms possessing a mitosome, a hydrogenosome, or a true mitochondrion, and some organisms evolved floridean starch as their storage carbohydrate. We hypothesized that interspersed within these evolutionary events, organisms maintained, gained, or lost laforin.
To trace the lineage of laforin, we generated a phylogeny derived from the small subunit ribosomal RNA gene of organisms belonging to diverse evolutionary niches and highlighted the organisms whose genome contains laforin (). This phylogenetic analysis revealed that each of the five protists containing a laforin orthologue is of red algal descent. However, the genome of some organisms of red algal descent lack laforin (). To determine why some organisms of red algal descent lack laforin, we analyzed the biology of each of the organisms in . We discovered that each organism of red algal descent that contained laforin also contained a true mitochondrion and produced floridean starch. Conversely, organisms of red algal descent lacking laforin either lacked a true mitochondrion or did not produce floridean starch. For example, is of red algal descent and possesses mitochondria; however, it does not produce floridean starch and, thus, lacks laforin (). Similarly, is of red algal descent and produces floridean starch, but it has mitosomes instead of mitochondria and, thus, lacks laforin (). Conversely, is a red alga that produces floridean starch and contains a single mitochondrion and, in agreement with our established criteria, contains laforin. Additionally, glaucophytes and green algae/land plants lack a laforin orthologue because they evolved as contemporaries of red algae and not as descendents (). Thus, our analyses generated three criteria to predict whether a protist's genome possesses laforin: the organism must be of red algal descent, possess a true mitochondrion, and produce floridean starch. To determine whether our criteria correctly predicted the presence of laforin, we investigated the biology of each organism from the 168 eukaryotic genomes we probed. We found that in each case, our criteria correctly predicted the presence or absence of laforin (Table S5, available at ).
Protists such as use insoluble floridean starch as an energy source when transitioning from inactive/hibernating life cycle stages to active/replicative stages (for review see ). Likewise, , a red alga that contains laforin, synthesizes insoluble floridean starch during the day and uses it as a source of energy at night. Plants have a similar diurnal cycle, producing insoluble carbohydrate in the form of starch during the day and catabolizing it during the night. Because Hs-laforin has been implicated in carbohydrate metabolism and we show that Cm-laforin binds and releases phosphate from amylopectin, we hypothesized that laforin plays a vital role in insoluble carbohydrate metabolism. Thus, we predicted that plants would also have a laforin-like activity; however, we were unable to identify a laforin orthologue in plants. Recently, several starch excess mutants that accumulate starch have been described in plants (; ; ); one of these is attributed to mutations in the () gene (At3g52180; ). and demonstrated that the gene (previously identified as a phosphatase and called ; ) encodes a protein containing a chloroplast-targeting peptide (cTP) and DSP domain at its amino terminus followed by a CBM-like domain at its carboxy terminus (), suggesting that SEX4 might be a laforin-like phosphatase ().
The DSP of SEX4 shares the key DXCXGXR catalytic residues with the DSP of Hs-laforin and is 24% identical to Hs-laforin (). Conversely, the CBM of SEX4 lacks many of the invariant CBM20 residues ( vs. ) and shares only 18% identity with the CBM of Hs-laforin (). Instead, a sequence search using the CBM of SEX4 shows that it is most similar to another class of CBM, the AMP-activated protein kinase β–glycogen-binding domain (AMPKβ−GBD) family (), and not to CBM20 (). Despite their structural differences, both CBM20 and the AMPKβ-GBD domains interact with individual glycan chains of carbohydrates (; ), suggesting that SEX4 could bind starch via its AMPKβ-GBD. Thus, SEX4 contains similar domains to laforin, but the domains are arranged in the opposite orientation ( vs. ). We next performed BLASTp searches of various databases (Table S1) and found that SEX4 is conserved in all land plants and in , a single-cell green alga closely related to the progenitor of land plants (). Thus, SEX4 likely evolved before or during the establishment of green algae and performs a kingdom-wide function in Plantae.
To ascertain whether SEX4 possesses biochemical properties similar to laforin, we cloned and assayed purified recombinant SEX4 protein (At-SEX4; Fig. S1 B). Because the cTP of SEX4 is highly hydrophobic and renders the protein insoluble, we deleted the first 52 amino acids and used purified recombinant HIS-tagged Δ52-SEX4 for our assays (Fig. S1 B). We found that Δ52-SEX4 has a similar specific activity and possesses similar kinetics as Hs-laforin against p-NPP ( and ) and efficiently liberates phosphate from amylopectin (). Conversely, mutation of the active site cysteine to serine abolished these activities (). Additionally, wild-type (Δ52-SEX4) and catalytically inactive SEX4 (Δ52-SEX4-C198S) bind amylopectin similar to Hs-laforin (). Importantly, mutations in key conserved AMPKβ-GBD residues that form essential hydrogen bonds with the sugar (, ) abolish this interaction () while minimally affecting the phosphatase activity of SEX4 (Fig. S4 A, available at ). These mutations considerably reduced the ability of SEX4 to release phosphate from amylopectin (Fig. S4 B). Thus, like Cm-laforin, SEX4 must also be positioned correctly via the CBM in order for the DSP domain to dephosphorylate amylopectin.
The locus was recently mapped in to At3g52180, and multiple mutations in this gene display a starch excess phenotype (; ). One characterized mutation is the allele that contains an transferred DNA (T-DNA) insertion in the sixth exon () and leads to the disruption of expression (). Because laforin and SEX4 are the only reported proteins in any kingdom that contain both functional CBM and DSP domains and because mutations in the gene expressing either protein results in aberrant carbohydrate accumulation, we postulated that SEX4 and laforin could be functional equivalents.
To test this hypothesis, we transformed plants to generate stable lines expressing SEX4, sex4-C/S, Hs-laforin, and Hs-laforin fused behind a cTP (cTP–Hs-laforin) to target Hs-laforin to the chloroplast (like SEX4) and monitored protein expression of the transgenes (). We then assayed starch accumulation in wild-type, , and transgenic plants. As per our prediction, transformants expressing SEX4 and cTP–Hs-laforin no longer displayed the starch excess phenotype, whereas the catalytically inactive sex4-C/S mutant and Hs-laforin transformants still accumulated excess starch (; and Fig. S5, available at ). Thus, the cTP–Hs-laforin fusion rescued the starch excess phenotype both qualitatively and quantitatively. Conversely, Hs-laforin lacking the cTP did not rescue any portion of the phenotype. Therefore, Hs-laforin is a functional equivalent of SEX4 that must be targeted to the chloroplast, just like SEX4, to perform the equivalent function.
Our studies probe the molecular mechanism of LD. We identified laforin orthologues in specific protists and further showed that Hs-laforin and plant SEX4 are functional equivalents. Our results provide compelling evidence that a laforin-like activity is required to regulate the metabolism of amylopectin-like material across multiple kingdoms. Additionally, they demonstrate the nature of this activity; that is, the dephosphorylation of the carbohydrate itself, thus providing a molecular explanation for LD. Although there are examples of DSPs that dephosphorylate nonproteinacious substrates (such as phosphate and tensin homologue, the myotubularin family, and Sac domain phosphatases that dephosphorylate the inositol head group of phospholipids; ; ; ; ; ; ), ours is the first example of a family of phosphatases that dephosphorylate complex carbohydrates.
We demonstrate that laforin is not merely restricted to the genomes of vertebrates but is well conserved in the protists , , , , and . Laforin's evolutionary lineage shows that it originated in a primitive red alga during early eukaryotic evolution. Despite its early origin, laforin was only maintained by organisms that synthesize floridean starch (such as the aforementioned five protists) and organisms that inhibit the production of insoluble carbohydrates (i.e., all vertebrates). Organisms that no longer performed either of these processes lost laforin. Conversely to laforin, we show that although SEX4 contains similar domains as laforin, its lineage differs in that SEX4 is conserved in all land plants as well as in , a close descendent of primitive green algae. Despite their different lineages, Hs-laforin performs the same function as the plant protein SEX4; thus, we propose that laforin and SEX4 are functional equivalents.
It must be noted that although laforin and SEX4 share a common function and similar domains, they are not orthologous proteins. They are not orthologues because although they share similar CBMs, the CBMs belong to different classes and differ considerably with respect to the primary amino acids that are important for binding carbohydrates, and the DSP and CBM of laforin and SEX4 are arranged in opposite orientations. Thus, it is likely that red and green algae independently evolved a phosphatase via convergent evolution that utilizes a similar mechanism to regulate insoluble carbohydrate metabolism.
Despite the independent means by which laforin and SEX4 evolved, they both dephosphorylate the same carbohydrate substrate and constitute a unique family of phosphatases. In addition, we demonstrate that endogenous Cm-laforin localizes around the floridean starch granules. Although most studies thus far suggest a carbohydrate substrate for laforin and SEX4, it is possible that they bind their respective amylopectin-like material (insoluble glycogen and starch, respectively) and dephosphorylate a proteinacious substrate. This proteinacious substrate would likely be involved in regulating carbohydrate metabolism, a process controlled by multiple levels of phosphorylation (). Although the overall carbohydrate machinery differs substantially between mammals and plants, both systems contain common phosphoproteins that share conserved functions (; ; ). These proteins would be likely substrate candidates for laforin and SEX4. To address this hypothesis, we tested the majority of the mammalian candidates, but none of them served as a substrate for laforin (; our unpublished data).
It is interesting that laforin and SEX4 are functional equivalents that dephosphorylate a complex carbohydrate and that the mutation of either gene results in the accumulation of insoluble carbohydrates in vertebrates and plants, respectively. Our understanding of the metabolism of insoluble carbohydrates in vertebrate systems is still in its infancy. In contrast, the plant community has made substantial progress in understanding the metabolism of starch (; ). In plants, it is clear that the phosphorylation of glucose residues within starch is required for its proper accumulation and degradation (; ; ). In , glucan water dikinase () and phosphoglucan water dikinase (; ) phosphorylate glucose monomers within amylopectin at the C6 and C3 position (), respectively. As with , mutations in the genes encoding glucan water dikinase and phosphoglucan water dikinase also yield a starch excess phenotype (; ; ). Phosphorylation is necessary for both starch accumulation and degradation; however, the timing of these phosphorylation and dephosphorylation events is unknown (; ). Intriguingly, although glycogen, the soluble storage carbohydrate in vertebrates, contains little to no phosphate, detrimental insoluble carbohydrates like LBs are highly phosphorylated, just like amylopectin in plant starch (; ). Therefore, it appears logical that laforin and SEX4 evolved to perform the critical role of dephosphorylating insoluble carbohydrates to allow their proper degradation.
This basic function of insoluble carbohydrate metabolism provides an intriguing explanation for both the existence of a laforin-like activity in protists and plants and the role of laforin in preventing LD. In protists and plants, carbohydrate dephosphorylation would be necessary for the utilization of insoluble carbohydrates as an energy source. When this activity is absent, these organisms accumulate unusable starch as in the mutants. In vertebrates, laforin would dephosphorylate nascent insoluble carbohydrates to inhibit the formation of detrimental LBs. In the absence of laforin, these nascent molecules increase in size and number and eventually cause LD.
Our work clearly demonstrates that a laforin-like activity is necessary for the proper metabolism of insoluble carbohydrates. This activity is required throughout multiple kingdoms and regulates an overlooked aspect of carbohydrate metabolism. It is striking that protists and plants have provided new insights into a human neurodegenerative disease involving aberrant carbohydrate metabolism that was described almost 100 yr ago by Lafora and Gluck (; ).
The complete open reading frame of Cm-laforin was cloned from cDNA provided by T. Kuroiwa (Rikkyo University, Tokyo, Japan) and from SSP Consortium clone U14967 (). Cm-laforin and were cloned into pET21a (Stratagene) according to standard protocols. A second pET21a construct was generated because the full-length protein is largely insoluble. We truncated the first 52 amino acids of SEX4 to generate pET21a Δ52-SEX4. pET21a VHR () and pET21a Hs-laforin () have been described previously. Hs-laforin, , and were cloned in frame of a triple HA tag into pCHF1 (), which is a modified version of pPZP221 (). pCHF1 contains the 35S cauliflower mosaic virus promoter, the Rubisco terminator from pea, and confers gentamicin resistance for selection in plants. Because and demonstrated that the cTP of SEX4 targets SEX4 to the chloroplast, we fused the cTP of SEX4 (nucleotides 1–213) in frame with Hs-laforin and the triple HA tag in pCHF to generate pCHF cTP–Hs-laforin. All point mutations were generated with the QuikChange Site-Directed Mutagenesis kit (Stratagene) according to the manufacturer's instructions. All constructs were verified by DNA sequencing. Recombinant proteins were expressed with a carboxy-terminal six-histidine tag in BL21 (DE3) CodonPlus cells (Stratagene). Fusion proteins were expressed and purified from soluble bacterial extracts using Ni-agarose affinity chromatography as described previously ().
Hydrolysis of p-NPP was performed in 50-μl reactions containing 1× phosphate buffer (0.1 M sodium acetate, 0.05 M bis-Tris, 0.05 M Tris-HCl, and 2 mM DTT at the appropriate pH), 50 mM NPP, and 100–500 ng of enzyme at 37°C for 1–5 min. The reaction was terminated by the addition of 200 μl of 0.25 M NaOH, and absorbance was measured at 410 nm. We tested the specific activity of each enzyme at pH 5.0, 5.5, 6.0, 6.5, 7.0, 7.5, and 8.0. The optimal pH for each enzyme was as follows: Hs-laforin, pH 5.0; Cm-laforin, pH 5.5; SEX4, pH 6.0; and VHR, pH 6.0. Malachite green assays were performed as described previously () with the following modifications: 1× phosphate buffer, 100–500 ng of enzyme, and ∼45 μg of amylopectin in a final volume of 20 μl. The reaction was stopped by the addition of 20 μl of 0.1 M -ethylmaleimide and 80 μl of malachite green reagent. Absorbance was measured at 620 nm. We tested the specific activity of each enzyme at the same pH units as above. The optimal pH for each enzyme was as follows: Hs-laforin, pH 7.0; Cm-laforin, pH 6.0; and SEX4, pH 8.0.
The sequences of laforin and SEX4 orthologues were obtained by performing tBLASTn searches using the GenBank dbEST database or BLASTp and PSI-BLAST () searches using GenBank eukaryote genome and nr databases, the genome project, Department of Energy Joint Genome Institute Resource, The Institute for Genomic Research, ToxoDB, GeneDB, Genoscope, and Genome Database. Accession numbers are listed in Tables S2 and S5 (available at ). The web address for each database is listed in Table S1. A list of each genome that we investigated and a reason why an organism's genome lacks laforin is listed in Table S4. Amino acid sequences of laforin orthologues were aligned by ClustalW () and refined manually using MacVector. Small subunit ribosomal RNA sequences were obtained by performing BLASTn using GenBank from all organisms and nr databases, and accession numbers are listed in Table S3. The phylogenetic tree was generated from a ClustalW () multiple sequence alignment using PROTDIST and FITCH from the PHYLIP 3.65 software package and was displayed using HYPERTREE 1.0.0 (Pfizer; ).
Homozygous plants (T-DNA insertion line SALK_102567; ) were isolated by PCR. Stable transgenic plant lines were generated by -mediated floral dipping (), and seeds were sterilized, plated on standard growth medium (), and selected for using 100 μg/ml gentamycin per standard protocols (; ). Plants were grown in Promix-HP soil at 22°C with supplemental lighting conditions of 16-h days. To stain starch in leaves, leaves were decolorized in 80% (vol/vol) ethanol, stained with an iodine solution, and destained in water. Starch content was quantified as previously described (). mRNA was obtained using an RNeasy Plant Mini kit (QIAGEN), and first-strand synthesis was performed using SuperScript III First-Strand Synthesis SuperMix (Invitrogen) according to the manufacture's recommendations. Four primer sets were used to test for the presence of transcripts in wild-type (Columbia) and plants. Three primer sets to the transcript and a positive control to , the primer set, was included in each PCR tube. Plant whole leaf lysate was obtained as described previously ().
The α–Hs-laforin and α–Cm-laforin antibodies were generated by immunizing rabbits with recombinant Hs-laforin or α–Cm-laforin, and antibodies were affinity purified from the serum with a HiTrap NHS-activated HP affinity column (GE Healthcare) of Hs-laforin or Cm-laforin protein, respectively. The α–Cm-laforin antibody was generated in a similar manner. Recombinant Hs-laforin and Cm-laforin were detected with their respective primary antibodies followed by goat α–rabbit HRP (GE Healthcare). Recombinant VHR and SEX4 were detected with α-HIS-HRP (Santa Cruz Biotechnology, Inc.). Protein expression of transgenes was monitored by Western analysis using rat anti-HA (clone 3F10; Roche) and goat α–rat HRP (Chemicon).
10D-14 () was provided by T. Kuroiwa and grown asynchronously at pH 2.5 in 2× Allens's medium at 42°C under continuous illumination as described previously (). For immunofluorescence, cells were fixed, washed, and blocked as previously described (). Cells were then probed with 1:100 preimmune serum or 1:1,000 α–Cm-laforin antibody followed by 1:1,000 AlexaFluor488 goat α–rabbit antibody (Invitrogen). Chloroplasts were visualized by their autofluorescence. Immunofluorescence was performed using a light microscope (DMR; Leica) with a PL APO 63× 1.32 NA oil objective (Leica) at room temperature, and images were captured with a CCD camera (C4742-95; Hamamatsu) using OpenLab 4.0.1 software (Improvision). For immunogold EM, cells were fixed, washed, sectioned, and blocked as previously described (). Sections were immunostained with 1:50 preimmune serum or 1:250 α–Cm-laforin antibody and with 10 nm of gold particle–conjugated goat α–rabbit antibody. Grids were viewed using a transmission electron microscope (1200EX II; JEOL), and images were collected using a digital camera (ORIUS SC600; Gatan) and Digital Micrograph software (Gatan). Photoshop (Adobe) and Illustrator (Adobe) were used to generate figures of all images.
Fig. S1 shows purified recombinant Cm-laforin and SEX4. Fig. S2 shows the phosphatase activity of Cm-laforin mutants using p-NPP and amylopectin as substrates. Fig. S3 shows immuno-EM of a cell probed with the secondary antibody alone. Fig. S4 shows the phosphatase activity of SEX4 mutants using p-NPP and amylopectin as substrates. Fig. S5 shows the quantitation of starch in wild-type, , and transgenic plants. Table S1 provides data about non-NCBI databases, Table S2 provides accession numbers for laforin orthologues, and Table S3 provides small subunit ribosomal RNA accession numbers. Table S4 provides data about the genomes investigated for the presence of laforin, and Table S5 provides accession numbers for AMPKβ-GBD proteins and SEX4 orthologues. Online supplemental material is available at . |
Coat proteins represent the core machinery by which transport from different intracellular membrane compartments is initiated. Coat proteins have two well-characterized functions: deformation of compartmental membrane in forming transport vesicles and cargo sorting that entails direct interaction with cargo proteins. Currently, three major coat complexes have been well characterized. The clathrin coat complex is composed of heavy and light chains that form a triskelion, which is coupled to different adaptors for transport from the plasma membrane and the trans-Golgi network. Coat protein I (COPI) and COPII complexes form vesicles that shuttle in the early secretory system that includes the ER and the Golgi complex (; ).
The ADP-ribosylation factor (ARF) family of small GTPases regulates the recruitment of coat proteins from cytosol to membrane in instigating vesicle formation. Their GTPase cycle is regulated by guanine nucleotide exchange factors that activate ARFs and GTPase-activating proteins (GAPs) that deactivate ARFs (; ). The better-characterized GAPs for ARF-related small GTPases, such as ARFGAP1 for ARF1 and Sec23p for Sar1p, function not only as key negative regulators of their small GTPases (; ) but also as their effectors by being core components of coat complexes (; ). Exploring whether other GAPs for ARF members may exhibit a similar behavior, we previously identified ACAP1 (ARFGAP with coiled coil, ANK repeat, and pleckstrin homology domains), a GAP for ARF6 (), to possess a novel function in cargo sorting by recognizing sorting signals in the cytoplasmic domain of the transferrin receptor (TfR) for its endocytic recycling (). Extending this finding, we have shown more recently that ACAP1 also functions in the cargo sorting of recycling integrin as an example of regulated recycling (). However, whether these elucidated roles of ACAP1 reflect its function as part of a coat complex remains unknown.
On a broader note, whether cargo sorting by ACAP1 represents an important mechanism of endocytic recycling also needs to be clarified. One notable view has been that the conventional mechanism of cargo sorting does not play a substantial role in endocytic recycling (; ). Instead, early endosomes have been proposed to use mainly lipid-based mechanisms along with compartmental retention for the selective recycling of proteins to the plasma membrane (). This view has been propagated to a large extent by investigations into TfR recycling in nonpolarized cells, for which evidence for a recycling sorting signal had been lacking () until recently (). However, it should be noted that studies in polarized cells have identified a variant of the clathrin adaptor protein 1 (AP1), which contains the μ1B subunit (), to mediate polarized TfR recycling to the basolateral surface of the plasma membrane (). Nevertheless, because other well-characterized examples of endocytic recycling have not revealed a role for the conventional mechanism of cargo sorting by coat proteins (), the extent that this mechanism is relevant for endocytic recycling remains to be defined.
Another well-characterized example of endocytic recycling has been the insulin-stimulated recycling of glucose transporter type 4 (Glut4; ; ; ). Myo1c, which binds to the actin filament, has been proposed to shuttle Glut4-containing transport vesicles to the plasma membrane along a cytoskeletal track (). Exo70, which is a component of the tethering complex, has been proposed to function in the docking of Glut4 vesicles at the plasma membrane (). VAMP2 and syntaxin 4 have been identified as SNAREs for Glut4 recycling, which mediate the final fusion step (; ). However, notably absent has been the identification of a coat complex for the early mechanistic steps of this recycling. Instead, only compartmental retention has been suggested by the identification of Tether, containing a UBX domain, for Glut4 (TUG), which has been shown to bind selectively to Glut4 and retain it at internal endosomal compartments until insulin stimulation disrupts this binding for Glut4 recycling to the plasma membrane ().
We now show that ACAP1 is part of a novel clathrin coat complex that mediates the stimulation-dependent recycling of integrin and insulin-stimulated recycling of Glut4. Our findings not only advance a basic understanding of how transport is accomplished in endocytic recycling but also provide mechanistic insights into these key physiological events for which endocytic recycling plays a critical role.
As we had previously found that knocking down ACAP1 by siRNA inhibited endocytic recycling of both TfR () and integrin (), we were surprised to find initially that overexpression of ACAP1 also inhibited the recycling of both TfR () and integrin (). Inhibition of integrin recycling was further confirmed by a quantitative biochemical recycling assay (Fig. S1 A, available at ), as we had done previously (; ). The observed inhibitions appeared specific, as overexpression of other GAPs for ARF6, such as ACAP2, that had the greatest sequence similarity to ACAP1 (), did not have similar effects on either TfR () or integrin ( and Fig. S1 A) recycling. Moreover, internalization of surface TfR () and integrin () to the internal perinuclear recycling endosome was unaffected. ACAP1 overexpression also did not affect other major intracellular transport pathways, such as the secretory pathway, as assessed by a temperature-sensitive mutant of the vesicular stomatitis virus G protein (VSVG-ts045; Fig. S1 B), and endocytic transport to the lysosome, by examining the EGF receptor (EGFR) upon ligand binding at the cell surface (Fig. S1 C).
In considering how overexpressed ACAP1 achieved an apparent specific inhibition on endocytic recycling, we initially entertained the possibility that its GAP activity, which had been shown to act on ARF6 (), might be enhanced, which might then inhibit ARF6-regulated endocytic recycling. A key prediction of this explanation was that overexpression of the catalytic-dead point mutant of ACAP1, previously generated by mutating residue 448 from arginine to glutamine (R448Q; ), should abrogate the observed inhibition induced by the wild-type form. However, its overexpression also inhibited both TfR (Fig. S1 D) and integrin (Fig. S1 E) recycling.
In search of an alternate explanation, we noted that the GAPs for ARF family members in the better-characterized intracellular transport pathway acted not only upstream of ARFs as their negative regulators but also as their effectors by being components of coat complexes (; ). Consistent with this possibility, we detected ACAP1 mainly on membranes that had an electron-dense coating by immunogold EM (). As control, overexpression of other GAPs for ARF6 did not exhibit a similar effect (for example, see ACAP2 overexpression in ). Moreover, overexpression of the catalytic-dead mutant of ACAP1 induced a similar membrane coating (). Thus, as coat complexes need to cycle dynamically on and off their target membrane to accomplish a round of transport (), we considered the possibility that overexpressed ACAP1 locked a coat complex onto endosomal membrane to prevent endocytic recycling.
As further evidence in favor of this possibility, we initially examined the ACAP1-induced coating in more detail by immunogold EM. Using antibodies against different components of the currently known coat proteins, we detected no substantial labeling for subunits of AP1, AP2, AP3, AP4, COPI, COPII, and GGAs (Golgi-localized, γ-ear–containing, ARF-binding protein); however, we detected labeling for the clathrin heavy chain (CHC; unpublished data). The CHC couples with the light chain to form a triskelion, which is known to couple to distinct adaptors, resulting in different types of clathrin coat complexes (). Consistent with this generalization, confocal microscopy revealed that overexpressed ACAP1 only showed a partial colocalization with endogenous CHC (). Thus, an intriguing implication was that ACAP1 functioned as part of a novel clathrin coat complex for endocytic recycling.
In favor of this possibility, we found that overexpressed ACAP1 coprecipitated with CHC using lysates derived from cells that overexpressed ACAP1, and in support of our hypothesis that overexpressed ACAP1 locked a coat onto membrane to inhibit endocytic recycling, we found that ACAP1 could be coprecipitated with CHC even when cells were first permeabilized to allow leakage of their cytosol before being subjected to the coprecipitation procedure (). Moreover, ACAP1 and CHC could interact directly, as assessed by a pull-down approach in which ACAP1 as a GST fusion protein was bound to beads and incubated with purified soluble clathrin triskelia ().
To assess whether ACAP1 acted in conjunction with the clathrin triskelion for endocytic recycling, we initially examined whether the recycling of a cargo protein known to be dependent on ACAP1 would be inhibited upon siRNA against CHC. Although we previously defined ACAP1 to be involved in TfR recycling (), knocking down CHC is known to inhibit TfR internalization (). Thus, as recycling could only occur after internalization from the plasma membrane, we overcame this confounding experimental hurdle by examining integrin recycling, as its internalization to the perinuclear recycling endosome was largely unaffected by siRNA against CHC (). In contrast, we found that integrin recycling was inhibited ().
We next sought to confirm these findings using additional approaches that would examine ACAP1 under more physiological conditions, specifically, when ACAP1 was not overexpressed. As one clue, we noted the precedence that some coat complexes could be better observed on intracellular membrane in certain cell types whose physiological function involved its extensive use, such as COPII for transport from the ER in hepatocytes because it is needed for a hyperactive secretory pathway (). Pursuing this possibility, we eventually found that endogenous ACAP1 could be readily visualized in differentiated 3T3-L1 adipocytes and showed considerable colocalization with endogenous CHC ().
The 3T3-L1 adipocytes have been a model system to study insulin-stimulated Glut4 recycling, and Glut4 has been shown to reside in an internal endosomal compartment at the basal condition, when no insulin stimulation is applied (; ; ). A chimeric Glut4 construct, such as HA-Glut4-GFP (), has been used extensively to study Glut4 recycling, because the position of the two tags enabled a quantitative assessment of Glut4 recycling by fluorescence microscopy. When this construct was stably expressed in adipocytes, we found that both endogenous ACAP1 and CHC showed considerable colocalization with its internal pool under the basal (no insulin) condition (). The specificity of the observed staining for endogenous ACAP1 was confirmed by siRNA against ACAP1 (Fig. S2 A, available at ). As both sortilin () and syntaxin 6 () had been shown to have colocalization with internal Glut4, we also examined these two proteins. ACAP1 showed considerable colocalization with sortilin and, to a lesser extent, with syntaxin 6 (). Moreover, providing further support for our hypothesis that ACAP1 overexpression inhibited endocytic recycling by locking a coat onto membrane, we found that endogenous ACAP1 and CHC could no longer be detected in permeabilized adipocytes (Fig. S2 B). In contrast, permeabilization of HeLa cells with overexpressed ACAP1 retained a compact perinuclear distribution (Fig. S2 C), as previously described ().
Intriguingly, we also found that endogenous ACAP1 detected on endosomal membrane was dependent on the differentiation state of the 3T3-L1 cells. Although it was readily visualized in the differentiated (adipocyte) state (), this staining was difficult to detect in the undifferentiated (fibroblast) state (Fig. S2 D). In contrast, staining for sortilin was not similarly affected (Fig. S2 E). The observed difference in ACAP1 staining could not be explained at the level of protein expression, as endogenous ACAP1 could be detected in either state by Western blotting (Fig. S2 F), implying a more subtle explanation for the observed phenomenon. Nevertheless, as our findings thus far on 3T3-L1 cells suggested the possibility that insulin-stimulated Glut4 recycling is an example of extensive physiological usage of ACAP1, we sought to further define the mechanistic relationship between ACAP1 and the clathrin triskelion in these cells, using Glut4 recycling as the context.
Initially, using quantitative microscopy to assess chimeric Glut4 recycling as previously described (), we found that siRNA against ACAP1 substantially inhibited the insulin-induced redistribution of internal Glut4 to the cell surface (). This result was further confirmed by an assay for the cellular uptake of glucose (), which had been shown to reflect Glut4 recycling (; ). Using the same assay, we also demonstrated the specificity of the siRNA against ACAP1, as three distinct targeting sequences all led to similar levels in the inhibition of glucose uptake (). Further specificity for the siRNA approach was reflected by the silencing of ACAP1 not having a considerable effect on the level of CHC, ARF6, and select signaling proteins (Fig. S3 A, available at ). Moreover, we found that the internalization of Glut4 to its internal perinuclear location at the basal condition was not substantially affected (Fig. S3 B). Silencing CHC, we found that Glut4 recycling was also markedly inhibited (). Notably, siRNA against CHC did not prevent the accumulation of internal Glut4 in the basal condition (Fig. S3 C), which was consistent with surface Glut4 previously shown to internalize by both clathrin and nonclathrin means ().
To provide more direct evidence that ACAP1 acted in conjunction with CHC for Glut4 recycling, we took multiple approaches. First, we assessed whether ACAP1 and clathrin required each other for localization to the Glut4 compartment, by examining whether the localization of one was affected upon silencing the other. In differentiated 3T3-L1 cells treated with siRNA against ACAP1, we found that endogenous CHC had reduced colocalization with Glut4, with quantitation revealing about a fivefold reduction (). In cells treated with siRNA against CHC, we found that endogenous ACAP1 also had reduced colocalization with Glut4, with quantitation again revealing about a fivefold reduction (). In contrast, siRNA against neither ACAP1 nor CHC had a considerable effect on the colocalization of Glut4 with sortilin (Fig. S4 A, available at ). Similarly, the colocalization of Glut4 and syntaxin 6 was not affected considerably by either siRNA treatment (Fig. S4 B).
Second, we found that ACAP1 associated with CHC in differentiated adipocytes through a coimmunoprecipitation approach (). As specificity, other transport factors known to participate in Glut4 recycling, such as Exo70 () and TC10 (), did not associate with this complex, and Rab11 () showed very weak association (). In contrast, we detected a complex of Glut4 with both ACAP1 and CHC (), for which the specificity of the immunoprecipitating antibody was verified (). Notably however, although we had found that the association of ACAP1 with integrin β1 was stimulation dependent (), insulin stimulation did not enhance the association of Glut4 with either ACAP1 or CHC ().
Third, because cargo sorting by coat proteins involves their direct interaction with cargo proteins (), we examined whether, and potentially how, the ACAP1-containing clathrin coat complex interacted with Glut4. Glut4 is a multispanning transmembrane protein that contains three prominent cytoplasmic domains, at the N terminus and at the C terminus, and also having a middle cytoplasmic loop (). Thus, we appended each domain to GST and initially examined which domains bound directly to soluble ACAP1 in a pull-down assay, as previously done (; ). ACAP1 was found to interact directly with the middle cytoplasmic loop, whereas the clathrin triskelion showed no substantial binding to any of the domains examined (). We also assessed the relationship between ACAP1 and the CHC in binding to the middle domain of Glut4. Although the sequential incubation of ACAP1 followed by the clathrin triskelion resulted in both being recruited to the Glut4 fusion protein, the converse sequential incubation prevented the recruitment of CHC (). Thus, these results suggested that ACAP1 acted like AP adaptors in bridging an interaction between cargo tails and clathrin triskelion.
We also determined residues in the middle domain of Glut4 critical for its binding to ACAP1. A systematic mutagenesis approach was undertaken that initially involved truncation mutants of this domain (). These constructs were expressed as GST fusion proteins on beads and then analyzed for binding to soluble ACAP1. Remarkably, similar to our previous finding for TfR (), we also found that ACAP1 bound to two distinct regions in the middle domain of Glut4 (). Subsequently, a more detailed analysis of these regions by alanine scanning mutagenesis revealed a requirement for two basic residues (KR) in one region (). For the other region, however, the approach of systematically mutating two tandem residues at a time only led to a mild reduction in binding to ACAP1 (). Focusing on residues where these mild reductions were detected, we found that a more extensive replacement with alanines at this subregion that consisted mostly of hydrophobic residues (PLSLL) resulted in a more dramatic reduction in the binding of ACAP1 to this second region in the middle domain of Glut4 (). When all these mutations were introduced into the entire middle domain of Glut4, we found that its binding to ACAP1 became reduced (). Confirming this result in the context of adipocytes using a coprecipitation approach, we also found that mutations introduced in the context of the entire chimeric Glut4 construct also led to reduced association with ACAP1 (). Importantly, this mutant construct also exhibited reduced recycling in response to insulin (). Thus, key residues in Glut4 needed for its binding to ACAP1 represented recycling sorting signals.
As ACAP1 has been shown to be a GAP for ARF6 (), a prediction was that the novel clathrin coat complex would be regulated by ARF6. Consistent with this prediction, we found that endogenous ARF6 in differentiated 3T3-L1 cells showed considerable colocalization with ACAP1, CHC, and internal Glut4 at the basal condition (). Remarkably, like endogenous ACAP1, we also found that endogenous ARF6 was also difficult to visualize in undifferentiated 3T3-L1 cells (Fig. S5 A, available at ). Moreover, consistent with the general paradigm that ARFs acts upstream of coat complexes (; ), we found that both endogenous CHC and ACAP1 showed reduced colocalization with Glut4 in differentiated 3T3-L1 cells treated with siRNA against ARF6, with quantitation revealing about a fivefold reduction for both cases (). In contrast, the fraction of ARF6 that colocalized with internal Glut4 was not considerably altered by the knockdown of either ACAP1 or CHC (). Confirming these results, we also found that silencing ARF6 disrupted the physical association between Glut4 and components of the novel clathrin coat complex ( and Fig. S5 B).
We also sought more direct evidence that ARF6 played a role in Glut4 recycling. First, the insulin-induced redistribution of internal Glut4 to the plasma membrane was dramatically reduced upon siRNA against ARF6 (). In contrast, this knockdown had no considerable effect on the accumulation of Glut4 internally (Fig. S5 C). Second, we found that silencing ARF6 reduced glucose uptake into adipocytes to an extent similar to that seen for silencing either ACAP1 or CHC (). Third, a biochemical fractionation approach had been used previously to track the formation of Glut4 vesicles from endosomal membrane, whereby Glut4 in vesicular membrane was distinguished from that on compartmental membrane by velocity sedimentation of cell homogenate (). Using this approach, we found that knocking down ACAP1, CHC, or ARF6 all redistributed Glut4, sortilin, and TfR (to a lesser extent) from a fraction that reflected its distribution in vesicular membrane to that in compartmental membrane ().
Finally, as GAP activity acts mechanistically upstream of ARF small GTPases, whereas coat complexes act downstream as their major effectors (; ), we considered a likely possibility that ACAP1 overexpression in HeLa cells had perturbed the ability of ACAP1 to act as coat component, thereby obscuring our ability previously (Fig. S1, D and E) to determine whether its GAP activity also played a role in endocytic recycling. To overcome this hurdle, we noted that the stable transfection of ACAP1 did not have a similar effect as its transient transfection (), likely because the former approach is known to express proteins at a more physiological level. Thus, to examine whether the GAP activity of ACAP1 played a role in Glut4 recycling, we sought to replace endogenous wild-type ACAP1 with the catalytic-dead mutant form by stable transfection. Moreover, as the sequence in mouse ACAP1 targeted by the siRNA showed considerable divergence with the human ACAP1, we knocked down endogenous ACAP1 in the mouse adipocytes and then stably transfected the human forms. Using this approach, we found that the catalytic-dead mutant did not restore glucose uptake to a level similar to that seen for the wild-type form (). Moreover, providing an explanation for why expression of the human wild-type form did not restore glucose uptake to the control condition (when no silencing was achieved), we found that its stable expression resulted in a lower level of ACAP1 than that seen for the endogenous situation (Fig. S5 D). Thus, we concluded that the GAP activity of ACAP1 also played a role in Glut4 recycling.
p
r
o
v
i
d
e
e
v
i
d
e
n
c
e
t
h
a
t
A
C
A
P
1
i
s
p
a
r
t
o
f
a
n
o
v
e
l
c
l
a
t
h
r
i
n
c
o
a
t
c
o
m
p
l
e
x
t
h
a
t
m
e
d
i
a
t
e
s
e
n
d
o
c
y
t
i
c
r
e
c
y
c
l
i
n
g
.
I
n
i
t
i
a
l
i
n
s
i
g
h
t
c
a
m
e
f
r
o
m
s
t
u
d
i
e
s
o
n
H
e
L
a
c
e
l
l
s
.
H
o
w
e
v
e
r
,
b
e
c
a
u
s
e
e
n
d
o
s
o
m
a
l
A
C
A
P
1
c
o
u
l
d
n
o
t
b
e
r
e
a
d
i
l
y
v
i
s
u
a
l
i
z
e
d
i
n
t
h
e
s
e
c
e
l
l
s
b
y
m
o
r
p
h
o
l
o
g
i
c
a
l
t
e
c
h
n
i
q
u
e
s
u
n
l
e
s
s
A
C
A
P
1
w
a
s
o
v
e
r
e
x
p
r
e
s
s
e
d
,
w
e
s
u
b
s
e
q
u
e
n
t
l
y
p
u
r
s
u
e
d
f
u
r
t
h
e
r
s
t
u
d
i
e
s
i
n
d
i
f
f
e
r
e
n
t
i
a
t
e
d
3
T
3
-
L
1
a
d
i
p
o
c
y
t
e
s
i
n
w
h
i
c
h
e
n
d
o
g
e
n
o
u
s
e
n
d
o
s
o
m
a
l
A
C
A
P
1
w
a
s
r
e
a
d
i
l
y
d
e
t
e
c
t
a
b
l
e
.
T
h
i
s
s
i
t
u
a
t
i
o
n
h
a
s
a
l
l
o
w
e
d
u
s
t
o
t
a
k
e
a
d
d
i
t
i
o
n
a
l
e
x
p
e
r
i
m
e
n
t
a
l
a
p
p
r
o
a
c
h
e
s
i
n
a
m
o
r
e
p
h
y
s
i
o
l
o
g
i
c
a
l
c
o
n
t
e
x
t
t
o
p
r
o
v
i
d
e
f
u
r
t
h
e
r
e
v
i
d
e
n
c
e
t
h
a
t
A
C
A
P
1
f
u
n
c
t
i
o
n
s
a
s
p
a
r
t
o
f
a
n
o
v
e
l
c
l
a
t
h
r
i
n
c
o
a
t
c
o
m
p
l
e
x
i
n
e
n
d
o
c
y
t
i
c
r
e
c
y
c
l
i
n
g
.
Alexa 594–labeled transferrin was obtained from Invitrogen. Draq5 to stain DNA was obtained from Biostatus. EGF, insulin, and other chemicals (unless specified otherwise) were obtained from Sigma-Aldrich. H-2-deoxy-D-glucose was obtained from PerkinElmer. HeLa and TRVb1 cells were cultured as previously described (). 3T3-L1 fibroblast was obtained and cultured following the manufacturer's instructions (American Type Culture Collection). 3T3-L1 cells stably expressing Myc-Glut4 or both Myc-Glut4 and sortilin-Myc-6xHis were cultured as previously described (). The differentiation of 3T3-L1 was performed as previously described (). Purified proteins that have been previously described are clathrin triskelia (obtained from W. Boll, Harvard Medical School, Boston, MA) and 6xHis-tagged ACAP1 ().
The following antibodies were described previously (; ; ; ): mouse TS2/16 against β1 integrin, mouse DM1α against α-tubulin, mouse 9E10 against the Myc epitope, mouse M2 against the Flag epitope, mouse 15E6 against the C-terminal HA epitope, rabbit anti-ACAP1, rabbit anti-cellubrevin, rabbit anti-ARF6, rabbit anti-Lamp1, mouse M3A5 against β-COP, rabbit anti-Akt, and secondary antibodies conjugated to Cy2, Cy3, or Cy5.
The following plasmids were used: Flag-tagged human ACAP1 wild-type and catalytic-dead mutant (R448Q), ACAP2 (), Myc-tagged human ACAP1 (), HA-Glut4-GFP (obtained from T. McGraw, Cornell University Medical School, New York, NY), and GFP-tagged VSVG-ts045 (obtained from J. Lippincott-Schwartz, National Institutes of Health, Bethesda, MD). GFP-tagged ACAP1 in pEGFP-C1 was generated by subcloning the coding sequence of ACAP1. To append ACAP1 or Glut4 cytoplasmic domains to the C terminus of GST, the coding sequences of ACAP1 or Glut4 cytoplasmic domains were amplified by PCR. ACAP1 was then subcloned into the BamH1 and EcoRI sites of pGEX-4T-3 vector (GE Healthcare), whereas the Glut4 constructs were subcloned into the EcoRI and NotI sites. Although GST fusion constructs of Glut4 expressed well using the bacterial system, GST-ACAP1 did not. Thus, we subsequently transferred it into the NotI site of pVL1392 for baculovirus expression. Point mutants of Glut4 were generated by using QuikChange II XL site-directed mutagenesis kit (Stratagene).
Transient transfections were performed using Fugene 6 (Roche Biochemicals) for HeLa cells or electroporation as previously described () for differentiated 3T3-L1. 3T3-L1 cell lines that stably express HA-Glut4-GFP, Myc-tagged human ACAP1 wild type, or Flag-tagged human ACAP1 R448Q were generated by transient transfection with selection in 1 mg/ml G418 (Life Technologies) and maintained in 0.2 mg/ml G418.
siRNAs against the sequences CGACATCATGGAATTCGTA, TAAGGACCCTGTAACCGTG, and AGACGTATCTCGACATATT for mouse ACAP1 (nucleotides 558–576, 903–921, and 2129–2147, respectively), and the sequence GAGCTGCACCGCATTATCA for human and mouse ARF6 (nucleotides 304–322; ) were obtained (Dharmacon). siRNAs against the sequence GCAATGAGCTGTTTGAAGA for human and mouse CHC (nucleotides 3182–3200; ) and scrambled sequences as control were obtained (Ambion). Transfection of siRNAs was achieved by using Oligofectamine (Invitrogen) for HeLa cells, and by using the DeliverX Plus delivery kit (Panomics) for 3T3-L1 adipocytes.
The different transport assays used have been described previously: insulin-stimulated redistribution of HA-Glut4-GFP (), TfR recycling (), integrin recycling (), transport of VSVG-ts045 through the secretory pathway (), and down-regulation of surface EGFR through endosomes to the lysosome (). The biochemical assay for integrin β1 recycling was done as previously described ().
Coprecipitation studies on whole cell lysates were performed as previously described (). Pull-down assays using GST fusion proteins were performed as previously described (). Cell permeabilization studies were done by treating intact cells with 0.2% saponin in PBS at 4°C for 5 min followed by PBS wash at 4°C for 5 min. Cellular uptake of glucose was performed as described previously ().
Localization studies by laser confocal microscopy were performed as previously described (). In brief, images were acquired on an inverted microscope (TE2000; Nikon) with C1 confocal system and the Plan Apochromat 40× oil (NA 1.00) and 60× oil (NA 1.40) objective lenses using EZ-C1 software (Nikon) at room temperature. Quantitation studies on images derived from confocal microscopy were performed through Photoshop CS (Adobe) and NIH image analysis software packages, Image J (v. 1.37a), using colocalization threshold plug-in (developed by the Wright Cell Imaging Facility, Toronto, Canada). Immunogold EM was performed as previously described ().
The experiment was performed essentially as previously described (). In brief, 3T3-L1 adipocytes were homogenized and subjected to centrifugation at 2,000 for 10 min at 4°C to remove nuclei and cell debris. The resulting supernatant was subjected centrifugation at 16,000 for 20 min at 4°C to obtain pellet that contains compartmental membrane and supernatant that contains cytosol and vesicular membrane.
Fig. S1 shows further characterization of ACAP1 overexpression. Fig. S2 presents further characterizations of ACAP1 localization and the specificity of ACAP1 antibody. Fig. S3 shows the effects of knocking down ACAP1 or CHC in adipocytes. Fig. S4 presents the relative distribution of Glut4 with either sortilin or syntaxin 6 upon perturbation of the novel clathrin coat complex. Fig. S5 shows further characterization of ARF6 in adipocytes. Online supplemental material is available at . |
Cell–cell adhesion is a fundamental feature of multicellular organisms and is involved in all aspects of tissue morphogenesis. In particular, cadherin-mediated cell–cell adhesion plays important roles in determining cell shape, movement, and sorting (; ), for example, during embryo compaction (; ), gastrulation (; ), and packing of photoreceptors in the retinal epithelium (). In addition to dynamic changes in the organization of cell–cell contacts, these complex cell movements require remodeling of the actin cytoskeletal network to effect global changes in cell shape. One of the keys to understanding tissue morphogenesis is to determine the interplay between cell–cell adhesion and activation of mechanical forces that control membrane dynamics and cell shape.
Initial contacts between cells involve interactions between opposing lamellipodia that initiate E-cadherin clustering and the subsequent expansion of the contact to form strong cell– cell adhesion (; ; ; ). These dynamic processes indicate diverse roles for the actin cytoskeleton in cell–cell adhesion. Lamellipodia activity is mediated by Rac1-controlled actin dynamics. Rac1 is activated upon E-cadherin adhesion (; ; ), and Rac1 protein localizes with E-cadherin during cell–cell adhesion (; ). However, the localization and dynamic regulation of Rac1 activity during cell–cell adhesion has not been followed, nor has the distribution of Rac1 activity been compared with the distribution and activities of its downstream effectors, the Arp2/3 complex and lamellipodia. It is generally thought that Rac1 activation induces interactions between the cortical actin cytoskeleton and cadherins, but recent studies testing binding of actin to the cadherin–catenin complex revealed that the interaction is not direct (; ). In light of these results, actin dynamics may be involved in other aspects of cell–cell adhesion, and therefore, the organization of actin during cell–cell adhesion needs to be reexamined.
Contraction of actin filaments by nonmuscle myosin II has been suggested to play a role during cell–cell adhesion in embryonic development (; ; ; ), stratification of keratinocytes (; ), and assembly of cell–cell junctions in epithelial monolayers (; ; ; ). It is thought that activated myosin II generates contractile forces at the cell periphery that expand or constrict cell shape during morphogenetic cell movements (). However, it is unclear how or where myosin II and contractile forces are locally activated and generated. Previous studies focused on the effects of disruption of actomyosin contraction on E-cadherin distribution and analysis of fixed cells (), cell–cell adhesion within confluent cell monolayers upon removal or readdition of extracellular Ca (; , ; ), or artificial spreading of cells on an E-cadherin substrate (). None of these studies, however, identified mechanisms regulating activation and location of, or the mechanical forces produced by, the actomyosin contractile apparatus during de novo cell–cell adhesion.
Here, we studied mechanisms coordinating different stages of de novo cell–cell adhesion between pairs of normal epithelial (MDCK) cells. Using high-resolution live-cell imaging, biosensors, and small molecule inhibitors, we show for the first time that Rac1 and RhoA activities and their downstream effectors are restricted to zones at the edges of the expanding contact and that rounds of activation and down-regulation of these GTPases are involved in the initiation, expansion, and completion of cell–cell adhesion.
High-resolution live-cell imaging of GFP-labeled E-cadherin (E-cadherin–GFP) in MDCK cells revealed that initial adhesion was established by extension of, and contact between, lamellipodia from opposing plasma membranes that resulted in local accumulations of E-cadherin–GFP at those sites. Subsequently, the overall level of E-cadherin–GFP at cell–cell contacts gradually increased as the contact began to expand (), but during the latter stages of cell–cell adhesion, much of the E-cadherin–GFP moved to the edges of cell–cell contact with relatively less remaining in the middle of the contact (see kymograph [] and intensity profile [] along the cell–cell contact; Video 1, available at ; ). The process of cell–cell adhesion was rapid (0.89 ± 0.15 μm/min), and the cell–cell contact maximized to a final length (40.3 ± 1.9 μm), similar to that of the cell diameter (). These results reveal two distinct stages of E-cadherin organization during de novo cell–cell adhesion formation and the role of active membrane processes in each stage: first, E-cadherin accumulation induced by contacts between opposing lamellipodia, and second, reorganization of E-cadherin to the periphery as the contact expanded.
Next, we examined the distribution of GFP-actin to determine whether changes in actin organization coincided with these two stages of E-cadherin reorganization during cell–cell adhesion. In individual cells, in the absence of cell–cell contact, actin filaments formed a thick cortical bundle that circumscribed the cell periphery and a more diffuse organization in extending lamellipodia ( and Video 2, available at ). During the initial stages of cell–cell adhesion, the cortical actin bundle appeared to disassemble in the immediate vicinity of the contact (), leaving a gap between the ends of the actin bundle that became wider as the contact expanded. This gap in the cortical actin bundle colocalized with accumulated E-cadherin (). Some diffuse actin remained at the cell–cell contact, but it was mostly associated with lamellipodia that intermittently swept over parts of the contact and particularly at the edges as the contact expanded laterally. Thus, both actin and E-cadherin undergo dramatic reorganization during the initial stage of adhesion and subsequent expansion of the contact, but their distributions are different: E-cadherin accumulates in a zone that expands outwards as the contact grows, whereas actin filaments are prominent at the edges of the expanding E-cadherin zone and are greatly reduced within the contact itself.
To examine mechanisms involved in the reorganization of the actin network during cell–cell adhesion, we identified sites of actin contractility and dynamic assembly. We used low concentrations of cytochalasin D (CD; 0.5 μM) to cap the barbed ends of actin filaments and examined effects on the actin network during cell–cell adhesion in live cells (); our goals were to see if the barbed ends of actin filaments were displaced from putative anchorage points on the membrane and to examine the consequences on the cortical bundle.
After CD addition, small actin asters appeared along and at the ends of the cortical actin bundle. Interestingly, actin asters did not form immediately adjacent to the contact itself ( and Video 3, available at ), supporting our earlier observation that the actin bundle disassembles immediately beneath the expanding cell–cell contact. Actin asters are thought to be generated by tension in the actin network through nonmuscle myosin II () and therefore provide fiduciary marks for changes in actin organization brought about by the release of tension in the cortical actin bundle. We observed that actin asters translocated away from the edges of the cell–cell contact along the remaining cortical bundle (Video 3); we interpret this movement as a consequence of the release of CD-capped barbed ends of actin filaments from anchorage points at the edges of the cell–cell contact. We also observed two distinct groups of actin asters, one from each edge of the cell–cell contact and the other from the opposite noncontacting end of the cell, presumably at the sites of focal adhesions, which translocated toward each other with a mean speed of 1.26 ± 0.3 μm/min (); we interpret this movement as evidence of the direction of contractile forces in the cortical bundle away from the edges of the cell–cell contact. Together, these results indicate that there is little or no actin tension along the cell–cell contact itself; the cortical actin bundle is under global tension around the cell perimeter; and anchorage points for the barbed end of actin filaments appear to be located at the edges of the expanding cell–cell contact and at the opposite, noncontacting end of cells.
To examine whether actin filament assembly occurred at the barbed ends located at the edges of cell–cell contacts as indicated from the CD experiment, we measured G-actin incorporation and actin turnover. In saponin-permeabilized cells, FITC-labeled G-actin preferentially incorporated into cortical actin structures at the edges of cell–cell contacts and some focal contacts close to the substratum (); note that there was little or no FITC-labeled G-actin incorporation along the cell–cell contact where E-cadherin had accumulated.
We examined actin turnover using FRAP. Localized photobleaching of GFP-actin in cortical actin bundles at the edges of a compacted contact revealed that actin was highly dynamic (τ = ∼1 min; ; and Video 4, available at ), consistent with the high level of actin polymerization at those sites shown by FITC-actin incorporation (). Actin in cortical bundles in single cells or at the noncontacting sides of adhering cells also exhibited a turnover rate similar to that shown at the edges of cell–cell contacts (unpublished data). However, in contrast to the edges of cell–cell contacts, we detected little or no FITC- actin incorporation in cortical bundles at free, noncontacting edges (), indicating that barbed-end actin polymerization is not the mechanism for observed actin turnover at those sites. Collectively, these results indicate that the cortical actin bundle is under global tension and undergoing polymerization at the barbed ends of filaments localized to the edges of the expanding cell–cell contact.
The results thus far indicate that the cortical actin bundle is generating mechanical forces from anchorage points at the edges of the cell–cell contact. Contraction of cortical actin bundle is mediated by nonmuscle myosin II; therefore, we examined the distribution of activated myosin II in adhering MDCK cell pairs. We found that total myosin II localized along cell–cell contacts, as reported previously (; ; ; ), and throughout the cortical actin bundle (; ). Using a phosphospecific (S19) myosin II antibody () to detect activated myosin II, we found that the highest levels of activated myosin II were concentrated in the ends of the cortical actin bundle located at the edges of expanding cell–cell contacts. Lesser amounts of phosphomyosin II were also located throughout the rest of the cortical bundle (, arrowhead), but little or none was detected within the cell–cell contact (). Even at the earliest stage of cell–cell contact (, arrow), we found that activated myosin II was preferentially localized at the edges of the expanding contact and excluded from the cell–cell contact itself.
Activation of actomyosin contractility requires phosphorylation of myosin light chains by Rho kinase (ROCK) and myosin light chain kinase (MLCK). Inhibition of ROCK by Y27632 resulted in less organized and more diffuse cortical actin bundle and inhibited the localization of activated myosin II at the edges of the cell–cell contact, although some diffuse phosphomyosin II staining was detected throughout the cell (). These results indicate that ROCK activity is required to both localize and activate myosin II to the edges of cell–cell contacts. Note that ROCK inhibition did not dissociate the cell–cell contact, indicating that the maintenance of cell–cell adhesion does not require activated myosin II and actomyosin contractility; this is also consistent with the lack of phosphomyosin II staining within the cell–cell contact ().
To determine roles of activated myosin II during completion of cell–cell contacts, we examined the effects of introducing the small molecule inhibitors ML-7, a MLCK inhibitor, and Y27632 into live cells. Addition of 25 μM ML-7 to MDCK cells expressing GFP-actin caused the immediate seizure of cell–cell contact expansion ( and Video 5, available at ), although prominent cortical GFP-actin bundles remained intact at the edges of the contact. Addition of ML-7 to newly established cell–cell contacts resulted in negative velocities of expansion because of retraction of lamellipodia and detachment of cell–cell contacts (; see Materials and methods for the measurements of expansion velocities).
Addition of 50 μM Y27632 had a lesser effect than ML-7 on the expansion of cell–cell contacts (). Although Y27632 inhibited myosin II activation (), it had little effect on lamellipodia activity, which appeared to be sufficient to induce additional contact formation and expansion but not compaction of the contact (). Note that in the presence of Y27632, E-cadherin–GFP remained along cell– cell contacts, but clusters did not form at edges of cell–cell contacts ( and Video 6, available at ), indicating that activation of myosin II is required for this stage in E-cadherin reorganization. This was confirmed by Y27632 washout, which resulted in the reformation of large E-cadherin–GFP clusters and the rapid movement of those clusters to the edges of expanding cell–cell contact ( and Video 6). Although cell–cell contact remained intact during Y27632 treatment, the cortical actin bundle completely disassembled ( and Video 7). Washout of Y27632 resulted in the rapid reassembly of the cortical actin bundle, which was, as before addition of inhibitor, particularly prominent at the edges of the cell–cell contact. Similar morphological effects were observed with blebbistatin (unpublished data), but no live-cell fluorescence images were collected because of the phototoxicity of blebbistatin. These results indicate that activation of myosin II and organization of the cortical actin bundle are locally controlled at the edges of the contact by the activity of ROCK and MLCK and that activation of myosin II by MLCK and ROCK is required for completion of cell–cell contact.
Members of the Rho family of small GTPase have been implicated in myosin II–mediated contraction (RhoA), lamellipodia activity (Rac1; ), and cell–cell adhesion (; ; ; ). During the formation of cell–cell contacts, we observed minimal filopodia activity and therefore did not focus on a role for Cdc42. Dominant-negative and constitutively active forms of Rho family small GTPases have been used to perturb the activities of endogenous proteins, but they do not show the distribution of GTPase activity and often cause pleiotropic effects on cell spreading and motility that interfere with the analysis of cell–cell adhesion. Therefore, we chose to use Raichu fluorescence resonance energy transfer (FRET)–based biosensors (; ) to follow the distributions of RhoA and Rac1 activities during compaction of cell–cell adhesion in live cells. These Raichu probes are chimeric proteins of Rac1 or RhoA and a Rho binding domain of an effector protein p21-activated kinase or protein kinase N, respectively. Upon activation by GTP loading, FRET efficiency increases as the result of an intramolecular conformational change. We chose to examine the distributions of active Rac1 and RhoA in adhering cell pairs in which only one of the cells expressed the Raichu probe, and therefore the fluorescence signal could be unequivocally localized; one representative example of each FRET pair from three to five independent experiments is shown.
Transient expression of Raichu-Rac1 in MDCK cells showed high FRET efficiency in active lamellipodia at the edges of the expanding cell–cell contact (, arrows; and Video 8, available at ). Only the Raichu-positive cell is visible (, left; cell–cell contact is located at the middle of each panel). In general, Raichu-Rac1 FRET was absent from older areas of the cell–cell contact that had formed previously, although occasionally a lamellipodium would transiently form in the middle of the contact, and Raichu-Rac1 FRET was prominent at the leading edge of the membrane (, asterisk).
The high FRET efficiency of Raichu-Rac1 at the edges of the expanding contact coincided with protruding lamellipodia, where we had shown that actin turnover was also high (). Actin and membrane dynamics initiated upon Rac1 activity are mediated by localized activity of the Arp2/3 complex. We examined the distribution of the Arp2/3 complex using Arp3-GFP () during cell–cell adhesion ( and Video 9, available at ). We found that during initiation of cell–cell adhesion, Arp3-GFP was localized at the tips of lamellipodia along the forming contact (, frames 1–5). Subsequently, as the cell–cell contact expanded, Arp3-GFP and lamellipodia were localized to the edges of the contact (, frame 8 and onward), with fewer incidences within the contact itself. The distributions of Arp3-GFP and lamellipodia at the edges of the expanding contact appear similar to that of active Rac1 (, compare A and B).
In migrating single MDCK cells, Raichu-RhoA FRET was highest around the leading edge and in the retracting uropod at the rear of the cell (), as previously described in fibroblasts (); these distributions are consistent with the role of RhoA in regulating actomyosin contraction at the edges of lamellipodia and in retraction of the plasma membrane during cell migration (). During expansion of cell–cell contacts, Raichu-RhoA FRET was most prominent at the most distal edges of the expanding cell–cell contact but was absent along the cell–cell contact and at the tip of lamellipodia ( and Video 10, available at ). Only the Raichu-positive cell is visible (, left; cell–cell contact is located at the middle of each panel). Note that membranes at the edge of cell–cell contact that had high RhoA FRET activity did not have dynamic lamellipodia-like membrane activity, like those with high Raichu-Rac1 FRET (), but instead exhibited persistent membrane growth. The distribution of highest RhoA FRET appeared similar to that of phosphomyosin II (compare with ).
We are unable to directly compare Rac1 and RhoA FRET in the same cell. However, we note that Rac1 FRET localized to the tips of transient lamellipodia at the edges of the contact, whereas RhoA FRET was observed to persist at the sides of the most distal edges of the expanding contact (, compare A and D), indicating that the distributions are different and that active RhoA localized more at the outside of the contact than active Rac1.
Our results show that RhoA and its downstream effectors ROCK and MLCK are locally active at the edge of cell–cell contacts and are critical for expansion of contacts, but the question is, how is RhoA specifically activated at that site on the plasma membrane? One pathway that activates RhoA is integrin-mediated adhesion and clustering (); it is interesting to note that integrin-mediated adhesion also regulates cadherin function during embryonic development and epithelial–mesenchymal transitions (). To explore possible cross talk between these two adhesion systems, we examined the spatial proximity of sites of E-cadherin–mediated adhesion to the basal cell surface, where integrin-mediated adhesion to the substratum occurs. Using total internal reflection fluorescence microscopy (TIRF-M) in live cells, we detected E-cadherin–GFP clusters at the edges of fully expanded cell–cell contacts that were in close proximity to the substratum (, arrowheads). By TIRF-M, E-cadherin–GFP in the middle of the cell–cell contact was only detectable at the early stage of cell–cell contact (, arrows), but not at the later stage and was only visible using widefield fluorescence microscopy (, arrowheads), indicating that the plasma membranes within the contact were bowed upward, away from the substratum, as the contact expanded (and hence out of focus for TIRF-M), but the contact remained anchored to the substratum at its edges. The proximity of E-cadherin–mediated adhesion to the substratum at the contact edges indicates that E-cadherin– and integrin-mediated adhesion sites are in very close spatial proximity to each other. Indeed, E-cadherin clusters marked by β-catenin staining at the edge of a cell–cell contact were closely associated with paxillin-positive integrin-mediated focal adhesions (). Although these two adhesion complexes form separate junctions, cortical actin bundles seem to be closely associated with both of them ().
We monitored focal adhesion formation during compaction of cell–cell contacts using GFP-paxillin as a maker for integrin-mediated adhesion. In contrast to the smooth, continuous movement of E-cadherin–GFP to the edge of the contact (), the translocation of GFP-paxillin to the edges of the expanding cell–cell contact occurred in steps ( and Video 11, available at ), perhaps as a result of uncoordinated disassembly and assembly of focal adhesions. Indeed, closer inspection showed that new paxillin-positive focal adhesion formed at the edges of cell–cell contact and older ones disassembled from the center of the cell–cell contact, leaving behind only a few small and transient paxillin-containing focal adhesions (). Note that the reorganization and clustering of integrin-based focal adhesions at the edges of cell–cell contacts correlated spatially and temporally with localized activation of RhoA and myosin II by ROCK and MLCK.
In this study, we used high-resolution live-cell imaging to show for the first time the distribution of active Rac1 and its downstream effectors the Arp2/3 complex and lamellipodia and active RhoA and its downstream effectors phosphomyosin II and actomyosin contraction during de novo contacts between stationary epithelial cells. Furthermore, we found that Rac1 and RhoA activation are under tight spatiotemporal control. Based on these novel findings, we define two stages of cell–cell adhesion driven initially by activation of Rac1 and lamellipodia, and then by activation of RhoA and actomyosin contraction ().
Initial adhesion appears to occur through opportunistic contacts between exploratory lamellipodia from opposing cells that result in the rapid accumulation of E-cadherin. At present, it is unclear whether E-cadherin accumulation is an active or passive process. Note that we did not detect synchronous actin accumulation with E-cadherin during de novo cell–cell adhesion, indicating that actin is not involved directly in initial E-cadherin reorganization. This is in agreement with binding studies with purified proteins showing that actin filaments do not bind directly to the cadherin–catenin complex (; ). Rapid diffusion of E-cadherin requires intrinsic plasma membrane activity that might mediate release of regional restrictions by local changes in actin organization.
The zone of E-cadherin accumulation spread outward as more contacts were formed by Rac1-induced lamellipodia at the periphery of the expanding contact (). Note, however, that active Rac1 was not detected within the E-cadherin zone, indicating that local Rac1 activity must be transient and down-regulated. Rac1 activation may be mediated directly by local activation of PI-3 kinase (; ) and accumulation of phosphoinositides that recruit guanine exchange factors (; ; ; ), although we () and others (; ) found that PI-3 kinase activity is not required for either Rac1 or E-cadherin accumulation at cell–cell contacts; further studies are needed to resolve this apparent conundrum.
During this initial stage of cell–cell adhesion, we found that actin underwent a dramatic reorganization beneath the site of contact and E-cadherin accumulation; the cortical actin bundle found in single migratory cells appeared to dissolve beneath the site of contact, leaving the ends of the bundle bracketing the edges of the expanding contact and only faint actin staining at the contact. The reorganization of actin filaments at cell–cell contacts could be the result of concentrated lamellipodial activity in the immediate vicinity of newly formed cell–cell contacts (; ) or the dramatic reorganization of integrin-based focal adhesion upon cell–cell adhesion (). In addition, we did not observe actin filaments directly inserting into the middle of cell–cell contacts at E-cadherin puncta at the ends of retracted thin filopodia close to the substratum or considerable amounts of actin polymerization around E-cadherin puncta, as reported in adhering keratinocytes (; ). The difference in actin organization between simple epithelial (MDCK) cells and keratinocytes and may be due to the different cell types, although we note that we find actin bundles closely associated with the cadherin–catenin complex at the edges of the MDCK cell contacts, where integrin-based focal adhesions are also localized; it would be interesting to examine the distribution of integrin focal adhesions during cell–cell adhesion between keratinocytes.
The second stage of cell–cell adhesion involves active expansion of the contact that results in the formation of a strong cell–cell adhesion (). Expansion of adhesive contacts requires E-cadherin () and is an active process requiring localized actomyosin activation and contractility. Although previous studies reported the location of activated (phospho-) myosin II at cell–cell contacts and that disruption of regulatory pathways controlling activation of myosin II affects the maintenance and reformation of disrupted cell–cell contacts in confluent cell monolayers (; , ; ), the present work is the first to analyze the spatiotemporal regulation of RhoA activity and actomyosin contractility during de novo cell–cell adhesion between pairs of cells.
We showed that active RhoA and phosphomyosin II were excluded from the center of the contact and restricted to the cortical actin bundle in a zone at the outside edges of cell–cell contacts, where G-actin also incorporated. Several opposing signaling cascades may regulate the highly localized zone of RhoA-induced actomyosin activity at the margins of the contact edges (). RhoA activation and actomyosin contraction could be induced by local clustering of integrin-mediated adhesions at the edge of the cell–cell contact (). Alternatively, or in combination, RhoA activity could be suppressed in the center of the expanding contact by p120 catenin localized with cadherin along the contact (; ) or active Rac1 at the periphery of the E-cadherin zone (). However, we have no direct evidence of these mechanisms at present, and further studies will be needed to test them directly.
Our studies indicate that actomyosin contractile forces were directed outward and backward from the cell–cell contact based on actin aster movements after CD-induced capping of actin filament barbed ends (). This activity would have the net effect of pulling the edges of the contacting membranes outward, to fully expand the contact to the width of the cells. Although the movement of cortical actin bundles at the edges of cell–cell contacts has been described and these cortical actin bundles are speculated as contractile bundles (; ), our results are the first to demonstrate the contraction and support previous speculation. This direction of contraction is opposite to that assumed to occur during the resealing of cell–cell contacts in cell monolayers (; , ; ), indicating that different contractile mechanisms are involved in initial cell–cell adhesion and contacts between cells in established monolayers; however, the direction of contraction in cell monolayers has not be shown and should be reexamined and defined directly. At present, it is not clear how the barbed ends of the cortical actin bundle are anchored at the edges of the contact during de novo cell–cell adhesion, although high-resolution TIRF-M indicates that they are closely localized with E-cadherin (β-catenin) and integrin-based focal adhesions. The E-cadherin–catenin complex does not bind actin directly (; ), which indicates that at these sites actin may be anchored by integrin-based focal adhesions (), but further studies on actin linkages to integrins and cross talk with the E-cadherin–catenin complex are required to resolve the mechanisms involved.
In summary, we propose that two zones of Rho family GTPase activity are restricted to the edges of the cell–cell contact as it expands laterally and that they have different roles in initiating adhesive contacts (Rac1) and expanding and completing the contact (RhoA; ). The zone of active Rac1 and its downstream effectors, the Arp2/3 complex and lamellipodia, is localized to de novo contacts between cells; these activities are transient and rapidly diminish as E-cadherin accumulates, but a new round of activation occurs at the periphery of the contacting membranes that would push the membranes together to initiate new E-cadherin adhesion. Diminished Rac1 activity, and hence membrane dynamics, in the newly formed cell–cell contact might allow the maintenance of weak trans-interactions between E-cadherin on opposing membranes. The zone of RhoA and its downstream effector actomyosin contractility is also restricted to the edges of the contact and is required to drive expansion and completion of cell–cell adhesion. Although we have not directly compared the distributions of active Rac1 and RhoA in the same cell, they appear different, with Rac1 localized over the tip of transient lamellipodia, whereas RhoA is localized continuously to the most distal sides of the edges of the expanding contact; we speculate, therefore, that the zone of RhoA activity may be on the outside of the Rac1 zone (). These sequential signaling zones comprising E-cadherin accumulation, Rac1-induced lamellipodia, and RhoA-induced actomyosin contraction coordinate the induction, initial stabilization, and expansion of the cell–cell contact ().
Formation of cell–cell contacts between stable MDCK cell lines expressing E-cadherin–GFP, GFP-actin, or GFP-paxillin was imaged with a custom confocal microscope () or the Marianas system (Intelligent Imaging, Inc.) equipped with a Xenon lamp (DG4 300W; Sutter Instrument) and camera (CoolSNAP HQ [Roper Scientific]; ). The cells were imaged in DME media supplemented with 25 mM Hepes, and the temperature was kept at 37°C using a custom environmental enclosure. FITC-labeled actin (Cytoskeletal, Inc.) was added to saponin-permeabilized MDCK cells to locate polymerizing ends of actin filaments (). GFP-actin turnover was measured with a FRAP module of Marianas system (). Images of myosin-stained MDCK cells were taken with an upright microscope (Axioplan; Carl Zeiss MicroImaging, Inc.) equipped with a mercury lamp; single band filter sets for FITC, Rhodamine, and Cy5; 100× Plan-NeoFluor 1.3 NA objective; and AxioCam
(Carl Zeiss MicroImaging, Inc.).
Raichu FRET probes for Rac1 (1026×) and RhoA (1298×) were variants of published probes that contained Venus instead of YFP and were gifts from M. Matsuda (Kyoto University, Kyoto, Japan). Using Lipofectamine 2000 (Invitrogen), MDCK cells were transiently transfected with the Raichu probes and imaged by Marianas system that configured with CFP/YFP filter set 86002v1(Chroma Technology). Pseudocolored ratio images were generated from images from CFP and FRET channels as described previously ().
All images were analyzed with ImageJ (). In , The expansion velocities were calculated by dividing the change in the contact length by duration of drug treatment, and plotted as a function of normalized contact length (L[t]/Lf, where L[t] is contact length and Lf is final contact length). Because the expansion velocity depends on the maturation of cell–cell contact (), the mean velocity profile from is plotted as a reference (line).
pEGFP-C1-paxillin and pEGFP-N1-Arp3 plasmids were gifts from C. Turner (State University of New York, Upstate Medical University, Syracuse, NY) and M. Welch (University of California, Berkeley, Berkeley, CA), respectively. Stable MDCK cell lines were generated as described previously (). Pharmacological agents were purchased from Sigma-Aldrich (CD and BDM) or Calbiochem (ML-7 and Y27632). Protein localizations in fixed cells were visualized with monoclonal myosin antibody (Beckman Coulter), polyclonal S19 phosphospecific myosin II antibody (generated by F. Matsumura, Rutgers University, New Brunswick, NJ), polyclonal anti–β-catenin (), and monoclonal anti-paxillin (BD Biosciences).
Video 1 is a time-lapse video of two E-cadherin–GFP–expressing MDCK cells making de novo cell–cell contact, shown in A. Video 2 is a time-lapse video of two GFP-actin–expressing MDCK cells making de novo cell–cell contact, shown in F. Video 3 is a time-lapse video of GFP-actin–expressing MDCK cells treated with 0.5 μM CD, shown in A. Video 4 is a time-lapse FRAP video of GFP-actin–expressing MDCK cells, shown in F. Video 5 is a time-lapse video of GFP-actin–expressing MDCK cells treated with 25 μM ML-7 for 1 h, shown in B. Video 6 is a time-lapse video of E-cadherin–GFP–expressing MDCK cells treated with 50 μM Y27632 for 1 h, shown in D. Video 7 is a time-lapse video of GFP-actin–expressing MDCK cells treated with 50 μM Y27632 for 1 h, shown in E. Video 8 is a time-lapse video of Raichu-Rac1–expressing MDCK cells, shown in A. Video 9 is a time-lapse video of two Arp3-GFP–expressing MDCK cells making de novo cell–cell contact, shown in B. Video 10 is a time-lapse video of Raichu-RhoA–expressing MDCK cells, shown in D. Video 11 is a time-lapse video of two GFP-paxillin–expressing MDCK cells making de novo cell–cell contact, shown in E. Online supplemental material is available at . |
Scratch-induced disruption of tissue culture monolayers has been used to study aspects of directed cell migration in a variety of cell types, including fibroblasts, astrocytes, endothelial cells, and epithelial cells. With most primary nontransformed cells, migration involves coordinated movement of the monolayer in a manner more similar to the morphogenetic movements seen during development, such as dorsal closure and convergent extension, than the movement of single cells such as in neutrophil chemotaxis. Disruption of the monolayer causes the loss of cell–cell contacts and a major consequence of this is to induce polarity in cells proximal to the scratch. One aspect of polarization is the formation of actin-rich protrusions, specifically at the front of the cell (). A second aspect of polarization involves the microtubule cytoskeleton and can be visualized as reorientation of the centrosome and Golgi to face the front of the cell. This involves the association of microtubule plus-end tips with plasma membrane complexes at the leading edge as well as movement of the nucleus to the back of the cell (; ; ). Numerous studies have now shown that the small GTPase Cdc42, or one of its close relatives, is required for polarization of the actin and microtubule cytoskeletons in astrocytes, primary fibroblasts, 3T3 fibroblasts, Vero epithelial cells, and endothelial cells (; , ; ; ; ; ; ).
Studies of the signaling pathways controlling microtubule polarization in different adherent cell types have identified a complex consisting of the scaffold protein Par6 and an atypical PKC (aPKC) downstream of Cdc42. Localized activation of Cdc42 leads to localized activation of the Par6/aPKC complex, and this has now been described in astrocytes (, ), primary rat fibroblasts (; ), 3T3 fibroblasts (), and endothelial cells (). The Par6/aPKC complex has at least two crucial activities in this process. First, it is required for the accumulation of the tumor suppressor protein adenomatous polyposis coli (APC) at the plus-end tips of microtubules, specifically at the leading edge. In primary astrocytes, GSK-3β is phosphorylated at Ser9 by PKCζ (), and this was assumed to be the likely mechanism for inhibition of kinase activity leading to APC accumulation. A second activity of the Par6/aPKC complex is to promote the accumulation of another tumor suppressor protein Dlg (Discs Large) in the plasma membrane at the leading edge. The subsequent association of microtubule-bound APC with membrane-bound Dlg is required for microtubule polarization and centrosome reorientation (). It is likely that many other cellular activities are required for reorganization of the microtubule cytoskeleton; for example IQGAP, another Cdc42 effector, is required both for protrusion polarity as well APC and microtubule polarity () and the dynein/dynactin complex is required for centrosome reorientation and mDia and EB1, regulated by Rho, also contribute to APC localization and stabilization (; ).
In this report, we reexamined the significance of GSK-3 phosphorylation using fibroblasts derived from knock-in mice in which the phosphorylation sites of both GSK3α and β isoforms (Ser21 and Ser9, respectively) have been replaced with Ala (). We find that GSK-3 phosphorylation is not required for Golgi/centrosome reorientation, but instead dishevelled (Dvl), axin, and Wnt ligands are required. It appears that a Cdc42/Par6/aPKC signaling pathway cooperates with a noncanonical Wnt signaling pathway to promote polarization of the microtubule cytoskeleton.
We previously reported that localized inhibition of GSK-3β is required for centrosome/Golgi reorientation and that GSK-3β is phosphorylated downstream of Cdc42/Par6/PKCζ in migrating astrocytes and fibroblasts (; ). To investigate whether phosphorylation is the mechanism of GSK-3 inhibition, we analyzed primary embryonic fibroblasts derived from double knock-in mice in which Ser21 of GSK-3α and Ser9 of GSK-3β have been replaced with nonphosphorylatable Ala residues (to be referred to as GSK-3) (). After scratching a monolayer, GSK-3 cells showed no defect in reorientation of the centrosome or the Golgi compared with littermate wild-type fibroblasts (GSK-3), both exhibiting ∼70% reorientation as early as 2 h after wounding (). The experiment is scored such that 33% corresponds to random orientation. Furthermore, small molecule inhibitors of GSK-3 blocked centrosome and Golgi reorientation in GSK-3 and GSK-3 cells to the same extent (). We conclude that although phosphorylation of GSK-3 occurs as a consequence of inducing directed cell migration, the spatially localized inhibition of GSK-3 occurs through a mechanism that does not involve GSK-3 phosphorylation.
An alternative mechanism of inhibiting GSK-3 activity is seen during Wnt signaling and involves protein–protein interactions mediated by dishevelled (Dvl). Wnt ligands induce the interaction of Dvl with the large scaffold protein axin, leading to dissociation of GSK-3 from a complex containing axin, APC, and β-catenin (). To determine whether Dvl is required for centrosome reorientation, all three Dvl isoforms were depleted in rat embryo fibroblasts (REFs) using specific siRNA (). Depletion of single Dvl isoforms had varying partial effects, but depletion of all three Dvl isoforms completely inhibited centrosome and Golgi reorientation (; red line represents random orientation). This suggests that localized inhibition of GSK-3 is mediated by Dvl.
To analyze which domains of Dvl are required for reorientation, full-length Dvl2 or different Dvl domain constructs were microinjected into the nuclei of leading edge cells immediately after scratching a REF monolayer (). Cells expressing full-length Dvl2 exhibited complete loss of reorientation (). Overexpression of Dvl also blocked centrosome reorientation in GSK-3 knock-in cells (unpublished data). This suggests that overexpression of Dvl interferes with the polarized inhibition of GSK-3, as has been observed in other situations (; ; ). Constructs containing the DIX domain (DIX and DIX+PDZ) were also able to block reorientation, while the PDZ alone, DEP domain, or ΔDIX had no effect on reorientation (), suggesting that Dvl regulates reorientation through its DIX domain.
The DIX domain interacts with Dvl itself and with axin (). To examine whether axin is required for reorientation, siRNA was used (). Interestingly, axin depletion blocked reorientation to a similar degree as Dvl2 or Dvl3 depletion (). Moreover, when axin was depleted together with Dvl2 and Dvl3, reorientation was completely inhibited, suggesting that Dvl2 and Dvl3 cooperate with axin, likely through a direct interaction. Axin overexpression after microinjection of an expression construct also blocked reorientation, whereas expression of axin lacking the DIX domain did not (), emphasizing the important role for the DIX domain in this process.
Because Dvl and axin are key components of Wnt signaling pathways and are also required for centrosome/Golgi reorientation, we tested whether Wnt ligands themselves might be involved. Wnt signaling was blocked using recombinant soluble Frizzled-related protein 1 (sFRP1), a naturally occurring antagonist of the Wnt pathway, which acts by binding to Wnt ligands and sequestering them away from frizzled receptors (). sFRP1 treatment led to substantial inhibition of centrosome/Golgi reorientation, suggesting that Wnts are involved in reorientation (). In contrast, recombinant Dickkopf1 (Dkk1) protein, another antagonist that specifically inhibits a subset of Wnt ligands that are required only for the canonical Wnt pathway, did not significantly block reorientation, implying that one or more Wnts belonging to the noncanonical family are responsible (). Importantly, both antagonists blocked β-catenin stabilization in L-cells treated with Wnt3a, confirming their activity as negative regulators of Wnt signaling (unpublished data).
Dvl activity has been reported to be controlled by multiple phosphorylation events (; ; ), which can be observed as a series of band shifts on Western blots (; ). To examine whether the block in centrosome reorientation by sFRP1 impacts Dvl phosphorylation, we examined Dvl2 and Dvl3 mobility on 7.5% polyacrylamide-SDS gels (). Dvl2 and Dvl3 appear as multiple bands even in unscratched, confluent REF monolayers, suggesting some constitutive phosphorylation (, top panel lanes 1 and 9). Addition of sFRP-1, but not Dkk-1 caused a significant increase in gel mobility (, lanes 5 and 13, respectively), suggesting that Dvl is phosphorylated in monolayers and that this is dependent on a constitutive, noncanonical-like Wnt activity. Scratching the monolayer resulted in no major changes in the mobility of Dvl.
Recent reports describe Wnt5a as a regulator of noncanonical Wnt signaling, specifically during events requiring cell migration in vertebrates (; ; ; ; ; ; ). This prompted us to examine a possible role for Wnt5a in polarized cell migration. Wnt5a protein could be detected by Western blot, though its levels remained constant throughout a 2-h time course after scratching, confirming that primary fibroblasts constitutively produce Wnt5a (). To address the role of Wnt5a in centrosome/Golgi reorientation, Wnt5a was depleted using two different specific siRNA oligonucleotides (). Both siRNA oligonucleotides led to substantial inhibition of reorientation, suggesting that Wnt5a is likely to be the major Wnt ligand regulating polarity in these cells (). siRNA-mediated depletion of two other Wnt ligands, Wnt1 and Wnt3a, did not block reorientation (unpublished data). Furthermore, Dvl mobility on gels was significantly increased by both Wnt5a siRNA oligos, suggesting that Dvl phosphorylation is dependent on Wnt5a (). Although we cannot exclude the possibility that other Wnts are expressed and involved, these data support an important role for Wnt5a in polarization of the centrosome/Golgi.
A key step in the pathway controlling centrosome/Golgi reorientation is APC recruitment to microtubule plus ends (; ). In control scratched monolayers, APC is localized along microtubules but is enriched at the leading edge (). In Dvl-depleted cells, APC is mislocalized; it is found along all microtubules at the back as well as the front of the cell and is enriched in the perinuclear region (). APC localization was also greatly affected by axin or Wnt5a depletion (, respectively), showing loss of microtubule tip accumulation and concentration of APC in the perinuclear region. APC localization in cells treated with sFRP1 was similar to that of cells treated with Wnt5a siRNA (unpublished data). Quantifying the effects of siRNA on APC localization demonstrates that Dvl, axin, and Wnt5a are all required for proper APC polarization ().
Dvl can be regulated by protein–protein interactions, leading to changes in its subcellular localization and subsequent complex formation (; ; ; ). Because we have previously shown that the Par6/aPKC/Cdc42 complex is also required for polarized APC accumulation, we looked for a biochemical interaction between these two signaling pathways. Dvl2 was immunoprecipitated from cells pretreated with an aPKC inhibitor (Gö6983), a Wnt inhibitor (sFRP1), or a Cdc42 inhibitor (Toxin B10463). In untreated cells, aPKC could be found in a complex with Dvl2 and Dvl3, and this was increased upon scratching (). Although the increase is modest, scratching induces localized changes in signaling activities at the front of the cell and this is very different, for example, from adding a soluble agonist. When Wnt signaling or Cdc42 activity was blocked, the induced, but not the basal level of this interaction was lost, suggesting both Wnt and Cdc42 activities are required to promote a Dvl2/aPKC interaction after scratching (). In contrast, aPKC inhibitors did not block this interaction, suggesting aPKC activity was not required for Dvl2/aPKC complex formation.
In conclusion, whereas Dvl, axin, APC, and GSK-3 participate in canonical Wnt signaling to promote a transcriptional response, the Wnt5a/dishevelled effects described here represent a noncanonical pathway that is independent of transcription and involves polarization of microtubules. Although Wnt5a is constitutively expressed in these cells, the scratch-induced generation of a new leading edge leads to spatially localized activation of the Cdc42/Par6/aPKC complex and these two pathways are both required to control polarity. The regulation of polarity by Wnt5a could therefore occur by (a) its polarized secretion, (b) polarized activation of its receptor, or (c) polarized activation of the downstream signal transduction pathway (, ; ). The biochemical link between Cdc42/Par6/aPKC and Wnt5a/Dvl is not clear, though we have identified an interaction between aPKC and Dvl2, which is increased soon after scratching. The results described here provide new insights into the role of noncanonical Wnt pathways in establishing microtubule polarity and identify a potential link between Cdc42 and Wnt signaling pathways, whereby polarization of Wnt signaling can occur in a Cdc42-dependent manner, in response to an external cue.
The following antibodies were used: mouse anti-Dvl1, rabbit anti-Dvl2 (used for Western blot), mouse anti-Dvl2 (used for immunoprecipitation), mouse anti-Dvl3, rabbit anti-aPKC, (Santa Cruz Biotechnology, Inc.), rabbit anti-axin (Zymed Laboratories), goat anti-Wnt5a (R&D Systems), mouse anti-p115 (Golgi marker), mouse anti-tubulin (used for IF), mouse anti-myc, (Cancer Research UK, London), rabbit anti-pericentrin (Covance), rat anti–α-tubulin (used for Western blot) (Harlan), and rabbit anti-APC (a gift from Inke Nathke, University of Dundee, UK). Secondary antibodies coupled to HRP were from Jackson ImmunoResearch Laboratories. Secondary antibodies used for immunofluorescence and fluorescent phalloidin were from Molecular Probes. The following reagents were used: recombinant human sFRP-1 and recombinant human Dkk-1 (R&D Systems), Gö6983, Toxin B10463 (Calbiochem), SB216763 (Tocris), and CHIR99021 (a gift from Philip Cohen, University of Dundee, UK).
Full-length cDNA of mouse Dvl2 was a gift from Trevor Dale (Institute of Cancer Research, UK). Deletion mutants of Dvl2 were made using PCR amplification of fragments of interest and subcloning into pRK5myc. All sequences were confirmed by sequencing (MWG Biotech). RFP-F, a farnesylated form of RFP that is targeted to the membrane, was described previously (). Axin constructs were a gift from Robert Kypta (Hammersmith Hospital, UK).
Primary rat embryonic fibroblasts (REF) were prepared as described previously (). Primary mouse embryonic fibroblasts (MEF) were provided by . Only cells from passage 2–3 were used in all experiments. Both cell types were maintained in DME supplemented with 10% fetal calf serum, streptomycin, and penicillin at 37°C. REFs were maintained in 10% CO and MEF in 5% CO. Microinjections were performed as previously described (). In brief, confluent cell monolayers grown on glass coverslips were scratched with a P2 tip. 30 min later, cells along the wound edge were microinjected with 0.4 μg/μl cDNA of interest plus 0.15 μg/μl pRFP-F, which served as an injection tracer and membrane marker. Cells were fixed 4.5 h later.
Polarization was determined 4.5 h or 2 h after scratching of REF or MEF monolayers, respectively. Golgi was detected using antibodies against p115 and centrosome was detected using antibodies against pericentrin. First row cells showing the Golgi/centrosome located in front of the nucleus and in the 120° sector facing the wound, were counted as oriented (; ). For APC polarization, cells exhibiting front accumulation of APC were counted as polarized. Cells exhibiting uniform distribution or perinuclear staining were counted as nonpolarized. For drug-treated or siRNA-treated cells at least 100 cells were counted per coverslip. For microinjected coverslips at least 50 cells were counted per coverslip. Each data point represents at least three independent experiments. Error bars represent SD and statistical tests were performed using a two-tailed test with equal variances.
Unless otherwise specified, all double-stranded predesigned HPLC-grade siRNA oligos were obtained from MWG Biotech. siRNA treatment was performed as previously described (; ). In brief, a total of 600 pmol of siRNA oligos were introduced into 1.5 × 10 cells by nucleofection (Amaxa) using solution V, program G-09. Cells were replated at a density of 5 × 10/cm and assayed 72 h later. Efficient depletion was observed 2–3 d after nucleofection, and was quantified by Western blot. The following siRNA oligos were used: rat Dvl1 (Ambion) GGAGGAGAUCUUCGACGACtt, rat Dvl2 GUACCAGCUUAGGAGAUUCtt, rat Dvl3 GGAAGAGAUCUCGGAUGACtt, rat axin GUAUCGUUGUGGCCUACUAtt, rat Wnt5a (#1) UAACACGUAUUCGAACUUAtt, rat Wnt5a (#2) GCACAGUGGACAACACUUCtt, nonspecific control AGGUAGUGUAAUCGCCUUGtt.
Immunoprecipitation of Dvl2 was performed using a monoclonal antibody, which preferentially recognizes the nonphosphorylated form of Dvl2 by Western blot (), although in our hands it can immunoprecipitate both phosphorylated and nonphosphorylated forms. Western blot for Dvl2 was performed using a rabbit antibody, which recognizes both forms. |
I actually liked organic chemistry in college (at UC Santa Cruz). That was the first time science really made sense to me, mostly because it involved learning some basic principles that allowed you to build more complex molecules. I liked learning the rules of molecules' interacting with each other rather than remembering what they are. And then it evolved from building molecules to applying them to understanding biology.
Yes. I did my Ph.D. with Carolyn Bertozzi. She's a very creative chemist who is mostly geared towards developing tools to understand glycobiology. In her lab, I learned about synthesizing organic molecules and asking biological questions with them.
Much like kinases, there are several isoforms of glycosyltransferases with similar catalytic activity but most likely with discrete functions. There were really no tools at the time to evaluate their functions with small molecules. The hope was that we would find selective inhibitors of each of the isoforms that add each individual carbohydrate onto a growing lipid or protein.
We also tried to develop methods to identify the proteins that carry specific types of modifications. Just like for kinases or the ubiquitin field, we asked what proteins were modified with specific glycans.
Over the last few years, I've been generally interested in how posttranslational events affect protein function. But for many modifications, we don't have very good detection methods to identify which proteins are modified.
The problem with these modifications is that they're transient and heterogeneous. They're often at substoichiometric levels. They come on and off at different cellular states. So, abundance is naturally an issue.
Heterogeneity is the other problem. It's not like one modification is on one protein. A given protein can have five of the same modifications on different sites. So that makes things kind of complicated.
Historically, all modifications have been detected by radioactivity: radioactive ATP, or nucleotide sugars, or fatty acids. The problem is, if you want to identify a protein selectively that carries those modifications, there's no intrinsic way to do affinity retrieval of radioactively labeled proteins. And often, radioactivity is—while effective—a bit cumbersome and slow.
Even for modifications which we have antibodies to detect selectivity, like phosphorylation of histones, for example, they are often context specific. I think the only antibody that's specific for a posttranslational modification is phosphotyrosine, and that's a rare exception that's very general.
So part of my interest is in developing general chemical tools for looking at different types of modifications. We started with glycosylation, and I ventured into protein lipidation, which we're continuing to work on.
That came from the fact that I got interested in host–pathogen interactions. I liked the idea of how viruses and bacteria are able to evade the immune response, how over years of evolution, two organisms coevolved to coexist.
It always seemed to me that nature was much more clever than scientists at manipulating cell biology. And studying that would actually give you some key insight into how cell biology worked.
Since we don't have very many tools to study posttranslational modifications, it's hard to analyze all the cellular effects of an infection. So we often find things that bacteria or viruses do to modulate phosphorylation or cytoskeletal rearrangements, just because we have good methods to analyze those pathways.
We've developed reagents for glycosylation and lipidation, and we're starting to think about other types of posttranslational modifications.
We take chemically reactive functional groups and install them onto the substrates that would be used by enzymes that add the modifications. For example, for the lipid work, we modified fatty acids with a chemical reporter, such as an azide or an alkyne group.
Then we take advantage of the fact that those two groups have some unique chemical properties that we can use to convert them into a fluorescent tag or an affinity handle. We essentially install on small molecules, like metabolites, the same thing people do for proteins. But instead of putting on an HA tag, we put a small chemical reporter that allows us to see them selectively.
The hope is that the derivitized substrates will be used by the cell's enzymes, which then gives us a signal to follow that's more robust than radioactivity. And it gives us the opportunity to enrich for things that are modified.
Many pathogens—particularly bacteria—inject enzymes directly into the host cell that modulate signaling pathways, which we have very little data on. Now that we have better methods to look at posttranslational modifications, we can ask, in the presence of these enzymes, are these pathways perturbed?
And we don't have to go in totally blind. For some, we match the chemical tool with the proposed enzymatic activity of a bacterial enzyme or toxin based on bioinformatics analysis. For example, there's a family of phospholipases from several gram-negative bacteria that we don't have substrates for. Now that we have tools to look at lipidated proteins, we can ask, are there potential substrates of the phospholipases that are lipidated proteins?
In Hidde's lab I got interested in how intracellular bacteria like avoid degradation by proteases in mammalian cells. We showed that active proteases were excluded from -containing vacuoles (). This might be one way bugs manage to survive in host cells.
When we get infected with or exposed to pathogens, we present antigens to the immune system, and that educates the body on what we've been exposed to. This essentially vaccinates us against a second round of infection.
For many bacterial pathogens, we don't know what the antigens are at the molecular level. This makes it difficult to design effective vaccines. So we are trying to directly purify the antigens from infected cells and characterize them by mass spectrometry.
One goal is to identify antigens that are presented to the immune system at different stages of infection, so that you can then identify antigen-specific immune responses in mouse models.
The corollary to that is, once we identify new antigens, they're potential vaccine candidates. You could test whether these specific antigens might be more effective versus what we already use, which are attenuated strains of bacteria for which we don't know the antigens. Hopefully, it will be a more precise method of designing vaccines.
It's a great place to start. I think the big advantage is the freedom of this place. The small but very diverse community encourages me to ask questions about general problems that we don't understand. It gives me the freedom to try different methods and approaches. I don't feel constrained, based on whether I should be a chemist, or a biologist, or whatever. Here, I can just ask scientific questions and not worry about whether my department chair is into that or not.
Definitely, just within the small campus itself—from immunologists, to neuroscientists, to other chemical biology people. Rockefeller also has a very rich tradition in bacterial pathogenesis. I think it was ideal for me, from a scientific point of view. And from a personal point of view, I enjoy New York City as well. |
Translational control and regulation of cell size are essential cellular processes that govern the development and homeostasis of cells and tissues (). The protein synthesis machinery has been largely considered an autonomous entity whose overall output is subject to a limited number of control mechanisms. However, several components of the translational machinery and, consequently, the process of protein biosynthesis are controlled by signaling pathways and transcriptional regulation (). In addition, changes in the control of translation are associated with carcinogenesis. Specifically, ribosome function can be modulated by tumor suppressors and oncogenes, whereas certain signaling pathways enhance the translational capacity of the cell. Deregulation of one or more steps that control protein biosynthesis has been associated with alterations in cell cycle progression and cell growth ().
Activation of mammalian target of rapamycin (mTOR) has emerged as a regulatory mechanism that is conserved from yeast to mammals in the control of protein biosynthesis and cell size (). mTOR is a large serine/threonine protein kinase that is found in two distinct multiprotein complexes: mTOR complex 1 (mTORC1; containing mLST8 and raptor), which has been implicated in translational regulation (), and mTORC2 (containing mLST8, mSin1, and rictor; ; ). Rapamycin in complex with FKBP12 interacts with mTOR and inhibits its activity when mTOR is part of mTORC1 (). mTOR activity is increased in many tumors, which is consistent with its pivotal role in protein biosynthesis, and specific inhibition of mTOR function through the use of rapamycin analogues is considered a promising avenue for cancer treatment (; ; ).
The best-characterized effectors of mTOR in the rapamycin-sensitive complex are S6 kinase 1 (S6K1) and the eukaryotic initiation factor 4E–binding protein 1 (4E-BP1). Phosphorylation of S6K1 by mTOR enhances the translational capacity by acting on translation initiation complex assembly (; ). Phosphorylation of 4E-BP1 by mTOR induces the dissociation of eukaryotic initiation factor 4E from 4E-BP1, which enhances the cap-dependent initiation of mRNA translation ().
mTOR serves as a sensor and integrator of multiple stimuli induced by growth factors, nutrients, energy, or stress. The best-characterized signaling pathway that regulates mTOR activity in mTORC1 is initiated by the activation of phosphatidylinositol 3-kinase (PI3K), which enhances the phosphorylation of Akt (also known as PKB; ; ). Akt phosphorylation inactivates the tuberous sclerosis complex (TSC) formed by hamartin (TSC1) and tuberin (TSC2; ), leading to accumulation of the GTP-bound form of the small G protein Rheb that activates mTOR (). The pathway from PI3K to mTOR is up-regulated in many cancers, as reflected by the increased phosphorylation of PI3K and Akt, which correlates with increased mTOR activity (; ). Growth factors that act through tyrosine kinase receptors have the ability to activate PI3K. Most prominent among these are insulin and insulin-like growth factor-1 (IGF-1; ). The up-regulation of IGF-1 expression and autocrine responses in many tumors may well be a major factor in the increased PI3K signaling and mTOR activity in cancers. Inhibitors of PI3K are accordingly pursued for targeted cancer therapy (; ).
TGF-β, a secreted cytokine, regulates a variety of processes in development and cancer and exerts many autocrine activities. The expression of TGF-β, specifically TGF-β1, is up-regulated in many, if not most, cancers and is thought to play a key role in cancer progression (). In early tumor progression, deregulation of TGF-β signaling leads to a loss of its autocrine antiproliferative effect, resulting in increased cell proliferation. However, the increased TGF-β expression by tumor cells provides distinct advantages for cancer progression (e.g., by inducing localized immunosuppression or enhanced angiogenesis; ). In addition, this increased TGF-β expression benefits cancer progression through autocrine effects on the tumor cells. Specifically, TGF-β can induce an epithelial to mesenchymal transition (EMT) of carcinoma cells, which leads to invasion and metastasis (; ; ). EMT is a complex process that involves cytoskeletal remodeling, cell–cell and cell–matrix adhesion, and transcriptional regulation, leading to the transition from a polarized epithelial phenotype to an elongated fibroblastoid phenotype (; ). A disassembly of cell–cell junctions, including the relocalization of E-cadherin and zonula occludens-1 (ZO-1), occurs during EMT and allows for an increase in cell motility. As a result of EMT, the fibroblastoid cells degrade the extracellular matrix and show invasive behavior. The key role of this TGF-β–mediated process in cancer is illustrated by the ability of TGF-β inhibitors to diminish cancer progression and metastasis ().
TGF-β signals through a complex of two types of serine/threonine kinase receptors. Upon ligand binding, the type II receptors activate the type I receptors, which recruit and activate the intracellular mediators Smad2 and Smad3. Smad2 and Smad3 combine with Smad4 to form complexes that translocate into the nucleus to activate or repress gene expression (; ). Increasing evidence shows that Smad signaling leading to transcription responses is complemented by non-Smad signaling mechanisms that are also activated by TGF-β (; ). TGF-β–induced EMT, which is largely studied using NMuMG epithelial cells as a model, integrates Smad as well as non-Smad signaling, including signaling through small GTPases and the Erk and p38 MAPK pathways (; ).
In our studies of TGF-β–induced EMT using NMuMG cells as a model, we observed that the typical loss of the epithelial phenotype with concomitant acquisition of the spindle-shaped fibroblastoid phenotype was accompanied by an increase in cell size and protein content. The TGF-β–induced increases in cell size and protein synthesis correlated with a rapid activation of mTOR and phosphorylation of its effectors S6K1 and 4E-BP1 through the PI3K–Akt pathway. Inhibition of mTOR by rapamycin blocked the TGF-β–induced increases in cell size and protein synthesis without affecting the EMT phenotype and inhibited cell migration, adhesion, and invasion. Although these results establish mTOR activation as a novel non-Smad TGF-β signaling pathway, they also reveal the direct regulation of protein synthesis by TGF-β as an effector pathway that complements the transcriptional regulation through Smads. The autocrine response of cells to TGF-β–induced mTOR activation and protein synthesis may represent a novel mechanism through which the increased TGF-β expression in tumor cells contributes to cancer progression.
Murine mammary epithelial NMuMG cells are known to undergo an EMT that is readily apparent at 36 h after TGF-β treatment (). As a result of EMT, the cells adopt a spindlelike shape with actin reorganization and E-cadherin delocalization from the adherens junctions (). To better characterize the EMT in NMuMG cells, we used time-lapse microscopy during the first 36 h after adding TGF-β. No major change in cell morphology was apparent during the first 24 h of TGF-β treatment, and the treated cells elongated between 24 and 36 h after adding TGF-β (Videos 1 and 2, available at ).
Concomitantly with the change in phenotype, the cells appeared to increase in size (). To assess whether this was indeed the result of an increase in cell size rather than a more spread phenotype, we evaluated the size of treated and untreated cells by flow cytometry using forward scatter as the parameter indicative of cell size. Furthermore, because cell size varies with cell cycle progression (), we measured the size of only the cells in the G1 phase. At 24 h after adding TGF-β, no difference in cell size was apparent between TGF-β–treated and untreated NMuMG cells (unpublished data). In contrast, after 48 h, TGF-β–treated cells consistently showed an increase in cell size of 10–30% (). We observed that the TGF-β–treated cells were also larger in the S and G2/M phases of the cell cycle (unpublished data). After EMT, the removal of TGF-β resulted in enhanced proliferation and mesenchymal to epithelial transition (unpublished data). This reversibility of the phenotype suggests that the increase in cell size did not result from senescence, which is consistent with the lack of senescence-associated β-galactosidase staining (; unpublished data).
Because changes in cell size often correlate with differences in protein content, we measured the protein content in untreated and TGF-β–treated NMuMG cells. TGF-β increased the protein synthesis in NMuMG cells from 20 to 60% after 24 h (), whereas no effect of TGF-β on protein synthesis was observed after 6 and 12 h (not depicted). To confirm the effect of TGF-β on protein synthesis, we measured the incorporation of S-labeled methionine into newly synthesized proteins at 24 h after TGF-β treatment. As shown in , we found that TGF-β induces an increase in the synthesis of new proteins.
mTOR signaling has been implicated in the control of protein synthesis through the phosphorylation of S6K1 and 4E-BP1 (). Because TGF-β increases protein synthesis, we explored the effect of TGF-β on S6K1 phosphorylation in NMuMG cells. Immunoblot analyses with antibodies specific to the phosphorylated form of S6K1 showed that TGF-β induced the phosphorylation of S6K1 within 1 h after adding TGF-β, reaching maximum phosphorylation at 24 h (). TGF-β also induced the phosphorylation of 4E-BP1 with a peak at 1 h of treatment, as apparent from immunoblot analyses using an antibody to phospho-Ser65 4E-BP1 (). The TGF-β–induced phosphorylation of 4E-BP1 appeared more transient than the S6K1 phosphorylation and was no longer apparent at 24 h.
In response to insulin or other inducers of protein synthesis, the phosphorylation of S6K1 and 4E-BP1 results from mTOR activation (). It has been shown that the phosphorylation of mTOR on Ser2448 correlates with its activation in response to insulin or growth factor stimulation (). Therefore, we examined whether mTOR is phosphorylated on Ser2448 in response to TGF-β using a phospho-Ser2448–specific antibody. As shown in , TGF-β induced the phosphorylation of mTOR with kinetics similar to the TGF-β–induced phosphorylation of S6K1 and 4E-BP1. To establish a causal relationship, we examined the effect of rapamycin, an inhibitor of mTOR activation, on S6K1 phosphorylation in response to TGF-β. As shown in , TGF-β did not induce S6K1 phosphorylation in the presence of rapamycin after 1 h, implicating mTOR activation in TGF-β–induced S6K1 phosphorylation. The effect of rapamycin on TGF-β–induced S6K1 phosphorylation was still apparent at 36 h ().
mTOR activation in response to tyrosine kinase receptors has been shown to result from the activation of PI3K, which, in turn, leads to the activation of Akt (). We investigated whether the activation of mTOR by TGF-β occurs through the activation of PI3K, which then activates Akt. Because PI3K activates Akt through phosphorylation on Ser473, we examined the level of Akt activation in response to TGF-β in NMuMG cells using an antibody against phospho-Ser473 Akt. We found that TGF-β led to Akt phosphorylation after 30 min with a peak at 1 h (, top). This activation was inhibited by LY294002 or wortmannin, which are two inhibitors of PI3K, indicating that the TGF-β–induced phosphorylation of Akt occurred through the activation of PI3K (, top). Stimulation of NMuMG cells with TGF-β also resulted in the phosphorylation of Smad3. A peak level for TGF-β–induced Smad3 phosphorylation was reached after 30 min of stimulation (, bottom). Thus, Akt activation in response to TGF-β occurred with similar kinetics, albeit possibly somewhat slower than Smad3 activation. The PI3K inhibitors did not affect the TGF-β–induced phosphorylation of Smad3 (, bottom).
TGF-β–induced Smad signaling results from activation of the type I TGF-β receptor (TβRI), which acts epistatically downstream from the type II receptor (a result of TβRII-mediated activation of the TβRI kinase). Whereas Smads are activated by TβRI, several responses have recently been directly linked to type II receptors without the functional requirement of the TβRI kinase activity (). Therefore, we examined whether the TGF-β–induced activation of Akt depended directly on the kinase activity of TβRI using the kinase inhibitor SB431542, which specifically binds the ATP-binding site of TβRI and does not inhibit TβRII. As shown in , SB431542 prevented the TGF-β–induced activation of Akt as well as Smad3 phosphorylation. We conclude that TGF-β induces Akt activation through activation of the TβRI kinase.
To further define the activating pathway leading to S6K1 and 4E-BP1 phosphorylation upon TGF-β treatment, we examined whether the phosphorylation of S6K1 and 4E-BP1 in response to TGF-β requires the activation of PI3K. To this end, we examined the effect of TGF-β on the phosphorylation of S6K1 and 4E-BP1 in the absence or presence of wortmannin, an inhibitor of PI3K, using insulin as a positive control to induce S6K1 and 4E-BP1 phosphorylation. Wortmannin inhibited the TGF-β–induced S6K1 and 4E-BP1 phosphorylation similar to its effect on insulin-induced phosphorylation and to the same extent as rapamycin, the inhibitor of mTOR (). Moreover, activation of Akt was required for TGF-β–induced S6K1 phosphorylation as for insulin-induced S6K1 phosphorylation because expression of an Akt mutant, in which both Ser473 and Thr308 have been replaced by alanines, blocked TGF-β–induced S6K1 phosphorylation (). This Akt-AA mutant has previously been shown to prevent Akt activation through dominant-negative interference (). Together, these data illustrate that TGF-β induces the activation of mTOR and phosphorylation of its downstream targets S6K1 and 4E-BP1 through the activation of PI3K and Akt. Rapamycin decreases the TGF-β–dependent increase of cell size and protein synthesis.
The increase in cell size and protein synthesis concomitantly with activation of the mTOR–S6K1 pathway in response to TGF-β raises the question of how much the activation of mTOR accounts for the TGF-β–induced increase in cell size and protein synthesis. Therefore, we examined the effect of rapamycin on the TGF-β–induced cell size increase. Rapamycin inhibited the TGF-β–induced increase in cell size, as assessed by forward scatter in FACS analysis (). Because autocrine TGF-β signaling occurs in most, if not all, cells in culture, including NMuMG cells, we set out to define the contribution of autocrine TGF-β signaling by assessing in parallel the cellular response to SB431542, which prevents the TGF-β–induced activation of Akt and is expected to block autocrine TGF-β signaling. As shown in , SB431542 treatment in the absence of added TGF-β resulted in a decreased cell size. These results suggest that autocrine TGF-β signaling contributes to the control of cell size.
Similar experiments were performed to evaluate the effects of rapamycin and SB431542 on protein synthesis. Consistent with the results on cell size, rapamycin and SB431542 both inhibited the TGF-β–induced increase in protein content (). Furthermore, SB431542 moderately decreased the protein content of NMuMG cells in the absence of added TGF-β. We conclude that similar to the control of cell size, autocrine TGF-β signaling through mTOR contributes to the control of protein synthesis.
We compared the effects of rapamycin and SB431542 on TGF-β–induced EMT. As expected, inhibition of the kinase activity of TβRI using SB431542 prevented TGF-β from inducing EMT, and the cells maintained their epithelial phenotype. In contrast, rapamycin did not affect TGF-β–induced EMT. Thus, the cells acquired their elongated shape with a loss of E-cadherin localization at cell–cell junction and actin reorganization (). EMT is accompanied by the disruption of ZO-1 from tight junctions and increased N-cadherin and fibronectin expression (; ). Rapamycin did not affect the changes in ZO-1, fibronectin, and N-cadherin subcellular localization that accompanied TGF-β–induced EMT as assessed by immunofluorescence nor did it affect the epithelial phenotype in the absence of added TGF-β (). Because rapamycin blocked the TGF-β–induced increase in protein content and size, the cells that underwent EMT in response to TGF-β and in the presence of rapamycin were smaller than in the absence of rapamycin (). Our results indicate that inhibition of the mTOR pathway does not affect the morphological changes associated with EMT but inhibits the TGF-β–induced increase in protein content and cell size that accompanies EMT of NMuMG cells.
To assess whether the data using murine NMuMG cells can be extended to other cell systems, we evaluated the human HaCaT keratinocytes, which also undergo EMT in response to TGF-β (). As shown in , TGF-β induced the elongated morphology characteristic of EMT after 72 h of TGF-β treatment. The idea that HaCaT cells underwent EMT in response to TGF-β was also apparent from the filamentous actin (F-actin) stress fiber formation and E-cadherin relocalization from the junctions ().
As in NMuMG cells, TGF-β induced the phosphorylation of S6K1 in HaCaT cells as assessed using a phospho-Thr389–specific antibody (). Rapamycin inhibited the TGF-β–induced phosphorylation of S6K1 (), confirming activation of the mTOR pathway by TGF-β during TGF-β–induced EMT. The rapamycin-induced decrease in S6K1 phosphorylation in the absence of TGF-β may reflect the autocrine TGF-β stimulation. Consistent with our results with NMuMG cells (), rapamycin did not inhibit the TGF-β–induced EMT phenotype in HaCaT cells (). It should be noted that the TGF-β–induced phosphorylation of S6K1 as assessed using a phospho-Thr389–specific antibody was not restricted to NMuMG and HaCaT cells, which undergo EMT in response to TGF-β, but also occurred in C2C12 mouse myoblasts, NRK-49F rat kidney fibroblasts, and 3T3-F442A mouse preadipocytes (unpublished data).
Consistent with the results of NMuMG cells, in HaCaT cells, TGF-β induced an increase in protein synthesis of ∼60% after 48 h, whereas no difference was observed after 24 h (), and induced an increase in cell size of ∼50% after 72 h (), whereas no difference was observed after 48 h (not depicted). The TGF-β–induced increases in protein synthesis and cell size were inhibited by rapamycin (). Thus, as in NMuMG cells, mTOR activation is required for the TGF-β– induced increases in protein synthesis and cell size but does not affect the TGF-β–induced EMT phenotype in HaCaT cells.
Besides inducing the phenotypic characteristics of EMT, TGF-β treatment of NMuMG cells also results in behavioral changes that are associated with EMT and are important aspects of TGF-β's stimulatory role in tumorigenesis, such as an increase in cell motility and invasion (; ; ). To evaluate whether the mTOR pathway plays additional roles in the EMT process of NMuMG cells in addition to its effect on protein content and cell size, we examined the effect of rapamycin on NMuMG cells treated with TGF-β to induce EMT.
We first examined the effect of rapamycin on the increased motility of cells in response to TGF-β using the monolayer wound assay (). Thus, NMuMG cells were treated with or without TGF-β in the presence or absence of rapamycin for 48 h, at which time all TGF-β–treated cells had undergone EMT. At that time, the cell monolayers were replenished with fresh medium and wounded with a pipette tip, and closing of the wound as a result of cell migration was analyzed by time-lapse microscopy ( and Videos 3 and 4, available at ). After 12 h, cells that had been treated with TGF-β in the absence of rapamycin and had undergone EMT showed a 77% wound closure. In contrast, cells that were not treated with TGF-β and thus had maintained their epithelial phenotype showed a 56% wound closure (). Exposure to rapamycin concomitantly with TGF-β treatment reduced the wound closure of cells that had undergone EMT from 77 to 42%. In cells that had not been exposed to TGF-β and thus had maintained their epithelial phenotype, rapamycin reduced the wound healing from 56 to 47% (). We also measured the migration speed of the cells in this experiment. We found that treatment with rapamycin concomitantly with TGF-β treatment and consequent EMT reduced the migration rate from 11.9 ± 1.5 to 6.5 ± 1.6 μm/h for cells that had been exposed to TGF-β.
We also measured the effect of rapamycin treatment with or without TGF-β on the migration of cells toward 10% FBS in transwell chambers (). Confirming the monolayer wound-healing results, rapamycin reduced the enhanced migration of cells that were treated with TGF-β to a level that is comparable with that of NMuMG cells that were not treated with TGF-β and therefore had maintained their epithelial phenotype. This effect was apparent at 24 h after initiation of the transwell migration assay but not at 6 h (). We conclude that activation of the mTOR pathway in response to TGF-β plays an essential role in the increased motility and migration of cells that have undergone EMT in response to TGF-β.
Migration involves discrete phases of adhesion and de- adhesion (). Therefore, we examined whether rapamycin treatment affected the adhesion of cells to the culture plates. In this assay, cells were treated with or without TGF-β in the presence or absence of rapamycin for 48 h (as for the wound-healing and migration assays), dissociated using trypsin, and seeded back in plastic culture plates. After 30 min, nonadherent cells were removed by washing, and the adhered cells were released from the plastic and counted. Cells that were treated with TGF-β and thus had EMT conversion showed increased adhesion. In contrast, cells that had been treated with rapamycin together with TGF-β showed reduced adhesion (). Thus, inhibition by rapamycin of the increased cell motility and migration in response to TGF-β is associated with (and likely, in part, is the result of) effects of rapamycin on the TGF-β–induced increase in cell adhesion.
Because rapamycin inhibited the migration of NMuMG cells that have undergone TGF-β–induced EMT, we also tested its effect on the invasive behavior of these cells. We used a standard in vitro invasion assay in which cells migrate through Matrigel and through the pores of a nitrocellulose filter toward 10% serum and are then counted (). Cells that had undergone EMT showed a much higher level of invasive behavior than those that had not been exposed to TGF-β and thus had not undergone EMT (). In contrast, the addition of rapamycin concomitantly with TGF-β treatment to induce EMT resulted in a much lower level of invasion than cells that underwent TGF-β–induced EMT in the absence of rapamycin. In fact, the invasiveness of the TGF-β/rapamycin-treated cells was similar to that of the NMuMG cells that were not exposed to TGF-β and did not undergo EMT. Therefore, we conclude that rapamycin inhibits the TGF-β–induced invasion associated with cells that have undergone EMT in response to TGF-β and, thus, that activation of the mTOR pathway by TGF-β plays an essential role in the increased invasive behavior of NMuMG cells with TGF-β–induced EMT.
#text
Cells were stimulated with 2–10 ng/ml TGF-β (PeproTech) for 10 min to 48 or 72 h. After incubation in a medium without insulin overnight, control cells were treated with 5 μg/ml insulin for 1 h to activate mTOR. To inhibit signaling effectors, the cells were treated with 100 nM rapamycin and 100 nM wortmannin (Calbiochem) or 5 μM SB431542 and 10 μM LY294002 (Sigma-Aldrich) for 1 h before TGF-β or insulin treatment, and DMSO was used as a control. To inhibit Akt signaling, cells were transfected using LipofectAMINE PLUS (Invitrogen) with a plasmid expressing an Akt mutant in which Ser473 and Thr308 were replaced by alanines ().
Cells were fixed with 4% PFA for 30 min, permeabilized in 2% PFA and 0.2% Triton X-100 for 10 min, and incubated in a PBS and 3% BSA blocking solution for 1 h. The slides were incubated for 2 h with anti–E-cadherin, anti–N-cadherin (BD Biosciences), anti–ZO-1 (Zymed Laboratories), or antifibronectin (Sigma-Aldrich) diluted (1:400, 1:500, 1:200, or 1:400, respectively) in PBS and 3% BSA and were stained for 1 h with FITC-conjugated secondary antibody (1:500; Invitrogen) or rhodamine-conjugated phalloidin (1:500; Invitrogen) to visualize actin filaments. The slides were incubated with DAPI (1:10,000; Sigma-Aldrich) for 10 min to stain nuclei. After mounting the slides (Cytoseal 280 mounting medium; Richard-Allan Scientific), the cells were viewed at room temperature by epifluorescence microscopy with a microscope (Axiophot; Carl Zeiss MicroImaging, Inc.) using a 100× NA 1.3 oil objective (Carl Zeiss MicroImaging, Inc.) or a microscope (Eclipse TE2000-E; Nikon) using a 100× NA 1.49 oil objective (Nikon), and images were analyzed using Spot (Diagnostic Instruments), NIS-Elements (Nikon), and Photoshop software (Adobe).
The cells, which were treated or not treated for 48 h with rapamycin or SB431542 with or without TGF-β, were trypsinized and resuspended in PBS containing 3% FBS, and 7 μg/ml Hoechst no. 33342 (Sigma-Aldrich) was added for 45 min. 1 μg/ml propidium iodide (Sigma-Aldrich) was added just before flow cytometry, which was performed with a sorter/analyzer SE three-laser system (FACSVantage; Becton Dickinson). The cell cycle and cell size were analyzed with CellQuest (BD Biosciences) and WinMDI 2.8 software.
For metabolic labeling of proteins, cells were preincubated in methionine–cysteine-free DME, and 150 μCi/ml [S]methionine–[S]cysteine (PerkinElmer) was added for 3 h. The cells were lysed in radioimmunoprecipitation assay buffer with protease inhibitors. After centrifugation for 10 min at 4°C, 1:10 of the supernatant was precipitated with 300 μl of 20% trichloroacetic acid on ice for 20 min. After filtration on 0.22-μm GS Millipore membranes, the precipitated S-labeled protein was quantified and normalized against cell number.
Cells were lysed in radioimmunoprecipitation assay buffer with protease and phosphatase inhibitors. Proteins were quantified using protein assay (Bio-Rad Laboratories), and 20 μg of protein was separated by SDS-PAGE and transferred to 0.2 μm nitrocellulose membrane. Membranes were blocked in TBS, 0.1% Tween 20, and 5% BSA for 2 h before overnight incubation with primary antibody diluted 1:1,000 in TBS, 0.1% Tween 20, and 5% BSA. Antibodies to phospho-S6K1 (Thr389), phospho–4E-BP1 (Ser65), phospho-mTOR (Ser2448), mTOR, phospho-Akt (Ser473), Akt, and phospho-Smad3 (Ser433/435)/Smad1 (Ser463/465) were obtained from Cell Signaling Technology. Antibodies to S6K1, 4E-BP1, and Smad3 were purchased from Santa Cruz Biotechnology, Inc., Bethyl Laboratories, and Zymed Laboratories, respectively. Antibody to α-tubulin was purchased from Sigma-Aldrich and diluted 1:4,000 in TBS, 0.1% Tween 20, and 5% BSA. The membranes were incubated for 1 h in HRP-conjugated secondary antibody diluted at 1:2,000–1:10,000 in TBS and 0.1% Tween 20. Immunoreactive protein was detected using ECL (GE Healthcare) and BioMax film (Kodak).
Cell monolayers were wounded with a plastic tip at 48 h after the initiation of treatment (). The migration was followed for 12 h at 37°C and photographed using a Spot RT slider camera (2.3.0; Diagnostic Instruments) mounted on a microscope (Axiovert S-100; Carl Zeiss MicroImaging, Inc.) with a 10× Hoffmann modulation objective and Openlab software (Improvision).
For cell migration in the transwell assay (), cells were labeled with 5 μg/ml of the fluorophore DiI (Invitrogen) for 12 h starting at 36 h after the initiation of treatment. The cells were then trypsinized, and 50,000 cells suspended in DME and 0.2% FBS were added to transwell inserts (8-μm pore size; Falcon HTSF Fluoroblok inserts; Becton Dickinson), held in 24-well companion plates with DME and 10% FBS, and incubated for 6–24 h. Migration of cells to the lower culture plate was assessed by the fluorescence of DiI using a fluorimeter (SpectraMax M5; Molecular Devices).
Cells were trypsinized and seeded onto multiwells for 30 min. Unattached cells were removed by washing twice with PBS, and the attached cells were counted after trypsinization.
Cells were trypsinized, and 50,000 cells resuspended in DME with 0.2% FBS were added to rehydrated Matrigel-coated inserts (BioCoat Matrigel Invasion Chamber; Becton Dickinson) and placed in 24-well companion plates with DME and 10% FBS (). After 24 h, the cells and Matrigel in the upper chambers were removed using a cotton tip. The filters were fixed in methanol for 5 min at −20°C, incubated in DAPI (1:5,000; Sigma-Aldrich) for 10 min, and mounted (Cytoseal 280 mounting medium; Richard-Allan Scientific). The invading cells were counted at room temperature using an epifluorescence microscope (Axiophot; Carl Zeiss MicroImaging, Inc.) with a 10× NA 0.3 air objective, and images were analyzed using Spot (Diagnostic Instruments) and Photoshop software (Adobe).
Time-lapse video microscopy of untreated or TGF-β–treated NMuMG cells is shown in Videos 1 and 2, respectively, to show changes in morphology and cell size during EMT. Images were captured at 20-min intervals for 36 h at 37°C in growth medium. Time-lapse video microscopy of a wound-healing experiment of TGF-β–treated NMuMG cells in the absence or presence of rapamycin is shown in Videos 3 and 4, respectively, to show cell migration after TGF-β–induced EMT. Images were captured at 15-min intervals for 12 h at 37°C in growth medium. Images were captured using a Spot RT slider camera (2.3.0; Diagnostic Instruments) mounted on a microscope (Axiovert S-100; Carl Zeiss MicroImaging, Inc.) with a 10× Hoffmann modulation objective and operated using Openlab software (Improvision). Online supplemental material is available at . |
Signaling by ligands of receptor tyrosine kinases (RTKs) is critical for the stimulation and regulation of neuronal differentiation. Neurotrophins, for instance, act through specific members of the Trk RTK family to activate multiple signaling pathways that alter protein activities and gene expression and thereby promote neuronal differentiation and function (for review see ). Targeted deletion of the neurotrophin nerve growth factor (NGF), or of its receptor TrkA, results in decreased innervation of the spinal cord, hippocampus, and cerebral cortex as well as of target organs such as the skin. In addition, NGF/TrkA signaling is essential for the survival of particular subgroups of neurons in trigeminal, sympathetic, and dorsal root ganglia (DRG) (; ).
Binding of neurotrophins to Trk receptors results in receptor dimerization, autophosphorylation, and in the subsequent activation of multiple signaling pathways such as the ERK/MAP kinase, phosphatidylinositol 3-kinase (PI-3K), and phospholipase C-γ (PLC-γ) pathways. Activation of the ERK1/2 MAP kinase pathway is directed through the recruitment of adaptor proteins such as Shc, Grb2, FRS2, SH2-B, and Gab1 to the receptor complex, through multiple redundant mechanisms (for review see ). In addition, the SH2 domain containing tyrosine phosphatase, SHP2, is necessary for the sustained, but not for the transient, activation of the ERK1/2 MAP kinase pathway and for the stimulation of neuronal differentiation downstream of NGF and other growth factors (; ; ; for review see ; ). SHP2 becomes activated through a conformational change induced by binding of the SHP2 SH2 domains to phosphorylated tyrosine residues on multiadaptor proteins (; ; ; ).
Another ERK MAP kinase, ERK5 (also known as Big MAP kinase), is also activated downstream of neurotrophins, although the exact mechanism involved is not known (; ). ERK1/2 and ERK5 have similar kinase domains and have been shown to share many downstream substrates, but there are also important differences (for review see ). Although both kinases are required for neuronal differentiation in , ERK1/2 appears to regulate neuronal induction while ERK5 regulates subsequent neuronal differentiation (; ; ). Moreover, ERK5 but not ERK1/2 promotes transcription and cellular survival after NGF-stimulation of distal axons in mature neurons ().
The Rho family GTPases Rac and Cdc42 become activated downstream of a variety of stimuli, including NGF, and regulate various aspects of neuronal differentiation, from neurite formation and axon outgrowth to growth cone function and dendritic maturation (; ; ). Cycling of these GTPases between GTP- and GDP-bound states is essential for their function and is regulated by binding to guanine nucleotide exchange factors (GEFs), GTPase activating proteins (GAPs), and guanine nucleotide dissociation inhibitors (GDIs). Loss-of-function mutations in certain of these regulators are associated with human diseases that affect the nervous system, including mental retardation and the degenerative motor neuron disease, amyotrophic lateral sclerosis (for review see ).
We have here identified a novel regulator of neuronal process extension, the neurite outgrowth multiadaptor GTPase-activating protein, NOMA-GAP, which belongs to a new family of multiadaptor proteins with RhoGAP activity. We show that NOMA-GAP is required for NGF-stimulated neurite outgrowth and extension and for the regulation of the ERK5 and Cdc42 pathways downstream of NGF. NOMA-GAP is directly associated with the NGF receptor, TrkA, and functions both as a GAP for Cdc42 and, upon tyrosine phosphorylation, as a multiadaptor protein, to recruit SHP2, Shc, and Grb2 proteins. We show that recruitment of SHP2 by NOMA-GAP is required for the extension of neuronal processes and for the activation of the ERK5 MAP kinase. Furthermore, we demonstrate that NOMA-GAP negatively regulates the activation of Cdc42 and PAK after stimulation by NGF and that this activity is also critical for the formation of neuronal processes.
We have searched for novel regulators of Rho family GTPases, which are involved in the control of neuronal differentiation, by expression profiling of E12-E14 mouse spinal cords and DRG (unpublished data). We identified a gene highly expressed in differentiating neurons that codes for a protein containing a RhoGAP domain, an incomplete Phox (PX) domain, and an SH3 domain. We have named this gene neurite outgrowth multiadaptor RhoGAP protein, NOMA-GAP (). NOMA-GAP has previously been entered in the human genome database (GenBank) under the name sorting nexin 26 (Gene ID 115703) due to the presence of the PX-like domain, which represents the only structural consensus between true sorting nexins and NOMA-GAP. An alternatively spliced form of the murine NOMA-GAP homologue, which is not present in humans, has been described as a regulator of insulin-stimulated glucose uptake in adipocytes (). As we have identified an alternative function, we will use the name NOMA-GAP ().
Sequence comparison of NOMA-GAP with human GenBank sequences revealed another gene sharing 47% overall sequence identity as well as structural similarity with huNOMA-GAP (). Partial sequences of the rat and mouse homologues, which lack the N-terminal region containing the PX and SH3 domains, had been previously named RICS, GC-GAP, or GRIT (; ; ). Given the high degree of structural and sequence identity between these two genes, we propose that they form a novel family of signaling molecules (; Fig. S1, available at ).
NOMA-GAP is first expressed in the neural tube of mouse embryos as early as E8.5, and during development of the central nervous system, expression increases along the anterior-posterior axis (, top left panels). Analysis of transverse sections shows that NOMA-GAP expression precedes terminal neuronal differentiation, as determined by staining for the post-mitotic neuronal marker neurofilament (NF; , bottom panels). Strong expression is also seen in the peripheral nervous system in the DRGs (). NOMA-GAP is also expressed in other neuronal tissues such as in the brain and in the eye (Fig. S2, available at ; and unpublished data). RICS is not expressed in the spinal cord during stages of neuronal differentiation, E8.5 to E14 (). We have also analyzed the location of NOMA-GAP protein in transverse sections of mouse embryos, using specific antibodies raised against the C terminus of NOMA-GAP (; see Materials and methods). NOMA-GAP protein is found in distinct punctate staining along, and at the ends, of axons projecting from the spinal cord (). To localize NOMA-GAP at the cellular level, we have used PC12 neuronal precursor cells, which differentiate upon the addition of NGF (). NOMA-GAP is evenly distributed in a fine punctate pattern around the cell periphery in unstimulated cells. Stimulation with NGF results in the extension of neuronal processes and the preferential accumulation of NOMA-GAP punctate staining at these early protrusions. NOMA-GAP is also present at the growth cones of differentiated cells (, indicated by arrows). We have controlled for the higher level of membrane at the growth cones by co-staining samples with a lipophilic styryl dye (in red; , third panel). In addition, co-staining of polymerized actin, shows that NOMA-GAP is concentrated in the body of the growth cone but absent in the microspikes emanating from the neurite tips (in red; , indicated by arrows in last panel).
We have analyzed the role of NOMA-GAP during NGF-stimulated neuronal differentiation using siRNA directed against rat NOMA-GAP in PC12 cells, which express both NOMA-GAP and RICS (Fig. S1 C). Transfection of siRNA to NOMA-GAP significantly reduced NOMA-GAP expression as visualized both in immunoprecipitates () and in fixed cells immunostained for endogenous NOMA-GAP (). NGF stimulates the transient cell flattening of PC12 cells followed by neurite outgrowth and extension over several days (). We determined the proportion of transfected cells forming neurites (>30 μm long; total processes) or containing long axonal-like processes (>100 μm long; long processes). Remarkably, down-regulation of NOMA-GAP expression by siRNA inhibited the extension of neuronal processes and resulted in cell flattening and spreading (, indicated by arrows). The formation of long neuronal processes was inhibited by over 50% in response to down-regulation of NOMA-GAP (). Down-regulation of RICS, on the other hand, did not inhibit neurite outgrowth and elongation (; ). These data suggest that NOMA-GAP is required for NGF-stimulated neurite outgrowth.
We then examined whether NOMA-GAP is part of the signaling complex that becomes recruited to the NGF receptor, TrkA. Immunoprecipitation of endogenous TrkA from PC12 cells resulted in the coimmunoprecipitation of endogenous NOMA-GAP and Shc proteins (). Unlike Shc, NOMA-GAP formed a constitutive, NGF-independent complex with the receptor. NGF-stimulation resulted in the strong tyrosine phosphorylation of the associated NOMA-GAP (). We confirmed that NOMA-GAP interacts directly with TrkA by expressing N- and C-terminal deletion mutants of huNOMA-GAP with the cytoplasmic domain of huTrkA in the yeast two-hybrid system (). Expression of a deletion mutant of NOMA-GAP containing the SH3 and RhoGAP domains (SH3-GAP) interacted with the TrkA cytoplasmic domain and enabled growth of the transfected yeast on selection plates (). No interaction was seen with the N-terminal 186 amino acids containing the PX domain (‘PX’), the C-terminal region of NOMA-GAP (Cterm), or with the isolated SH3 domain of NOMA-GAP (). Together, these data show that NOMA-GAP interacts directly with the cytoplasmic domain of TrkA, through an N-terminal region that includes residues 279–782.
We then dissected the function of the different domains of NOMA-GAP in PC12 cells (; Fig. S3, available at ). Transfected cells were stimulated for a short period (48 h) with NGF, during which time control cells start to differentiate, but have not yet formed long neuronal processes (, top left column). Samples were then scored for the proportion of transfected cells bearing short neurites (30–100 μm), long processes (>100 μm), or no neurites. Expression of NOMA-GAP alone did not affect cellular morphology during this time period. However, NOMA-GAP strongly synergized with NGF in the formation and extension of neuronal processes (, top middle column; quantified in ). NOMA-GAP mutants, where the RhoGAP domain (delRhoGAP), the whole N-terminal region (containing the PX, SH3, and RhoGAP domains; Cterm), or the C-terminal region (Nterm) had been deleted, failed to induce the extension of long neuronal processes. The formation of shorter neurites was not affected (; quantified in ). This suggests that both the RhoGAP domain, and sequences in the C-terminal region of NOMA-GAP are required for the stimulation of neurite extension. The PX domain, on the other hand, negatively regulates NOMA-GAP function, as deletion of this domain (delPX) stimulates NOMA-GAP–induced process outgrowth (, bottom right column; quantified in ). PX domains have been reported to negatively regulate protein activity in other proteins with tandem PX-SH3 domains ().
Next we examined the signaling function of the C-terminal region of NOMA-GAP (residues 965–1287), which is rich in tyrosine residues that lie within consensus sites for the binding of SH2 domains. We performed a modified yeast two-hybrid screen using this portion of NOMA-GAP and a human brain cDNA library (HY4004AH). An activated mutant of the tyrosine kinase Src (Y416F Y527F) was coexpressed from the same vector to enable tyrosine phosphorylation of NOMA-GAP (see also Materials and methods). We identified several clones encoding the N-terminal region of the tyrosine phosphatase SHP2 as well as full-length clones of the adaptor proteins Grb2 and Shc (). The C terminus of NOMA-GAP contains a consensus binding site for the Grb2 SH3 domain (PXXPXR; ), which may explain weak binding of Grb2 to NOMA-GAP in the absence of tyrosine phosphorylation (). Both full-length and the N terminus of SHP2 interacted with the C terminus of NOMA-GAP in a Src-dependent fashion (). The N terminus of SHP2 contains two SH2 domains, and binding to NOMA-GAP is abrogated by mutation of the conserved arginine residues in the phosphotyrosine binding pockets of these domains (RK mutant, ; ). This indicates that interaction with SHP2 is mediated through phosphorylated tyrosine residues on NOMA-GAP. To identify the SHP2-binding site on NOMA-GAP, we mutated all tyrosine residues in the C-terminal region of NOMA-GAP found in the consensus YXXI/V/L sequence for SHP2 binding (for review see ; ). Mutation of a single tyrosine in the C terminus of NOMA-GAP, residue 1169, abolished the direct interaction of NOMA-GAP with SHP2 but had no effect on the binding of Grb2 or Shc (). Mutation of other tyrosine residues, for example tyrosine 1119, had no effect on SHP2 binding to NOMA-GAP.
SHP2 is activated through a conformational change induced by binding of tyrosine-phosphorylated proteins to its SH2 domains (). Because NOMA-GAP binds to SHP2 through the SHP2 SH2 domains, it should lead to the activation of this phosphatase. We confirmed this by incubating recombinant wild-type GST-tagged SHP2 with lysates from serum-stimulated cells expressing full-length or the C terminus of NOMA-GAP. Recombinant SHP2 was recovered by incubation with glutathione-Sepharose beads, and phosphatase activity assayed on the artificial substrate pNPP. Indeed, expression of both full-length and the C terminus of NOMA-GAP stimulated SHP2 phosphatase activity ().
To test whether the SHP2 binding site on NOMA-GAP, residue 1169, is a major phosphorylation site downstream of NGF, we analyzed tyrosine phosphorylation of Myc-tagged wild-type and Y1169A NOMA-GAP immunoprecipitated from NGF-stimulated PC12 cells. Wild-type NOMA-GAP is weakly phosphorylated on tyrosine residues in cells grown in reduced serum and becomes strongly phosphorylated upon stimulation with NGF (). Mutation of residue 1169 significantly reduced tyrosine phosphorylation of NOMA-GAP, indicating that this is a major tyrosine phosphorylation site downstream of NGF. Expression of dominant-negative SHP2, which contains a 31-amino acid deletion in the phosphatase domain (DNSHP2; ), did not significantly alter tyrosine phosphorylation on NOMA-GAP (), suggesting that NOMA-GAP is not a major substrate for SHP2.
We confirmed that NOMA-GAP associates with SHP2 downstream of NGF signaling in two-way immunoprecipitations of endogenous proteins from NGF-stimulated PC12 cells. Immunoprecipitation of NOMA-GAP resulted in the coimmunoprecipitation of SHP2 only upon stimulation with NGF (). NOMA-GAP also associated with Grb2, and this was enhanced by stimulation with NGF. Immunoprecipitation of endogenous SHP2 confirmed that NGF stimulates the association of NOMA-GAP with SHP2 in PC12 cells ().
We interfered with SHP2 function downstream of NOMA-GAP by either mutating the SHP2 binding site on full-length NOMA-GAP (Y1169A, ) or by coexpressing DNSHP2. Interfering with SHP2 function inhibited NOMA-GAP-induced extension of neuronal processes in NGF-stimulated PC12 cells (; quantified in ). We then analyzed whether wild-type and mutant human NOMA-GAP could rescue the inhibition of NGF-induced neurite outgrowth caused by down-regulation of endogenous NOMA-GAP (). Expression of wild-type human NOMA-GAP, but not of Y1169A or N- and C-terminal deletion mutants, rescued NGF-stimulated neurite outgrowth in PC12 cells treated with NOMA-GAP siRNA (; quantified in ).
We also analyzed the function of NOMA-GAP in the spinal cord by using in ovo electroporation of chick embryos. Using this technique, cells in the proliferative ependymal layer on one side (; right-hand side, top row) of the spinal cord are electroporated at stages preceding endogenous neuronal differentiation. These cells then populate that entire half of the spinal cord. Expression of wild-type or delPX human NOMA-GAP resulted in the premature differentiation of cells in the proliferative ependymal layer (as shown by staining for the post-mitotic neuronal marker NeuN; quantified in ) and the projection of neuronal processes from these cells (neurofilament staining; indicated by arrows, second column, ). This was abrogated by mutation of the SHP2 binding site in NOMA-GAP (Y1169A, last column, ; quantified in ). Together, the above data indicate that recruitment of SHP2 through the C-terminal domain is required for the function of NOMA-GAP in neuronal process extension and differentiation.
SHP2 has been shown to regulate the activation of the Ras/ERK1/2 MAP kinase pathway (for review see ; ). Indeed, NOMA-GAP–stimulated neurite extension in PC12 cells was inhibited by the MAP kinase pathway inhibitor U0126 or by a dominant-negative mutant of Ras (N17Ras; , quantified in ). Moreover, overexpression of wild-type NOMA-GAP resulted in the strong activation of the Elk1 reporter in epithelial cells (). Activity was dependent on interaction with SHP2, as mutation of the SHP2 binding site Y1169 strongly inhibited transcriptional activation of Elk1. We also analyzed whether NOMA-GAP could stimulate MAP kinase activation in PC12 cells. PC12 cells expressing wild-type or Y1169A NOMA-GAP were stimulated with NGF either for 24 h () or for shorter time periods (). To our surprise, NOMA-GAP did not significantly alter the phosphorylation (and thus activation) of the classical MAP kinases, ERK1 and 2, but resulted in the sustained elevated activation of ERK5. Cells expressing Y1169A NOMA-GAP, on the other hand, showed an elevation in the transient activation of ERK5 but then a rapid loss in ERK5 phosphorylation (, quantified in ). A similar result was observed upon coexpression of DNSHP2 with NOMA-GAP (unpublished data). Down-regulation of endogenous NOMA-GAP in PC12 cells, on the other hand, resulted in a decrease in the sustained (1 h), but not in transient (5 min), activation of ERK5 and had no effect on the activation of ERK1/2 (). Furthermore, ERK5 activation in NOMA-GAP down-regulated PC12 cells could be rescued by coexpression of wild-type but not of Y1169A NOMA-GAP (). NOMA-GAP thus regulates sustained activation of ERK5 downstream of NGF, through recruitment of SHP2.
Binding assays of the RhoGAP domain of NOMA-GAP with dominant-active and -negative forms of Cdc42 and other RhoGTPases showed that NOMA-GAP preferentially binds to active Cdc42 (Fig. S5, available at ). We studied whether NOMA-GAP regulated the Cdc42 signaling pathway downstream of NGF in PC12 cells. We used antibodies to the autophosphorylation sites on the S/T kinase PAK 1/2, which lies downstream of Cdc42 in NGF-stimulated PC12 cells (). Down-regulation of NOMA- GAP resulted in the inhibition of ERK5 as observed before, but also in a strong elevation in the autophosphorylation, and thus activation, of PAK (). Furthermore, we could rescue this elevation by coexpression of hNOMA-GAP (). A partial rescue was seen upon coexpression of the SHP2 binding mutant, Y1169A NOMA-GAP (), suggesting that SHP2 may play a small role in the regulation of PAK downstream of NOMA-GAP. We also analyzed the levels of GTP-Cdc42 in NOMA-GAP siRNA-treated PC12 cells, using a pull-down assay with the Cdc42/Rac-interactive binding (CRIB) domain of PAK (). Down-regulation of NOMA-GAP resulted in increased Cdc42-GTP in NGF-stimulated and unstimulated PC12 cells. Thus, NOMA-GAP is an important negative regulator of Cdc42/PAK signaling downstream of NGF.
Expression of dominant-active Cdc42 (V12 Cdc42) in PC12 cells is inhibitory to neurite outgrowth, and induces instead cell flattening and spreading as well as filopodia formation (; ; ). This could be reversed by coexpression of NOMA-GAP (). PC12 cells treated with NOMA-GAP siRNA exhibit a similar, although less extensive, phenotype ( and ). We asked whether the inhibition of neurite outgrowth caused by down-regulation of NOMA-GAP is due to the elevated activation of Cdc42 observed in these cells. We therefore titrated dominant-negative Cdc42 (N17Cdc42) into NGF-stimulated NOMA-GAP or control siRNA-treated PC12 cells (). As has been previously observed (), expression of N17Cdc42 in controlsi-treated NGF-stimulated cells resulted in inhibition of neurite outgrowth and the cells showed a round undifferentiated phenotype (, first column; quantified in ). Down-regulation of NOMA-GAP leads to cell flattening. Expression of low levels of N17Cdc42 (but not N17Rac) in these cells, however, resulted in a decrease in cell spreading and stimulated the extension of neurites (). Thus, NOMA-GAP enables neurite outgrowth downstream of NGF signaling by tempering Cdc42 activation. At higher levels of N17Cdc42, neurite outgrowth is also inhibited in these cells, indicating that this GTPase has a function, albeit at controlled levels, in neurite outgrowth.
h
a
v
e
h
e
r
e
i
d
e
n
t
i
f
i
e
d
a
n
o
v
e
l
r
e
g
u
l
a
t
o
r
o
f
n
e
u
r
o
n
a
l
p
r
o
c
e
s
s
e
x
t
e
n
s
i
o
n
,
t
h
e
n
e
u
r
i
t
e
o
u
t
g
r
o
w
t
h
m
u
l
t
i
a
d
a
p
t
o
r
R
h
o
G
A
P
p
r
o
t
e
i
n
,
N
O
M
A
-
G
A
P
,
w
h
i
c
h
h
a
s
b
o
t
h
m
u
l
t
i
a
d
a
p
t
o
r
a
n
d
e
n
z
y
m
a
t
i
c
f
u
n
c
t
i
o
n
s
t
h
a
t
a
r
e
e
s
s
e
n
t
i
a
l
f
o
r
t
h
e
t
r
a
n
s
d
u
c
t
i
o
n
o
f
s
i
g
n
a
l
s
a
n
d
b
i
o
l
o
g
i
c
a
l
a
c
t
i
v
i
t
y
o
f
N
G
F
/
T
r
k
A
.
I
n
p
a
r
t
i
c
u
l
a
r
,
w
e
d
e
m
o
n
s
t
r
a
t
e
t
h
a
t
N
O
M
A
-
G
A
P
p
l
a
y
s
t
w
o
d
i
s
t
i
n
c
t
a
n
d
c
r
u
c
i
a
l
r
o
l
e
s
d
o
w
n
s
t
r
e
a
m
o
f
N
G
F
s
i
g
n
a
l
i
n
g
:
p
r
o
m
o
t
i
o
n
o
f
S
H
P
2
/
E
R
K
5
s
i
g
n
a
l
i
n
g
a
n
d
t
e
m
p
e
r
i
n
g
t
h
e
a
c
t
i
v
a
t
i
o
n
o
f
t
h
e
C
d
c
4
2
/
P
A
K
s
i
g
n
a
l
i
n
g
p
a
t
h
w
a
y
.
Full-length human NOMA-GAP was subcloned from ESTs AW166303, AL137579, and BG772651 (RZPD Deutshces Ressourcenzentrum für Genomforschung GmbH, Berlin, Germany) into the mammalian expression vector pEFmyc.2. The N and C termini of NOMA-GAP and full-length human SHP2 were subcloned into the yeast expression vectors pGBT9, pGADT7 (CLONTECH Laboratories, Inc.), and pSrcBridge. pSrcBridge was generated by subcloning nonmyristoylated (G2E) active (Y416F Y527F) chicken Src into the second multiple cloning site of pBridge (CLONTECH Laboratories, Inc.). NOMA-GAP deletion and point mutants were subcloned into pEFmyc.2. NOMA-GAP C terminus was subcloned into pcDNA3Flag (Invitrogen). Wild-type and dominant-negative (delP mutation; provided by Dr. A.M. Bennett, Yale University School of Medicine, New Haven, CT; ) SHP2 were subcloned into pEF HA.6. Point mutations were inserted by PCR using oligonucleotide primers containing respective base substitutions. Y527F Src pEF and all pEF vectors have been described previously (). Control siRNA and NOMA-GAP siRNA were purchased from Dharmacon (siCONTROL Non-Targeting siRNA pool D-001206-13-20 and siGENOME SMARTpool D-050940, respectively). A pool of the following NOMA-GAP siRNAs was used: GUACAGAAAUGGAGGACAUUU, GGACAGACCAGAAGUUACUUU, and GAGGUCCUGUUCAGCGAUAUU. The siRNAs were also tested singly with similar effects. RICS siRNA has been previously described (). The following antibodies were used: Anti-Neurofilament 68, anti-α-tubulin, and anti-phospho-ERK1/2 (from Sigma-Aldrich); anti-HA tag (Roche); anti-NeuN antibody (Chemicon International); anti-Myc (A14), anti-SHP2 (C18), and anti-Trk (C14) antibodies from Santa Cruz Biotechnology, Inc.; Fast-Track anti-NOMA-GAP antibody (AbCam); anti-GFP antibody (Invitrogen); anti-Shc antibody (Upstate Biotechnology); anti- phospho-ERK5 (Biosource International); anti-ERK1/2, anti-phospho-tyrosine, anti-PAK1/2, and anti-phospho PAK1/2 from Cell Signaling Technology; and anti-Grb2, anti-Cdc42, and anti-phospho tyrosine PY20 from BD Transduction Laboratories. Polymerized actin was stained with Phalloidin-Texas red (Sigma-Aldrich), the lipophilic styryl dye FM 1-43FX (Invitrogen) was used to stain cell membranes and TOTO-3 (Invitrogen) was used to stain nuclei for counting NeuN staining in chick embryos. All fluorescent secondary antibodies (Cy2, Cy3, and Cy5) were from Jackson Laboratories. Phospho-specific and total protein immunoblotting for a given protein was performed simultaneously on duplicate Western blots. U0126 was from Calbiochem.
PC12 cells were grown at 37°C and 5% CO in DME (Invitrogen) supplemented with 10% horse serum and 5% fetal calf serum. Stimulations were carried after overnight culture in media with reduced serum (2.5% horse and 1.25% fetal calf serum) with 100 μg/μl (or 25 μg/μl for low levels) of NGF (Promega). DNA transfections were performed according to the manufacturer's instructions using LipofectAMINE 2000 (Invitrogen). 40 ng of pEGFP-C1 (CLONTECH Laboratories, Inc.) were used as a transfection marker. For transfections, PC12 cells were seeded onto poly--lysine/laminin-coated coverslips (BD Biosciences) or on poly--ornithine/laminin (Sigma-Aldrich) coated tissue culture plates and transfected 24 h later. Fertilized chick eggs (Charles River Laboratories) were incubated at 38°C in a humidified incubator and unilateral in ovo electroporations were performed at room temperature using a T820 electro-squareporator (BTX, Inc.).
For immunofluorescence, PC12 cells were prepared as described previously (). siRNA-transfected PC12 cells were stimulated with NGF 4 h after transfection and analyzed 72 h after transfection. PC12 cells overexpressing NOMA-GAP constructs were stimulated with low levels of NGF 24 h after transfection and analyzed 72 h after transfection. In PC12 differentiation assays, at least 150 GFP-positive cells were scored from random fields per condition and per experiment for the presence of short (30–100 μm) or long (>100 μm) processes. Mean pixel intensity of immunofluorescent staining for endogenous NOMA-GAP was determined for GFP-positive cells from jointly stained samples using LSM 5 Pascal version 3.2 software. Immunofluorescent staining of chick embryos was performed on 12-μm cryosections. Samples were analyzed at room temperature using a confocal microscope (LSM 510 META) equipped with Plan-Neofluar 40× 0.75 objective (all from Carl Zeiss MicroImaging, Inc.) and the software package LSM 5 Pascal version 3.2. In situ hybridizations were performed at room temperature using digoxygenin-labeled (Roche) RNA probes as previously described (). Photos were taken at room temperature with an AxioCam HRC camera mounted on an Axioskop microscope with 10× and 20× Plan-Neofluar objectives (all from Carl Zeiss MicroImaging, Inc.). Images were adjusted for contrast and image size using Adobe Photoshop 7.0.
Results were analyzed using the statistical package SPSS. The multiple test of ANOVA was used to test the null hypotheses of equality of means of different groups in experiments with multiple conditions. The error probabilities of particular pairwise tests were corrected by Bonferroni. A test was used to compare the means of experiments with only two conditions. Probabilities are summarized as asterisks above the graphs as follows: *, P < 0.05; **, P < 0.01.
For endogenous coimmunoprecipitations and for whole-cell lysates, PC12 cells were lysed on ice in IP buffer (50 mM Tris, 100 mM NaCl, 1 mM EDTA, and 1% vol/vol Triton X-100, pH 7.4) containing protease and phosphatase inhibitors (10 μg/ml Leupeptin, 10 μg/ml Aprotinin, 5 μg/ml Pepstatin A, 1 mM Benzamidine, 0.5 μg/ml Microcystin LR, and 1 mM NaVO; all from Sigma-Aldrich). For immunoprecipitations of Myc-tagged NOMA-GAP, lysates were prepared on ice in RIPA buffer (50 mM Tris, 150 mM NaCl, 1% vol/vol NP-40, 0.1% wt/vol SDS, and 0.5% wt/vol deoxycholate, pH 8.0) containing protease and phosphatase inhibitors. Immunoprecipitations were performed at 4°C for 1–2 h using equal amounts of protein (determined by Bradford assay) and using antibodies precoupled to protein G–Sepharose 4 Fast Flow (GE Healthcare) or, for Myc immunoprecipitations with anti-Myc agarose beads (Sigma-Aldrich), and subsequently washed thoroughly with lysis buffer. Cdc42 pull-downs were performed as described previously () using recombinant GST-PAK CRIB (see Fig. S4). PC12 lysates for pull-downs were prepared on ice in pull-down buffer (50 mM Tris, 100 mM NaCl, 1% vol/vol Triton X-100, 0.1% wt/vol SDS, 0.5% wt/vol deoxycholate, and 1 mM DTT, pH 7.4) containing protease and phosphatase inhibitors, incubated for 30 min at 4°C with GST-PAK CRIB precoupled to Glutathione-Sepharose 4B beads (GE Healthcare), and then washed thoroughly in IP buffer containing 1 mM DTT and protease inhibitors. Images were processed and quantified using Adobe Photoshop 7.0.
For SHP2 phosphatase assays, NIH3T3 cells were lysed on ice 24 h after transfection in Hepes buffer (25 mM Hepes, 100 mM NaCl, 1% vol/vol Triton X-100, 40% Glycerol, and 1 mM EDTA, pH 7.0) containing protease inhibitors, clarified by centrifugation, and incubated with recombinant GST-tagged wild-type SHP2 at 4°C (see Fig. S4). GST-SHP2 was recovered by incubation with Glutathione-Sepharose beads. SHP2 activity was assayed at room temperature on para-Nitrophenylphosphate (pNPP) as described previously (). Elk1 transactivation assays were performed using the PathDetect Elk1 trans-Reporting System (Stratagene). Firefly and renilla (for normalization) luciferase activities were determined using the Dual-Luciferase Reporter Assay System kit (Promega) 24 h after transfection of RK13 epithelial cells.
Plasmids were cotransfected into yeast ( Y190, AH109, and L40 strains; CLONTECH Laboratories, Inc.) as previously described ().), and plated on selection (−His-Leu-Trp+/− Met) or on total growth plates (−Leu-Trp) and grown at 32°C in a humidified incubator. A human brain cDNA library (HY4004AH; CLONTECH Laboratories, Inc.) was used for screening.
Fig. S1 shows NOMA-GAP family of multiadaptors. Fig. S2 shows expression of NOMA-GAP in the murine brain and eye. Fig. S3 shows expression of huNOMA-GAP deletion and mutant proteins in PC12 cells. Fig. S4 shows recombinant GST-SHP2 and GST-PAK CRIB proteins. Fig. S5 shows binding of RhoGTPase proteins to the NOMA-GAP RhoGAP domain. Online supplemental material is available at . |
Before every cell division, organelles are duplicated and segregated between mother and daughter cells. Peroxisome segregation is a regulated process in . After duplication, about half of the peroxisomes are retained in the mother cell, and the others are transported along actin cables to the growing bud in a class V myosin (Myo2p)-dependent manner (; ). The peroxisomal integral membrane protein Inp2p has been identified as the peroxisomal Myo2p receptor ().
The longstanding question of how peroxisomes multiply has been addressed by several groups whose findings have given rise to several models of peroxisome biogenesis (). In the first model, peroxisomes are derived from the ER and mature into functional peroxisomes (for reviews see ; ; ). In the second model, peroxisomes multiply by the growth and division of existing peroxisomes, with the ER providing the lipids (; ; ). Thus, the main difference between these two models of peroxisome multiplication is that according to the maturation model, peroxisomes are continuously formed de novo from the ER, whereas in the growth and division model, peroxisome numbers are maintained by the division of preexisting peroxisomes. There is evidence supporting both of these models, and a third model has been postulated that incorporates features of the first two. According to this model, ER-derived preperoxisomal structures mature into peroxisomes that subsequently divide (for reviews see ; ).
Recent studies have shown that the ER plays an essential role in peroxisome formation. cells lacking the peroxin Pex3p are devoid of any peroxisomal structures (). When Pex3p is reintroduced, peroxisomes are formed de novo from the ER. Pex3p expression is first detected in the ER, where it concentrates in an ER subdomain called the peroxisomal ER. Subsequently, peroxisomal ER structures are severed from the ER and fuse with each other to form a precompartment, which matures into import-competent peroxisomes (; ; ). To date, only Pex3p has been shown to reach peroxisomes via the ER in . Most other membrane proteins are thought to be imported after ER and matrix protein import, which relies on several integral membrane proteins and occurs at a later stage in the maturation pathway (for reviews see ; ).
The ER to peroxisome pathway is evolutionarily conserved and has been shown to give rise to peroxisomes de novo in mammalian cells (). A defect in this pathway results in Zellweger syndrome and perinatal death (; ; ; ; ).
Electron micrographs suggestive of the fission of peroxisomes in the yeasts () and () have been published previously. In some cell types, a complex peroxisomal reticulum is observed. This has been postulated to comprise a dynamic network that undergoes continuous fission and fusion (). The balance between these two processes would then determine whether peroxisomes are found as single entities or as networks. It was only with the introduction of live cell imaging that peroxisome fission was shown to occur unequivocally (; ). The fusion of peroxisomes has not been described.
A role in peroxisome fission has been suggested for the yeast dynamin-related proteins (Drps) Vps1p and, more recently, Dnm1p based on the low abundance and morphology of peroxisomes in cells lacking these proteins (; ). Mammalian and plant orthologues have also been implicated in the regulation of peroxisome abundance (; ; ; ). Drps and dynamin are large evolutionarily conserved GTPases implicated in the budding of transport vesicles and organelle fission/fusion (). Whether they act as mechanoenzymes or regulatory GTPases is unresolved (). Cells lacking Vps1p and Dnm1p display a single elongated peroxisome with a beads-on-a-string–like appearance. Mammalian and plant cells lacking the Vps1p orthologue display similar elongated peroxisomal structures. It has thus been suggested that these Drps play a role in peroxisome fission (; ). This is in line with the observation that both Vps1p and Dnm1p have been found associated with the peroxisomal membrane (; ). However, no direct evidence has been provided that Drps are required for peroxisome fission. Indeed, even in cells lacking Vps1p and Dnm1p, existing peroxisomes can divide, albeit at a later stage in the cell cycle, around the time of cytokinesis (; ).
In the absence of direct evidence for a role of Drps in peroxisome fission and with the existence of a de novo peroxisome formation process, it has been suggested that Drps are required for the de novo pathway rather than for the fission of existing peroxisomes. Drps were suggested to act either early in the de novo pathway at the point where Pex3p exits the ER (for reviews see ; ) or at the last step, fission to release mature peroxisomes, in the model whereby peroxisomes form de novo as an elongated and constricted structure (; for review see ; ). The questions that we have addressed in this paper are what is the contribution of peroxisome fission versus de novo peroxisome formation to the total number of peroxisomes and what role do the Drps play in peroxisome maintenance?
To address these questions, we have analyzed peroxisome multiplication in and the role of the Drps in this process. We have developed pulse-chase and mating assays to follow the fate of existing and de novo–formed peroxisome populations with time using fluorescence microscopy and live cell imaging. We show that in wild-type (WT) cells grown on a nonfermentable carbon source, the only mode of peroxisome multiplication is fission. We have developed an assay that follows the trafficking of Pex3-GFP to peroxisomes after the release of a block in transport out of the ER. We show that the ER-derived Pex3-GFP–containing structures do not give rise to de novo–formed peroxisomes in WT cells but instead fuse readily with existing peroxisomes. This delivery of ER-derived material to mature peroxisomes constitutes the only fusion event in WT cells. Surprisingly, it is only in cells that lack peroxisomes as a result of a defect in inheritance that peroxisomes arise de novo out of the ER. The process of forming peroxisomes de novo is much slower than peroxisome multiplication by fission and, in contrast to fission, does not require Drps: Drps are not required for exit of peroxisomal proteins from the ER. However, they are required for fission of existing peroxisomes. Our data support a peroxisome multiplication model whereby the ER provides essential membrane components allowing peroxisomal membrane growth and subsequent fission by Drps.
Expression was under control of the promoter. In the absence of galactose and in the presence of glucose, expression is undetectable both by Western blot analysis () and fluorescence microscopy (not depicted). In the presence of galactose, GFP-PTS1 is expressed and is imported efficiently into structures that are concluded to be peroxisomes, as they are absent in the peroxisome assembly mutant (not depicted).
We routinely induced expression for 3 h on galactose medium and shifted the cells to glucose medium to shut down expression. Analysis of the level of GFP-PTS1 in equal culture volumes at 2-h intervals after shutdown showed that GFP-PTS1 initially increased, after which it remained stable for another 4 h (). We are confident that the stable level of GFP-PTS1 is a result of tight shutdown of the promoter and not leaky expression balanced out by GFP-PTS1 breakdown. First, intraperoxisomal proteins are extremely stable, and no intraperoxisomal proteases have been described in . Second, the only known way to degrade intraperoxisomal proteins is via pexophagy, whereby complete peroxisomes are broken down. Deletion of the gene that blocks pexophagy (; ) does not affect GFP-PTS1 levels after shutdown ().
The number of peroxisomes per cell under a given condition is relatively stable, with rapidly growing cells on glucose medium containing 5–10 peroxisomes per cell. This means that with every cell division, the number of peroxisomes doubles. Peroxisomes have been suggested to multiply either by de novo formation from the ER or by fission of existing peroxisomes. We argued that if peroxisomes multiply by de novo formation from the ER, the fluorescence intensity of prelabeled peroxisomes will remain constant over time, but the number of fluorescent peroxisomes per cell will decrease with each cell division. On the other hand, if peroxisomes multiply by the fission of existing peroxisomes, the number of fluorescent peroxisomes per cell will remain constant, but the fluorescence intensity per peroxisome will halve with every cell division. In the experiment described in , the cells were dividing at a rate of once every 2 h. Analysis of cells 2 and 6 h after shutdown revealed that the number of fluorescent peroxisomes per cell remained constant, but the fluorescence intensity of individual peroxisomes markedly decreased (). These observations support the model whereby peroxisomes multiply by fission of preexisting peroxisomes.
However, if peroxisomes fuse and divide continuously, a decrease of fluorescence intensity per peroxisome does not rule out the possibility that peroxisomes form de novo, fuse with existing peroxisomes, and divide (thus diluting out the GFP). In mitochondria, fusion and fission are balanced in a dynamic equilibrium. Mitochondrial fusion was shown convincingly by the demonstration of contents mixing after cells with differentially labeled mitochondria were mated (). We have developed a similar assay to study peroxisome dynamics. has two haploid mating types: and . After mixing, cells of opposite mating type fuse with each other, upon which cytoplasmic contents, including organelles, are exchanged, and a diploid zygote buds from the fused cells. It takes ∼3 h between the fusion of haploid cells and budding of the zygote ().
In the first experiment, cells containing GFP-PTS1–labeled peroxisomes were mated with cells containing HcRed-PTS1– labeled peroxisomes. Mating cells were easily identified by their distinct dumbbell shape and the presence of both red and green peroxisomes. We found peroxisomes from each of the parental cells in the zygote, with some remaining in the parental cells. Also, some peroxisomes were exchanged between the two parental cells. Although an occasional coincidence was observed in the merged image, analysis of individual z stacks showed the peroxisomes to be in different planes. Therefore, no overlap between red and green peroxisomes was observed (), indicating there was no mixing of peroxisomal contents. Analysis of >100 mated cells at later time points, when large zygotes were observed (, fourth row), or even in diploid cells (not depicted) still did not reveal any contents mixing. We conclude from this that peroxisomes do not fuse. Furthermore, these results confirm that fluorescent reporter protein expression is shut down properly; had there been any residual expression, a low level of red would have been observed in the green peroxisomes and vice versa. We conclude that we are able to label distinct peroxisome populations and follow their fate with time.
In the next experiment, we tested whether all peroxisomes in a cell are import competent. WT cells expressing GFP-PTS1 were grown on galactose medium for 3 h, transferred to glucose medium for 2 h to shut down expression, and mated with cells expressing HcRed-PTS1. Peroxisomes formed after the shutdown of GFP-PTS1 will not contain GFP-PTS1 but will be identified after mating with cells by their ability to import HcRed-PTS1. 2 h after mating, once cytoplasmic mixing had occurred, all peroxisomes in mated cells contained both GFP and HcRed. We did not detect any newly formed peroxisomes (red only) or any peroxisomes that had lost import competence (green only; ).
Subsequently, we tested whether we could use our mating assay to study de novo peroxisome formation by mating a mutant with a mutant constitutively expressing HcRed-PTS1. Both mating partners are devoid of peroxisomes. Formation of peroxisomes by the complementation of cells has revealed that this process is slow. We obtained similar kinetics with our mating assay (). 5 h after mating, less than half of the mating cells displayed cytoplasmic labeling indicative of the absence of import-competent peroxisomes. However, after 7 h, 80% of mating cells had formed peroxisomes de novo.
We seeded the cells thinly onto a glucose medium–containing agarose pad on a microscope slide and incubated them at 30°C to allow colony formation. In this way, we can follow descendants of a single budding cell. Colonies were photographed after 6–8 h (8–10 h after shutdown). The GFP signal in WT cells was weak and had to be enhanced to be made visible. As seen in , all cells in the WT control colony contain peroxisomes that label with both GFP and HcRed. This indicates that peroxisomes have divided and segregated during cell division. However, GFP and HcRed do not overlap completely, and because the colonies could not be fixed, we were unable to determine whether any peroxisomes had formed de novo. Therefore, we repeated the aforementioned experiment in liquid culture and fixed the cells before imaging. All HcRed-labeled peroxisomes also labeled with GFP (), indicating that all peroxisomes are derived from the peroxisomes that were present before GFP expression was shut down.
We also used a mating experiment to test for the de novo formation of peroxisomes (). GFP-PTS1 expression was induced for 3 h in WT cells followed by a 5-h chase on glucose. Preperoxisomal structures, as proposed by the maturation model, would have ample time to mature into import- competent peroxisomes, and these de novo–formed peroxisomes would import HcRed-PTS1 supplied by the mating partner. They would not contain GFP as they become import competent after shutdown. No red-only peroxisomes were observed (). Therefore, we conclude that peroxisomes do not form de novo in WT cells.
Strikingly, the situation is completely different in cells, which are deficient in peroxisome segregation: in these cells, peroxisomes do not move from the mother cell into the bud, as they lack the Myo2p receptor (). In most of the colonies, there were only one or two cells that contained both GFP and HcRed-PTS1. None of the other cells in the colony contained GFP, which confirms that the original GFP-containing cells (from which the colony was derived) failed to segregate their peroxisomes. However, almost half of the cells in the colony had multiple peroxisomes that contained HcRed (). We have quantified this in , where we show that after 6–8 h on the agarose pad, 38% of cells in colonies had formed peroxisomes de novo, with 47% of cells still without peroxisomes.
Deletion of or results in a complete absence of peroxisomal structures (). When is reintroduced, peroxisomes are formed de novo from the ER. Pex3p is first detected in the ER, where it concentrates in an ER subdomain called the peroxisomal ER. Subsequently, peroxisomal ER structures are thought to be severed from the ER and to fuse with each other to form a precompartment, which matures into import-competent peroxisomes (). Exit of Pex3p from the ER requires Pex19p (; ). Pex3p has also been shown to pass through the ER in WT cells before ending up in peroxisomes. The de novo synthesis of peroxisomes is a slow process. However, the kinetics of Pex3p association with peroxisomes in WT cells is much faster (; ). This has been attributed to a rapid flux through the maturation pathway in WT cells and a slow flux during complementation of the mutant because in this mutant, the entire pathway has to be resurrected (). Our data support an alternative explanation for their findings. We propose that the reason for the different kinetics of Pex3p arrival in peroxisomes is that peroxisomes do not arise de novo from the ER in WT cells but that Pex3-GFP is delivered to existing peroxisomes.
We tested this hypothesis by labeling peroxisomes constitutively with HcRed-PTS1 and inducing the expression of Pex3p-GFP. Initially, we observed a very faint ER-labeling pattern (unpublished data), as has been seen previously (; ). At later time points, we found Pex3-GFP in all peroxisomes present in the cell (). This is in agreement with Pex3p being transported to existing peroxisomes.
Because it is difficult to visualize Pex3p passing through the ER, we developed an alternative assay whereby we first trap and accumulate Pex3-GFP in the ER and subsequently release the block to follow its trafficking. To this end, we performed a mating experiment. Because Pex3p accumulates in the ER in cells, we anticipated that upon mating with a WT cell, the soluble Pex19p in the WT mating partner will diffuse into the cell and initiate the exit of Pex3p-GFP from the ER. First, we pulse-labeled peroxisomes in WT cells with HcRed-PTS1 and shut down expression so that only the existing peroxisome population is labeled. In parallel, we pulse-labeled cells with Pex3p-GFP and shut down expression so that the Pex3-GFP that is trapped in the ER before mating is followed. As expected, before cell fusion, Pex3p-GFP labeling displayed a typical ER pattern in cells (). In addition to the ER labeling, there are some faint dots, which, during the 2-h chase, become more pronounced, whereas the typical ER labeling becomes weaker (). The dots have been seen previously () and are thought to comprise the peroxisomal ER. Upon cell fusion, Pex3p-GFP left the ER and associated with the (prelabeled) peroxisomes in the WT mating partner. All of the prelabeled peroxisomes acquired Pex3p-GFP, although to a varying extent. This experiment shows that ER-localized Pex3-GFP can be transported to existing peroxisomes soon after cell fusion (before zygote formation). At later time points after fusion, as indicated by the size of the zygote, the extent of Pex3-GFP association with peroxisomes increases (, second row) until it almost completely overlaps with the preexisting peroxisomes (, third row). These experiments show that Pex3-GFP is sorted from the ER to existing peroxisomes.
Mutants lacking the Drp Vps1p contain reduced numbers of peroxisomes (). A further reduction in peroxisome number is found in cells lacking both Vps1p and Dnm1p (). This could be caused by a lack of fission of existing peroxisomes, as has been suggested previously, but definitive evidence is still lacking. Because Vps1p is partially localized to the cytosol (; ), we hypothesized that we could use the mating assay to test whether Vps1p is involved in the fission of existing peroxisomes. For this purpose, we used cells devoid of peroxisomal structures ( cells) as a source of Vps1p. These cells were labeled with HcRed-PTS1. The mating partner was a mutant expressing GFP-PTS1.
Cells were pulse labeled by the expression of reporter constructs on galactose medium for 3 h and glucose medium for 2 h, mixed to initiate mating, and fixed after 2 h. Mated cells were easily identifiable by the presence of both HcRed and GFP and by their distinct morphology. In mated cells, HcRed colocalized with GFP-labeled peroxisomal structures. Hardly any of the fused cells showed the typical elongated peroxisomal structures found in cells, but, instead, the peroxisomes were small and numerous, as found in WT cells (compare mated with nonmated cells; ). In contrast, when cells were used as the mating partner with cells, peroxisomal structures remained elongated but still labeled with both PTS1 reporter proteins (). Time-lapse microscopy of a mating experiment clearly showed that the elongated structures undergo fission into multiple peroxisomes (). These experiments show that upon cell fusion, existing peroxisomes divide rapidly in a Vps1p-dependent process.
We wanted to test whether Dnm1 is involved in the fission of existing peroxisomes in a process analagous to that of Vps1p. Because Dnm1p is mainly membrane bound and no large cytoplasmic pool is available (; ), we reasoned that fission is unlikely to occur unless Dnm1p levels in the cell are increased. We overexpressed Dnm1p in cells and mated them with cells containing prelabeled peroxisomes. Upon mating, existing peroxisomes were divided into multiple small peroxisomes (). This shows that Dnm1p can substitute for Vps1p in peroxisome fission.
Furthermore, Dnm1p overexpression increases peroxisome number in a haploid mutant, showing that Dnm1p can substitute for Vps1p in haploid cells as well (unpublished data). We conclude that the reduced number of peroxisomes in cells is caused by a decrease in the fission of peroxisomes.
The molecular mechanisms involved in exit of Pex3p from the ER are still poorly defined. However, it has been postulated that the peroxisomal ER is severed from the rest of the ER and, via homotypic fusion and maturation, can form new peroxisomes. However, our data show that Pex3-GFP travels from the ER to existing peroxisomes. To investigate whether Drps are required for the transport of Pex3-GFP from the peroxisomal ER to peroxisomes, we analyzed the trafficking of Pex3-GFP in a mutant. We induced the expression of Pex3-GFP and saw it accumulate temporarily in structures different from peroxisomes in WT and cells. Whereas it remained in the ER in cells (), Pex3-GFP accumulated in prelabeled peroxisomes in WT cells () and in the typical elongated peroxisomes in cells (). Again, it was difficult to record the ER staining of Pex3-GFP in WT or cells because the residence time of Pex3p-GFP in the ER is so short and its level in the ER is very low at any point in time.
We conclude that the ER to peroxisome pathway is operational in cells and that Drps are not essential for the transport of Pex3p to peroxisomes. This is in line with the observation that a block in the ER to peroxisome pathway results in a complete lack of peroxisomes, whereas cells contain peroxisomes.
To investigate whether Drps are required for de novo peroxisome formation, we constructed a / mutant and compared it with the mutant using the same methodology as described in . First, GFP-PTS1 was induced in cells constitutively expressing HcRed-PTS1. Cells were seeded on a microscope slide and grown into colonies. As can be seen in , GFP-labeled peroxisomes were observed throughout the colony, indicating that the single peroxisome had segregated during cell division. However, in 17% of the dividing cells, either the bud or the mother cell was temporarily devoid of a peroxisomal structure. Compare the green and red panels in . This has been observed previously. In most cells, the single peroxisome extends from the mother cell into the daughter cell and is split in two around the time of cytokinesis (; ). The molecular basis of this fission event is unknown. As the peroxisomal structures moved during imaging, the overlap is not complete. However, cells grown in liquid culture followed by fixing and imaging show a complete overlap between GFP- and HcRed-labeled peroxisomes (unpublished data). These results indicate that peroxisomes segregate in cells and that peroxisomes do not form de novo, as no red-only peroxisomes were observed.
In the / mutant, however, the result is different. Only one peroxisomal structure in the whole colony was labeled with GFP, whereas half of the cells were showing multiple HcRed-labeled peroxisomes. We have quantified this in , where we show that after ∼8 h on the agarose pad, only 35% of / cells had formed peroxisomes de novo, with 53% of the cells still without peroxisomes. As expected, some of the other cells showed very weak cytoplasmic labeling (, red-only panel).
This shows us three things. First, multiple peroxisomes have formed de novo in those cells that failed to inherit peroxisomes. Second, in the cell with the single GFP-labeled peroxisome, no peroxisomes were formed de novo. Third, the preexisting peroxisomal structure failed to divide, showing that Inp2p is required for fission of the peroxisomal structure present in cells. We conclude that Drps are not required for the de novo formation of peroxisomes and that peroxisomes only form de novo if no peroxisomes are already present in the cell.
#text
Yeast strains were derivatives of BY4741 () or BY4742 () obtained from the EUROSCARF consortium. Double or triple gene deletions were made by replacing the entire coding sequence of the mutated genes with a marker ( or the hygromycin B phosphotransferase gene cassette that confers resistance to hygromycin B; ). The double mutants and were made by replacing the entire reading frame with the hygromycin B cassette. The gene was replaced by the cassette in the mutant to create . The mutants and and its derivatives are , whereas the other strains are . The strain used in was . and centromere plasmids were derived from Ycplac33 and Ycplac111 (). GFP-PTS1 is a peroxisomal luminal GFP marker protein appended with the well-characterized PTS1 (). A far-red peroxisomal luminal marker was made by appending a variant of the chromoprotein (HcRed) with the PTS1. As a source of HcRed, we used HcRed-Tandem with optimized yeast codon usage (Evrogen). As a marker for the ER to peroxisome pathway, Pex3-GFP was used ().
The constitutive expression of GFP-PTS1 and HcRed-PTS1 was under the control of the promoter and promoter, respectively. Dnm1p overexpression was achieved using the promoter. All constitutive expression constructs contained the terminator. Conditional expression constructs contained the promoter. To reduce the half life of the transcript, we replaced the terminator with the terminator (; ).
Cells were grown overnight in selective glucose medium and transferred to selective galactose medium at an OD of 0.1 to allow the induction of reporter proteins for 3 h. Depending on the experiment, cells were either prepared for mating or were grown on selective glucose medium for the time indicated in the figures and text.
For mating, after induction for 3 h on selective galactose medium, expression was shut down by switching cells to selective glucose medium for 2 h (unless indicated otherwise). The cells were collected by filtration onto a 0.22-μm nitrocellulose filter (type GS; 25-mm diameter; Millipore), and this filter was incubated cells side up on a prewarmed YPD plate at 30°C. 10 cells of each strain were collected per 25-mm filter.
After 2 h (or longer when indicated in text and figures), cells were harvested by vortexing the filter in selective glucose medium and fixed for 5 min by adding formaldehyde to 3.6%. Free aldehyde groups were quenched in 0.1 M ammonium chloride/1× PBS. Cells were imaged within 1 h of fixing, as loss of fluorescence intensity and increase of autofluorescence was seen in fixed cells left for extended periods. Growth of cells into colonies on an agarose pad was performed as described previously (). For each experiment, >100 cells were examined, and images are representative of the findings.
Live and fixed cells were analyzed with a microscope (Axiovert 200M; Carl Zeiss MicroImaging, Inc.) equipped with an Exfo X-cite 120 excitation light source, band pass filters (Carl Zeiss MicroImaging, Inc. and Chroma Technology Corp.), an α plan-Fluar 100×/1.45 NA or A-plan 40×/0.65 NA Ph2 objective lens (Carl Zeiss MicroImaging, Inc.), and a digital camera (Orca ER; Hamamatsu). Image acquisition was performed using Openlab software (Improvision) at 21°C. Live cells were imaged in minimal medium. Fluorescence images were collected as 0.2-μm z stacks, merged into one plane after contrast enhancing in Openlab, and processed further in Photoshop (Adobe) except when stated differently in the text or figure legends. Brightfield images were collected in one plane. In
, the brightfield image was added into the blue channel in Photoshop (Adobe). The level of the brightfield images was modified, and the image was blurred, sharpened, and blurred again before one more round of level adjustment so that only the circumference of the cell was visible.
Yeast glass bead lysates were prepared using a bead beater and 50-μm glass beads at full speed for 45 s in the presence of a protease inhibitor cocktail (Roche Diagnostics). Equal culture volumes were analyzed using Western blotting with anti-GFP (Roche Diagnostics). Standard methods were used for genetic manipulations. |
The Notch signaling pathway is of central importance for the regulation of developmental processes by mediating direct cell-to-cell communication between cells in a wide variety of developmental contexts in different species (; ; ; ; ). Products of the Notch, Delta, and Serrate (called Jagged in mammals) genes are crucial for this direct interaction between neighboring cells. Notch genes encode large transmembrane proteins that, at the extracellular surface of a cell, act as receptors for proteins encoded by the Delta and Serrate genes. Like Notch, Delta and Serrate are transmembrane ligands with a variable number of EGF-like repeats in their extracellular domains (; ; ). In addition, the Delta and Serrate proteins contain in their extracellular portion a conserved cysteine-rich region known as DSL domain (Delta, Serrate, lag-2), which is essential for ligand binding (, ). Generally, vertebrates contain several copies of genes encoding particular Notch pathway components. In the mouse, there are four genes encoding Notch proteins, three genes coding for Delta (, , and ), and two for Serrate (Jagged1 and Jagged2) proteins, respectively. Little is known about how these ligands interact with various Notch receptors in vivo, and whether the signals elicited by these interactions are quantitatively or qualitatively different.
Studies in zebrafish, , chicken, and mouse embryos have demonstrated an essential requirement for Notch signaling during somite formation and patterning. In mice, mutational analyses have shown that two ligands, () and (; ), are essential for normal somite formation and patterning. and are coexpressed throughout most of the presomitic mesoderm (PSM) and differentially expressed in the anterior and posterior compartments of newly formed somites (). However, despite the overlapping mRNA expression in the PSM, loss of or function leads to clearly distinct phenotypes. Somites in mutant embryos lack any detectable anterior–posterior (A-P) polarity, as indicated by the loss of expression. Loss of segment polarity is already evident in the anterior PSM. Somites are not fully epithelialized, their borders are not maintained, and expression in the PSM is down-regulated to barely detectable levels (; ). In contrast, null alleles of (; ) disrupt somite polarity such that expression appears randomized throughout somites instead of being restricted to the posterior compartment, and expression is readily detected, although transcriptional oscillation appears abnormal (; ). These qualitatively different phenotypes suggest that these ligands are not functionally equivalent in vivo. Support for this notion comes from a recent study that showed that DLL3 cannot activate Notch in vitro and suggested that DLL3 acts as an antagonist to DLL1 on the cell surface ().
Here, we demonstrate that the DLL1 and DLL3 proteins are not equivalent in mouse embryos. Replacement of the Notch ligand with in mice resulted in a phenotype indistinguishable from the -null phenotype, although expressed from the locus is functional. Similarly, in transgenic flies, , as well as , acted as a bona fide activator of Notch in the wing imaginal disc, whereas did not. Changing the ratios of and in vivo in mice or ectopic expression in flies did not provide genetic evidence for antagonism between DLL1 and DLL3 or repression of Notch activity by DLL3. Also, NICD was not up-regulated in the PSM of embryos lacking DLL3. In vitro analyses using chimeric DLL1-DLL3 proteins showed that differences in the DSL domains, EGF repeats, and intracellular domains (ICDs), respectively, contribute to their biochemical nonequivalence. In contrast to DLL1, DLL3 protein was predominantly detected inside the cell, including in the Golgi network. Our data prove that DLL1 and DLL3 have distinct functions in vivo; under physiological conditions, the proteins are differentially localized in the cell, and DLL3 might not act simply by antagonizing DLL1.
To express the coding region from the locus and simultaneously eliminate function, we generated mice that carried a chimeric “minigene” fused in frame into the ATG of the endogenous gene, analogous to the -null allele generated previously (). In the minigene, the coding sequence, either with or without a C-terminal HA tag, was linked at the 3′ end to genomic sequences of the gene containing exons 9–11 (). After processing of the primary transcript, the coding sequence is thus flanked by the 5′ and 3′ UTRs, which should generate transcripts with stability and properties similar to those of the genuine mRNA. As a control to ensure that this structural alteration at the locus had no adverse effects on expression of the minigene, we also generated mice that carried an analogous minigene version of targeted to the locus ().
The knockin ( and ) and the control alleles (
) were passed through the germ line of ZP3∷Cre females () to remove the floxed neo cassette that was included in the targeting vectors (). Heterozygous mice carrying either knockin allele did not show obvious external phenotypes or obvious malformations of the axial skeleton (; and not depicted).
allele were viable and fertile without any apparent phenotype (, m-o; and not depicted), indicating that the minigene was sufficient to compensate for the disrupted endogenous gene. In contrast, no homozygous or offspring were obtained from matings of heterozygous or mice, respectively, although the HA-tagged DLL3 protein was expressed (). This suggested that was unable to functionally replace .
To test whether the and cDNAs expressed from the locus generate functional proteins, we crossed heterozygous and mice, respectively, to mice carrying a null mutation of , (). disrupts A-P somite patterning and leads to severe malformations of the axial skeleton (). Compound heterozygous mice carrying one copy of the and the -null allele, respectively, i.e., mice that have only one functional copy of either gene are normal (unpublished data). Mice that are homozygous mutant for but are heteroallelic for the wild type and the knockin alleles also have one functional copy of and one copy of or expressed from the locus. Therefore, we reasoned that functional DLL3 protein generated from the knockin alleles should at least partially rescue the phenotype of homozygous mice. Indeed, homozygous mice that carried one copy of the or allele ( = 2 and 4, respectively) were normal or showed only subtle defects of the vertebral column (; and not depicted), in contrast to homozygous mice (). Consistent with normal axial skeleton development, cyclic expression and stripy expression of .1 was restored in
; embryos (). This unambiguously demonstrated that the and cDNA expressed from the locus generated fully functional DLL3 protein and indicates that the C-terminal HA tag does not interfere with the physiological functions of DLL3.
To address whether the DLL3 protein can rescue some aspects of the loss of DLL1 function, we analyzed homozygous and embryos. Embryos homozygous for the -null allele die around embryonic day (E) 11.5 and can be readily identified at E10.5 by large hemorrhages (). Homozygous embryos with either knockin allele were virtually identical to null mutants (), and A-P somite patterning, as well as cyclic gene expression, was similarly disrupted in homozygous and embryos (, compare j–l with p–r), indicating that cannot functionally replace in vivo. Collectively, our analyses demonstrate unequivocally that the DLL1 and DLL3 proteins are biochemically not equivalent and have divergent functions in vivo.
Based on in vitro experiments and overexpression in embryos, it was suggested that functions as an inhibitor of Notch signaling in a cell-autonomous manner (). If this reflects the physiological role of in the PSM of mouse embryos, where and are coexpressed in the majority of cells (), changes in the ratio of to might lead to either reduced or enhanced Notch signaling and defects in somite patterning and axial skeleton development. Indeed, about one third of mice heterozygous for the -null allele showed defects in individual vertebrae, such as minor fusions or reduction of laminae, split vertebral bodies, and reduced pedicles (). This indicates that somite patterning is sensitive to dosage (i.e., Notch activity) and raises the possibility that the increased gene dosage ratio of (2:1) in heterozygotes contributes to the haploinsufficiency phenotype. We therefore analyzed whether penetrance and expressivity of axial skeleton defects are increased in embryos, which carry three copies of but only one copy of , and thus have a ratio of 3:1. In 10 out of 28 skeletons, we found minor vertebral malformations similar to those observed in heterozygotes (unpublished data), indicating that the increase of dose did not enhance expressivity or penetrance of defects found in heterozygotes.
allele that still contained the neo cassette.
allele (
) attenuates expression and generates a hypomorphic allele that is lethal at birth in homozygotes (). Heterozygous mice, which have only one functional copy of , and thus most likely half the level of mRNA and protein, survive.
hypomorphs is below this level, and one hypomorphic
allele thus generates less than half of one wild-type allele (i.e., <25% of the total amount in wild type). We compared the phenotypes of mice that were heteroallelic for the hypomorphic and the null allele (
; <25% Dll1; two copies of ) with mice heteroallelic for the hypomorphic and the knockin allele (; <25% Dll1; three copies of ) and observed no obvious enhancement of the phenotype (, compare p–r with s–u). Thus, over the range of gene doses that we tested, we obtained no genetic evidence for antagonism of and during somitogenesis under physiological conditions.
To address more directly how DLL3 affects Notch activation in vivo, we analyzed the formation of the activated ICD of Notch1, NICD, in mouse embryos by immunohistochemistry. In wild-type embryos, Notch1 is activated in a sharp band in the anterior PSM, and posterior in variable patterns that reflect cyclic Notch activity (; ). Loss of DLL1 function abolishes formation of NICD in the PSM ( = 4; and not depicted; ), indicating that DLL1 is the major activator of Notch1 in the PSM. If DLL3 acts as an antagonist of DLL1, one would expect increased levels or expression of activated Notch1 throughout the PSM in mutants, similar to embryos without function (). In embryos lacking DLL3 ( = 14), NICD was detected in a fuzzy stripe in the anterior PSM and at the posterior end, but not throughout the PSM, and levels appeared reduced rather than increased (; and not depicted).
To further analyze the activities of and in vivo, we generated transgenic flies carrying UAS constructs of both genes and expressed them with help of the Gal4 system during wing development using Gal4. The activity of and was monitored by analyzing the activation of the Notch target Wg. During wing development, Wg is expressed in a stripe of cells along the dorsoventral boundary under control of the Notch pathway (). Expression of with Gal4 induces ectopic Wg expression along the A-P boundary () and inhibited Notch activation in the region of highest expression (, arrowheads), consistent with the known inhibition of Notch by coexpression of high levels of (). Similarly, expression induced Wg along the A-P boundary (), indicating that the mouse DLL1 protein can activate Notch in imaginal discs in vivo. This was also observed in the absence of endogenous (unpublished data). Likewise, DLL4, another mammalian ligand known to activate Notch (; ), induced Wg expression (), suggesting that mammalian DSL ligands that activate mammalian Notch in general can activate the Notch receptor. In contrast, Wg expression was not induced in wing discs expressing or (five independent lines tested for each construct; ; and not depicted), consistent with the inability of to substitute for in mice. In addition, a chimeric DLL1-DLL3 ligand that did not activate mammalian Notch (, construct A; see the following paragraph) did not activate Notch in the wing disc (), supporting the idea that the inability of DLL3 to activate Notch does not simply reflect species differences, although formally we cannot exclude this possibility. did not affect the normal expression of Wg induced along the dorsoventral border by endogenous and in the domain overlapping with the dorsoventral border (, arrowheads), indicating that expression of does not block Notch activation by the endogenous ligands. Identical results were obtained with HA-tagged or untagged versions of and (unpublished data).
The DLL1 and DLL3 proteins show considerable differences in their amino acid sequences. Compared with DLL1, the DLL3 protein has a divergent DSL domain, fewer EGF repeats, and altered spacing between some EGF repeats, and it lacks Lysine residues and a PDZ binding domain in its ICD (see Discussion). To analyze which of these structural differences contribute to the functional divergence, we generated various C-terminally flag-tagged chimeric cDNAs () and cell lines stably expressing the chimeric proteins. Cell lines were analyzed for expression by Western blotting to identify clones expressing similar levels of different protein variants for further analyses. Chimeric ligands A–H but not DLL3 (see the following section) were readily detected on the cell surface after surface biotinylation, followed by immunoprecipitation and Western blotting ().
A chimeric ligand, in which the N-terminal portion of DLL3, including the DSL domain, was replaced by the corresponding DLL1 sequence (, construct A) did not activate Notch neither in vitro () nor in wing discs (), indicating that the N terminus and DSL domain of DLL1 are not sufficient to confer Notch activating properties on DLL3. The inability of construct A to activate Notch could be due to the presence of the ICD of DLL3, or specific EGF repeats, or both. To distinguish between these possibilities, we tested a chimeric ligand consisting of the extracellular domain of DLL1 and the transmembrane domain (TM) and ICD of DLL3 (, construct B). This construct also did not activate Notch (), indicating that the ICD of DLL1 is essential, which was further supported by the inability of a DLL1 variant that lacked the ICD (, construct H) to activate Notch (). To further define features in the extracellular domain of DLL1 that are essential for Notch activation, we tested various constructs that contained the ICD of DLL1 and combinations of portions of the extracellular domains of DLL1 and DLL3.
EGF repeats 3–6 of DLL3 closely resemble repeats 5–8 in DLL1. We first tested a DLL1 variant that had EGF repeats 5–8 replaced by EGF 3–6 from DLL3. This ligand (, construct C) effectively activated Notch in vitro () and in wing discs (), indicating that the four proximal EGF repeats of DLL1 and DLL3 are functionally equivalent. We thus focused on the distal EGF repeats for further analyses. A DLL1 version that contained EGF repeats 1 and 2 from DLL3 (, construct D) did not activate, suggesting that either the spacing of DLL3 EGF repeats or their sequence is important. To distinguish these possibilities, we inserted the DLL3 spacer sequence between EGF repeat 1 and 2 of DLL1 (, construct E) or deleted the spacer between EGF repeat 1 and 2 of DLL3 (, construct F). Both ligands did not activate Notch, indicating that both spacing and sequence of the first EGF repeats are critical for DLL1 function. To test whether the DSL domain and distal two EGF repeats of DLL1 are sufficient to activate Notch, we replaced the six proximal EGF repeats of DLL1 by the EGF repeats of DLL3 (, construct G). This ligand activated Notch in vitro, though not at maximal levels (), and in wing discs (), suggesting that EGF repeats 1 and 2 of DLL1, or the spacing between these EGF repeats, are also essential for full activity.
In the course of analyzing chimeric ligands, we also generated CHO cell lines stably expressing DLL3. Surprisingly, it proved difficult to obtain cells that expressed DLL3 efficiently on the surface, as determined by surface biotinylation and subsequent immunoprecipitation and Western blotting, and several clones expressing DLL3 had only minor amounts on the surface. Even when DLL3 was detected on the surface, relative amounts of DLL3 were always significantly lower than those of DLL1 (, compare lane a with lanes b and c). Likewise, we detected no or minor amounts of biotinylated DLL3 on the cell surface of C2C12, HEK293, and CHO cells after transient transfection, although DLL3 protein was readily detected in cell lysates (Fig. S1, available at ). To further analyze the apparent difference in surface presentation, we transiently expressed DLL3 in cells stably expressing DLL1, and vice versa, and analyzed the surface expression of both proteins. In either case, DLL1 was readily detected on the surface, whereas DLL3 was not or only at minor amounts (, lanes d–i). Consistent with the surface biotinylation data, DLL1 was readily detected by immunohistochemistry on the cell surface of CHO cells stably expressing DLL1 (), as well as on the surface of transiently transfected CHO, HEK293, and C2C12 cells (Fig. S2, a–d; and not depicted), and on the apical surface of wing disc cells (), although surface expression patterns were variable, and DLL1 protein was also detected in vesicular structures in the cytoplasm. In contrast, DLL3 staining was mostly perinuclear ( and Fig. S2, e–h). In the perinuclear region, DLL3 colocalized with GM130, a marker for the cis-Golgi network (Fig. S2, i–k). When coexpressed, DLL1 and DLL3 colocalized in perinuclear structures but essentially not at the membrane ( and Fig. S2, n and q). Collectively, our data suggested that in cells expressing DLL1 and/or DLL3, DLL1 is largely on the cell surface and DLL3 is largely intracellular, including the Golgi apparatus. Similarly, in embryos in PSM cells, endogenous DLL1 was present on the surface and colocalized with membrane proteins detected by an anti-pancadherin antibody (), in addition to some intracellular DLL1 that colocalized mainly with GM130 (). Importantly, endogenous DLL3 was not detected on the membrane of PSM cells () but in intracellular punctae largely overlapping with GM130 (), indicating that the localization of DLL3 in the Golgi network occurs under physiological conditions. Both DSL proteins colocalized in some areas but were otherwise essentially nonoverlapping (Fig. S2, r–t). We also observed in PSM cells colocalization of DLL1 but not DLL3 with clathrin heavy chain that marks clathrin-coated vesicles and early endosomes (Fig. S2, u–z), further supporting differential subcellular localization of DLL1 and DLL3. Collectively, our data suggest that in vivo DLL3 accumulates in the Golgi network and only minor amounts, if any, are present on the surface of PSM cells.
The surface biotinylation results suggested that chimeric ligands differ with respect to their propensity to localize to the surface. To analyze in more detail how different portions of DLL1 and DLL3 affect the distribution of stably overexpressed chimeric ligands in the cell, we studied their localization on the cellular level by indirect immunofluorescence (). Chimeric ligands that contained the TM and ICD (TM-ICD) of DLL1 and at least the DLL1 N-terminal portion including the DSL domain fused to extracellular DLL3 sequences were detected on the cell surface, in addition to variable intracellular expression (). In contrast, the DLL1 N terminus alone (chimera J) was not sufficient to direct detectable surface expression (), similar to the extracellular domain of DLL3 fused to DLL1 TM-ICD, (chimera K; ). Because DLL1 lacking the ICD was also detected predominantly on the surface (), the DSL domain of DLL1 appears to be necessary to direct surface expression of these chimeric ligands. A chimera containing the DLL1 extracellular domain juxtaposed to the DLL3 TM-ICD (chimera B) was predominantly on the surface, although not evenly distributed (). Chimeras that contained the DLL1 N-terminal portion, including the DSL domain, and the DLL3 TM in the context of juxtaposed DLL3 intra- and extracellular sequences were found predominantly intracellular (). In contrast to cells expressing DLL1, or construct H or B, that consistently showed clear expression on the surface, surface presentation of most chimeric ligands was variable and not detected in all expressing cells. As expected, a chimera containing the DLL1 N-terminal portion without the DSL domain fused to DLL3 (chimera M) showed no detectable membrane localization (). Collectively, it appears that the DLL3 TM and adjacent extra- and intracellular sequences contribute to retention of chimeric ligands in intracellular compartments and localization of DLL3 in the Golgi network.
xref
#text
The ORF (with or without a C-terminal HA tag) was fused to a genomic SacI–EcoRI fragment containing part of exons 9, 10, and 11. A PGK-neomycin expression cassette flanked by loxP sites was introduced 3′ to the Dll3-Dll1 fusion. A 4.6-kb BamHI–KpnI fragment of genomic DNA upstream of the ATG fused in frame to , and ∼3 kb of genomic DNA downstream of the SalI site in exon 2 were included as regions of 5′ and 3′ homology, respectively. A diphtheria toxin A expression cassette was cloned upstream and downstream of the homology arms, respectively. The analogous Dll1ki targeting vector was cloned as previously described (). Linearized vector DNA was electroporated into 129Sv/ImJ embryonic stem cells and selected as described previously (). Correctly targeted clones were identified by PCR using primers derived from the neo sequence (TGTCACGTCCTGCACGACG) and genomic sequences downstream of the targeting vector (GGTATCGGATGCACTCATCGC). PCR-positive clones were verified by Southern blot analysis using external probes located 3′ and 5′ to the regions of homology in the vector and used to generate chimeric mice. The cassette was removed in the female germ line using ZP3∷Cre mice (; backcross generation N6 to 129Sv/ImJ).
Genomic DNA isolated from tail biopsies or yolk sacs was genotyped by PCR.
allele, neoF (TGGATGTGGAATGTGTGCGAG) and Dll1h3′B6 (AAGGGGAGAAGATGCTTGATAACC); for allele, Dll3pu1 (ACGAGCGTCCCGGTCTATAC) and Dll3pu2 (AGGTGGAGGTTGGACTCACC). After amplification, PCR products were cleaved with HaeIII and separated on 3% agarose gels (, 100 bp; and wild type, 65 bp).
E18.5 embryos were eviscerated and skinned, and skeletons were stained as described previously () with slightly longer incubation periods. Stained skeletons were stored and photographed in ethanol/glycerol (1:1) using a microscope (M420; Leica) with Apozoom 1:6 and Photograb-300Z version 2.0 software (Fujiflm).
Whole-mount in situ hybridization was performed following a standard procedure with digoxygenin-labeled antisense riboprobes () with minor modifications. Pictures were taken using the Leica M420 microscope with Apozoom 1:6 and Photograb-300Z version 2.0 software.
E10.5 embryos were collected in PBS, immediately fixed in MeOH/DMSO/30% HO (1:1:1) for 1 h on ice, and washed 3× 10 min and 2× 1 h in 50 mM NHCl at room temperature, followed by an incubation in TS-PBS (PBS, 10% FCS, and 1% Triton X-100) for 3× 10 min and 2× 1 h at 4°C. Embryos were then successively incubated with anti–cleaved Notch1 antibody (Val1744; Cell Signaling), biotinylated anti-rabbit antibody (Vector Laboratories), and streptavidin-HRP (NEL750; Perkin-Elmer) at a dilution of 1:100 in TS-PBS overnight at 4°C, respectively. Between antibody incubations, embryos were washed repeatedly with TS-PBS during the day at room temperature. For the color reaction, embryos were incubated 2× 10 min in solution A (100 mM Tris-HCl, pH 7.5, 0.1% Triton X-100, and 0.04% 4 Chloro-1 naphthol), 2× 5 min and 1× 10 min in solution B (solution A without Triton X-100), 1× 10 min in solution C (2 parts of 0.125% 4 Chloro-1 naphtol in 100% ethanol mixed with 3 parts of distilled water) followed by incubation in solution D (solution C with 0.006% HO), and stopped with 4% paraformaldehyde. Pictures were taken using the Leica M420 microscope with Apozoom 1:6 and Photograb-300Z version 2.0 software.
Embryos were dissected at E9.5, fixed in 4% paraformaldehyde overnight at 4°C, and stored in methanol at −20°C. Rehydrated embryos were washed three times in antigen unmasking solution (Vector Laboratories), heated to 100°C for 10 min, and allowed to cool to room temperature. Embryos were washed in water and cracked for 8 min in 100% acetone prechilled to −20°C and then rehydrated in water. Embryos were blocked overnight in 1% BSA dissolved in PBS-TR (PBS containing 0.1% Triton X-100) at 4°C. Primary antibodies diluted in block were incubated with embryos at 4°C for 2–3 d with gentle agitation. Embryos were washed six times in PBS-TR for 30 min each and then reblocked for 1–2 h at room temperature. Fluorochrome-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories and Invitrogen) were diluted in block and incubated with embryos at 4°C overnight with gentle agitation. The embryos were washed six times in PBS-TR for 30 min each, cleared by successive 10-min washes in 25% glycerol, 50% glycerol, and 70% glycerol. The posterior third of the embryos was dissected and flat-mounted sagittally in ProLong Gold antifade (Invitrogen). Fluorochromes used were Texas red, Alexa Fluor 488, FITC, Cy2, Cy3, and Cy5. Labeled cells were analyzed at room temperature by confocal laser-scanning microscopy using the LSM 510 Meta (Carl Zeiss MicroImaging, Inc.) connected to the inverted microscope (Axiovert 200M; Carl Zeiss MicroImaging, Inc.) with a Plan Apochromat 63×/1.4 oil differential interference contrast objective or using a TCS SP confocal microscope (Leica) using a PL APO 100×/1.4 objective (Leica). For image acquisition, LSM 510 and Leica Confocal Software v2.5 were used, respectively. For images acquired on the TCS SP confocal microscope, ImageJ was used to add scale bars. Pictures were processed and assembled using Photoshop and Illustrator CS (Adobe).
Immunocytochemistry was performed as described by and visualized at room temperature using a TCS SP confocal microscope or as follows. Cells grown on gelatin-coated coverslips were rinsed twice with PBS and fixed with methanol for 10 min at 4°C. After three washes with PBS, the cells were blocked with 5% donkey serum in PBS for 30 min at room temperature. Cells were incubated with the primary antibody for 1 h at room temperature and, after three washes with PBS, with the fluorochrome-conjugated secondary antibody (Dianova; Invitrogen). After washing, the coverslips were mounted in Gel/Mount (Biomeda) or ProLong Gold antifade. Texas red–, FITC- and/or Alexa Fluor 488–labeled cells were analyzed at room temperature by confocal laser-scanning microscopy using the LSM 510 Meta connected to the inverted microscope Axiovert 200M with a Plan Apochromat 63×/1.4 oil differential interference contrast objective. Images were processed using LSM 510 software. Pictures were processed and assembled using Photoshop and Illustrator CS.
The pTracer-Dll1Flag plasmid (a gift from S. Chiba, University of Tokyo, Tokyo, Japan) was modified by inserting an IRES-neo cassette after the Dll1Flag ORF and served as a vector (pTracer-IRESneo) for expression of flag-tagged Dll1, Dll3, and chimeric ligands. Chimeric ligands were generated by conventional cloning methods. Junctions between the and sequences were created without changing the amino acid sequence by PCR mutagenesis using primers with a restriction site–containing overhang. In the case of the chimeric ligands D and E, two gene fragments containing a deletion or an insertion between EGF1 and -2 were synthesized (GenScript). In addition, HA-tagged versions of and were cloned into pTracer. The integrity of all constructs was verified by sequencing. The junctions of and sequences in chimeric ligands are shown in Table S1 (available at ).
L-Dl19 cells stably expressing rDll1HA were provided by G. Weinmaster (University of California, Los Angeles, Los Angeles, CA). CHO cells stably expressing chimeric ligands were generated by transfection of CHO cells using Jetpei (BIOMOL Research Laboratories, Inc.) according to the manufacturer's instructions followed by neomycin selection. HeLa cells stably expressing Notch1 () were provided by A. Israël (Institut Pasteur, Paris, France).
HeLaN1 cells were transiently transfected with the Rbp-J luciferase reporter construct (Rbp)6-luc () using Jetpei, following the manufacturer's instructions. 10 transfected HeLaN1 cells were cocultivated on 6-well plates for 24 h with 10 CHO cells expressing ligands. Each CHO cell line was cocultivated four times in two independent experiments. Luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized to cotransfected Renilla luciferase activity (pRL-TK; Promega). Expression of chimeric ligands was verified by Western blot analysis.
Biotinylation-streptavidin pull down was performed essentially as described previously () or as described below. Cells were plated on 6-cm dishes and grown to confluence. Plates were washed three times with cold PBS (137 mM NaCl, 2.7 mM KCl, 4.3 mM NaHPO, 1.4 mM KHPO, pH 7.3, 1 mM MgCl, and 0.1 mM CaCl) and placed on ice with 500 μl PBS. 10 μl Sulfo-NHS-LC-Biotin solution (5 mg/ml in 0.1 M sodium phosphate buffer, pH 7; Pierce Chemical Co.) were added three times in 10-min intervals. After 30 min, the biotin solution was aspirated, and the plates were washed once with 50 mM glycine in DME and incubated for 30 min to quench the biotinylation reaction. Cells were washed twice with PBS and lysed with 400 μl RIPA (50 mM Tris/HCl, pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% DOC, and 0.1% SDS, supplemented with 2.8 μg/ml aprotinin, 0.15 mM benzamidine, 2.5 μg/ml leupeptin, and 2.5 μg/ml pepstatin A). Lysates were incubated for 30 min on ice and centrifuged for 10 min at 12,000 to remove cellular debris. The biotinylated proteins were precipitated with streptavidin agarose (Sigma-Aldrich) overnight at 4°C. The streptavidin agarose beads were washed three times with RIPA before resuspension in 2× sample buffer. Equivalent amounts of lysates and precipitates were subjected to SDS-PAGE and analyzed by Western blotting as described.
Antibodies used were as follows: HA (rat; clone 3F10; Boehringer), Wg (mouse; clone 4D4; Developmental Studies Hybridoma Bank), β-gal (rabbit, Cappel Research Products), Flag (mouse; clone M2; Sigma-Aldrich), PKCζ C20 (rabbit; Santa Cruz Biotechnology, Inc.), DLL1 (rabbit; Santa Cruz Biotechnology, Inc.), GM130 (mouse; clone 35; BD Biosciences), pancadherin (mouse; clone CH-19; Sigma-Aldrich), and clathrin heavy chain (BD Biosciences). Monoclonal antibodies against DLL1 were generated by immunization of rats with a peptide comprising amino acids 524–540 (PGPMVVDLSERHMESQG) of mouse DLL1 coupled to KLH or ovalbumin (Peptide Specialty Laboratories) subcutaneously and intraperitoneally with a mixture of 50 μg peptide-KLH, 5 nmol CPG oligonucleotide (Tib Molbiol), 500 μl PBS, and 500 μl IFAs. After a 6-wk interval, a final boost without adjuvant was given 3 d before fusion of the rat spleen cells with the murine myeloma cell line P3X63-Ag8.653. Hybridoma supernatants were tested in ELISA using the specific peptide or an irrelevant peptide coupled to ovalbumin. Peptide-specific mAbs were further characterized in Western blotting. mAb PGPM-1F9 reacted specifically with the DLL1 protein and was used for this study. In addition, guinea pig antisera were raised against the peptide CSPEHGYCEEPDE mapping to residues 222–234 of mouse DLL3 and affinity purified according to the manufacturer's instructions (Peptide Specialty Laboratories).
All constructs were cloned into the pUAST vector () and used to generate transgenic flies by -element–mediated transformation of embryos (). mDLL1-Flag was cloned through EcoRI + XbaI digestion from pTracerCMVDll1. mDLL3-Flag was cloned with EcoRI + NotI digestion from pTracerCMVDll3. pUAST–construct G plasmid was generated by a three-fragment ligation with the EcoRI–NotI and NotI–NotI fragments from pTracerCMV construct G into EcoRI + NotI digested pUAST vector. pUAST–construct C was cloned by an EcoRI + XbaI digestion from the pTracerCMVconstruct C. ratDLL1HA was cloned with XbaI from the pEF-Bos vector (pEF-Bos-ratDLL1HA vector was provided by G. Weinmaster). mDLL4-cDNA was received from Amgen and cloned by EcoRI digestion from pCR2.1 (Invitrogen) into the pUAST vector. DLL4HA was generated by replacing the PshAI–KpnI fragment of the DLL4-cDNA with a 351-bp PCR product generated by using following oligonucleotides: A-5′-CCAGCTCAAAAACACAAACCAGAAG-3′ and B-5′ AATTCTCTAGATCAAGCGTAATCTGGCACATCGTATGGGTAAGCTACCTCTGTGGCAATCACACA-3′. Activity of the Notch signaling pathway was revealed by monitoring the expression of Wg and the synthetic reporter construct Gbe+Su(H)m8-lacZ with antibody staining (). Immunostainings of wing imaginal discs were performed as described by using Alexa 488, 568, and 647 goat anti–mouse, Alexa 568 and 647 goat anti–rat, and Alexa 568 and 647 goat anti–rabbit antibodies, respectively (Invitrogen). Discs were mounted in VectaShield H-1000 (Vector Laboratories), and fluorochromes were visualized using an Axioplan2 with ApoTome (Carl Zeiss MicroImaging, Inc.), 10×/0.30 Plan-NEOFLUAR, 25×/0.80 Imm Korr Plan-NEOFLUAR, and 63×/1.4 Oil Plan-Apochromat lenses (Carl Zeiss MicroImaging, Inc.) at 25°C. Pictures were taken with an AxioCam HRm camera (Carl Zeiss MicroImaging, Inc.) and AxioVision (versions 4.4 and 4.6) software (Carl Zeiss MicroImaging, Inc.) and processed using Photoshop CS.
Fig. S1 shows surface biotinylation data of DLL3 expressed in HEK293, CHO, and C2C12 cells. Fig. S2 shows immunofluorescence detection of DLL3 and DLL1 in cell lines and PSM cells. Table S1 shows the junctions of and sequences in chimeric ligands. Online supplemental material is available at . |
High voltage–activated Ca channels in neurons consist of several subunits, a pore-forming α subunit (Caα), and several auxiliary subunits, including αδ and β (Caβ; ). Caβ subunits are involved in transport of the pore-forming α subunit to the plasma membrane (; ). Caβ subunits shield an ER retention signal on the α subunit, thereby guiding the pore-forming subunit to the target membrane ().
Caβ subunits also determine the biophysical properties of the Ca channel. The effects of the Caβ subunit family members on the biophysical properties are complex. Four family members have been described (Caβ). P/Q-type channels assembled with Caβ and β subunits in heterologous expression systems are fast inactivating in comparison with Caβ- and β-assembled channels (; ; ). Caβ has the most dramatic effects on the channel properties, causing the channel to inactivate very slowly. In addition, the Caβ subunit is unique because this subunit can be attached to the plasma membrane via its palmitoylated N-terminal protein domain ().
Several studies also suggest that at least certain Caβ subunit family members can target and function independently of the Caα subunits at the plasma membrane and other intracellular structures such as the nucleus. For example, these subunits may be involved in gene transcription () and the regulation of Ca oscillations and insulin secretion ().
Recently, the crystal structures of the Caβ core domains and the interaction domain between Caβ and Caα have been determined (; ; ). These studies revealed that the Caβ subunits belong to the membrane-associated guanylate kinase family containing Src homology type 3 and guanylate kinase domains (; ; ). A mutagenesis study of the Src homology type 3 and guanylate kinase domains showed that these domains regulate the inactivation of these Ca channels () but also suggested that Caβ subunits are involved in scaffolding and in the precise localization of Ca channel complexes to defined subcellular domains. Indeed, deletion of the nonconserved N and C termini of the Caβ subunit results in a loss of synaptic localization and presynaptic function (). In addition, the isolated N terminus of Caβ is capable of interacting with proteins of the vesicle release machinery ().
All Caβ subunits are expressed in the brain. Their subcellular distribution within neurons reveals that they are localized to neuronal cell bodies and dendrites. In addition, Caβ has been suggested to be localized to synaptic terminals (). However, its precise function for determining synaptic transmission and, in particular, synaptic plasticity is unclear. Therefore, the goal of this study is to analyze the distribution of endogenously and exogenously expressed Caβ subunits in hippocampal neurons and to correlate their distribution with their effects on synaptic transmission. Our results suggest that Caβ and Caβ subunits are targeted to presynaptic terminals, where they determine whether synapses facilitate or depress.
We first investigated whether hippocampal neurons in culture express endogenous Caβ subunits, as would be predicted by the presence of the endogenous high voltage–activated Ca channels (; ). We produced a peptide-derived antibody, which recognizes all β-subunit family members (pan-β antibody). As indicated in , the antibody recognized specifically Caβ subunits in hippocampal neurons, as demonstrated by antagonistic action of the epitope peptide (not depicted). Many, but not all, of the puncta colocalize with the synaptic markers synaptobrevin 2 () and synapsin 1 (). The subunits are expressed throughout the neuron with high and uniform staining detected in the soma and proximal dendrites, with more clustered distribution in synaptic areas. We next analyzed whether we could detect Caβ subunit–specific mRNAs in these neurons and whether we could see quantitative differences among the four different Caβ mRNAs. As a positive control, we used 18S RNA. Real-time PCR revealed the highest mRNA levels for the Caβ subunits and lower mRNA levels for Caβ (Caβ ≥ Caβ ≥ Caβ; ). The results indicate that all four Caβ subunits are expressed in hippocampal neurons in culture, which localize to the soma and to synapses.
We next analyzed whether the exogenous expression of the Caβ members resembles the endogenous distribution of the Caβ subunits as determined in and whether Caβ subunits can target to synaptic sites when expressed alone in neurons (). We found that the Caβ and Caβ subunits reveal a more homogenous distribution, whereas the Caβ and Caβ subunits are highly clustered (). When cells expressing these exogenous subunits were immunostained with the synaptic marker synaptobrevin-II or synapsin-1, we found that Caβ subunits revealed a higher degree of colocalization with synaptic markers than Caβ. Association of the Caβ subunits with the Caα subunits predicts that both proteins should be distributed in cytoplasmic as well as membrane regions, which we confirmed by Western blots from cytosolic and membrane fractions of whole rat brain using the pan-Caβ antibody (). Exogenous expression of the Caβ subunits revealed a similar distribution, with subtype-specific enrichment within either the cytoplasmic or the membrane fraction (). Caβ subunits are highly enriched in the membrane fraction, whereas Caβ was mostly concentrated in the cytoplasm (). Caβ and Caβ subunits were equally distributed in both fractions (). Because Ca channel Caβ and Caβ subunits reveal a mainly punctuate distribution within the neurons, we wanted to know whether we can detect Caβ subunits in presynaptic terminals on vesicles or vesicular structures (). The high expression levels of the GFP-tagged subunits allowed us to study their localization by immunoelectron microscopy. As a negative control, we used the untagged GFP overexpressed in hippocampal neurons. As shown in , Caβ and Caβ subunits were detected on vesicular structures () and close to presynaptic terminals (). We also observed that both Caβ and Caβ were attached to the plasma membrane (). In contrast, GFP was found only in the nucleus and outside of the nucleus but was not associated with vesicles or transported to the presynapse (unpublished data). The results suggest that both Caβ and Caβ subunits are transported to synaptic sites and to the plasma membrane, where they most likely associate with the Caα subunits to form channel complexes.
Caβ subunits determine the biophysical properties of the Ca channel. When expressed with the P/Q-type channel in oocytes or HEK293 cells, Caβ subunits determine the time course of inactivation in a subunit-specific manner. Caβ- and Caβ-assembled channels inactivate rapidly, whereas Caβ- assembled channels inactivate slowly (). Caβ-assembled channels inactivate with a time course that lies between Caβ and Caβ. The gating properties of the presynaptic Ca channels determine Ca influx into the presynaptic terminal and, therefore, determine transmitter release and synaptic plasticity, such as facilitation and depression.
We wanted to know how P/Q-type channels assembled with different Caβ subunits open and closed during action potential (AP) waveforms, which we obtained from cultured hippocampal neurons. We expressed Caα2.1 subunits together with the Caαδ and the various Caβ subunits in HEK293 cells and applied 30 APs to analyze how many channels would be opened during AP trains. To determine the proportion of open channels, we used the following protocol. Based on the voltage dependence of the activation of P/Q-type channels, we applied a 10-ms depolarizing test pulse to a test potential in which ∼100% of channels within the cells were open (, , ; ). This value is given by the amplitude of the tail current. We then compared the tail current elicited by the AP to the tail current elicited by the 10-ms depolarization to +100 mV. We were interested in three values. We wanted to know whether activation with the AP waveforms would reveal differences in the opening of the channels when assembled with different Caβ subunits. The results indicated that the AP opens between 55 and 65% of the channels. No considerable differences were observed between channels assembled with the different Caβ subunits ().
We next compared the ratio between the amount of channels opened by the first and the second AP (). By comparing this value, we gain information on differences on the influx of Ca through Ca channels into the presynaptic terminal, which may determine whether synapses facilitate or depress during paired pulses. No differences were detected between channels assembled with different Caβ subunits. We next analyzed whether a 20-Hz train of 30 APs leads to a decrease in channel opening, as would be expected from the inactivation of Ca channels during long, constant depolarizations (; ). When comparing the proportion of channels opened by the first AP relative to the amount of channels opened by the 30th AP, we found that currents mediated by Caβ- and Caβ-assembled channels are reduced by 10–15% (). In contrast, currents mediated by Caβ- assembled channels are reduced by 2% (), and currents mediated by Caβ-assembled channels increased by 5% (). Thus, P/Q-type channels assembled with different β subunits reveal substantial differences in the amount of channel opening during long AP trains.
It has been shown that the biophysical properties of P/Q-type channels depends on the cellular environment in which the pore-forming Caα subunit is expressed (). We found that the maximal current elicited by a 500-ms-long voltage ramp is shifted to more negative potentials (around 20 mV) in neurons expressing non–L-type channels in comparison with HEK293 cells expressing P/Q-type channels encoded by the Caα2.1, Caαδ, and Caβ subunits (). Therefore, Caβ subunit–mediated effects on presynaptic Ca channel (non–L type) inactivation may be shielded in neurons by, for example, neuronal-specific channel-interacting proteins. To show that the Caβ subunits (i.e., Caβ and Caβ) also change the biophysical properties of non–L-type channels in hippocampal neurons, we analyzed the Ca channel inactivation of somatic neuronal non–L-type channels. As shown in , the exogenous expression of Caβ and Caβ subunits reduce non–L-type channel inactivation in a subunit-specific manner. Caβ subunit expression leads to an increase in the non–L-type current during a 100-ms test pulse from −60 to 0 mV, whereas neuronal non–L-type currents in the presence of Caβ subunits do not change in size ().
Our results on the recombinant P/Q-type channels and endogenous neuronal Ca channels suggest that Ca influx into the presynaptic terminal should be altered during long 20-Hz AP trains but not for paired-pulse responses. We analyzed the effect of the Caβ subunits on paired-pulse facilitation (PPF) by comparing the first and second excitatory postsynaptic current (EPSC; defined as the paired-pulse ratio [PPR]) and analyzed the effect on synaptic depression by comparing the first and last EPSCs (averaged 27–30 EPSCs) within a 20-Hz stimulation protocol when 30 pulses were elicited in 4 mM of extracellular Ca (). Because we did not observe effects on synaptic transmission when Caβ and Caβ subunits were expressed in our initial studies (unpublished data), we only analyzed Caβ and Caβ subunit effects on synaptic transmission in the following experiments. According to our results regarding the effects of Caβ subunits on the inactivation properties of Ca2 channels, we found that Caβ subunits did not change the PPR as expected from the aforementioned biophysical analysis. However, Caβ subunits increased the PPR, leading to facilitation ().
We next analyzed whether the Caβ and Caβ subunits influence synaptic transmission during longer AP trains. The biophysical analysis predicts that in the presence of Caβ and Caβ subunits, Ca influx into the presynaptic terminal should be increased as a result of the noninactivating properties of the presynaptic Ca channels in comparison with Caβ and Caβ subunits. The increased Ca influx may cause more vesicle depletion (depression) and may influence asynchronous transmitter release, which has been shown to be proportional to the residual [Ca] (). Analysis of the synaptic responses during 30 20-Hz AP trains revealed that Caβ- and Caβ-expressing neurons show larger depression in comparison with wild-type neurons () Note that the amount of depression is related to the largest EPSC compared with the minimal EPSC at the end of the stimulus train. The largest EPSC in noninfected neurons and Caβ-expressing neurons is the EPSC elicited by the second pulse. Therefore, depression is significantly larger for Caβ (0.34 ± 0.01; = 14) as well as for Caβ (0.49 ± 0.03; = 15) in comparison with noninfected neurons (0.7 ± 0.01; = 15). To determine whether Caβ- and Caβ-expressing neurons reveal more vesicle depletion during AP trains, we compared the readily releasable pool size before and after 20-Hz train stimulations. As shown in (E and F), the pool size is substantially reduced in Caβ-expressing neurons (12 ± 3.5%) and is slightly reduced in Caβ-expressing neurons (9 ± 2.7%) in comparison with control neurons (3 ± 2.2%). However, the Caβ effects were not substantial. To further verify that Caβ- and Caβ-expressing neurons may increase the Ca influx into the presynaptic terminal, we analyzed the asynchronous release. We found that onset of the asynchronous release was much faster and the amount of asynchronous relative to the phasic release at the beginning of the AP train was increased in Caβ- and Caβ- expressing neurons in comparison with control neurons (). Although the total amount of phasic and asynchronous release () as well as the mean EPSC amplitude () were slightly increased in Caβ-expressing neurons in comparison with control and Caβ-expressing neurons, the differences were not substantial.
These aforementioned results support the idea that during AP trains, the Ca influx into the presynaptic terminal is larger in the presence of Caβ and Caβ subunits. A larger Ca influx into the presynaptic terminal during AP trains in Caβ and Caβ subunit–expressing neurons should also result in faster vesicle recycling (). To test this hypothesis, we repeated the experiments described in . We first analyzed the recovery of the readily releasable vesicle pool (RRP) after RRP depletion without 20-Hz stimulation trains applied during depletion. No differences were found for the recovery of the RRP regardless of whether Caβ or Caβ subunits were expressed in the neurons (). Also, recovery of the EPSC after RRP depletion was not different between neurons expressing or not expressing Caβ and Caβ subunits (), suggesting that exogenously expressed Caβ and Caβ subunits most likely do not interfere with the vesicle recycling. We next analyzed the RRP recovery after 20-Hz stimulation trains were applied during the initial sucrose application (). We confirmed the observation described by that the RRP recovery for all neurons analyzed (regardless of whether Caβ subunits were expressed or not) was accelerated by the 20-Hz stimulus train (). Interestingly, RRP recovery was faster in Caβ and Caβ subunit–expressing neurons in comparison with control neurons (τ without 20-Hz train stimulation: control = 11.4 s, Caβ = 9.1 s, and Caβ = 12.6 s; τ after 20-Hz train stimulation: control = 4.1 s, Caβ = 1.8 s, and Caβ = 1.6 s), suggesting again that Ca influx into the presynaptic terminal is increased during 20-Hz stimulation trains in Caβ and Caβ subunit–expressing neurons.
Although the exogenous expression of Caβ subunits determines the synaptic responses during long AP trains according to the biophysical properties of the assembled presynaptic Ca channels, the induced facilitation by Caβ during paired pulses cannot be explained by the biophysical properties of presynaptic Ca channel assembled with the Caβ subunit. However, this may suggest that in the presence of Caβ subunit, the Ca dependence of the vesicle release is altered, which may result in a reduced release probability (). Therefore, we compared the release probability of noninfected and Caβ- as well as Caβ-expressing neurons. The release probability can be examined by comparing the RRP with the number of vesicles elicited by a single AP. The RRP size is determined by application of a hypertonic sucrose solution (). As shown in , we found that the release probability in the presence of Caβ was reduced in comparison with noninfected and Caβ-expressing neurons. No differences in the mean RRP and EPSC size were detected between the neurons expressing different Caβ subunits, probably because only a small number of cells were analyzed.
The relationship between the Ca influx into the presynaptic terminal and the vesicle release is approximately given by the following equation: vesicle release ∝ [Ca], where the Hill coefficient is defined as the Ca cooperativity. The Ca cooperativity in many synapses is high (three to four), indicating that a small change in Ca influx can result in drastic changes in transmitter release. Thus, in our experiments, the Hill coefficient gives an indirect measure of the Ca influx through presynaptic Ca channels relative to the transmitter release. This means that a change in the number, localization, or organization of the presynaptic Ca channels most likely results in a change in Ca dependence of the transmitter release. Interestingly, in the presence of the Caβ subunits, the Ca-dependent transmitter release dose-response curve became more shallow, with a small change in the half maximal [Ca] concentration (EC) when compared with wild-type neurons or neurons exogenously expressing Caβ subunits (). Because the Caβ subunit particularly changed the cooperativity of the transmitter release, this result may suggest that Caβ is involved in organization of the Ca channel domains necessary for efficient vesicle release. For example, Caβ-assembled channels may be further apart from the release machinery. If this is the case, synaptic transmission in Caβ-expressing neurons should be more sensitive to the slow Ca buffer EGTA. Indeed, we found that when 10 mM EGTA was applied intracellularly or 50 μM EGTA-AM was applied extracellularly, the EPSC amplitude was substantially more reduced in Caβ-expressing neurons to 60 and 93%, whereas in Caβ-expressing neurons and control neurons, the EPSC amplitude was reduced only by 50 and 87–89% ().
sup
#text
Microisland and continental cultures of hippocampal neurons were prepared according to a modified version of published procedures (). In brief, hippocampal CA1-CA3 neurons from newborn rats (postnatal day 0–3) were enzymatically dissociated in 2 U/ml DME plus papain (Worthington) for 60 min at 37°C. Dissociated neurons were either plated onto astrocyte-covered poly--lysine/collagen (Sigma- Aldrich)-treated microislands that were prepared 3–5 d before plating (autaptic cultures) or were plated onto poly--lysine/collagen-treated coverslips that were placed invertly over astrocyte feeder cells (continental cultures). Neuronal cultures were grown in Neurobasal-A media (Invitrogen) supplemented with 4% B-27 (Invitrogen) and 2 mM Glutamax (Invitrogen) for 12–15 d.
Continental hippocampal cultures were prepared as described in the previous section and were infected with GFP-tagged Caβ subunits. 12–18 h after infection, neurons were fixed with 4% PFA and permeabilized with 0.2% Triton X-100 in PBS. Anti–synaptobrevin-II (SYSY) and antisynapsin (Invitrogen) antibodies were used to label the synaptic markers. Neurons were incubated with the primary antibody overnight at 4°C, washed, and incubated with AlexaFluor568-conjugated secondary antibody (Invitrogen) for 30 min at room temperature. Cells were embedded in Prolong Gold Antifade (Invitrogen). Images were acquired with a confocal microscope (LSM 510; Carl Zeiss MicroImaging, Inc.) mounted on an inverted microscope (Axiovert 200M; Carl Zeiss MicroImaging, Inc.). Images were acquired with a 63× oil plan Apo NA 1.4 objective at room temperature, processed with the built-in LSM 510 software (version 3.5; Carl Zeiss MicroImaging, Inc.), and analyzed by using VOLOCITY software (Improvision).
Polyclonal anti–pan-β antibody was raised by Harlan Bioproduct for Science according to a published procedure (). In short, a highly conserved peptide sequence presented in all β subunits (CESYTSRPSDSDVSLEEDRE) was synthesized, and a standard 112-d protocol was used for polyclonal antibody production (Harlan Bioproduct for Science). The specificity of the product was documented with Western blots using rat brain homogenate as well as homogenates of HEK293 cells expressing Caβ, Caβ, Caβ, or Caβ subunit and resulted only in bands with desired molecular weights.
For HEK293 cell recordings, HEK293 cells (tsA201 cells) were transfected with the Ca channel subunits Caα2.1 and Caαδ, with Caβ, Caβ, Caβ, or Caβ, and with GFP to identify positively transfected cells (molar ratio of 2:1:1:0.25). Whole cell recordings were performed as described previously (). For EPSC recordings, only dots containing a single neuron forming excitatory synapses (autapses) were used using an EPC-9 amplifier (HEKA). Recordings were performed at room temperature.
For EPSC measurements as well as for recordings of Ca currents in HEK293 cells, the extracellular recording solution contained 172 mM NaCl, 2.4 mM KCl, 10 mM Hepes, 10 mM glucose, 4 mM CaCl, and 4 MgCl, pH 7.3; the internal solution contained 145 mM potassium gluconate, 15 mM Hepes, 1 mM potassium-EGTA, 4 mM Na-ATP, and 0.4 mM Na-GTP, pH 7.3. For EGTA experiments (), the internal solution contained 10 mM potassium EGTA (Sigma-Aldrich), or 50 μM EGTA-AM (Invitrogen) was applied 15 min before recording to the extracellular recording solution. Currents were elicited by a 2-ms-long test pulse to 10 mV and recorded and analyzed as described previously (). For recordings using various extracellular Ca concentrations (extracellular [Ca]), solutions containing different extracellular [Ca] were applied directly onto the recorded neurons by using a fast-flow perfusion system (ALA Scientific). Non–L-type channel recordings in cultured hippocampal neurons were performed as previously described (; ). The internal recording solution contained 120 mM -methyl--glucamine, 20 mM tetraethylammonium-Cl, 10 mM Hepes, 1 mM CaCl, 14 mM phosphocreatine (Tris), 4 mM Mg-ATP, 0.3 mM NaGTP, and 11 mM EGTA, pH 7.2, with methanesulfonic acid. The external solution contained 145 mM tetraethylammonium, 10 mM Hepes, 10 mM CaCl, and 15 mM glucose, pH 7.4, with methanesulfonic acid. In addition, 1 μM tetrodotoxin (Sigma-Aldrich) and 5 μM nimodipine (Sigma-Aldrich) were added to the external solution to block voltage-dependent Na channels and L-type Ca channels. Non–L-type currents were elicited by 500-ms voltage clamp ramps from –60 to 90 mV with 1-min intervals and by 100-ms-long voltage pulses from −60 to 0 mV (). Here, capacitative and tail currents were subtracted after the experiment. The sizes of RRPs were measured according to published procedures (; ). In short, 500 mM sucrose was applied directly onto the recorded autaptic neurons for 4 s by using a fast-flow perfusion system (ALA Scientific). The EPSC and RRP charge was calculated by integrating the currents elicited by the single AP or the sucrose application.
The asynchronous and phasic release was calculated as described in . In brief, we estimated the phasic release by integrating the EPSC after each pulse within the 20-Hz stimulation protocol after subtraction of a baseline value measured 1 ms before each test pulse. The asynchronous release was calculated by subtracting the phasic release from the total integrated current for each EPSC. The holding current was subtracted before integration in every experiment. Statistical significance throughout the experiments was evaluated with analysis of variance using Igor Pro software (Wavemetrics). Standard errors are mean ± SEM.
10 cells of acutely dissociated hippocampal neurons were plated on poly--lysine–collagen–coated plates for continental culture as described in the Cell culture section. The total RNA was subtracted from 14-d in vitro–cultured neurons with the RNeasy Mini kit (QIAGEN) and purified with on-column DNase digestion using the RNase-Free DNase Set (QIAGEN). For RT-PCR, 1 μg RNA was used for reverse transcription with the Advantage RT-for-PCR kit (BD Biosciences) to generate 100 μl cDNA, and 3 μl of the final RT product was used for real-time PCR of each Caβ subunit. Real-time PCR quantification was performed on the iCycler Iq Detection System (Bio-Rad Laboratories) with CYBR green assay (Bio-Rad Laboratories). The DNA fragments of Caβ, Caβ, Caβ, and Caβ were amplified from cDNA with the following primer pairs: Caβ forward (GGCTGTGAGGTTGGTTTCAT) and Caβ backward (TGTCACCTGACTTGCTGGAG); Caβ forward (CATGAGACCAGTGGTGTTGG) and Caβ backward (CAGGGAGATGTCAGCAGTGA); Caβ forward (CAGGTTTGATGGCAGGATCT) and Caβ backward (GTGTCAGCATCCAACACCAC); Caβ forward (GAGAGCGAAGTCCAAACCTG) and Caβ backward (TCACCAGCCTTCCTATCCAC); and 18S forward (AAACGGCTACCACATCCAAG) and 18S backward (CCTCCAATGGATCCTCGTTA).
The specificity of RT-PCR products was documented with gel electrophoresis and resulted in a single product with desired length. The melt curve analysis showed that each primer pair had a single product- specific melting temperature. All primer pairs have at least 95% of PCR efficiency, as reported from the slopes of the standard curves generated by iQ software (version 3.1; Bio-Rad Laboratories). The PCR reactions used a modified two-step profile with initial denaturation for 3 min at 95°C, 40 cycles of 95°C for 15 s, and at 57°C for 25 s. Relative gene expression data were analyzed with the 2-ΔΔCT method ().
For immunoelectron microscopy of the cultured hippocampal neurons, 14-d in vitro neurons were infected with GFP-tagged Caβ or Caβ subunits with the Semliki Forest virus (SFV) expression system for 12 h before fixing with 4% PFA in 1× PBS for 20 min at 4°C. Cells were washed with 1× PBS containing 0.05% (vol/vol) Triton X-100, blocked with 10% goat serum (Invitrogen), and incubated with polyclonal rabbit anti-GFP antibody (Invitrogen) at 4°C overnight. The neurons were then rinsed five times with PBS/0.05% Triton X-100 for 3 min and incubated with goat anti–rabbit IgG conjugated with 10-nm gold particles (Electron Microscopy Sciences) for 2 h at room temperature on a shaker. After rinsing, neurons were fixed with 2% glutaraldehyde and 4% PFA in 0.1 M cacodylate buffer at 4°C overnight. After postfixing with 1% osmium tetroxide and staining with 1% uranyl acetate, neurons were dehydrated through an ethanol series from 50 to 100% ethanol and were transferred to propylene oxide, infiltrated with Embed 812 (Electron Microscopy Sciences) for 12 h, and hardened for 24 h at 60°C. Coverslips were removed, and 60-nm sections were cut on an ultramicrotome (Ultracut E; Reichert-Jung. The slices were recovered on Formvar-coated single slot copper grids and examined in a electronic microscope (JEM-1200EX; JEOL) at 80 kV.
For the brain slice immunoelectron microscopy, 100-nm-thick adult rat brain slices were prepared on a vibrating blade microtome (VT 1000S; Leica) and immediately infected with GFP-tagged Caβ or Caβ subunits with the SFV expression system for 12 h in an incubator with 5% CO at 37°C. The expression of the subunits was verified by the GFP fluorescent signals before the slices were fixed with 4% PFA in 1× PBS at 4°C overnight. The slices were rinsed with 1× PBS containing 0.05% (vol/vol) Triton X-100 five times for 3 min, blocked with 10% goat serum (Invitrogen), and incubated with a polyclonal rabbit anti-GFP antibody (Invitrogen) overnight at 4°C. Procedures and conditions for the second antibody, postfixation, and embedding, etc., were the same as for cultured neurons.
Rat Caβ, Caβ, Caβ, and Caβ were gifts from T. Snutch (University of British Columbia, Vancouver, Canada) and E. Perez-Reyes (University of Virginia, Charlottesville, VA). They were cloned in frame into pEGFP-C vectors (CLONTECH Laboratories, Inc.) and then into the Semliki forest virus vector pSFV1 (Life Technologies) for virus production. Thus, the GFP tag is located on the N terminus of the Caβ subunits.
About 8 × 10 hippocampal neurons were cultured on four collagen/poly--lysine–coated 100-mm culture dishes for 14 d and infected with GFP-tagged Caβ, Caβ, Caβ, and Caβ carrying virus for 13–16 h. Infected or noninfected cells were scraped in 0.32 M sucrose-TBS (0.15 M NaCl and 0.05 Tris, pH 7.4) containing 1× Complete Mini protease inhibitor (Roche) and were homogenized for 50 strokes with Dounce tissue grinder (Wheaton Millville) before promptly being loaded on top of freshly prepared 0.8 M/1.2 M sucrose-TBS gradient for centrifugation. Centrifugation was performed in a J-2-21 M/E ultracentrifuge (Beckman Coulter) at 3 × 10 rpm with a SW25.1 rotor for 45 min at 4°C. Equal volumes of the cytosol and membrane fractions were used for Western blots, which were performed according to standard procedures ().
Fig. S1 shows that the AP amplitude during 20-Hz stimulations is not reduced in noninfected or Caβ and Caβ subunits expressing hippocampal neurons. Fig. S2 shows that Caβ subunits expressed in hippocampal neurons do not change the relative contribution of N- and P/Q-type channels to non–L-type currents and EPSCs. Fig. S3 shows that the N terminus of Caβ interferes with synaptic transmitter release in hippocampal neurons. Online supplemental material is available at . |
Cell–cell contacts are dynamic structures. This is evident during the biogenesis of cultured epithelial monolayers, in which discontinuous early contacts are characteristically replaced by continuous, linear cell–cell adhesions as the epithelia mature (; ). The maturation of contacts is often accompanied by the appearance of specialized apical epithelial junctions and the establishment of apical–basal polarity (). This implies that the transition to coherent linear contacts may represent an important stage in epithelial biogenesis. Cadherin cell adhesion molecules are key features of cellular contacts at all stages of epithelial biogenesis (), and changes in cadherin function are often invoked to account for the dynamic regulation of cell–cell interactions. However, the precise molecular mechanisms responsible for this are not yet thoroughly characterized.
Classic cadherins such as epithelial cadherin (E-cadherin) exert their morphogenetic impact in close cooperation with the actin cytoskeleton (). Actin integrity is essential both for cadherin surface adhesiveness and for cells to make and maintain cadherin-based contacts. Moreover, the capacity of the actin cytoskeleton to support diverse mechanical events, notably surface protrusion, contractility, and anchorage, makes it an ideal partner for cadherin adhesion to mediate dynamic changes at cell–cell contacts. However, the precise molecular mechanisms responsible for this functional interaction are far from clear. Passive scaffolding of cadherin adhesion complexes onto cortical actin filaments, although conceptually simple, has not been empirically verified (). It is also not evident how any single molecular mechanism for cadherin–actin interactions can readily encompass the diverse functional states that may occur at contacts during epithelial morphogenesis and maturation. Instead, there is increasing evidence that cadherins may interact with a range of different actin effector machinery, including actin nucleators and cross-linkers ().
Actin-based motors are also major determinants of cytoskeletal organization and function. Indeed, to date, three members of the myosin superfamily—myosins II, VII, and VI—have been implicated in cadherin biology. Conventional nonmuscle myosin II has commonly been observed at cell–cell contacts (). Its impact is complex (), but, in epithelial cells, myosin II can be recruited and activated in response to E-cadherin ligation, where it serves to promote adhesion and the local accumulation of cadherin at cell–cell contacts (). Myosin VII was reported to associate indirectly with E-cadherin through the transmembrane protein vezatin (). Myosin VII is involved in dynamic adhesive events in (), but its physiological role in mammalian cadherin adhesion remains to be elucidated.
A role for myosin VI in cadherin function was first identified in the egg chamber of . demonstrated that myosin VI was necessary for border cell migration, a morphogenetic process that requires E-cadherin (DE-cadherin). Moreover, myosin VI and armadillo ( β-catenin) coimmunoprecipitated from ovarian extracts, and myosin VI and DE-cadherin each stabilized the expression of the other protein. Thus, myosin VI and DE-cadherin appeared to be biochemically and functionally interdependent. Myosin VI is unusual among myosins, as it is a minus end–directed motor (; ). Accordingly, it has often been thought to participate in vesicle trafficking, both in endocytosis, in which its direction of movement is postulated to promote the internal movement of cargo (; ; ), and in exocytosis, especially protein sorting in the biosynthetic pathway (). Membrane anchorage coupled with minus end–directed movement may also be a mechanism for myosin VI to exert protrusive force at the cell surface (). However, myosin VI shows limited processivity, and its stepping behavior in vitro is stalled by resistant force, causing it to remain bound to actin filaments for prolonged periods (). Thus, it has been alternatively suggested that myosin VI might act as an actin-based anchor under certain circumstances (). Despite these interesting possibilities, it remains unclear how myosin VI may contribute to the biological activity of DE-cadherin or what its role might be in mammalian cells.
We have now examined the contribution of myosin VI to E-cadherin function in mammalian epithelial cells. We report that myosin VI is recruited to E-cadherin adhesions at a late stage in the maturation of cultured epithelial monolayers. There, it supports cadherin adhesive strength, the morphological integrity of epithelial apical junctions, and the perijunctional actin cytoskeleton. We further demonstrate that myosin VI exerts this effect by recruiting and cooperating with the actin regulator vinculin.
We began by analyzing the subcellular localization of myosin VI and E-cadherin in MCF7 mammary epithelial cells, which underwent a characteristic morphological transition during culture. 24 h after plating, cells were confluent but displayed E-cadherin staining that was punctate and serrated at the cell–cell contacts, which then became consistently linear and continuous by 48 h (). We quantitated this by counting the number of discontinuities in the contacts, which decreased between 24 (14.01 ± 0.81 [±SEM]; = 30) and 48 h (3.00 ± 0.32; = 30). Strikingly, myosin VI was preferentially detected at cell–cell contacts after 48 h, when it coaccumulated with linear E-cadherin (, 48 h), whereas relatively little coaccumulated at 24 h. Preferential coaccumulation at these linear (cohesive) contacts was also detected in cells transiently expressing GFP-tagged porcine myosin VI (Fig. S2, available at ). Similarly, in MDCK monolayers, myosin VI became detectable only when cells formed extensive, continuous contacts (unpublished data), although this process was more rapid than in MCF7 cells.
The preferential localization of myosin VI in older, cohesive contacts was not caused by changes in the total cellular expression of this protein (), nor did it simply reflect increased local amounts of E-cadherin at contacts. Although fluorescence intensity at contacts of both E-cadherin and myosin VI increased with time (Fig. S3 a, available at ), ratiometric analysis revealed that 48-h-old cultures possessed substantially more myosin VI at contacts relative to E-cadherin than did 24-h-old cultures (). Therefore, the comparatively late recruitment of myosin VI to cadherin contacts occurred independently of changes in the levels of E-cadherin at those contacts.
We then used coimmunoprecipitation analysis to test whether myosin VI formed a biochemical complex with E-cadherin. Myosin VI was detected in E-cadherin immunoprecipitates, and E-cadherin was identified in myosin VI immune complexes (), confirming a previous study of (). This interaction persisted when actin filament integrity was disrupted by 100 μM latrunculin A (). Moreover, a biochemical complex between these two proteins was only detected after 48 h of culture, which is coincident with the appearance of myosin VI at cadherin contacts (). In contrast, Mena and WAVE, which are other actin regulators found at cell–cell contacts (; ), did not coimmunoprecipitate with E-cadherin (unpublished data). These findings indicate that myosin VI is preferentially recruited into E-cadherin complexes as cells form linear, cohesive contacts with one another.
To test whether E-cadherin adhesion was responsible for recruiting myosin VI to cellular contacts, we treated 48-h-old MCF7 monolayers with a function-blocking anti–E-cadherin antibody (SHE78-7). As shown in , myosin VI staining was displaced from cell–cell contacts as early as 15 min after exposure to the antibody, before the integrity of the cell–cell contacts was overtly disrupted. This was confirmed by the measurement of myosin VI and E-cadherin fluorescence, which showed that myosin VI intensity fell rapidly relative to E-cadherin staining in cells treated with SHE78-7 antibody ( and S3 b). Consistent with these immunofluorescence results, less myosin VI coimmunoprecipitated with E-cadherin 30 min after the addition of SHE78-7 antibody (). Together, these findings demonstrated that productive adhesion was necessary for E-cadherin to recruit myosin VI.
We then used RNAi to test the functional impact of myosin VI on E-cadherin. Two different siRNAs consistently reduced myosin VI expression by 80–100% for up to 7 d (). Because these duplexes yielded similar effects, data are only shown for one siRNA. Myosin VI protein expression was not affected by either scrambled oligonucleotides or a control duplex (, Cont) bearing a single nucleotide mismatch from the knockdown (KD) sequence, nor were β-tubulin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels affected by the siRNA ().
Depletion of myosin VI dramatically altered the morphology of cell–cell contacts. Whereas control cultures retained E-cadherin with a continuous linear distribution at the contacts, myosin VI KD cells showed punctate and serrated E-cadherin staining () similar to that seen in 24-h-old control cultures before the appearance of myosin VI (). Quantitation confirmed an increased number of discontinuities in contacts from myosin VI KD cells (14.50 ± 0.77; = 30) compared with control cells (2.60 ± 0.34; = 30). Similarly, the transient expression of a dominant-negative myosin VI mutant lacking the motor head domain (p–myosin VI–tail-GFP) markedly disrupted cell morphology and cell–cell contacts (unpublished data). This suggested that myosin VI might critically influence the morphological transition of cell–cell contacts over time.
Consistent with this, the reconstitution of myosin VI using porcine myosin VI (p–myosin VI–GFP), which is resistant to KD (), rescued the phenotype of myosin VI KD cells. Myosin VI KD cells expressing p–myosin VI–GFP regained linear E-cadherin staining at their cell–cell contacts compared with the punctate E-cadherin distribution of untransfected neighboring cells (). However, the transient expression of p–myosin VI–tail-GFP failed to rescue the myosin VI KD phenotype (), indicating that the actin-binding motor head was necessary for myosin VI to affect junctional maturation.
We then directly tested the adhesive impact of myosin VI using laminar flow adhesion assays that measure the cellular resistance to detachment from substrata coated with recombinant cadherin ligand (hE/Fc). Myosin VI KD resulted in an ∼40% reduction in adhesive strength compared with cells transfected with control duplexes (). In contrast, integrin-based adhesion to fibronectin was unaffected by myosin VI depletion (Fig. S4, available at ), suggesting that myosin VI preferentially affected cadherin adhesion. Myosin VI KD did not affect the total or surface levels of E-cadherin ( and Fig. S5, a and c), the cellular expression of catenins, or their binding to E-cadherin (Fig. S5, a and b).
Filamentous actin (F-actin), like E-cadherin, undergoes a distinct morphological transition as epithelial monolayers mature in culture. After 24 h, F-actin staining appeared relatively loosely organized in the perijunctional regions, whereas by 48 h, F-actin was densely accumulated at the cell–cell contacts (). This dense actin organization was lost in KD cells but not in control cells, most evidently at 48 h (). Expression of porcine myosin VI restored the dense perijunctional F-actin staining (). Interestingly, the total cellular levels of F-actin were not altered in KD cells (unpublished data), suggesting that myosin VI regulates the perijunctional organization of F-actin rather than affecting the global actin filament content of the cells.
E-cadherin adhesion is necessary for biogenesis of the specialized intercellular structures that comprise the apical junctional complex of epithelia (). Therefore, we next examined the impact of myosin VI depletion on the integrity of the junctional complex (). In control MCF7 monolayers, the tight junction marker ZO-1 displayed crisp, linear immunofluorescent staining throughout the apical-most regions of cell–cell contacts. In contrast, ZO-1 staining was discontinuous and fragmented in myosin VI KD monolayers. Similarly, staining for the desmosomal marker desmoplakin was distributed unevenly at contacts between myosin VI KD cells but was homogenous throughout contacts between control cells. However, despite this fragmentation, both ZO-1 and desmoplakin showed persistent staining in regions of contacts that retained E-cadherin (unpublished data), suggesting that the disruption of both tight junctions and desmosomes might reflect the loss of cohesive integrity between the cells.
In contrast, staining for vinculin, a marker of the zonula adherens (), was nearly completely lost from cell–cell contacts in myosin VI KD cells, even where E-cadherin persisted (). Vinculin also marks focal adhesions, but myosin VI depletion did not displace vinculin from these cell–matrix adhesive junctions, which is consistent with our observation that integrin-based adhesion was unaffected by myosin VI KD. Total cellular levels of vinculin were also not affected by myosin VI KD (Fig. S5 a), implying that myosin VI regulated the recruitment of vinculin to cadherin adhesions. As vinculin is reported to interact with the E-cadherin complex (; ), we investigated whether myosin VI participated in this biochemical interaction. Indeed, coimmunoprecipitation analysis identified an apparent complex containing E-cadherin, vinculin, and myosin VI in 48-h-old control MCF7 cells. Immune complexes generated with antibodies to any of these three proteins also contained the other two proteins (). However, the ability of vinculin to coimmunoprecipitate with E-cadherin was lost in myosin VI KD cells (). Therefore, myosin VI was necessary for vinculin to stably interact with E-cadherin in mature cell–cell contacts. The ability to bind both vinculin and E-cadherin resides in the myosin VI tail, as transiently expressed p–myosin VI–tail-GFP was able to coimmunoprecipitate these proteins (). Moreover, cross-linking by actin filaments cannot readily account for this interaction, as this myosin VI fragment lacks the actin-binding head domain.
Interestingly, vinculin-deficient F9 embryonal carcinoma cells were reported to show defects in cell–cell cohesion reminiscent of those we observed upon the depletion of myosin VI (). This led us to postulate that vinculin might be responsible for the impact of myosin VI on cell–cell integrity. To test this, we first examined the impact of depleting vinculin itself (). siRNA effectively reduced cellular vinculin levels (), most comprehensively at cell–cell contacts (not depicted). Importantly, junctional continuity was disrupted by vinculin KD (), with E-cadherin staining becoming punctate and serrated (14.40 ± 0.81 discontinuities/contact; = 30) compared with control transfections (2.77 ± 0.45 discontinuities/contact; = 30). Thus, the impact of vinculin KD on E-cadherin organization resembled that of myosin VI KD. However, neither myosin VI levels (Fig. S5 d) nor the ability of myosin VI to coimmunoprecipitate with E-cadherin were affected in vinculin KD cells (), nor were total cellular levels of E-cadherin or catenins perturbed by vinculin KD (Fig. S5 d).
Then, we asked whether the restoration of vinculin activity might affect the myosin VI KD phenotype (). Because full-length vinculin readily adopts an autoinhibited conformation (), we used membrane-directed fragments of vinculin that restored tight junction integrity in F9 cells (). We transiently expressed chimeric constructs bearing the head (α-cat/vinHead) or tail (α-cat/vinTail) region of vinculin () fused to the β-catenin–binding domain of α-catenin to target them to cadherin contacts and not to focal adhesions. Both proteins consistently localized to cell–cell contacts (). Strikingly, we found that both of these transgenes also restored the continuous linear organization of E-cadherin at contacts between myosin VI KD cells (). The α-cat/vinHead construct rescued contact integrity somewhat more efficiently than did the α-cat/vinTail construct, but both rescued much more effectively than did full-length α-catenin alone ().
Furthermore, we reasoned that if vinculin acts downstream of myosin VI, vinculin would be necessary for myosin VI to support the cohesive integrity of cell–cell contacts. As KD and reconstitution provides a strategy to isolate the activity of a specific protein, we tested whether exogenous myosin VI could rescue E-cadherin organization when vinculin as well as endogenous myosin VI were depleted (). Importantly, the transient expression of porcine myosin VI–GFP was unable to rescue linear cadherin contacts in double vinculin/myosin VI KD cells (). This indicated that vinculin was necessary for the reconstitution of myosin VI to be effective. Finally, we tested whether myosin VI could compensate for the loss of vinculin and found that the overexpression of myosin VI did not restore the integrity of contacts in vinculin single KD cells (). Together, these findings identify vinculin as a critical effector for myosin VI at cohesive cell–cell contacts.
Cadherin-based cell–cell contacts undergo distinct morphological transitions both in vivo and in vitro, even after cells have established contact with one another. The challenge is to define specific molecular mechanisms responsible for the morphogenesis of these cell–cell interactions. Our current work identifies myosin VI as an important factor that interacts with cadherin adhesion to control the process by which early discontinuous interactions become organized into cohesive, linear contacts as epithelial monolayers mature in culture. We found that myosin VI binds to E-cadherin and localizes at cell–cell contacts as they undergo the transition from discontinous to coherent and continuous. This appears to reflect a regulated recruitment process, as cellular levels of myosin VI do not change during this period. Importantly, the disruption of myosin VI function by RNAi or expression of a dominant-negative mutant perturbed the cohesive integrity of cell–cell contacts and reduced cadherin adhesion. Moreover, myosin VI depletion perturbed the integrity of tight junctions and desmosomes, which is consistent with the central role for E-cadherin function in junctional biogenesis (). In contrast, myosin VI had no effect on integrin-based cell adhesion to fibronectin. Thus, in cultured mammalian cells, myosin VI acts at a relatively late stage in epithelial maturation to generate cohesive cell–cell interactions.
Our findings support and extend earlier evidence from that implicated myosin VI in cadherin-based cell–cell interactions. Myosin VI cooperates with DE-cadherin to support morphogenetic movement in the fly egg chamber (). Specifically, myosin VI is expressed and forms a complex with DE-cadherin and armadillo in migrating border cells. Importantly, myosin VI is necessary for this cadherin-dependent form of cell-on-cell migration, which is consistent with our finding that myosin VI supports cadherin adhesion and aspects of cellular morphogenesis in mammalian epithelia. Similarly, the disruption of myosin VI perturbed intercellular cohesion, DE-cadherin localization, and dorsal closure during early fly morphogenesis (). Collectively, our findings in mammalian cells, taken with these earlier precedents in invertebrates, suggest an important conserved contribution of myosin VI to cadherin function.
How might myosin VI regulate cadherin contacts and adhesion in mammalian epithelia? This unconventional motor has been implicated in two broad cellular processes: membrane transport and actin filament organization. Myosin VI is recruited to clathrin-coated pits and persists on the subsequent uncoated vesicles, likely through interaction with several adaptor proteins (; ). This, taken with its minus end–directed movement, has suggested a role in endocytosis (). Myosin VI also associates with the Golgi apparatus and can support exocytotic transport (). Indeed, in MDCK cells, myosin VI was necessary for the basolateral delivery of proteins that contain tyrosine-based motifs sorted by the AP-1B clathrin adaptor complex (). However, myosin VI did not affect protein sorting that depended on dileucine motifs, including E-cadherin (). Similarly, we found no substantial change in either the total or surface levels of E-cadherin in myosin VI KD cells, nor could we detect changes in the transport of E-cadherin to the cell surface (unpublished data). Thus, although myosin VI may play a more subtle role in cadherin trafficking, this pathway does not readily account for our results.
Instead, we favor the notion that myosin VI participates in cadherin–actin cooperation. We found that the perijunctional actin cytoskeleton was clearly disrupted by myosin VI depletion. As epithelial monolayers matured, reorganization of the perijunctional actin cytoskeleton accompanied the appearance of linear, cohesive cadherin contacts, an initially loose distribution of F-actin being replaced by dense staining concentrated in the immediate vicinity of cell–cell contacts. In myosin VI KD cells, the dense perijunctional packing of F-actin was replaced by a looser organization, without any concomitant change in total cellular F-actin levels. Similarly, earlier studies reported that myosin VI can stabilize actin filament networks () and potentially filaments themselves (), characteristically promoting the dense packing and accumulation of filaments. For example, during spermatid individualization in , myosin VI is necessary to organize the actin cones that separate the syncytial spermatids. Notably, the density of filaments in the actin cones is substantially reduced in myosin VI mutant testes without apparent changes in filament turnover, suggesting that myosin VI participates in packing and organizing actin filament networks (). The impact of myosin VI on organization of the perijunctional actin cytoskeleton implies that a similar contribution may occur in epithelial cells.
Importantly, several lines of evidence identify the actin-binding protein vinculin as a downstream effector for myosin VI at cadherin adhesions. Myosin VI was necessary for vinculin to stably associate with E-cadherin in mature epithelial monolayers, which was assessed by both coimmunoprecipitation and immunofluorescence analysis. Vinculin was also selectively lost from cell–cell contacts but not focal adhesions in myosin VI KD cells. However, the ability of myosin VI to interact with E-cadherin was not affected by vinculin KD, implying that vinculin recruitment to E-cadherin is downstream of myosin VI. Vinculin KD disrupted the integrity of cohesive E-cadherin contacts in a manner similar to myosin VI KD. Membrane-targeted vinculin fragments effectively restored the cohesive integrity of cadherin contacts in myosin VI KD cells. This implied that reconstitution of vinculin function at E-cadherin adhesions could compensate for the global loss of myosin VI. Vinculin was necessary for myosin VI to regulate junctional integrity. Exogenous myosin VI could not rescue contact integrity in cells depleted of vinculin as well as endogenous myosin VI, indicating that vinculin was necessary for the reconstitution of myosin VI to be effective. Myosin VI could not compensate for the loss of vinculin, as the overexpression of myosin VI did not rescue contact integrity in vinculin KD cells. This argues against the possibility that vinculin and myosin VI are in parallel pathways. Collectively, these findings indicate that vinculin is an important effector for myosin VI at cadherin contacts.
Vinculin has long been recognized to accumulate at cell–cell contacts, where it is thought to incorporate into adherens junctions (). However, it is notable that previous studies implicated α-catenin in recruiting vinculin to cadherin adhesions (; ). We, too, found that vinculin stains in early cell–cell contacts (unpublished data) before myosin VI is readily detected, a process that therefore may entail α-catenin. Yet, in our experience, myosin VI was necessary for vinculin to stably incorporate into mature cadherin adhesions. This suggests that two mechanisms participate in localizing vinculin to cadherin adhesions, with myosin VI being dominant at a later stage than α-catenin.
Thus, our findings identify myosin VI and vinculin as part of a molecular apparatus responsible for generating the cohesive cell–cell contacts that distinguish epithelial biogenesis in vitro. We postulate that myosin VI and vinculin cooperate to reorganize the perijunctional actin cytoskeleton, leading to the generation of cohesive, linear cadherin contacts. Precisely how vinculin participates in this process has yet to be defined at a molecular level. Vinculin can both bind and bundle actin filaments (; ; ; ), so may provide a mechanism to organize and compact actin filament meshworks at cadherin adhesions. The actin-binding site has been mapped to the tail region of the vinculin molecule (). Therefore, it was interesting to note that the head fragment could also rescue contact integrity. The mechanism for this effect remains to be determined.
However, our data further indicate that myosin VI does not act only by recruiting vinculin into a cadherin-based complex. Although the myosin VI tail alone can coimmunoprecipitate both E-cadherin and vinculin and localize to cell–cell contacts (unpublished data), this fragment did not rescue the cohesive integrity of those contacts in myosin VI KD cells. Therefore, the actin-binding activity of the myosin VI head domain must functionally cooperate with vinculin for myosin VI to support cohesive cell–cell contacts. The nature of this cooperation remains to be determined. Our data do not exclude the possibility that other proteins participate in this myosin VI–based effector pathway. One speculative possibility is that myosin VI may serve as an actin-based anchor (), cooperating with vinculin's ability to organize actin in order to link cadherin complexes onto perijunctional actin filaments. Perhaps this cooperative interaction with actin also serves to stabilize vinculin at contacts, accounting for the role of myosin VI to stably localize vinculin in mature contacts.
Irrespective of the deep molecular mechanism, our findings highlight the concept that multiple actin regulators operate at cadherin adhesive contacts, and they reinforce the notion that the functional expression of individual mechanisms is likely to be tightly regulated by cellular context. Understanding the regulated recruitment of myosin VI will be important if we are to elucidate the varied molecular mechanisms that cadherin adhesion uses to regulate the actin cytoskeleton.
MCF-7 cells and CHO cells have been described previously (; ). For transient expression, cells were transfected with LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions and analyzed 24–48 h after transfection. hE/Fc was prepared and used as previously described ().
hE/Fc-based laminar flow adhesion assays were performed as described previously (). Integrin-based adhesion assays were performed using longer term adhesion assays () with alterations to substrata coating. In brief, poly--lysine–coated six-well plates were incubated overnight at 4°C with 10 μg/ml fibronectin (Sigma-Aldrich) in PBS and were incubated with HBSS containing 5 mM CaCl for 30 min at 37°C. Cells were isolated by incubation for 10 min in 0.01% (wt/vol) crystalline trypsin (Sigma-Aldrich) in HBSS containing 5 mM CaCl. Freshly isolated cells were allowed to attach to substrata for 90 min at 37°C in a CO incubator and were subjected to detachment by systematic pipetting. For this, five regions in each well (the four quadrants and center) were washed three times with 200 μl HBSS/CaCl delivered using a stand-mounted pipette. Cells remaining adherent to the wells were then incubated with 10 mg/ml MTT dissolved in dimethylsulfoxide and read at OD in a microplate reader. Cellular content in wells after pipetting was compared with the cellular content of wells prepared under identical conditions but not subjected to pipetting (yielding the total number of cells plated in each well).
Porcine myosin VI–GFP and p–myosin VI–tail-GFP constructs were both gifts from T. Hasson (University of California, San Diego, La Jolla, CA; ). Plasmids encoding α-cat/vincHead, α-cat/vincTail, and α-catenin–HA were gifts from M. Watabe-Uchida and M. Takeichi (Kyoto University, Kyoto, Japan; ). To construct α-catenin bearing a C-terminal T7 tag, the HA tag of pCA-huaEcat-HA-pA was replaced by cloning annealed oligonucleotides encoding the T7 tag into a unique Sal1 site. GFP-C1 control empty vector was purchased from CLONTECH Laboratories, Inc.
Primary antibodies used in this study are listed as follows: a rabbit pAb (pAbMVI) directed against the whole tail domain of myosin VI was generated using a GST fusion protein (provided by F. Buss, University of Cambridge, Cambridge, UK; ). Approximately 250 μg of protein was used with Freud's adjuvant per injection to immunize rabbits. Immunoblotting cell lysates from several cell lines, including MCF-7 cells, with anti–myosin VI sera detected an intense band at the expected molecular mass of 140 kD as well as several other minor bands at lower molecular masses (that are possible degradation products), whereas naive rabbit sera from the same rabbit did not detect this band (Fig. S1, available at )). Additionally, RNAi depletion of myosin VI resulted in loss of the band at 140 kD, and the reexpression of pMVIGFP restored this band at the expected molecular mass of 170 kD (). The other antibodies were obtained as follows: mouse (mAb) HECD1 against human E-cadherin (a gift from P. Wheelock, University of Nebraska, Omaha, NE; with permission from M. Takeichi); mouse mAb against GFP (Roche); mouse mAb against human β-tubulin (Transduction Laboratories); rabbit pAb against the ectodomain of E-cadherin (pAbE-cad; ); mouse mAb SHE78-7 against human E-cadherin (Zymed Laboratories); mouse mAb against the cytoplasmic domain of E-cadherin (Transduction Laboratories); rabbit pAb against human GAPDH (R&D Systems); mouse mAb against human α-catenin (Transduction Laboratories); mouse mAb against human p120-catenin (Transduction Laboratories); rabbit pAb against human desmoplakin (a gift from K. Green, Northwestern University School of Medicine, Chicago, IL); rabbit pAb against human ZO-1 (Zymed Laboratories); mouse mAb against human vinculin (hVIN-1; Sigma-Aldrich); mouse mAb against T7 tag (Novagen); and rabbit pAb against GFP (Invitrogen). Secondary antibodies were species-specific antibodies conjugated with AlexaFluor350, 488, or 594 (Invitrogen) for immunofluorescence or were conjugated with HRP (Bio-Rad Laboratories) for immunoblotting.
To identify myosin VI, cells were incubated with prepermeabilization buffer (0.5% Triton X-100, 10 mM Pipes, pH 6.8, 50 mM NaCl, 3 mM MgCl, and 300 mM sucrose containing 1× complete protease inhibitors [Roche]) for 5 min on ice followed by fixation with 4% PFA in cytoskeletal stabilization buffer (100 mM KCl, 300 mM sucrose, 2 mM EGTA, 2 mM MgCl, and 10 mM Pipes, pH 7.2) for 10 min on ice. Otherwise, cells were fixed in PFA on ice for 10 min and were permeabilized. Fixed and stained specimens were mounted in either Mowiol (Calbiochem) or polyvinylalchohol mounting medium containing -propylgallate (Sigma-Aldrich).
MCF7 monolayers were incubated with function-blocking antibody SHE78-7 directed against the E-cadherin ectodomain at a 2-μg/ml concentration for 15–30 min. The surface expression of E-cadherin was measured by the sensitivity to surface trypsinization as described previously (). Surface biotinylation was performed as previously described ().
Cells were lysed in 1 ml of lysis buffer (50 mM Tris, pH 7.4, 2 mM CaCl, 150 mM NaCl, and 0.5% NP-40). Protein complexes were immunoprecipitated with anti–E-cadherin pAb, anti–myosin VI pAb, anti-GFP pAb, or a monoclonal vinculin mAb bound to protein A–agarose beads and were separated by SDS-PAGE. To coimmunoprecipitate α- and β-catenin with E-cadherin, the lysis buffer was 50 mM Tris, pH 7.4, 2 mM CaCl, 150 mM NaCl, and 1% NP-40; for p120-catenin, the buffer was 20 mM Hepes, 50 mM CaCl, 0.1 mM EDTA, and 1% NP-40.
MCF7 monolayers seeded to 25% confluency were transfected with validated human myosin VI RNAi duplexes (5′-GGUUUAGGUGUUAAUGAAGtt-3′) and control duplexes (5′-GGUUUAGGUGUGAAUGAAGtt-3′; Ambion) at a 50-nM concentration using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's guidelines. Cells were then replated at 25% confluency on day 3 after transfection and were allowed to incubate overnight before retransfection with 50 nM RNAi duplexes to ensure sustained KD. Cells were studied 24–48 h after the second plating. Vinculin RNAi was performed similarly with 100 nM of validated human vinculin RNAi duplexes (5′-GGCAUAGAGGAAGCUUUAAtt-3′) and mismatched controls (5′-GGCAUAGACCTTGCUUUAAtt-3′; Ambion).
Fig. S1 shows a characterization of the myosin VI antibody. Fig. S2 depicts the localization of myosin VI–GFP at cell–cell contacts. Fig. S3 quantitates the fluorescence intensity of myosin VI and E-cadherin at cell–cell contacts as cultures mature or when treated with cadherin-blocking antibodies. Fig. S4 shows the effect of myosin VI RNAi on cell adhesion to fibronectin. Fig. S5 depicts the effects of myosin VI KD on the expression of cadherin, catenins and vinculin, the cadherin–catenin complex, and the surface expression of cadherin. It also shows the effect of vinculin KD on levels of cadherin, catenins, and myosin VI. Online supplemental material is available at . |
Hypoxia is a hallmark of diverse human malignancies, including breast cancer, head and neck cancer, prostate cancer, pancreatic cancer, brain tumors, and malignant melanoma (). The extent of hypoxia in a tumor may represent an independent indicator of poor prognosis (); however, the mechanisms by which hypoxia affects cancer progression remain incompletely understood. Under decreased O tension, the transcription factor, hypoxia-inducible factor 1α (HIF-1α), is stabilized. HIF-1α dimerizes with HIF-1β to activate a complex genetic program that is responsible for many hypoxia-associated changes in cell physiology, including expression of vascular endothelial growth factor (VEGF) and induction of angiogenesis (). Constitutive expression of HIF-1α renders tumor cells resistant to hypoxia and/or nutrient deprivation (). Thus, clonal selection of HIF-1α–overexpressing cells may increase the likelihood that tumor cells survive at implantation sites during metastasis ().
In hypoxic cancer cells, cell signaling pathways that support invasion and metastasis may be activated downstream of the hepatocyte growth factor receptor/c-Met () or the erythropoietin receptor (EpoR; ; ). Hypoxia also may induce expression of the urokinase-type plasminogen activator (uPA) receptor (uPAR; ; ). In breast cancer, uPAR expression is associated with a poor prognosis (; ; ). uPAR may promote cancer progression by multiple mechanisms (; ). By binding uPA at the leading edge of the migrating cell, uPAR organizes a cascade of extracellular proteases that facilitate cellular penetration of tissue boundaries (). uPAR also laterally associates with integrins in the plasma membrane, regulating the state of integrin activation (, ; ). Furthermore, uPAR functions in association with coreceptors, including integrins, receptor tyrosine kinases, and G protein–coupled receptors, to initiate signal transduction (; ; ; ; ).
uPAR-dependent cell signaling is regulated by ligands. Binding of uPA to uPAR activates FAK, c-Src, H-Ras, Akt, extracellular signal-regulated kinase (ERK)/MAPK, and myosin light chain kinase (; ; ; ). uPAR also binds directly to vitronectin, and this interaction promotes activation of Rac1 (; ). A second mechanism by which uPAR controls Rac1 activation is by regulating the interaction of αβ with fibronectin (). These uPAR-dependent cell signaling events impact cell migration and survival (; ); however, the role of uPAR-initiated cell signaling in cell physiology remains incompletely understood.
demonstrated that in kidney epithelial cells, α3 integrins and uPAR cooperate to induce phenotypic changes consistent with epithelial–mesenchymal transition (EMT). In cancer, EMT may be an important step leading to invasion and metastasis (). At the molecular level, EMT is characterized by loss of epithelial cell markers, including the cell adhesion protein, E-cadherin (), and acquisition of mesenchymal markers, such as vimentin (). A major mechanism by which E-cadherin is down-regulated in EMT is transcriptional repression by Snail (). Overexpression of Snail is sufficient to induce EMT and is associated with highly invasive tumors both in mice and humans (). In ovarian cancer cells, tumor hypoxia increases Snail levels and decreases E-cadherin expression (). Absence of the von Hippel–Landau tumor suppressor in renal carcinoma cells increases the level of HIF-1α and is associated with decreased expression of E-cadherin (). These studies suggest a relationship between tumor hypoxia and EMT. Our objective was to further elucidate this linkage.
In this study, we demonstrate that hypoxia induces diverse molecular, phenotypic, and functional changes in breast cancer cells that are consistent with EMT. We also show that cell signaling factors, which are thought to be involved in EMT, are activated in hypoxic cancer cells, including ERK/MAPK, Rac1, Akt, and glycogen synthase kinase-3β (GSK-3β). The EMT-associated phenotypic changes and changes in cell signaling are due to hypoxia-induced uPAR expression and activation of uPAR- dependent cell signaling. Hypoxia-induced EMT is blocked by uPAR gene silencing and mimicked by uPAR overexpression in normoxia. Using a newly developed chick chorioallantoic membrane (CAM) model system, we show that hypoxic conditions promote cancer cell dissemination in vivo. We conclude that EMT may be induced in hypoxic cancer cells, by a mechanism that involves activation of uPAR-dependent cell signaling.
The concentration of O in conventional cell culture is 21%, corresponding to a O of 150 mm Hg, well above the O of well-perfused tissues in vivo (40 mm Hg; ). To better model the tumor microenvironment, particularly at the hypoxic core, we cultured breast cancer cells in an incubator adjusted to 1.0% O. shows that after 50 h in hypoxia, MDA-MB-468 cells, which typically appear epithelial with well-developed cell junctions, acquire a spindle shape and generally lose cell contacts. Immunoblot analysis shows that MDA-MB-468 cells express vimentin when cultured in 1.0% O.
EMT has been defined as a three-part process in which cells acquire a fibroblast-like morphology, down-regulate epithelial-specific proteins such as E-cadherin while simultaneously expressing mesenchymal proteins such as vimentin, and ultimately, digest and migrate through ECM (). To further test whether the morphologic changes in hypoxic MDA-MB-468 cells represent EMT, we performed immunofluorescence microscopy, examining the epithelial cell marker E-cadherin. As shown in , after 50 h at 1.0% O, E-cadherin was lost from the cell surface in most cells. Residual E-cadherin remained localized primarily in intracellular pools.
Localization of the transcription factor, Snail, was examined by costaining cells with DAPI. Under normoxic cell culture conditions (21% O), immunostaining for Snail was localized diffusely throughout the cytoplasm (). In contrast, in 1.0% O, Snail translocated to the nucleus. shows that vimentin immunostaining was substantially increased in MDA-MB-468 cells cultured in hypoxia, confirming the results of our immunoblot analyses. In control experiments, none of the described antigens were detected when primary antibody was omitted.
EMT is thought to promote cancer cell migration and invasion (). To study the effects of hypoxia on cell migration, we examined cells in Transwell chambers maintained in 1.0 or 21% O. The underside of each Transwell membrane was coated with FBS so that vitronectin was the primary protein adsorbed to the surface in a haptotactic gradient (). 10% FBS also was added to the bottom chamber as a chemoattractant. MDA-MB-468 cells were added to each Transwell in a single-cell suspension. As shown in , cell migration was increased 2.2 ± 0.1-fold (P < 0.001; = 21) in 1.0% O. Similarly, invasion of MDA-MB-468 cells through reconstituted 3D Matrigel matrices was increased 2.3 ± 0.2-fold (P < 0.001; = 10) in 1.0% O.
To confirm these results using a second experimental protocol, we treated MDA-MB-468 cells with CoCl, which stabilizes HIF-1α and mimics many of the cellular changes observed in hypoxia (; ). 100 μM CoCl increased MDA-MB-468 cell migration 1.8 ± 0.1-fold (P < 0.005; = 3) compared with vehicle-treated control cultures (). These results support a model in which hypoxia-induced EMT, in MDA-MB-468 cells, is associated with more aggressive cell behavior. To further test this model, we compared the ability of MDA-MB-468 cells to remodel fluorescein-labeled type I collagen in 1.0 and 21% O. The collagen was adsorbed to microscope coverslips and labeled with fluorescein before adding cells. As shown in , remodeling of the fluorescent collagen was substantially increased when the coverslips were maintained in 1.0% O.
We previously demonstrated that activation of EpoR-dependent cell signaling promotes migration of MCF-7 breast cancer cells in hypoxia (). Therefore, we tested whether Epo or EpoR are up-regulated in hypoxic MDA-MB-468 cells; however, this was not the case (unpublished data). Instead, MDA-MB-468 cells demonstrated substantially increased uPAR expression, as determined by immunoblot analysis (). uPAR expression also was increased when MDA-MB-468 cells were treated with CoCl (unpublished data). These results, taken in the context of our previous study (), suggest that receptor activation to promote cell migration and invasion in hypoxia may occur in a cell-specific manner.
We hypothesized that activation of uPAR-dependent cell signaling may be responsible for the molecular and morphological changes observed in MDA-MB-468 cells under hypoxic conditions. To test this hypothesis, first we examined the basal level of activation of ERK/MAPK and Rac1 in cells that were transferred to 1.0% O. shows that both signaling proteins were activated in 1.0% O or when the cells were treated with 100 μM CoCl. Because CoCl was less effective in activating ERK/MAPK and Rac1, compared with hypoxia, we also assessed the relative abundance of HIF-1α under both sets of conditions and determined that HIF-1α levels were increased more substantially in hypoxia. The extent of Rac1 activation in 1.0% O was statistically significant (1.7 ± 0.2-fold; P < 0.05; ). Higher levels of Rac1 activation were not anticipated, as we analyzed whole cell extracts and Rac1 activation is known to occur locally, at the subcellular level, in the migrating cell ().
To test whether uPAR is responsible for the increase in cell migration observed in hypoxia, we treated MDA-MB-468 cells with antibodies that block uPA binding to uPAR and inhibit uPAR-initiated cell signaling (). shows that the antibodies inhibited the increase in cell migration observed in hypoxia by 90 ± 7% (P < 0.005; = 6) without affecting cell migration in 21% O. Next, we applied a gene silencing approach. Stable subclones of MDA-MB-468 cells, in which uPAR is silenced, were generated using the pSUPER vector system to express short hairpin (sh) RNA. Two cell lines, sh-uPAR6 and sh-uPAR12 cells, demonstrated 86 ± 2 and 70 ± 9% uPAR knockdown, respectively, as determined by quantitative PCR (qPCR), compared with parental cells that were transfected with empty vector (). In 1.0% O, the degree of knockdown was 93 ± 1 and 78 ± 2% in sh-uPAR6 and sh-uPAR12 cells, respectively.
Under normoxic conditions, migration of sh-uPAR6 and sh-uPAR12 cells was not significantly affected compared with pSUPER cells; however, when the sh-uPAR6 and sh-uPAR12 cells were transferred to 1.0% O, the hypoxia-induced increase in cell migration was inhibited by >70% (). In Matrigel invasion experiments performed under normoxic conditions, uPAR gene silencing had a modest effect. sh-uPAR12 cells demonstrated a 47 ± 10% decrease in Matrigel invasion, compared with uPAR-expressing cells (P < 0.05; = 6). The decrease in invasion was less pronounced with sh-uPAR6 cells. However, once again, when the experiment was performed in 1.0% O, the effects of uPAR gene silencing were more profound. The hypoxia-induced increase in Matrigel invasion was inhibited by >80% in the sh-uPAR6 and sh-uPAR12 cells ().
Next, we tested the effects of uPAR gene silencing on Rac1 activation in hypoxia. MDA-MB-468 cells that were transfected with empty vector demonstrated increased Rac1 activation when cultured in 1.0% O, as was observed with the parental cells; however, Rac1 was not activated by hypoxia in the sh-uPAR6 and sh-uPAR12 cells (). These results support a model in which uPAR plays an essential role in the regulation of Rac1 activation in hypoxia.
To further test the role of uPAR in hypoxia-induced MDA-MB-468 cell EMT, we examined the effects of uPAR gene silencing on cell morphology and E-cadherin immunostaining. When cultured in 1.0% O, the control cells, which were transfected with empty vector, adopted a fibroblast-like morphology and demonstrated loss of cell surface E-cadherin, duplicating the changes observed with the parental cells (). In contrast, sh-uPAR6 and sh-uPAR12 cells, which were cultured in 1.0% O, demonstrated unchanged epithelial cell–like morphology with abundant cell contacts. E-cadherin immunostaining was unchanged ().
We also performed studies to compare collagen remodeling by the uPAR-expressing and gene-silenced cells. The three cell lines were plated on coverslips coated with fluorescein- labeled type I collagen. Under normoxic conditions, considerable remodeling of the fluorescent collagen was not observed with any of the cells (). In 1.0% O, the uPAR-expressing pSUPER cells demonstrated substantial collagen remodeling, as was observed with parental cells. In contrast, the sh-uPAR6 and sh-uPAR-12 cells failed to demonstrate collagen remodeling. Collectively, these results demonstrate that uPAR is required for hypoxia-induced EMT in MDA-MB-468 cells.
Although activation of ERK/MAPK and Rac1 is associated with increased cell migration and with the other properties observed in hypoxic MDA-MB-468 cells, the effects of these cell signaling pathways on cell physiology are cell type specific. Factors that may determine whether activation of ERK/MAPK and/or Rac1 regulate cell migration include but are not limited to the integrin expression pattern of the cell and the composition of the substratum (; ). In MCF-7 cells, activation of endogenous EpoR cell signaling in hypoxia promotes cell migration by a pathway that requires ERK/MAPK. Dominant-negative (DN) MEK1 and PD098059 block the response (). However, as shown in , in MDA-MB-468 cells, DN-MEK1 failed to inhibit the increase in cell migration induced by hypoxia. Similarly, PD098059 failed to inhibit the increase in cell migration caused by CoCl (). We also studied the effects of DN-MEK1 expression on Matrigel invasion by MDA-MB-468 cells. Again, no inhibition was observed ().
Next, we tested the effects of DN-Rac1 on hypoxia-induced cell migration and invasion. As shown in , under normoxic cell culture conditions, DN-Rac1 did not affect cell migration and modestly decreased invasion (45 ± 16%; P < 0.05). In contrast, DN-Rac1 almost entirely neutralized the effects of hypoxia on cell migration and Matrigel invasion. We confirmed these results in separate experiments in which we treated cells with 100 μM CoCl (instead of hypoxia). DN-Rac1 inhibited the CoCl-induced increase in cell migration by 87 ± 11% (P < 0.05; = 3; unpublished data). We also showed that cells that were transfected to express DN-Rac1 and GFP demonstrate decreased collagen remodeling compared with cells that were transfected to express GFP alone (). Thus, DN-Rac1 inhibited changes in three separate functional properties that were associated with EMT in MDA-MB-468 cells under hypoxia.
Diverse cell signaling factors have been implicated in EMT, including Rac1, c-Src, Ras, FAK, and PI3K–Akt (; ; ). These factors may operate in concert and possibly in a cell- specific manner. Because recent studies indicate that the PI3K–Akt pathway is activated by uPA binding to uPAR (; ), we explored the role of Akt in hypoxia-induced EMT. As shown in , when MDA-MB-468 cells were cultured in 1.0% O, the level of phosphorylated Akt (p-Akt) was substantially increased.
In EMT, an important function of Akt is regulation of GSK-3β activity. Akt-induced GSK-3β phosphorylation results in GSK-3β inactivation. Otherwise, GSK-3β stabilizes epithelial cell morphology by targeting Snail. GSK-3β regulates Snail activity by inhibiting Snail expression, promoting Snail degradation, and decreasing nuclear localization (; ). shows that in hypoxia, the level of phosphorylated GSK-3β in MDA-MB-468 cells was increased. Snail expression also was increased at both the protein and mRNA levels. Snail mRNA was increased 2.9 ± 0.6-fold (P < 0.05; = 4) after exposure to 1.0% O for 15 h, as determined by qPCR ().
To determine whether uPAR is necessary for Akt activation in hypoxia, we treated MDA-MB-468 cells with synthetic peptide, α325, which disrupts uPAR–integrin interactions and blocks uPAR-initiated cell signaling (; ). shows that α325 blocked hypoxia-induced Akt phosphorylation. The scrambled version of α325 (scα325) had no effect. shows that 10 μM of the PI3K inhibitor, LY294002, blocked the increase in Snail protein level seen in hypoxia. LY294002 also preserved epithelial cell morphology and cell contacts in MDA-MB-468 cells that were exposed to 1.0% O for 50 h, as determined by phase-contrast imaging (). Equivalent results were obtained with 5 μM Akt inhibitor (unpublished data). Furthermore, LY294002 preserved cell surface E-cadherin in MDA-MB-468 cells in hypoxia (). These results suggest that uPAR-dependent Akt activation is essential for hypoxia-induced EMT. To further test this conclusion, we examined Snail mRNA levels in cells that were treated with LY294002 or vehicle for 50 h. LY294002 decreased Snail mRNA by 45 ± 8% in 21% O and by 75 ± 2% in 1.0% O (P < 0.05; = 3), completely neutralizing the effects of hypoxia on Snail mRNA expression ().
Hypoxia activates complex gene expression programs downstream of HIF-1α and other transcription factors (). Although the data presented thus far indicate that uPAR is necessary for EMT in hypoxic MDA-MB-468 cells, we wished to test whether uPAR alone induces EMT in MDA-MB-468 cells. To accomplish this goal, we overexpressed uPAR in MDA-MB-468 cells and studied these cells under normoxic cell culture conditions. Initially, we performed transient transfection experiments, introducing a full-length human uPAR expression construct in pcDNA together with pMAX-GFP, which encodes GFP. Control cells were transfected with pMAX-GFP alone. Migration and Matrigel invasion were studied. shows that transient transfection with uPAR increased cell migration by 2.1 ± 0.3-fold (P < 0.05; = 9) and Matrigel invasion by 2.0 ± 0.2-fold (P < 0.005; = 8).
To examine the effects of uPAR on morphologic and molecular parameters of EMT, stable cell lines that overexpress uPAR were cloned. shows two cell lines (C14 and C18) in which uPAR expression was substantially increased. Control cells that were transfected with empty vector (pcDNA) demonstrated equivalent levels of uPAR compared with parental cells; the uPAR is not apparent in the figure because of exposure time. uPAR overexpression in the C14 and C18 cells induced expression of vimentin and decreased E-cadherin expression, as determined by immunoblot analysis (). Phase-contrast imaging of the C14 and C18 cells in 21% O (normoxia) revealed loss of cell contacts (). Furthermore, in immunofluorescence microscopy studies, E-cadherin was substantially lost from the cell membranes of C14 and C18 cells and localized more diffusely in the cytoplasm (). These results indicate that uPAR overexpression independently induces EMT in MDA-MB-468 cells, without exposure to hypoxia.
To determine whether hypoxia may promote cancer cell dissemination in vivo, we modified the chicken egg CAM model of . MDA-MB-468 cells were stably transfected to express GFP. To promote tumor growth on the CAMs, the cells were suspended in Matrigel and applied to a 1-cm sterile cotton gauze that was placed on the CAM. shows a stereomicroscope image of a tumor that developed after 11 d. The GFP-expressing cells are shown in .
To model hypoxia, 25 μl of 100 μM CoCl vehicle was applied to the tumor-bearing gauze daily. qPCR studies, performed 4 d after initiating treatment, demonstrated a 3.0 ± 0.7-fold increase in VEGF mRNA expression (P < 0.05; = 3) by the CoCl-treated cells compared with controls (). The primers and probes used for these studies were specific for human VEGF. Human uPAR mRNA also was increased 2.6 ± 0.4-fold in the MDA-MB-468 cells that were treated with CoCl (P < 0.01; = 3). These results suggest that CoCl treatment induces changes in MDA-MB-468 cells on CAMs that are equivalent to those observed in cell culture.
To determine whether CoCl treatment modifies the propensity of MDA-MB-468 cells to disseminate in vivo, the eggs were killed 11 d after inoculation. The heart–lung block was isolated from each chick embryo, and the number of fluorescent cells or cell clusters was determined by fluorescence microscopy. In chick embryos that were treated with vehicle, the mean number of fluorescent cells/cell clusters detected was 9 ± 2 ( = 13; ). In chick embryos that were treated with CoCl, the mean was 25 ± 5 ( = 9). Thus, CoCl treatment induced a 2.7 ± 0.6-fold increase in cancer cell dissemination to the heart–lung block (P < 0.01). The complete process of cancer metastasis includes implantation, survival, and growth at the secondary site. Our CAM assays were not designed to assess these latter stages of the metastasis cascade because of the relatively short length of the experiments. Nevertheless, our results still suggest that conditions that model hypoxia in vivo may increase the propensity for cancer cell metastasis.
Increased uPAR expression was previously reported in MDA-MB-231 breast cancer cells under hypoxia (). To test whether uPAR expression is increased in hypoxia in diverse cancer cells, we screened a series of cell lines by qPCR. Significantly increased uPAR mRNA was observed when MDA-MB-468 cells were cultured in 1.0% O () as anticipated (P < 0.01; = 5). MDA-MB-231 cells also demonstrated significantly increased uPAR mRNA in hypoxia (P < 0.05; = 4), consistent with the study of . Other cell lines that demonstrated significantly increased uPAR mRNA when exposed to hypoxia included the following: A431 epidermoid carcinoma cells (P < 0.05; = 3), ZR-75-1 breast cancer cells (P < 0.005; = 4), and SCC15 squamous cell carcinoma cells (P < 0.05; = 3). Increased uPAR mRNA was not observed under hypoxic conditions in MDA-MB-435 cells or SK-BR3 cells. MCF-7 cells express extremely low levels of uPAR in 1.0 or 21% O, possibly explaining the predominant role of the EpoR in hypoxia in this cell line (). The comparisons shown in are accurate to the extent that expression of the housekeeping gene, hypoxanthine phosphoribosyltransferase 1 (HPRT-1), is conserved in the various cells under the conditions of the experiment.
Next, we sought to determine whether EMT is induced in the cells in which uPAR expression is increased in hypoxia. As MDA-MB-231 cells already exhibit a mesenchymal phenotype, we excluded this cell line from our analysis. A431 cells did not demonstrate EMT-associated changes in hypoxia. In contrast, both the ZR-75-1 and SCC15 cells demonstrated phenotypic and molecular changes consistent with EMT. shows that the ZR-75-1 cells lost cell contacts and grew more independently in 1.0% O. The vimentin level was increased in hypoxia, as determined by immunoblot analysis. The morphology of SCC15 cells in normoxic cell culture was slightly different than that demonstrated by the breast cancer cells, probably because this cancer cell line was derived from squamous cell epithelium, as opposed to columnar epithelium. Again, in 1.0% O, cell contacts were lost and the cells grew more independently. Vimentin expression was increased at the protein level, as determined by immunoblot analysis. Thus, hypoxia increased uPAR expression, and this was correlated with EMT-like changes in many, but not all, of the cancer cells studied.
Solid tumor growth is limited by the extent of vascularization. Hypoxia generally develops within 100–200 μm from blood vessels (). Without compensatory adaptations, severe hypoxia causes tumor cell death. This is evident in many human cancers and in xenografts, which typically demonstrate large areas of necrosis. Cellular adaptations observed in local hypoxia include changes in cell signaling and gene expression, which promote not only cell survival, but also cell migration, invasion, metastasis, and resistance to chemotherapy (). Hypoxia-induced changes in cell physiology occur downstream of HIF-1α () and other global transcription factors, such as cAMP regulatory element binding protein (), and are thus polygenic. Understanding the overall effects of hypoxia on cancer cell physiology is an important goal; however, it is also important to identify individual gene products that may be targeted to counteract the adverse effects of hypoxia in cancer.
In this study, we show that hypoxia induces EMT in breast cancer cells, building on previous studies in which aspects of EMT were reported in hypoxic ovarian carcinoma cells and in renal cell carcinoma (; ). For the first time, we report that uPAR is essential for hypoxia-induced EMT in breast cancer cells. uPAR expression is increased in MDA-MB-468 cells that are cultured in 1.0% O, and silencing of uPAR prevents hypoxia-induced EMT. Furthermore, when uPAR is overexpressed in MDA-MB-468 cells, these cells undergo EMT in normoxic cell culture. demonstrated EMT in kidney epithelial cells that were transfected to express α3 integrin and uPAR. Our studies extend their work by showing the relationship of uPAR to EMT in cancer cells, by demonstrating the importance of uPAR-dependent cell signaling, and by linking uPAR to EMT in hypoxia.
EMT refers to a series of phenotypic and molecular changes that occur in various steps of normal development and in cancer cells. When EMT occurs in cancer, the prognosis may be adversely affected (). Several cell signaling factors have been implicated. Akt may play a central role (; ; ). A constitutively active mutant of Akt induces EMT in carcinoma cell lines (). Rac1 also has been implicated in EMT. Initially, an alternatively spliced variant of Rac1, called Rac1b, which exists primarily in the GTP-loaded form, was shown to promote EMT in cancer cells (; ); however, there is also evidence that pathways that result in activation of the predominant form of Rac1 may have the same effect (). Other oncogenes, including c-Src and Ras, have been implicated in EMT (). Rac1, Akt, c-Src, and Ras are all activated downstream of uPAR (). Thus, we favor a model in which hypoxia-induced uPAR expression activates cell signaling down diverse pathways that are complementary in inducing the full spectrum of cellular changes observed in EMT.
To test our model, we studied two separate uPAR-dependent cell signaling pathways in MDA-MB-468 cells. We showed that Akt is activated in hypoxia and this response is blocked by peptide α325, implicating uPAR (; ). The PI3K inhibitor, LY294002, and Akt inhibitor prevented loss of cell contacts in hypoxia. LY294002 also inhibited hypoxia-induced Snail expression and preserved cell surface E-cadherin. These results are best explained by the activity of Akt as a negative regulator of GSK-3β, which controls Snail by suppressing NF-κB–dependent Snail expression, by targeting Snail for degradation, and by inhibiting Snail nuclear localization (; ). These GSK-3β activities limit the ability of Snail to function as an E-cadherin transcriptional repressor (). The linkage of uPAR to Akt activation in hypoxia provides one mechanism by which uPAR may regulate Snail and thus promote EMT. In the same cell culture model system, antagonizing Rac1 blocked the increase in cell migration, Matrigel invasion, and collagen remodeling, which were observed in hypoxia. In our hands, it was not possible to generate stable cultures of DN-Rac1–expressing MDA-MB-468 cells because the cells were not sustainable. Therefore, it was not possible to accurately assess morphologic indices of EMT in cells transfected with DN-Rac1. Nevertheless, our evidence indicates that targeting either Rac1 or the PI3K–Akt pathway blocks cellular properties associated with uPAR-induced EMT in hypoxic breast cancer cells.
Rac1 has emerged as an important cell signaling factor downstream of uPAR in diverse experimental systems (; ; ; ). uPAR-dependent Rac1 activation does not require uPA, although ERK/MAPK activation by uPA increases uPAR expression (), and this may indirectly lead to an increase in Rac1 activation (). Vitronectin is a provisional ECM protein, which accumulates at sites of inflammation (), and thus should be available to activate the uPAR–Rac1 pathway in many malignancies. A recently described pathway in which uPAR supports integrin αβ in activating Rac1 () expands the continuum of contexts in which uPAR may control Rac1 in vivo.
Because hypoxia and CoCl promoted breast cancer cell migration and Matrigel invasion in vitro, we wished to study whether hypoxia promotes metastasis in vivo. To accomplish this goal, we established a new model system in which GFP- expressing MDA-MB-468 cells were inoculated onto chicken egg CAMs and treated daily with 100 μM CoCl. After 11 d, we assessed the number of fluorescent cells/cell clusters in the heart–lung block. We interpreted the presence of cancer cells in the heart–lung block as evidence that the cancer cells had penetrated into the vascular compartment of the chick embryo. In control in vitro experiments, we examined the effects of CoCl on MDA-MB-468 cell growth. The CoCl, at different concentrations, had no effect or was slightly growth inhibitory (unpublished data).
CoCl induced changes in MDA-MB-468 cell gene expression on CAMs that were anticipated (increased uPAR and VEGF mRNA). CoCl also significantly increased tumor cell dissemination. Thus, we have evidence that the molecular changes that occur in hypoxic breast cancer cells are correlated with an increased propensity for these cells to disseminate in vivo. The increase in cancer cell dissemination may be explained by activation of uPAR-dependent pathways that also result in increased cell migration and Matrigel invasion. However, given that this is a new model system, our results should be interpreted with caution because additional work will be necessary to evaluate the effects of CoCl on the CAM surface, which constitutes the tumor cell microenvironment. Possible changes in the microenvironment that could contribute to the increase in cancer cell dissemination include increased vascularization and changes in host cell expression of chemoattractant proteins. Additional studies also will be necessary to confirm the role of uPAR in CoCl- induced cancer cell dissemination on the CAMs.
In hypoxia, HIF-1α alone may regulate >70 gene products (); however, the spectrum of regulated genes is probably cell type specific. In one study, which compared gene expression in response to HIF-1α in MCF-7 cells and MDA-MB-231 breast cancer cells, 26 genes were up-regulated in both cell types, whereas 63 and 117 genes were up-regulated in only MCF-7 or MDA-MB-231 cells, respectively (). This result may be explained by multiple factors, including but not limited to the karyotype of the cell, gene amplification, posttranslational histone modification, CpG methylation, factors that affect transcript stability, and transcription regulatory factors present in the cell before inducing hypoxia (; ; ). Cell specificity in the response to hypoxia emerged in our studies when we probed various cell lines for uPAR mRNA expression by qPCR. Our results and prior results reported by others (; ) indicate that increased uPAR expression may be observed frequently but not uniformly in hypoxia. Therefore, consideration of other receptors, such as the EpoR and c-met, which may be activated in hypoxia (; ), remains important. Although these receptors activate overlapping signaling pathways, the complete activity of each receptor is unique. In ZR-75-1 and SCC15 cells, hypoxia induced significantly increased uPAR mRNA expression and cellular changes consistent with EMT. Thus, the correlation of uPAR expression with EMT in hypoxia was extended beyond the MDA-MB-468 cell system. However, we have not yet determined whether the increase in uPAR expression is responsible for hypoxia-induced EMT in these other two cell lines. Understanding how uPAR and other hypoxia- activated receptors function as regulators of cancer cell physiology in hypoxia remains an important goal.
Snail-specific polyclonal antibody and E-cadherin–specific monoclonal antibody (HECD-1) were obtained from Abcam. uPAR-specific antibodies 399R and 3932 and uPA-specific antibody 3471 were obtained from American Diagnostica, Inc. Rac1- and HIF-1α–specific monoclonal antibodies were purchased from BD Biosciences. Rac/Cdc42 assay reagent, which includes residues 67–150 of p21-activated kinase fused to glutathione--transferase and coupled to glutathione–Sepharose (PAK-1 PBD) was purchased from Millipore. Polyclonal antibody that recognizes total ERK/MAPK was obtained from Millipore. Antibodies that detect phosphorylated ERK/MAPK, phosphorylated Akt, and phosphorylated GSK-3β were obtained from Cell Signaling Technologies. Vimentin-specific monoclonal antibody, tubulin-specific monoclonal antibody, nonspecific murine IgG, and rabbit IgG were obtained from Sigma-Aldrich. The MEK inhibitor, PD098059; the PI3K inhibitor, LY294002; and Akt inhibitor were obtained from EMD Biosciences. Type I collagen was purchased from Southern Biotechnology Associates, Inc. NHS-Fluorescein was obtained from Pierce Chemical Co. The Alexa Fluor 594 protein-labeling kit was obtained from Invitrogen. qPCR reagents, including primers and probes for human uPAR, Snail, VEGF, and HPRT-1 were obtained from Applied Biosystems. Synthetic peptide α325 and its scrambled version, scα325, were obtained from AnaSpec.
The expression construct encoding DN-MEK1 (S217→A) in pBABE was previously described (). pKH3-N17Rac1, which encodes DN-Rac1, was provided by I. Macara (University of Virginia, Charlottesville, VA). The expression construct encoding full-length human uPAR (pcDNA-uPAR) was previously described (). pMAX-GFP, which encodes GFP, was obtained from Amaxa.
MDA-MB-468, MDA-MB-435, ZR-75-1, SCC15, MDA-MB-231, and A431 cells were obtained from American Type Culture Collection. SK-BR3 cells and low-passage MCF-7 cells were provided by S.J. Parsons (University of Virginia, Charlottesville, VA). SK-BR3 cells were cultured in DME (HyClone) supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. MDA-MB-231 cells were cultured in Leibowitz-15 medium (HyClone) with 10% FBS, penicillin, and streptomycin. MCF-7 cells were cultured in RPMI (HyClone) with 10% FBS, penicillin, and streptomycin. SCC15 cells were cultured in a 1:1 mixture of DME/Ham's-F12 with 10% FBS. ZR-75-1 cells were cultured in RPMI with 2 mM -glutamine, supplemented with 18 mM sodium bicarbonate, 25 mM glucose, 10 mM Hepes, 1.0 mM sodium pyruvate, 10% FBS, penicillin, and streptomycin.
MDA-MB-468 cells were plated on sterile coverslips. After 18 h, the cells were washed and incubated in serum-free medium under hypoxic (1.0% O) or normoxic (21% O) conditions for 50 h. Cultures were fixed in 4% formaldehyde, permeabilized with 0.2% Triton X-100, and incubated with antibodies specific for E-cadherin, vimentin, or Snail for 18 h, followed by secondary antibody conjugated with Alexa Fluor 488 and phalloidin conjugated with Alexa Fluor 568 for 1 h (Invitrogen). Control cultures were treated equivalently except for the omission of primary antibody. Preparations were mounted on slides using ProLong Gold with DAPI (Invitrogen) and examined using a fluorescent microscope (DMIRE2; Leica). Images were obtained using a 63× oil-immersion objective (NA 1.4) and a Hamamatsu digital camera with Simple PCI software at room temperature. Deconvolution was performed using Simple PCI software.
Cell migration was studied using 6.5-mm Costar Transwell chambers with 8-μm pores (Corning), as previously described (). The Transwell membranes were precoated with 20% FBS on the underside only for 2 h at 37°C. Cells in suspension were treated with 25 μg/ml uPAR- specific antibody 399R, 25 μg/ml uPA-specific antibody, 25 μg/ml nonspecific murine IgG, 25 μg/ml rabbit IgG, 50 μM PD098059, or with vehicle for 15 min before adding the cells to the top chamber (10 cells per Transwell). The bottom chamber contained 10% FBS. Cells were allowed to migrate in 21 or 1.0% O. In separate studies, the indicated concentration of CoCl was added to both chambers. Studies were allowed to proceed for 24 h. At that time, nonmigrating cells were removed from the top surface of each Transwell using a cotton swab. Transwell membranes were then stained with Diff-Quik (Dade-Behring). Cells that migrated through the membrane to the lower surface were counted by light microscopy.
To study cell invasion, Biocoat Inserts containing reconstituted, growth factor–reduced Matrigel (BD Biosciences) were purchased. Cells that were treated with antibodies or PD098059 and untreated cells were added above the Matrigel matrix. FBS was added to the lower chamber as described for the cell migration experiments. Matrigel invasion was allowed to progress for 24 h in 21 or 1.0% O. The Matrigel and cells on the top membrane surface were removed with a cotton swab. Penetration of cells to the underside of the membrane was determined as described for the cell migration experiments.
100 μg/ml type I collagen was adsorbed to glass coverslips by incubation for 24 h at 4°C and labeled with fluorescein or Alexa Fluor 594. MDA-MB-468 cells and subclones thereof were allowed to adhere to the coverslips and cultured for 24 h in 21 or 1.0% O. The preparations were fixed in 4% formaldehyde, permeabilized with 0.5% Triton X-100, and mounted on slides using ProLong Gold with DAPI. Images were obtained using a fluorescent microscope (DMIRE2; Leica) on a Hamamatsu digital camera and Simple PCI software.
MDA-MB-468 cells and subclones thereof were cultured in 60-mm dishes. The cells were transferred to serum-free medium for 6 h and exposed to 1.0% O or 100 μM CoCl for 15 h. Control cultures were maintained in 21% O. Cell extracts were prepared in ice-cold RIPA buffer (20 mM sodium phosphate, 150 mM NaCl, pH 7.4, 1% NP-40, 0.1% SDS, and 0.5% deoxycholic acid) containing complete protease inhibitor cocktail (Roche) and 1 mM sodium orthovanadate. The protein concentration in each extract was determined by bicinchoninic acid assay (Sigma-Aldrich). Cell extracts were subjected to SDS-PAGE on 10% slabs. Proteins were transferred to polyvinylidene difluoride membranes and probed with antibodies that detect phosphorylated ERK/MAPK, human uPAR, vimentin, phosphorylated Akt, phosphorylated GSK-3β, and snail. The same membranes also were probed to detect tubulin or total ERK/MAPK as a loading control.
Affinity precipitation of active Rac1 was performed using the fusion protein PAK-1 PBD, which binds specifically to the active, GTP-bound forms of Rac1 and Cdc42 (). MDA-MB-468 and subclones thereof were cultured in 10-cm plates in serum-free medium for 6 h and exposed to 1.0% O or 100 μM CoCl for 15 h. Control cultures were maintained in 21% O. Cell extracts were prepared in ice-cold RIPA buffer containing protease inhibitor cocktail and 1 mM sodium orthovanadate. The extracts were incubated with 20 μg PAK-1 PBD coupled to glutathione–Sepharose for 60 min at 4°C. The glutathione–Sepharose was washed three times and treated with SDS sample buffer to dissociate the PAK-1 PBD and associated proteins. Cell extracts were subjected to SDS-PAGE on 12% slabs, and immunoblot analysis was performed to detect Rac1. Samples of each cell extract were also subjected to immunoblot analysis before incubation with PAK-1 PBD to determine total Rac1 and tubulin, as a loading control.
Total RNA was extracted using the RNeasy kit (QIAGEN). cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad Laboratories, Inc.). qPCR was performed using a System 7300 instrument (Applied Biosystems) and a one-step program: 95°C for 10 min, 95°C for 30 s, and 60°C for 1 min, for 40 cycles. HPRT-1 gene expression was measured as a normalizer. Results were analyzed by the relative quantity (ΔΔC) method. Experiments were performed in triplicate with internal duplicate determinations.
The original assay for studying tumor metastasis from tumors grown on CAMs () was modified to facilitate analysis of breast cancer cells. In brief, fertilized white leghorn eggs (McIntyre Farms) were incubated for 9 d before tumor cell inoculation in a rotary incubator at 38°C. GFP-expressing MDA-MB-468 cells were suspended at 4°C in Matrigel at a concentration of 4 × 10 cells/ml. A window was cut in the eggshell above the dropped CAM and a 1.0-cm sterile cotton gauze with a 4-mm center cutout was placed on the membrane. 2 × 10 MDA-MB-468 cells were inoculated onto the CAM through the center of the gauze. To model hypoxia on the CAMs, 25 μl OptiMEM (Invitrogen) medium, with or without 100 μM CoCl, was applied to the gauze daily for 11 d. Because the gauze was rewet every day, and the tumor was always growing underneath the gauze, we assured saturation of the tumor with CoCl. At developmental day 20, necropsy was performed on the developing chick embryo. The heart and lungs were isolated, and cell dissemination was quantified under fluorescent microscopy by counting the cells/cell clusters. In parallel experiments, tumors were harvested after 4 d of growth on the CAMs. RNA was isolated and subjected to qPCR analysis to determine levels of human uPAR and VEGF mRNA. |
Cyclin E, an activator of cyclin-dependent kinase (Cdk) 2, accumulates at the G1/S boundary of the cell cycle, where it stimulates functions associated with entry into and progression through S phase (; ; ; ). Normally, cyclin E levels are tightly regulated so that peak cyclin E–Cdk2 kinase activity occurs only for a short interval near the G1/S boundary (). This is accomplished primarily by the periodic E2F-dependent transcription of cyclin E during late G1 phase (; ) and its subsequent phosphorylation-dependent ubiquitin-mediated proteolysis in S phase when cyclin E–Cdk2 complexes become active (). Cyclin E expression and activation of Cdk2 at the G1/S boundary are compatible with the known roles of cyclin E in the promotion of replication-associated functions (; ; ; ).
The overexpression of cyclin E has been observed in a broad spectrum of human malignancies, suggesting that proper regulation of cyclin E is important for the preservation of normal cellular functions (; ; ; ; ). Under such circumstances, cyclin E is often expressed at levels higher than those observed in normal tissues, but also the periodic expression at the G1/S boundary is frequently lost, with cyclin E levels maintained throughout the cell cycle. This can occur through a single mutation in Cdc4 (also known as Fbw7), the F-box component of an SCF (Skp1-Cul1–F-box protein) ubiquitin ligase that targets cyclin E for ubiquitin-mediated degradation (; ; ). Furthermore, the overexpression of cyclin E in some cancers has been associated with aggressive disease and poor outcome (; ; ; ; ). A direct causal link between cyclin E overexpression and tumorigenesis is supported by a transgenic mouse model in which the ectopic expression of cyclin E in the mammary epithelium induces mammary carcinogenesis (; ).
The basis for cyclin E–induced tumorigenesis remains controversial. Because cyclin E, by virtue of its ability to activate Cdk2, is a positive regulator of the cell cycle, it has been proposed that the role of cyclin E in tumorigenesis is to stimulate cellular proliferation (). This is consistent with the role of cyclin E–Cdk2 in phosphorylating and inactivating the retinoblastoma protein Rb, a negative regulator of proliferation (; ; ). Furthermore, the ectopic expression of cyclin E was shown to drive cultured cells from G1 into S phase with accelerated kinetics (; ). However, in these studies, the overall rate of proliferation was not altered, arguing against a direct link between cyclin E overexpression and increased proliferation in the context of tumorigenesis. On the other hand, cyclin E–mediated mammary hyperplasia was observed in mice carrying a mammary epithelium–specific cyclin E transgene, which is consistent with a link between cyclin E overexpression and cellular proliferation in vivo ().
An alternative view of the role of cyclin E overexpression in tumorigenesis comes from a study of genomic instability (). In cultured nontransformed cells, the overexpression of cyclin E led to both chromosome instability and polyploidy (). Chromosome instability was a relatively infrequent event under moderate levels of cyclin E overexpression but was manifest reproducibly as both chromosome losses and gains, suggesting nondisjunction or other mitotic aberrations as possible mechanisms. Polyploidy was a more frequent event that was easily scorable under the experimental conditions. Polyploid cells, which are themselves unstable, can readily give rise to aneuploidy (). Cyclin E overexpression in a tissue culture model has also been associated with the formation of micronuclei, which is suggestive of the generation of aneuploid cells (). Genomic instability in the forms of both chromosome instability and polyploidy, leading to aneuploidy, could easily explain the link between cyclin E overexpression and tumorigenesis.
Genomic instability has been shown to promote tumorigenesis on several levels, most notably by facilitating the loss of heterozygosity at tumor suppressor loci and amplification of oncogenes (). Consistent with this interpretation, early loss of the heterozygosity of the tumor suppressor gene encoding p53 was observed in the mouse mammary cyclin E transgenic model (). Importantly, both aneuploidy and polyploidy have been observed in human tumors in which cyclin E is overexpressed ().
In terms of the mechanism whereby cyclin E overexpression promotes genomic instability, cell cycle analysis has provided several clues. Early studies indicated that the ectopic expression of cyclin E, although accelerating entry into S phase, caused a substantial prolongation of S phase (; ). A more recent analysis revealed that the overexpression of cyclin E specifically impairs the assembly of prereplication complexes upon mitotic exit, leading to slow and inefficient DNA replication and most likely replicative stress (). It is easy to envisage how replication defects can lead to genomic instability. In another study in which the cyclin E–targeting F-box protein Cdc4 was mutated in cultured cells, impairing cyclin E turnover, mitotic aberrations in some cells were observed by time-lapse videomicroscopy (). However, this study could not distinguish between direct effects on mitosis and indirect effects accruing as a result of replicative stress, and because SCF also targets other cellular regulatory proteins, the effects could not be definitively attributed to cyclin E overexpression.
In this study, we investigate whether cyclin E overexpression can have a direct impact on the progression through and the quality of mitosis. By expressing cyclin E in a manner that precludes effects on prereplication complex assembly, we demonstrate that the overexpression of cyclin E expression both delays progression through early phases of mitosis and causes mitosis to be executed aberrantly in many cells. Furthermore, we demonstrate that these effects are mediated by the direct action of cyclin E–Cdk2 on protein complexes that regulate mitotic progression.
To study the acute effects of cyclin E overexpression, a recombinant adenovirus expressing wild-type (WT) cyclin E from the constitutive cytomegalovirus promoter was constructed (). As shown by Western blotting (), KB cells 24 h after transduction with cyclin E adenovirus (, ad cyclin E) have increased cyclin E expression compared with cells transduced with an empty vector control adenovirus (, ad empty vector [EV] control). Although this ectopically expressed cyclin E exhibits some cell cycle–dependent regulation, levels are substantially higher than WT at all cell cycle phases (Fig. S1, available at ). To confirm that the adenoviral expression of cyclin E resulted in the elevation of cyclin E–Cdk2 kinase activity with respect to the cell cycle, kinase assays were performed on extracts from mitotic cells, where cyclin E is not normally expressed. KB cells were synchronized using a single thymidine block at the G1/S boundary and were released for 8 h into nocodazole (see ). Cyclin E–Cdk2 kinase activity in immunoprecipitates was detectable well above background in mitotically enriched extracts after transduction with the cyclin E–expressing adenovirus (). The level of cyclin E expression throughout the cell cycle in the adenovirally transduced cell lines used in the current study is equivalent to that characteristic of a subpopulation of breast cancer–derived cell lines that were shown previously to express high levels of cyclin E (; ).
Flow cytometry was used to screen for cell cycle abnormalities resulting from cyclin E overexpression. Previous studies have shown that cells overexpressing cyclin E accumulate in S phase (; ; ). FACS profiles of asynchronous cyclin E and control adenovirus-transduced cells confirmed this observation but also revealed an increase in G2/M cells (; ad cyclin E = 17.3% and ad control = 12%).
The aforementioned flow cytometry experiments revealed an accumulation of cells with 4C DNA content. However, these experiments were unable to distinguish G2 from mitotic populations. Immunostaining cells with an antibody reactive with phosphorylated histone H3 (H3-P) produces a distinct punctate pattern beginning at centromeres in G2 cells and brightly covering entire chromosomes in mitotic cells (). demonstrates the nuclear pattern associated with G2 cells. The onset of histone H3 phosphorylation in U2OS cells was confirmed to begin in late S phase to early G2 by labeling cells with both anti–H3-P and anti-BrdU antibodies after a short BrdU pulse (unpublished data). Scoring interphase U2OS cells for positive H3-P staining demonstrated a decreased G2 population in cyclin E adenovirus-transduced cells (; ad control = 12% and ad cyclin E = 6%). This decrease in the fraction of G2 cells indicates that the 4C (G2 + M) accumulation, which was previously shown by FACS analysis in cells overexpressing cyclin E, must be the result of an accumulation of cells in mitosis rather than in G2.
To determine whether cyclin E overexpression leads to the accumulation of cells at a particular stage of mitosis, the distribution of mitotic phases was scored in asynchronous populations using immunofluorescence microscopy. KB cells were transduced with cyclin E or control adenovirus for 24 h, fixed, and immunostained with anti–H3-P to highlight mitotic cells and anti–α-tubulin to label mitotic spindles. Mitotic phases were then scored based on the positions of chromosomes and the orientation of the mitotic spindle. Six mitotic phases were classified and scored in the analysis: prophase, prometaphase, unaligned metaphase, metaphase, anaphase/telophase, and cytokinesis ().
Results from such mitotic counts indicate that the overexpression of cyclin E leads to the accumulation of KB cells in prometaphase (, ad cyclin E = 31% and ad control = 24% of the mitotic cells counted) and unaligned metaphase (, ad cyclin E = 20% and ad control = 13%) compared with control cells. The accumulation of cells at early stages of mitosis appears to include later stages of prometaphase up until the interval of alignment of chromosomes on the metaphase plate but not including the point of complete chromosome alignment. Consequently, the aligned metaphase population was decreased as a percentage of the entire mitotic population (adenovirus cyclin E = 13% and adenovirus control = 19%). However, once cells overexpressing cyclin E recover from the delay and proceed into anaphase, they traverse the later phases of mitosis without delay relative to controls.
The recordings of IME and U2OS cells provided direct real-time evidence that the overexpression of cyclin E confers an early mitotic delay. In addition, they provided insight into the dynamics of the prometaphase delay. For live cell recordings, unaligned metaphase could not be accurately determined as a result of the lack of spindle labeling. Therefore, prometaphase was scored as the time from chromosome congression toward the midzone until all chromosomes were completely aligned at the metaphase plate. shows the mean time in minutes of prometaphase, metaphase, and anaphase/telophase based on recordings of control or cyclin E adenovirus-transduced cells undergoing mitosis. Cyclin E–overexpressing cells spent on average almost twofold longer in prometaphase than controls (adenovirus cyclin E = 26–29 min and adenovirus control = 11–16 min, depending on cell type). In some abnormal mitoses characteristic of cyclin E–overexpressing cells, chromosomes oscillated near to the metaphase plate for an extended period before finally progressing into anaphase with either aligned or unaligned chromosomes. illustrates mitotic division in a control U2OS cell. Prometaphase is ∼15 min, and metaphase begins at 20 min (indicated by the red asterisk in ). (b and c) shows mitotic divisions in cells overexpressing cyclin E with delayed and abnormal prometaphases. In the second series, a cyclin E–transduced IME cell () shows an abnormal prometaphase (>30 min) during which chromosomes never progressed to a successful metaphase and began anaphase after 38 min despite a failure of chromosome alignment. In the third series, a cyclin E–transduced U2OS cell (; also shown in Video 1, available at ) delayed in prometaphase for almost 40 min, during which chromosomes nearly aligned at the metaphase plate but proceeded to oscillate near the plate for a prolonged period until the recording ended. Another demonstration of a cell overexpressing cyclin E expression that delays before complete metaphase alignment is shown in Video 2.
In addition to the early mitotic delays, mitotic failures were observed in which cells overexpressing cyclin E did not progress into anaphase but instead began to decondense chromosomes, resulting in a polyploid cell (Video 3, available at ).This was consistent with our previous observation of an accumulation of polyploid cells in cyclin E–overexpressing populations (). Indeed, flow cytometric analysis of cyclin E–overexpressing cells indicates a dose-dependent failure of mitosis at levels ≥1–2% per cell division (Fig. S2). Together, analysis of mitosis in both fixed and live cells suggests that a primary consequence of cyclin E overexpression is mitotic impairment, which, in extreme cases, can lead to mitotic failure and polyploidy.
To investigate the cause of the cyclin E–mediated mitotic delay, levels of mitotic regulatory proteins were determined. Single cell analysis of cyclin B1 expression was used to compare peak levels of cyclin B1 in early mitosis and its degradation in metaphase in cells overexpressing cyclin E and controls. KB cells were released into mitosis from a single thymidine block. Representative images of mitosis in cells transduced with either control or cyclin E adenovirus are shown in . Cyclin B1 (, green), a centromere-binding protein (, red), and DNA (, blue) were labeled, and cyclin B1 levels were measured in early prometaphase cells, unaligned metaphase cells, and metaphase cells. Cyclin B1 staining intensities of each cell were divided into low, medium, and high categories (). The overexpression of cyclin E led to elevated levels of cyclin B1 in prometaphase and unaligned metaphase. Furthermore, in metaphase, all control cells expressed low levels of cyclin B1, as expected as a result of its degradation by the mitotic ubiquitin ligase anaphase-promoting complex (APC). However, the degradation of cyclin B1 was delayed in 45% of metaphase cells with cyclin E overexpression, which continued to express medium and high levels of cyclin B1 (, metaphase). These results suggest that overexpression of cyclin E can delay the normal degradation of cyclin B1. However, cyclin B1 degradation was not completely inhibited, as anaphase cells from both control and cyclin E–overexpressing cells contained low levels of cyclin B1 (unpublished data).
Next, protein levels of other important APC/C substrates were assessed during mitosis. For these experiments, cells were transduced with adenovirus and simultaneously arrested in a single thymidine block overnight. As adenoviral cyclin E is not expressed at substantial levels before 8 h after transduction (unpublished data), the majority of nonthymidine-blocked cells will have passed through mitosis and early G1 before cyclin E accumulation. Therefore, this protocol precludes S-phase defects conferred by cyclin E–mediated interference with prereplication complex assembly in late telophase. In addition, the single thymidine block and release protocol ensures approximately equal percentages of cells in G2/M for each time point (). In KB cells, the peak of mitosis based on microscopic observation occurs from 8 to 10 h after release from a thymidine block. Western blot analysis of protein expression up to and through mitosis in cyclin E–overexpressing and control cells revealed an accumulation of cyclin B1, Cdc20, and securin (). In cells overexpressing cyclin E, mitotic protein levels were elevated relative to controls at each time point up to and through the peak mitotic samples. These data indicate that defects in mitotic protein expression are not cyclin B specific but rather suggest that the APC/C ubiquitin ligase is inhibited under conditions of cyclin E overexpression, as all of the proteins exhibiting elevated expression are APC/C substrates targeted for ubiquitin-mediated proteolysis.
To determine whether the observed protein accumulations were caused by decreased degradation, we examined protein stability by carrying out cycloheximide chase experiments under conditions of cyclin E overexpression. Cells transduced with cyclin E or control adenovirus and arrested using a single thymidine block were then released for 4 h. Cycloheximide was added to terminate protein synthesis, and samples were collected, as indicated, for Western blot analysis (). As can be seen, cyclin B1, securin, and Cdc20 were degraded much more rapidly in control cells compared with cells overexpressing cyclin E, suggesting that the cyclin E–mediated accumulation of key mitotic regulatory proteins occurs by decreasing their rate of ubiquitin-mediated proteolysis.
The data presented up to this point suggest that elevated cyclin E–Cdk2 inhibits the APC/C. APC/C activity requires one of two activating regulatory subunits, Cdc20 or Cdh1 (; ). Cdc20 is thought to activate the APC/C during mitosis, whereas Cdh1 is the activating subunit from mitotic exit to the G1/S boundary (; ; ; ; ; ). Because Cdc20 itself was found to accumulate in response to cyclin E overexpression, it was unlikely to be the relevant target of cyclin E–Cdk2. On the other hand, whereas Cdh1 is inactivated by phosphorylation, it is thought to already be inactive during early mitosis as a result of the phosphorylation by cyclin A–Cdk2 (; ). However, recent studies have suggested that Cdh1 does contribute to early mitotic APC/C activation (, ), raising the possibility that it is the target of cyclin E overexpression. Furthermore, although under normal circumstances, it was shown that cyclin E–Cdk2 does not contribute substantially to Cdh1 phosphorylation and inactivation (; ; ), overexpressed cyclin E might be able to interact with and phosphorylate Cdh1.
We first determined whether cyclin E–Cdk2 could bind directly to Cdh1 because protein kinases often form stable complexes with substrates. Cyclin A binding to Cdh1 has previously been described and was used as a positive control (; ). Three expression plasmids coding for N-terminally myc-tagged WT or mutants Cdh1 and Cdh1 were used to characterize the binding of cyclin E and cyclin A to Cdh1 (). Cdh1 is mutated in the WD-40 repeat domain at the RVL motif, which has been shown to be necessary for cyclin A binding, and Cdh1 is mutated at four phosphorylation sites shown to be targeted by cyclin A–Cdk2 that are essential for Cdh1 inhibition. Expression plasmids for cyclin E or cyclin A were cotransfected with expression plasmids encoding Cdh1, Cdh1, Cdh1, or an empty vector into 293T cells. Cdh1 was immunoprecipitated from cell lysates using anti-myc antibody. The immunoprecipitation data show that exogenous cyclin E binds efficiently to Cdh1 as does the positive control, cyclin A (). In addition, weak binding is detectable between exogenous cyclin A and both Cdh1 mutants. Exogenous cyclin E also bound both WT and, to a lesser degree, both mutant Cdh1 mutants. However, no binding was observed at endogenous levels of cyclin E (shown as a band with lower mobility than exogenous cyclin E), which is consistent with previous results ().
To determine whether Cdh1 is a substrate of cyclin E–Cdk2, in vitro phosphorylation assays were performed. GST-Cdh1 () and GST alone were purified from . In vitro kinase assays were performed using cyclin E– and cyclin A–Cdk2 purified from baculovirus-transduced insect cells on histone H1, GST, and GST-Cdh1 as substrates (). Robust phosphorylation of histone H1 by cyclin E– and cyclin A–Cdk2 confirmed that both kinase complexes were active. GST alone (∼28 kD) was not phosphorylated by either cyclin–Cdk2 complex. However, purified GST-Cdh1 (75 kD) as well as several degradation products were phosphorylated by both cyclin E– and cyclin A–Cdk2.
Phosphorylation of Cdh1 was also confirmed in cell culture. In , reduced mobility was shown for Cdh1 and Cdh1 when coexpressed with either cyclin E or cyclin A. To ensure that the reduced mobility was a result of the phosphorylation of Cdh1, parallel samples using Cdh1 were treated with λ-phosphatase to dephosphorylate Cdh1 before analysis (). Cdh1 exhibited reduced mobility upon the cotransfection of either cyclin E or cyclin A, which was reversed with phosphatase treatment, but not when phosphatase inhibitors were present, indicating that ectopically expressed Cdh1 can be phosphorylated by either cyclin E– or cyclin A–Cdk2.
Next, we assessed the ability of cyclin E–Cdk2 to phosphorylate endogenous Cdh1 under conditions in which the cyclin E–mediated accumulation of mitotic APC/C substrates was observed (as in ). Samples were collected up to and through mitosis after a thymidine block and release (). Overexpression of cyclin E reduced the mobility of endogenous Cdh1 (), which is consistent with cyclin E–Cdk2 contributing substantially to the phosphorylation and inactivation of Cdh1 under these conditions. To determine whether the phosphorylation of Cdh1 by cyclin E–Cdk2 under these conditions affected the association of Cdh1 with APC, cells were transfected with a plasmid expressing myc-Cdh1 and were transduced with cyclin E–expressing or control adenovirus. APC immunoprecipitates (using Cdc27 antibody) were then prepared 4h after release from a thymidine block. Cdh1 association with the APC was reduced to ∼60% of control values as a consequence of cyclin E overexpression ().
To assess the effects of cyclin E–Cdk2 on the activity of APC directly, in vitro ubiquitylation assays were performed (). Cdh1 was phosphorylated in vitro before addition to the ubiquitylation reaction. APC activity was determined by the conversion of unmodified cyclin B1 to multiubiquitylated cyclin B1 derivatives.
Cyclin B1 was efficiently ubiquitylated when Cdh1 was added to the reaction in the absence of cyclin E–Cdk2 (). Upon the addition of purified cyclin E–Cdk2, the ubiquitylation of cyclin B1 was dramatically reduced, which is consistent with the phosphorylation-dependent inhibition of Cdh1 (; and quantified in b). The addition of purvalanol, a Cdk inhibitor, partially reversed cyclin E–mediated inhibition. These data suggest that cyclin E–Cdk2 inhibits Cdh1 via phosphorylation. Furthermore, a phosphorylation site mutant of Cdh1, where all possible Cdk sites were mutated to alanines, was completely resistant to cyclin E–Cdk2-mediated inhibition (; and quantified in b). These data confirm that cyclin E–Cdk2 can inactivate APC through the inhibitory phosphorylation of Cdh1.
If cyclin E–Cdk2 inhibits APC in vivo, reducing cellular levels of Cdh1 should mimic the phenotype associated with the overexpression of cyclin E. Using Cdh1-specific siRNA, the effects of Cdh1 reduction were compared with those associated with cyclin E overexpression. A strong knockdown of Cdh1 protein (using 20 nM of Cdh1-specific siRNA) inhibited cell proliferation, which had previously been reported (; ). To only partially ablate Cdh1, the time of incubation and the concentration of Cdh1-specific siRNA were reduced. The accumulation of cyclin B1, securin, and Cdc20 in cells targeted by Cdh1-specific siRNA was compared with that in cells transduced with cyclin E or control adenovirus (). A comparable up-regulation of cyclin B1, securin, and Cdc20 protein occurs as a result of both Cdh1 reduction and cyclin E overexpression.
We next determined the effects of reducing Cdh1 on mitotic progression. Fixed cells were scored as previously discussed (see section Cyclin E overexpression leads to an early mitotic delay) in asynchronous populations after the transduction of EV control or cyclin E adenovirus for 24 h and incubation with Cdh1- or GFP-specific siRNA for an additional 24 h (). As shown previously, cyclin E overexpression caused an accumulation in early mitosis before metaphase. This same trend, but with a more pronounced effect, was observed in cells partially silenced for Cdh1 (). These data confirm that Cdh1 has a role in the regulation of progression through early mitosis and that cyclin E–mediated inhibition of Cdh1 could, in principle, account for the impairment of early mitotic events.
Based on the aforementioned data, the direct reduction of Cdh1 or overexpression of cyclin E leads to the accumulation of mitotic APC/C substrates and delay in progression through early mitosis. However, whether the mitotic delay is caused by the accumulation of APC/C substrates and, if so, which APC/C substrates contribute to the mitotic delay remained to be determined. Based on our observations, it appeared that most chromosomes become bioriented but delay in aligning at the metaphase plate and progressing into anaphase. This phenotype is consistent with a defect in sister chromatid separation possibly caused by elevated levels of cyclin B1 and securin as cells progress into and through mitosis. Degradation of securin and cyclin B1 is needed to activate separase and thereby promote the cleavage of cohesin (; ), which is required for separation of the sister chromatids at anaphase, allowing the poleward migration of chromosomes (, ; ; ). Therefore, we tested whether slightly reducing securin and cyclin B1 expression in cells overexpressing cyclin E could override the cyclin E–induced delay of early mitosis.
Cells were transduced with EV control or cyclin E adenovirus for 24 h and transfected with low concentrations of siRNA (10 pM) targeting securin and/or cyclin B1 or GFP for an additional 24 h. Expression of cyclin B1 and securin in cells released into mitosis from a thymidine block was compared by Western blot analysis (). Cells transduced with cyclin E adenovirus and transfected with cyclin B1– and securin-specific siRNA resulted in levels of cyclin B1 and securin comparable with those found in control cells, whereas cells transfected with control GFP-specific siRNA continued to exhibit elevated levels of cyclin B1 and securin. As expected, Cdc20 levels were not affected by siRNA targeting cyclin B1 and securin.
Mitotic progression was assessed in cells with reduced levels of securin and cyclin B1, securin alone, or cyclin B1 alone mediated by siRNA transfection (). Reducing both cyclin B1 and securin eliminated the delays in prometaphase and unaligned metaphase along with the resulting decrease in the percentage of metaphase cells mediated by cyclin E overexpression. However, reducing securin or cyclin B1 alone did not completely eliminate the prometaphase or unaligned metaphase delay associated with cyclin E overexpression. Therefore, the depletion of securin or cyclin B1 alone is not sufficient to alleviate the mitotic inhibitory effects of cyclin E overexpression and suggests that stabilization of both proteins contributes to cyclin E–mediated mitotic delay.
Although Cdh1 can activate both mitotic and interphase APC/C, models of cell cycle regulation have proposed that the ubiquitin ligase ACP is maintained in an inactive state from the G1/S boundary until mitotic exit as a result of phosphorylation by Cdks active during this interval (; ; ; ; ; ). Although there is evidence that cyclin E–Cdk2 is primarily responsible for the accumulation of G2/M cyclins and Cdh1 phosphorylation in (; ), in mammalian cells, it was shown that cyclin A–Cdk2 specifically phosphorylates Cdh1 on a number sites and that this phosphorylation prevents association with the APC/C core (; ; ). Because cyclin A–Cdk2 becomes active at the G1/S transition and persists until early mitosis, it has been assumed that Cdh1 does not contribute to APC/C ubiquitin ligase activity during this interval and that only an alternative APC/C cofactor, Cdc20, carries out mitotic functions. However, a recent study suggests that APC does have a role in the ubiquitylation of securin during early mitosis in that mice carrying mutations that misregulate Cdh1 nuclear transport exhibit mitotic defects (). Consistent with this finding, in the present study, we show that APC activity must contribute to the regulation of mitotic proteins up to and through early mitosis. Specifically, the partial silencing of Cdh1 by RNAi led to a hyperaccumulation of APC substrates and impaired progression through early phases of mitosis. Therefore, APC must retain residual activity before mitosis even in the presence of cyclin A–Cdk2. This conclusion, along with the observed ability of cyclin E–Cdk2 to phosphorylate Cdh1 in vivo and inhibit APC in vitro, is consistent with a model in which cyclin E–mediated effects on mitosis are caused by the inhibition of residual APC activity.
This proposal is seemingly at odds with previous reports that cyclin E–Cdk2 does not bind to Cdh1 or contribute to its phosphorylation in mammalian cells (; ). Indeed, we could not detect an association between endogenous cyclin E and Cdh1. However, when cyclin E was overexpressed by transfection, the association of cyclin E with Cdh1 and the cyclin E–mediated hyperphosphorylation of Cdh1 were readily detectable. These data suggest that when normally expressed at the G1/S boundary, cyclin E is unlikely to have a major impact on Cdh1 regulation. However, overexpressed cyclin E, either in the experimental contexts reported here or in the course of tumorigenesis, can influence Cdh1 activity in a manner that has functional consequences. Whether the temporal deregulation of cyclin E expression, simply the deregulation of cyclin E levels, or both allows cyclin E to have an impact on Cdh1 activity remains to be determined.
Overexpression of cyclin E causes cells to delay in mitosis before anaphase. This phenotype could be explained by a cyclin E–mediated failure to activate the mitotic protease separase. Indeed, it has been reported that one defect caused by the RNAi-mediated depletion of separase is a prometaphase delay similar to that reported here in the context of cyclin E overexpression (). In addition, the depletion of separase induces polyploidy (). Separase specifically cleaves the cohesin subunit Scc1, resulting in a loss of centromeric sister chromatid cohesion in mammalian cells, potentiating anaphase (, ; ; ). The primary mode of separase regulation is through the binding of the separase inhibitor securin (; ). Securin ubiquitylation by the APC/C targeting it for proteasomal degradation during the early phases of mitosis is thought to constitute the principal mitotic regulatory role of the APC/C. However, the role of securin ubiquitylation during early mitosis has been attributed to the APC/C cofactor Cdc20 rather than Cdh1, as the latter cofactor has been assumed to be inactive during this interval (; ; ; ). Indeed, in yeast, the inactivation of Cdc20 leads to metaphase arrest, whereas the inactivation of Cdh1 has no obvious effects on early stages of mitosis (; ; ).
Nevertheless, we show that in mammalian cells, partial silencing of Cdh1 by RNAi leads to the accumulation of securin and an early mitotic delay, and it has been reported that mammalian APC shares the capacity to target securin with APC (). Therefore, it was reasonable to consider securin as the relevant anaphase inhibitory target stabilized by the cyclin E–mediated inhibition of Cdh1. However, when securin expression was adjusted by partial RNAi-mediated silencing so that cyclin E overexpression restored securin to normal levels, early mitotic delay was not abrogated. Recent studies have suggested that cyclin B1–Cdk1 also inhibits separase, raising the possibility that the abnormal accumulation of cyclin B1 during early mitosis may also contribute to cyclin E–mediated mitotic delay (; ). However, as with securin, RNAi- mediated restoration of cyclin B1 levels alone was not sufficient to completely abrogate mitotic delay. Only when the simultaneous restoration of both securin and cyclin B1 was performed were the mitotic effects of cyclin E overexpression neutralized. The contribution of two distinct and possibly synergistic mitotic inhibitors explains why seemingly modest elevation of levels of these proteins has a profound effect on progression through mitosis.
Based on observations made from fixed cells, the overexpression of cyclin E causes an accumulation of cells either in a prometaphase-like state or a state in which chromosomes are almost but not completely aligned on the metaphase plate. However, this categorization was a formalism. Because these are static observations, we cannot describe the sequence of events in real time that led to the accumulation of these cell types. It is possible that both of these phenotypes reflect the same population resulting from an inability to lose sister chromatid cohesion. Normally, centromeric cohesion is lost when bioriented chromosomes are aligned at metaphase. However, the inability to lose cohesion in a timely manner may promote an unusually dynamic situation in which chromosomes under tension cyclically move toward and away from the metaphase plate, potentially accounting for both of the observed populations that accumulate as a result of cyclin E overexpression. Indeed, in time-lapse series of such cells, chromosomes appeared to be unusually dynamic, undergoing substantial oscillatory movements for long periods of time. However, because spindles and centromeres were not visible under these circumstances, it was not possible to evaluate the attachment or tension status of chromosomes. The observation that mitotic chromosomes in cells overexpressing cyclin E appear hypercondensed (unpublished data) is also consistent with a prolonged prometaphase or metaphase caused by the inability to lose cohesion in a timely fashion (; ).
Immortalized human fibroblasts and IME cells, two cell lines immortalized by stable transduction with a pBABEpuro retrovirus expressing the human telomerase reverse transcriptase gene (gift from J. Shay, The University of Texas Southwestern Medical Center, Dallas, TX), were used in initial experiments for chromosome painting, flow cytometry, and live cell microscopy analysis. U2OS cells, an osteosarcoma cell line, were used for immunofluorescence and live cell microscopy experiments. KB cells, a human nasopharyngeal epidermoid carcinoma, and breast cancer–derived cell lines MDA-MB-157, -436, and -468 (American Type Culture Collection) were used for immunofluorescence microscopy and biochemical experiments. IME cells were cultured in MCDB media (Invitrogen) and supplemented with 1% FCS, holotransferrin, EGF, bovine pituitary extract, insulin, -glutamine, 100 U/ml penicillin, and 1.0 mg/ml streptomycin. All other cells were grown in DME (Invitrogen) supplemented with 10% FCS or 20% normal calf serum and 100 U/ml penicillin, 1.0 mg/ml streptomycin, and -glutamine.
Recombinant retroviruses expressing a hyperstable form of cyclin E (T380A) or a control empty vector were prepared using the 293T-derived packaging cell line Phoenix-Ampho and the pBABE retroviral plasmid vector () containing a hygromycin resistance marker for the selection of long-term stable expression mediated by the constitutive retrovirus long terminal repeat promoter. Viral preparation steps are listed as follows: Phoenix-Ampho cells were transfected overnight by calcium phosphate precipitation; 24 h after transfection, medium was changed; 48 h after transfection, virus-containing medium was harvested; and 0.2 μm was filtered and frozen in aliquots for future use. Viral infection steps are listed as follows: frozen viral aliquots were thawed at RT, and retrovirus was added to cells with an equal amount of medium (10% FCS) for 24 h, after which fresh media was added. Cells carrying the retrovirus construct were selected after 48 h using 200 ng/ml hygromycin.
Recombinant adenovirus expressing WT cyclin E or a control empty vector under the constitutive cytomegalovirus promoter were purified by CsCl ultracentrifugation. Cells were transduced with ∼1,000 virus particles/cell (comparable with the levels used in ) in a low volume of medium with 0% serum for 2 h, and then additional medium plus 10% FCS was added for 24 h. Cyclin E expression in IME, U2OS, and KB cells was assayed by Western blotting and immunofluorescence microscopy.
The histone H2B-GFP retrovirus was constructed from an H2B-GFP expression plasmid (BOS H2BGFP-N1; a gift from K. Sullivan, National University of Ireland, Galway, Ireland; ). H2B-GFP was cut and ligated into a pBabe backbone containing a neomycin resistance marker for selection. Viral preparation and transduction were performed as for the cyclin E retrovirus.
Lysates were extracted in cold lysis buffer (50 mM Tris, 150 mM NaCL, 1 mM EDTA, 1% Triton X-100 detergent, and phosphatase/protease inhibitors), sonicated, and cleared for 10 min at 10,000 at 4°C, and protein concentrations were determined using a Bradford solution (Bio-Rad Laboratories) read at 595 nM. For Western blotting analysis, samples were separated by SDS-PAGE and transferred onto Immobilon P membranes (Millipore). Filters were then blocked for 1 h at room temperature in blocking buffer (5% nonfat milk in 1× TBS plus 0.1% Triton X-100 [TBS-T = 10 mM Tris, 500 mM NaCl, and 0.1% Triton X-100, pH 8]). Primary antibody was added in blocking buffer overnight at 4°C. HRP-conjugated secondary antibodies (1:5,000) diluted in blocking buffer were incubated with the membrane for 1 h at RT. Primary antibodies used were anti–cyclin E monoclonal antibody (HE12; Santa Cruz Biotechnology, Inc.) at 1:1,000, anti–cyclin B1 monoclonal antibody (BD Biosciences) at 1:1,000, anti–cyclin A polyclonal antibody (Santa Cruz Biotechnology, Inc.) at 1:1,000, anti-Cdc20 polyclonal antibody (Abcam) at 1:300, anti-Cdh1 monoclonal antibody (Calbiochem) at 1:200, antisecurin monoclonal antibody (Abcam) at 1:200, anti-myc (9E10; Sigma-Aldrich), and anti–β-actin polyclonal antibody (Sigma-Aldrich) at 1:10,000.
Cells were plated onto acid-washed glass coverslips. Cells were fixed for 10 min at room temperature in 2% PFA in 1× PBS and for 10 min in methanol at −20°C or in cold methanol alone for 10 min (for the detection of cyclin E). Cells were then permeabilized with PBS-TX (1× PBS + 0.1% Triton X-100) for 15 min followed by incubation in blocking solution (PBS-TX + 1% BSA) at 37°C for 30 min. Coverslips were incubated with primary antibodies diluted in blocking solution overnight at 4°C in a humidity chamber. Secondary antibodies were used at a dilution of 1:250 (Cy2, Rhodamine red-X, or Cy5; Jackson ImmunoResearch Laboratories) and incubated for 1 h at RT. Cells were washed and dehydrated with sequential washes of 70, 90, and 100% ethanol. Coverslips were mounted using SloFade with DAPI (Invitrogen). Primary antibodies used were anti–cyclin E monoclonal antibody (HE12) at 1:1,000, anti–cyclin B1 monoclonal antibody (BD Biosciences) at 1:1,000, antiphosphorylated histone H3 (mitotic marker; Upstate Biotechnology) at 1:400, anti–α-tubulin (Dm1a; Sigma-Aldrich) at 1:1,000, anti-Sm human autosera (centromere-binding protein; gift from K. Sullivan) at 1:100, and sheep polyclonal anti-BrdU (Research Diagnostics, Inc.).
Images were collected at intervals of 0.3 μm in the z direction on a wide-field optical sectioning microscope system (DeltaVision; Applied Precision) based on an epifluorescence microscope (IX70; Olympus). A 60× or 100× 1.35 NA neofluor oil immersion lens (Olympus) was used. Images were processed using a constrained iterative deconvolution algorithm. All images shown are projections of multiple focal planes, which were created using SoftWoRx analysis software (Applied Precision) and contain information from 3D image stacks.
IME or U2OS cells expressing H2B-GFP were grown directly on glass coverslips. Cells were maintained during imaging in phenol red–free DME plus 10 mM Hepes, pH 7.5, buffer in a closed chamber system (Bioptechs) on a heated stage at 37°C. 5 3-μm sections of differential interference contrast and FITC fluorescence images were acquired at 1-min intervals with a cooled CCD camera mounted on a deconvolution microscope system (DeltaVision; Applied Precision).
For flow cytometry, cells were collected by trypsinization, washed with 1× PBS, and fixed with 70% ethanol (added dropwise with agitation) overnight at −20°C. Samples were then washed and resuspended in 1 ml PBS-TX plus 10 μg/ml RNase solution and 2 μg/ml propidium iodide overnight at 4°C. For BrdU staining, cells were pulsed with 10 μM BrdU for 20 min before collection. Cells were treated with 1 ml HCL + 0.5% Triton X-100 for 30 min at RT to denature DNA. The cell suspension was neutralized with 2 ml of 0.1 M Borax. Cells were washed with 1× PBS and resuspended in 50 μl of 1× PBS/1% BSA/0.5% Tween 20 and incubated in 2.5 μg/ml FITC-conjugated anti-BrdU antibody for 1 h at RT. Cells were then washed and incubated with propidium iodide overnight. Before analysis, cells were filtered through a 74-μm mesh (Small Parts) to remove cell clumps before detection on a flow cytometer (FACSCalibur II; Becton Dickinson). For each cell line, 30,000 cells were counted per sample. Data were analyzed using CellQuest (Becton Dickinson). All cell lines were analyzed using the same user-defined Gates and Regions to obtain percentages of cells with a G1, S, or G2/M DNA content.
KB cells were transduced with EV control or cyclin E adenovirus for 2 h and incubated in medium with 1.5 mM thymidine overnight (∼18 h). Cells were released into S phase and through mitosis by adding 10% FCS medium (with or without 100 nM nocodazole as indicated) after three washes in warm medium. Cells were collected by trypsinization at various time points and split into samples for Western blotting and flow cytometry.
siRNAs (21-mers) targeting GFP, Cdh1, securin, and cyclin B (siGENOME; Dharmacon) were used to reduce protein expression in KB cells. Cells were transfected using LipofectAMINE 2000 (12 μl/10-cm dish; Invitrogen) with siRNA (10 pM cyclin B1/securin-specific siRNA, 5 nM Cdh1-specific siRNA, and GFP-specific siRNA up to 20 nM total siRNA concentration or 20 nM GFP-specific siRNA alone). Complexes were incubated for 20 min at RT before adding to cells for 24 h.
Cell lysates were subjected to immunoprecipitation followed by histone H1 kinase assay or Western blotting. IME or KB cells were lysed in lysis buffer (50 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100 detergent, and phosphatase/protease inhibitors) on ice, sonicated, and cleared for 10 min at 10,000 at 4°C, and protein concentrations were determined using a Bradford solution (Bio-Rad Laboratories) read at 595 nM. For kinase assays, 7.5 μg anti-Cdk2 or IgG1 antibody was added to the lysates with γ-bind G Sepharose beads (100 μl of a 50% slurry; GE Healthcare) for 2 h at 4°C. Beads were washed three to five times with lysis buffer and twice with 1× reaction buffer (without ATP). Beads were resuspended in 2× reaction buffer (40 mM Tris-HCl, pH 7.5, and 15 mM MgCl), 20 μM ATP, 20 μg histone H1, and 10 μCi γ-[P]ATP and incubated for 30 min at 37°C. Samples were prepared and run on an 11% SDS-PAGE gel, and autoradiography was performed or phosphorylation was quantified using a phosphorimager (Storm 840; Molecular Dynamics).
Cdh1 kinase assays were performed with 2 μg –purified GST-Cdh1 as a substrate in the kinase reaction. 10 μg GST peptide alone and 2 μg histone H1 kinase assays were used to compared cyclin E/Cdk2 kinase activity.
GST-Cdh1-WT (a gift from C. Sorensen, Danish Cancer Society, Institute of Cancer Biology, Copenhagen, Denmark) had been cloned into the GST-2TK plasmid (GE Healthcare). The fusion protein was expressed in bacteria after induction with IPTG, bound to glutathione–Sepharose beads, and eluted with reduced glutathione. The washes and eluted fractions were run on an SDS-PAGE gel and stained with Coomassie to evaluate purity. GST-Cdh1 was observed running at 75 kD with degradation bands running below.
Myc-tagged Cdh1 constructs (WT, AAA, and 4A or empty vector pX-myc; gifts from C. Sorensen), which were described previously (), were expressed in 293T cells along with purified cyclin E or cyclin A using calcium-phosphate transfections and were collected after 48 h of transfection. 9E10 anti-myc antibody (Sigma-Aldrich) was used for immunoprecipitation. Beads were boiled in SDS-running buffer and analyzed by Western blotting. Whole cell lysates and IPs were immunoblotted for cyclin E, cyclin A, or Cdh1.
For kinase inhibition of Cdh1, 2 μl Cdh1 proteins generated with the TNT Quick Coupled Transcription/Translation kit (Promega) with or without ∼40 ng cyclin–Cdk2 complexes were diluted in kinase buffer (50 mM Tris, pH 7.5, 10 mM MgCl, 2 mM DTT, 200 μM ATP, 5 μM okadaic acid, and 1 mM EGTA) to a final reaction volume of 10 μl and incubated at 30°C for 1 h. 100 nM purvalanol A was added to the kinase reactions either before the addition of Cdh1 (with a 15-min preincubation at 30°C) or at the end of the reaction with an additional 15-min incubation to quench the kinase activity.
Anti-Cdc27 polyclonal antibodies coupled to an Affiprep support (Bio-Rad Laboratories) were used to precipitate the APC from interphase egg extracts. The beads were washed four times in buffer XB (20 mM Hepes, pH 7.7, and 100 mM KCl) supplemented with 0.5% NP-40 and 400 mM KCl and then were washed twice in XB. Cdh1 proteins were then added to the APC beads and incubated at 4°C for 30 min. The beads were washed twice in XB and resuspended in 5 μl of reaction cocktail (20 ng E1, 100 ng UbcH10, and 1.25 μg ubiquitin)/μl in XB supplemented with an energy regenerating system and 1 μl S-labeled cyclin B1 N-terminal fragment. The reactions were incubated at RT, and samples were taken at the indicated times, resolved by SDS-PAGE, and imaged using a phosphorimager (Typhoon; Molecular Dynamics). Densitometry was performed on the nonmodified substrate using ImageQuant software (Molecular Dynamics). Data was normalized to input and control reactions.
Video 1 shows the cyclin E–mediated early mitotic delay presented in c. Video 2 presents cyclin E–mediated early mitotic delay shown by live cell microscopy in a U2OS cell. Video 3 shows cyclin E–mediated mitotic failure in a U2OS cell. Fig. S1 shows cell cycle analysis of adenovirally expressed cyclin E. Fig. S2 shows a comparison of cyclin E levels during the cell cycle between adenovirally transduced cell lines used in this study and breast cancer–derived cell lines expressing high levels of cyclin E. Fig. S3 shows an analysis of the cyclin E–mediated accumulation of polyploidy cells during one cell cycle equivalent. Online supplemental material is available at . |
For epithelial cells to function, they must polarize into apical and basolateral membranes. The basolateral membrane is defined by lateral cell–cell adhesion and basal interactions with the extracellular matrix. In mammalian epithelia, the tight junction acts as a landmark separating the apical from the basolateral surface. Work over the last 20 yr has begun to elucidate the molecular mechanisms that contribute to the formation of apicobasal polarity in epithelial cells (). Important to this progress were studies in and that elucidated key proteins that were necessary for cell polarization (; ). These included a scaffold protein called Stardust and an apical transmembrane protein called Crumbs. Crumbs is a protein that defines the apical membrane, as overexpression leads to an expansion of the apical membrane in epithelia (). In mammals, there are three isoforms of Crumbs. Crumbs1 was first identified as RP12, a gene mutated in a subset of patients with retinitis pigmentosa and Leber congenital amaurosis (). Crumbs2 is found in the brain, eye, and kidney, but its function is unclear ().
We and others have extensively characterized Crumbs3 (CRB3; ; ; ). The expression of CRB3 is much broader than the other mammalian Crumbs isoforms. CRB3 has been shown to be important for epithelial polarity and tight junction formation. Recently, we have also shown an important role for CRB3 in ciliogenesis by epithelial cells, and similar results have been obtained in zebrafish (; ). Although CRB1 and CRB2 isoforms as well as Crumbs have a large extracellular domain with EGF and Laminin repeats, CRB3 has only a small extracellular domain. However, all Crumbs proteins have a highly conserved intracellular domain that ends with the sequence ERLI. This sequence binds to at least two scaffold proteins that are important for cell polarization, Stardust/PALS1 (protein associated with Lin-7) and Par6 (; ; ; ). The binding of Crumbs to these scaffold proteins is crucial for it to act as an effector in polarity determination and tight junction formation (; ; ). However, it has been clear since the identification of mammalian CRB3 that there is an alternate splice form that ends with the C-terminal sequence, CLPI. In this paper, we describe an important role for this CRB3 isoform in ciliogenesis as well as cell division and report its interaction with importin β-1
A splice form of CRB3 (CRB3 isoform b) is generated by alternate splicing within the fourth exon of the CRB3 gene, leading to a divergent 23-amino-acid sequence at its C terminus ending in the sequence CLPI (). Inspection of expressed sequence tags (ESTs) indicate that the CRB3-CLPI splice form can be found in human, mouse, rat, cow, and dog, but apparently not in or zebrafish. EST prevalence indicates a wide tissue expression for both CRB3-ERLI and CRB3-CLPI in rodent and human.
We generated and purified polyclonal antibodies against the last 20 amino acids of CRB3-CLPI. Using immunoblotting, we detected expression of CRB3-CLPI in multiple cell lines (). As we had previously found with CRB3-ERLI (), multiple forms of CRB3-CLPI were seen on blotting, which is due, at least in part, to differential glycosylation. Our previous work indicated that CRB3-ERLI is localized to the apical surface, tight junction, and cilia in MDCK cells (; ). We were not able to detect specific localization of CRB3-CLPI in newly polarized MDCK cells that did not have cilia. However, once the cells fully differentiated, we could detect endogenous CRB3-CLPI in cilia (, top). To confirm this result, we expressed a full-length Myc–CRB3-CLPI construct in MDCK cells. In this construct, the Myc tag was placed in the extracellular domain near the signal peptide. We detected this transfected Myc–CRB3-CLPI using two methods. One used CRB3-CLPI antibody at a high dilution of 1:1,000 that could not detect endogenous CRB3-CLPI (, middle); the other used anti-Myc 9E10 monoclonal antibody (, bottom). With both these reagents, we could detect cilia staining of the transfected CRB3-CLPI. We further verified the specificity of the CRB3-CLPI antibody by adding the CRB3-CLPI antigenic peptide to the immunostaining. This peptide blocked the anti–CRB3-CLPI staining of both endogenous and transfected proteins (Fig. S1 a, available at ).
Cilia have been the focus of recent studies because of the association of cilia-localized proteins with many human diseases, including polycystic kidney disease (; ). It has been suggested that defects in spindle pole polarity contribute to the genesis of polycystic kidney disease (; ). Accordingly, we studied the localization of CRB3-CLPI during the cell cycle. As we previously stated, we could not detect the localization of endogenous CRB3-CLPI during interphase in MDCK cells. However, at prophase, CRB3-CLPI was concentrated around the centrosomes (). As the cells progressed through metaphase and anaphase, there was a close colocalization of CRB3-CLPI and centrosomes. This continued through anaphase, but at telophase, CRB3-CLPI appeared diffuse throughout the cell and could not be clearly localized.
We also tested the localization of the Myc–CRB3-CLPI protein during the cell cycle staining both with the diluted 1:1,000 CRB3-CLPI antibody or Myc antibody (). A sharp localization of the transfected CRB3-CLPI protein to a pericentrosomal location in metaphase could be detected using either antibody. In some of these cells, CRB3-CLPI was also detected at the cell cortex, but this was felt to be due to overexpression, as this was never seen when staining the endogenous protein. This result indicates that CRB3-CLPI marks a pericentrosomal membrane component, as the Myc tag is in the extracellular domain and the CRB3-CLPI antibody epitope is in the intracellular domain. We obtained similar CRB3-CLPI staining results in Cos-7 cells (Fig. S1 b). Cos-7 cells express more endogenous CRB3-CLPI than MDCK cells, and during interphase, the CRB3-CLPI could also be detected in a pericentrosomal region, consistent with Golgi localization. During mitosis, the CRB3-CLPI in Cos-7 cells localized to a tight dot surrounding the centrosome, as was seen with MDCK cells. Interestingly, we also found that CRB3-CLPI localized to the midbody during cytokinesis (Fig. S1 b). The localization of the CRB3-CLPI was markedly different from that seen with CRB3-ERLI. At interphase in MDCK cells, CRB3-ERLI was seen apically and at tight junctions as previously described (Fig. S1 c). During metaphase, anaphase, and telophase, however, CRB3-ERLI was diffuse throughout the MDCK cells. There was no localization around the spindle poles, as was seen with CRB3-CLPI.
Our previous studies, as well as studies in zebrafish, have demonstrated that removal of Crumbs affects ciliogenesis (; ). However, in our previous studies, we used short hairpin RNA (shRNA) constructs that could have eliminated both the CRB3-ERLI and CRB3-CLPI isoforms. Accordingly, we transfected MDCK cells with a pSilencer shRNA specifically directed toward CRB3-CLPI or CRB3-ERLI and selected stable cell lines. We were able to obtain a considerable knockdown of the CRB3-CLPI protein (). In CRB3-CLPI, no defects in tight junctions were detected, unlike what was seen with CRB3-ERLI–specific knockdowns (). However, we noted that many cells with CRB3-CLPI knockdown displayed a multinuclear phenotype that was not seen in the CRB3-ERLI knockdown (). This was seen in both clones 1 and 2 MDCK knockdown cell lines, which were generated using different shRNA constructs (see Materials and methods). Because the expression of CRB3-ERLI was slightly increased (, middle) in cells with CRB3-CLPI knockdown, we needed to exclude the possibility that the multinuclear phenotype of CRB3-CLPI knockdown was a consequence of the increased level of CRB3-ERLI expression. We studied stable MDCK cell lines that overexpress CRB3-ERLI (). However, the multinuclear phenotype was not detected in these cells, indicating that the overexpression of CRB3-ERLI did not contribute to this phenotype (Fig. S2, available at ). This multinuclear phenotype was associated with markedly abnormal mitotic spindles (). The most common defect seen was multiple spindle poles (∼80%), but misaligned and disorganized bipolar spindles were also seen.
CRB3-CLPI knockdown cells often contained supernumerary centrosomes (), and this likely contributed to these cells having multiple spindle poles. We found that the presence of supernumerary centrosomes coincided with multinuclei in CRB3-CLPI knockdown cells (). Almost all the multinucleated cells had more than two centrioles (supernumerary centrosomes), whereas only 12% of mononuclear cells had more than two centrioles (). We found that once the CRB3-CLPI knockdown cells grew to confluence, the multinuclear cells with supernumerary centrosomes were markedly reduced and mononuclear cells predominated. This was not due to loss of the CRB3-CLPI knockdown, as when these cells were diluted and replated, the multinuclear cells reappeared (Fig. S3, available at ). We hypothesize that as cells reached confluence, the multinuclear cells disappeared, possibly as a result of apoptosis (; ; ). The cell division defect appears to be stochastic, as a certain fraction of the cells are mononuclear, divide properly, and predominate in the confluent monolayer. We next examined the role of CRB3-CLPI in ciliogenesis. We allowed cells to grow on filters for 7 d to achieve confluence to the point where multinuclear cells were rare (Fig. S3). At this stage, we found that CRB3-CLPI knockdown MDCK cells had a defect in ciliogenesis (). The loss of cilia could be due, in part, to a polarity defect, as many knockdown cells failed to localize centrosomes and Golgi during polarization to a subapical localization ().
To determine if these phenotypes were specific for CRB3-CLPI knockdown cells, we reexpressed Myc–CRB3-CLPI in these knockdown cells using a cDNA that was resistant to the shRNA. We were able to demonstrate that reexpression of the Myc–CRB3-CLPI but not vector alone could reverse the multinuclear phenotype (). In addition, we noted that the abnormal localization of centrosomes in growth-arrested CRB3-CLPI knockdown cells could be reversed (). Reexpression of CRB3-CLPI not only reversed the abnormal centrosomal phenotype but also restored Golgi localization to the apical region of the MDCK cells. However, we were still not able to detect cilia in these rescued cells, perhaps because of the level of overexpression of Myc–CRB3-CLPI or as an effect of the Myc tag.
The CRB3-ERLI protein can interact with the polarity proteins PAR6 and PALS1 (; ). However, we were not able to demonstrate such interactions with CRB3-CLPI (unpublished data). Accordingly, we performed large-scale anti-Myc immunoprecipitations from MDCK cells expressing Myc–CRB3-CLPI, Myc–CRB3-ERLI, or vector alone and looked for differences in interacting proteins. A specific band of ∼100 kD detected by Myc–CRB3-CLPI immunoprecipitation was excised and sent for liquid chromatography/mass spectrometry (MS; ). Analysis yielded 19 matching peptides and 28% coverage for mouse importin β-1. MS/MS analysis yielded two peptides that matched mouse importin β-1, AAVENLPTFLVELSR and WLAIDANAR.
Importin β-1 directly or via its interactions with importin α isoforms and Ran GTPase facilitates trafficking of proteins to the nucleus (; ). However, recent studies have also suggested an important role for these proteins in mitotic spindle generation and centrosome maintenance (for review see ); thus, it appeared that importin β-1 was a good candidate for a CRB3-CLPI binding partner. Indeed, we were able to show that importin β-1 colocalized with CRB3-CLPI during mitosis (). This was in agreement with previous reports on importin β-1 targeting spindle assembly factors during mitosis (; ; ). In addition, we were able to show that importin β-1 colocalized to the cilia with CRB3-CLPI (). We next examined the coimmunoprecipitation of CRB3-CLPI with importin β-1. Myc–CRB3-CLPI, Myc–CRB3-ERLI, or MDCK wild-type (wt) cells were transfected and immunoprecipitated with the Myc antibody. We found that endogenous importin β-1 would coimmunoprecipitate with Myc–CRB3-CLPI but not with Myc–CRB3-ERLI or in control MDCK cells (). This result was confirmed by coexpressing Flag–importin β-1 and Myc–CRB3-CLPI in Cos-7 cells. We were able to immunoprecipitate Flag–importin β-1 with anti-Myc antibodies only in cells expressing both tagged proteins (). Similarly, Flag immunoprecipitation brought down Myc–CRB3-CLPI but not Myc–CRB3-ERLI (). Similarly, full-length importin β-1 GST fusion protein was also able to precipitate Myc–CRB3-CLPI from lysates (Fig. S4, available at ). The ability of the importin β-1 GST fusion protein to bind importin α is used as a control in Fig. S4. We noted that the colocalization of Myc–CRB3 CLPI and importin β-1 was strongest during cell division and found that Myc–CRB3-CLPI more strongly interacted with importin β-1 shortly after release of a mitotic block ().
Importin β-1 often interacts with cargo via importin α, and this interaction is regulated by the Ran small GTPase. However, we were not able to detect importin α in the CRB3-CLPI immunoprecipitates (unpublished data). This may not be surprising, as it has been demonstrated that importin β-1 can bind cargo proteins in the absence of importin α (; ). However, we were able to demonstrate that Ran regulates the interaction of importin β-1 with CRB3-CLPI (). Transfection of GTP-Ran (Q69L) but not GDP-Ran (T24N) blocked the interaction of CRB3-CLPI with importin β-1.
To further assess the functional importance of the interaction, we generated a dominant-negative importin β-1 missing the N terminus and transfected it into MDCK cells. This dominant-negative form of importin β-1 is missing the Ran GTPase binding motif but is still able to interact with Myc–CRB3-CLPI () and target to spindle poles (Fig. S5 a, available at ). We were able to show that overexpression of this dominant-negative importin β-1 closely phenocopied CRB3-CLPI shRNA with multinuclear cells (), and abnormal spindle poles () with supernumerary centrosomes (). In addition, we also saw the loss of cilia in these cells ().
Finally, we looked at targeting of importin β-1 in cells missing CRB3-CLPI. Although many of these cells showed abnormal spindles with multiple centrosomes, importin β-1 was seen concentrated around the spindle poles, suggesting that the CRB3-CLPI was not essential for this targeting of importin β-1 (). Next, we examined the effects of importin β-1 knockdown on CRB3-CLPI targeting. We were unable to obtain clonal cells with sustained knockdown of importin β-1, presumably because of toxicity induced by loss of this protein. However, we were able to transiently express double-stranded shRNA and study knockdown cells within 48 h of transfection (). As might be expected, cells lacking importin β-1 demonstrated abnormal mitotic spindles (for review see ), and these spindles consistently lacked CRB3-CLPI staining (). This indicates that importin β-1 played an important role in targeting CRB3-CLPI but not vice versa.
Studies to date have begun to reveal the role of the Crumbs proteins in multiple developmental systems from to zebrafish. Crumbs proteins have a conserved ERLI motif at their C terminus that binds to PALS1 and Par6 and is crucial for their function (; ; ). However, there is an alternate splice form of mammalian CRB3 that adds 23 unique amino acids to the C terminus. The CRB3-CLPI isoform concentrates in a membrane compartment that localizes around the centrosome. Loss of CRB3-CLPI leads to defects in spindle assembly, cilia formation, and cell division. CRB3-CLPI interacts in a Ran-regulated fashion with importin β-1, and this interaction appears important for CRB3-CLPI targeting to the pericentrosomal region.
The defect that leads to the multinuclear phenotype most likely represents a cytokinesis defect, and indeed we found CRB3-CLPI localized to the midbody in Cos-7 cells. Studies have pointed to an evolutionarily conserved role for the centrosome in cytokinesis (for review see ). For example, work from , ) described an essential process in which the centrosomal protein Centriolin anchors the exocyst and SNARE complexes and guides vesicle transport to the midbody in the final stages of cytokinesis. Others have shown an important role for the centrosomal Bardet-Biedl syndrome proteins in cytokinesis ().
In addition to a cytokinesis defect, it is also possible that correct localization of this CRB3-CLPI–containing membrane is necessary for mammalian cells to complete cell division, as is seen with members of the Golgi matrix (). Inheritance of Golgi membranes is perhaps the best-studied example of membrane organelle inheritance and is due to vesiculation and dispersion of the membrane (). It has been argued that this diffuse distribution in the cytoplasm of mitotic cells ensures equal inheritance; however, recent studies indicate that mitotic Golgi fragments also align with astral microtubules at the spindle poles in certain cell types (). Indeed, members of the Golgi matrix can regulate cell cycle progression, perhaps ensuring proper Golgi inheritance before cell division can be completed (, ). However, the localization of GM130, a membrane Golgi marker, and CRB3-CLPI was not identical during cell division, indicating that CRB3-CLPI marks a different compartment (Fig. S5 b). Early endosomes are another membrane compartment that exists in a pericentrosomal distribution early during cell division and contributes to cytokinesis (), but we saw no colocalization between early endosome markers and CRB3-CLPI (Fig. S5 b).
In addition to a role in cytokinesis, the CRB3-CLPI–containing pericentrosomal membrane appears to contribute to the formation of the cilia. It has long been known that a pericentrosomal ciliary vesicle covers the centrosome during early ciliogenesis (), and other cilia membrane components localize near the centrosome during cell division (). We hypothesize that CRB3-CLPI also exists in this early cilia membrane. Loss of cilia was seen with two different shRNAs directed against CRB3-CLPI; however, we could not rescue the cilia defect with shRNA-resistant CRB3-CLPI despite rescuing the cell division defect. It should be noted that we have not been able to rescue the cilia defect seen in CRB3-ERLI knockdowns either, but the cilia defect with the CRB knockdown has been seen both in mammalian cells and zebrafish (; ). Our rescue studies lead to the overexpression of CRB3, and it is well known that overexpression of Crumbs proteins can affect cellular phenotypes (; ; ). A membrane defect is not the only possible mechanism for the lack of cilia in the CRB3-CLPI knockdown cells. We have also seen striking defects in centrosomal and Golgi targeting within the CRB3-CLPI knockdown cells; the exact mechanism of these defects is unclear, as we did not detect interactions of this CRB3 isoform with other polarity proteins, such as PALS1 or Par6. Thus, the exact role of CRB3-CLPI in cilia formation will require additional studies.
Another major finding in our studies is an interaction between CRB3-CLPI and importin β-1. In nuclear translocation, importin β binds cargo directly or indirectly through importin α. Upon entering the nucleus, the cargo is released when importin β binds to the Ran GTPase (; ; ; ). CRB3-CLPI may directly bind to importin β-1, as importin α did not immunoprecipitate with CRB3-CLPI. We find that GST–importin β-1 can precipitate CRB3-CLPI from cell lysates (Fig. S4); however, we have not been able to demonstrate that a GST–CRB3-CLPI intracellular domain can precipitate importin β-1 under similar conditions. Thus, it is not yet clear whether the interaction is direct or additional proteins are involved. It is interesting to note that we detected increased binding of CRB3-CLPI to importin β-1 after release of mitotic arrest, suggesting that a posttranslational modification such as phosphorylation might be involved.
We also observed that the interaction of CRB3-CLPI with importin β-1 was regulated by Ran GTP. As has been found with other importin β-1 interactions, Ran GTP weakened the interaction between importin β-1 and this cargo. Ran–importin β complexes play a fundamental role during mitosis, including targeting spindle assembly factors (; ; ; ; ; ). Ran is also concentrated at centrosomes and is thought to regulate centrosome cohesion, as overexpression of RanBP1 leads to abnormal centriole splitting (). The Ran network also regulates centrosome duplication and spindle assembly (; ; ). These defects in centrosome duplication and cohesion can lead to the multiple spindle poles seen in cells with perturbed Ran signaling. Consistent with these results are the findings that importin β-1 overexpression also leads to abnormal spindles, possibly because of defects in centriole cohesion (). Multiple spindle poles and supernumerary centrioles were also seen with loss of CRB3-CLPI from cells by shRNA, an effect that was rescued by the reexpression of CRB3-CLPI. These results suggest that CRB3-CLPI can be delivered to spindle poles by importin β-1 during mitosis, and this delivery may be important for centrosome maintenance, in addition to concentrating specific membrane components near the centrosome. Although it is intriguing to implicate complex mechanisms of centrosome maintenance as the cause of the supernumerary centrosomes, it is also likely that many of the cells had supernumerary centrosomes as a result of cytokinesis defects. Indeed, there was a strong correlation between multiple nuclei and supernumerary centrosomes in our studies, suggesting that cytokinesis defects could have played a large role in the centrosome abnormalities. In fact, we saw multiple centrosomes in almost all multinuclear cells. However, there were ∼12% of cells that had a single nucleus and supernumerary centrosomes, suggesting that CRB3-CLPI knockdown might have a direct effect on centrosomes in addition to the cytokinesis defect.
We also detected importin β-1 with CRB3-CLPI in the cilia by immunostaining. Indeed, proteomic studies have identified importin family members in the centrosome and cilia (; ). The finding of a connection between nuclear proteins, cilia, and centrosomes described in this paper is not unique (). It was especially interesting to see the loss of cilia in cells expressing dominant-negative importin. Recently, a hypothesis was generated suggesting that there may be similarities between the nuclear pore complex proteins and intraflagellar transport proteins (). Importins that interact with the nuclear pore complex might also have similar types of interactions with the intraflagellar transport complex delivering cilia proteins such as CRB3-CLPI.
The finding of importin proteins in the cilia is also of great interest because of recent data demonstrating the signaling pathways that lead from the cilia to the nucleus. The best documented of these is the hedgehog pathway that leads to processing of gli transcription factors (). It has been suggested that this processing may occur in the cilia, and the processed gli products would need to be sent to the nucleus (). Several other cilia to nuclear signaling pathways have been described necessitating the need for the trafficking of proteins from the cilia to the nucleus (; ). Therefore, importins may have a role transporting proteins from the cilia directly to the nucleus. In summary, our findings describe a unique membrane compartment containing CRB3-CLPI that lies close to centrosomes during cell division and ciliogenesis. They also indicate an important role for this membrane compartment not only in ciliogenesis but also in cell cycle control and possibly polarity determination. Finally, they point to a new role for the multipurpose importin family in delivering cellular components to the centrosome.
CRB3-ERLI constructs were previously described (; ). For expression of Myc–CRB3-CLPI, full-length CRB3-CLPI was amplified from a human embryo cDNA library and cloned into pcDNA3.1 Zeo (+) vector via BamH and Not1 sites. Then, using single primer mutagenesis, a single Myc tag was placed behind the signal peptide (). To rescue CRB3-CLPI shRNA clones in MDCK cells, we deleted the CMV promoter of pcDNA3.1 Myc–CRB3-CLPI to decrease the expression level of the transfected construct. For expression of Flag–importin β-1 wt and Flag–importin β-1 N-deletion, we amplified human full-length importin β-1 from an EST clone (American Type Culture Collection) and subsequently ligated the amplified product to p3xFLAG-CMV-9 vector and pGSTag vector (Sigma-Aldrich) via a BamH1 site. Then, we deleted the first 360 amino acids of importin β-1 full length to generate Flag–importin β-1 N-deletion. Full-length human GTPase Ran was amplified from an EST clone (American Type Culture Collection) and subsequently ligated to pcDNA3.1 Zeo(+) vectors via BamHI and Xho1 sites. Ran Q69L and Ran T24N were generated by mutagenesis using single primers ().
MDCK II, HeLa, and COS-7 cells were cultured as described previously (; ). MDCK cells were transfected with Myc–CRB3-CLPI or Myc–CRB3-ERLI (FuGENE 6 transfection Reagent; Roche) and cultured in DME complete media supplemented with 200 μg/ml Zeocin (Invitrogen) for 10–14 d, and clones were selected. Flag–importin β-1 N-del stable cell lines were cultured in DME media with 600 μg/ml G418 to obtain clones.
Double-stranded oligonucleotides corresponding to canine CRB3-CLPI 3′ nontranslated sequences TAGCAGGGAAGAAGGTACT and GAAGGTACTTCAAAGACTC were selected for CRB3-CLPI shRNA targeting sequences and inserted into the pSilencer vector (Ambion). Stable knockdown clones were selected in 200 μg/ml Hygromycin B. CRB3-ERLI shRNA stable knockdown clones were selected as described for CRB3-CLPI shRNA clones using the canine targeting sequence of CCTCAAGCTGCCACCCGAG. Double-stranded oligonucleotides corresponding to canine importin β-1 sequences ACCCCAACAGCACAGAGCA and GAGGATGCCCTGATAGCAG were selected as importin β-1 shRNA targeting sequences using the pSilencer vector. Importin β-1 transient knockdown was induced by importin β-1 shRNA transfection using Lipofectamine 2000 (Invitrogen) for 48 h.
We performed immunostaining as described previously (). Rabbit anti–CRB3-CLPI was made against peptides of NHAAEARAPQDSKETVRGCLPI. Mouse anti–Flag M2, mouse anti–acetylated tubulin, mouse anti–α-tubulin, mouse anti–γ-tubulin (Sigma-Aldrich), rat anti–α-tubulin (Chemicon), mouse anti–importin β-1 (ABR Affinity BioReagents and BD Biosciences), mouse anti–importin α/Rch-1, mouse anti-Ran, mouse anti-EEA1, mouse anti-Rab11, mouse anti-GM130 (BD Biosciences), rabbit anti-Giantin, and rabbit anti-pericentrin (Covance) were used for immunofluorescence or immunoblots. Rabbit anti–CRB3-ERLI was as previously described (). All images were obtained using a meta laser-scanning confocal microscope (LSM 510; Carl Zeiss MicroImaging, Inc.). Samples were scanned with appropriate lasers and filter sets, and images were collected at 0.5-μm intervals on an inverted microscope (Axiovert 100M; Carl Zeiss MicroImaging, Inc.) using a 63× water objective (C-Apochromat) with 1.2 NA. LSM 510 meta software (Carl Zeiss MicroImaging, Inc.) was used to collect images. Images were analyzed with LSM image browser (Carl Zeiss MicroImaging, Inc.), and subsequent preparation was performed using Creative Suite software (Adobe). 2D images were taken using a 60× oil objective with 1.4 NA (Plan Apo) on an inverted microscope (Eclipse TE2000U; Nikon). Image acquisition was performed with MetaMorph software and a charge-coupled device camera (Carl Zeiss MicroImaging, Inc.).
Lysis buffer (50 mM Hepes, 150 mM NaCl, 1.5 mM MgCl, 1 mM EGTA, 1% Triton, and 10% glycerol) with protease inhibitor cocktail tablets (Roche) and phosphatase set I and II (EMD Bioscience) was used to extract cells. Antibodies to Myc 4A6 (Upstate Biotechnology), Flag M2 (Sigma-Aldrich), or importin β-1 (BD Biosciences) were added to Cos-7, HeLa, or MDCK cell extracts overnight at 4°C. 50 μl of 50% protein A/G beads (Zymed Laboratories) was added to the lysate for 2 h to bind the antibodies. After washing, the immunoprecipitates were eluted with sample buffer, separated by Bis-Tris PAGE, transferred to nitrocellulose, and immunoblotted (). Large-scale anti-Myc immunoprecipitation of Myc–CRB3-CLPI and Myc–CRB3-ERLI MDCK stable cells was performed as described previously (). The specific bands that coimmunoprecipitated with Myc–CRB3-CLPI were cut from the gel and analyzed at the Michigan Proteome Consortium using a 4800 Proteomic Analyzer (Applied Biosystems).
MDCK II, CRB3-CLPI, and CRB3-ERLI shRNA MDCK stable cells were grown on transwell filters until confluent. After washing with cold PBS (without calcium) three times, low calcium media (5 μM Ca) was added to the cells overnight. The next day, DME complete media (2 mM Ca) was added to the cells that were then fixed and stained at the time points indicated ().
For synchronizing HeLa cells, 100 ng/ml Nocodazole (Sigma-Aldrich) in DME complete media was added to the cells for 12 h. After washing three times with ice-cold PBS, cells were placed in warm DME complete media and lysed 30 min later.
Fig. S1 shows that the CRB3-CLPI antigenic peptide blocks the anti–CRB3-CLPI staining of both endogenous and transfected proteins, CRB3-CLPI localizes to spindle poles and the midbody during mitosis in COS-7 cells, and CRB3-ERLI does not localize to the spindle poles during mitosis. Fig. S2 shows that overexpression of Myc–CRB3-ERLI does not induce the multinuclear phenotype in MDCK cells. Fig. S3 shows that after growth arrest, CRB3-CLPI knockdown cells did display multinuclei and supernumerary centrosomes. Fig. S4 shows that an importin β-1 GST fusion protein is able to precipitate Myc–CRB3-CLPI. Fig. S5 shows that the Flag–importin β-1 N-deletion mutant protein colocalizes with CRB3-CLPI in spindle poles during mitosis in MDCK cells, and CRB3-CLPI does not colocalize with GM130, EEA-1, or Rab11. Online supplemental material is available at . |
xref
italic
#text
To find Rabs involved in primary cilium formation, the 39 predicted human RabGAPs were tested for their ability to prevent primary cilium formation in telomerase-immortalized retinal pigmented epithelial (hTERT-RPE1) cells ( and Fig. S1 A, available at ). This revealed that cells expressing TBC1D7, EVI5like, and (available from GenBank/EMBL/DDBJ under this accession no.) were compromised in their ability to form primary cilia (). Further support for a role of TBC1D7 and XM_037557 at primary cilia comes from the observation that they overlap with γ-tubulin to the basal body and that catalytically inactivate XM_037557 is present on the cilia (). Other GAPs either had no effect on primary cilia or, in the case of TBC1D3, caused a reduction in primary cilia accompanied with increased levels of cell death, and this is therefore unlikely to represent a specific effect (). Notably, GAPs that block Rab1-dependent secretion or Rab5-dependent endocytosis, TBC1D20 and RabGAP5 (; unpublished data), respectively, did not have any effect on primary cilium formation (). General perturbation of membrane trafficking is therefore unlikely to explain the effects of TBC1D7, EVI5like, and XM_037557 on primary cilia formation. EVI5like, TBC1D7, and XM_037557 therefore represent good candidates for GAPs controlling specific Rabs involved in primary cilium formation.
Strikingly, these GAPs showed great specificity toward single Rabs when tested in biochemical assays (). This approach showed that EVI5like acts on Rab23, whereas XM_037557 acts on Rab8a, and TBC1D7 acts on Rab17 (). Consistent with these biochemical data and the effects of GAP expression ( and ), dominant-negative forms of Rab8a, -17, and -23, but not the other Rabs tested, including Rab8b, prevented primary cilium formation (Fig. S1 B). Intriguingly, Rab23 has previously been implicated as a downstream component in the Hedgehog signaling pathway (, ; ), components of which localize to and function at primary cilia (). The function of Rab23 in Hedgehog signaling may therefore be due to a previously unknown requirement in primary cilium formation ( and ). Rab17 has been previously reported to be induced during cell polarization and to be involved in the function of apical sorting endosomes in polarized epithelial cells (; ). Its identification here () may indicate that sorting to the primary cilium is analogous to apical-basolateral sorting in polarized epithelial cells (). Further support for this proposal comes from the identification of Rab8a as the target of XM_037557 (), as Rab8 is known to be involved in polarized trafficking from recycling/sorting endosomes in epithelial cells (, ).
To further define the steps at which Rab8a, -17, and -23 might act, their localization was then examined in hTERT-RPE1 cells induced to form primary cilia by serum starvation. Screening the human Rabs revealed that Rab8a was the only Rab that could be detected on primary cilia when expressed as a GFP-tagged protein ( and Fig. S2 A, available at ). This localization was then confirmed using specific antibodies to Rab8a (; ). Strikingly, cells overexpressing Rab8a typically showed significantly (P < 0.05) longer cilia than control cells, as defined by the extent of both acetylated tubulin staining and the GFP-Rab8a–positive ciliamembrane (), suggesting that it is a limiting factor for primary cilium formation. In contrast, none of the other Rabs tested (Fig. S2 A), including Rab8b (), were found at primary cilia or had any obvious effect on cilium formation. Confirming previous reports (; ), Rab23 was found predominantly at the plasma membrane (), whereas Rab17 was present in punctate structures showing partial overlap with endocytic markers ( and Fig. S2 B). In addition, Rabl4 and GTPases of the Arl family implicated in microtubule and cilia formation (; ; ; ) were also absent from primary cilia (Fig. S2 A). Consistent with previous reports, the only other GTPases tested showing cilium targeting were Rabl5 and Arl13b (Fig. S2 A); however, although both are important for cilium function, neither is essential for cilium formation (; ). Therefore, Rab8a is the sole Rab on primary cilia and, by analogy, with the function of other Rabs, may serve to define this membrane domain (; ; ).
Because of its potential role in defining the membrane domain of the primary cilium, Rab8a was investigated further. To test the role of endogenous Rab8a in primary cilium formation, conditions were established for gene silencing using siRNA duplexes in hTERT-RPE1 cells using the Golgi apparatus and primary cilium protein IFT20 as a positive control () and the Golgi protein GMAP210 as a negative control (). As expected, IFT20-depleted cells were unable to form primary cilia (, A and C; ). Consistent with its unique localization, Rab8a-depleted cells showed strongly reduced primary cilium formation (). Cells depleted of the Golgi protein GMAP210 showed the same degree of primary cilium formation as control cells (). Along with the GAP-mediated Rab8a inactivation (), these results show that endogenous Rab8a plays an important role at the primary cilium.
In general, Rabs are thought to function by promoting tethering interactions between membranes, and between membranes and the cytoskeleton (). Although several effector proteins for Rab8 have been reported in the literature, they are mainly associated with actin function, and none of them provide an obvious link between microtubule function and membrane traffic (; ). Furthermore, it is not known if they show any specificity for Rab8a or -8b. We reasoned that because of its unique localization, Rab8a should have specific effector proteins at the primary cilium. We therefore performed two-hybrid screening using Rab8a and counterscreened the positive clones obtained against Rab8b using established methods (). Using this approach, we found that the cenexin/ODF2 splice variant 3 (cenexin 3) but not variant 1 (cenexin 1) interacted specifically with Rab8a, whereas the previously described Rab8 effector optineurin bound to both Rab8a and -8b ( and Fig. S3, available at ; ). Deletion of the last 20 amino acids of cenexin 3 abolished the interaction with Rab8a, suggesting that this region forms part of the Rab8a binding domain ( and Fig. S3). Confirming these findings, the interaction of Rab8a with cenexin 3 could be reconstituted using purified proteins, and this interaction was lost when the last 20 amino acids of cenexin 3 were deleted (). Importantly, cenexin 3 did not bind to other related Rabs, including Rab8b (), and therefore has the properties expected of a Rab8a-specific effector protein.
Cenexin/ODF2 is a basal body protein with three splice variants, differing in their N and C termini, that function in the nucleation or anchoring of microtubules at the centriole (). In light of the finding that cenexin 3 is an effector for Rab8a, we reinvestigated cenexin 1 and 3 to determine the relationship between primary cilium targeting and their Rab8a interaction properties. As previously reported, cenexin 1, which lacks the Rab8a binding domain, was present on the centrioles in the basal body both before and after serum starvation, but not on the microtubules of the primary cilium (). Strikingly, cenexin 3, although found on the basal body before induction of cilia, like Rab8a, relocated to the primary cilium on induction by serum starvation (). Furthermore, expression of the Rab8a binding domain of cenexin 3 (Cen3R8BD) had a dominant-negative effect on primary cilium formation (), supporting the idea that the Rab8a binding properties of cenexin 3 are important for its function. In contrast, full-length cenexin 1 and 3, or cenexin 3 lacking the Rab8a binding domain, had no inhibitory effect on primary cilium formation when overexpressed (). The interaction of Rab8a with cenexin 3 may therefore provide a link between membrane trafficking and tethering reactions and microtubule function at the primary cilium.
Based on the findings presented here and observations from the literature (; , ; ), we propose a working model for the function of Rab8a, -17, and -23 at primary cilia (). This model draws a parallel to the function of Rab8 in polarized epithelial cells, where it is needed for transport from a sorting endosomal compartment to the cell surface (, ). At primary cilia, we propose that Rab8a defines the primary cilium membrane domain by two mechanisms: first, by controlling the delivery of material to this specific region of the plasma membrane from a Rab17-positive endosomal compartment, and second, by linking the plasma membrane with cilia microtubules through the basal body and ciliary microtubule binding protein cenexin 3. In this model, Rab23, reported to act as a downstream component in Hedgehog signaling from primary cilia (; , ), is required for retrograde transport away from primary cilia. The identification of specific Rab GTPases acting at primary cilia and their GAP regulators provides a basis for future work on membrane trafficking at primary cilia and may also be relevant for other multiciliated cells, such as those found in lung epithelia (). Finally, these findings may also prove useful for understanding diseases where signaling pathways associated with primary cilia have become aberrantly activated (; ; Singla and Reiter, 2006; ).
Rabbit antibodies to Rab8a were a gift from J. Peränen (Institute of Biotechnology, University of Helsinki, Helsinki, Finland; ). Other antibodies were as follows: sheep anti-GM130 (), mouse anti-EEA1 (clone 14; Becton Dickinson); mouse anti-human transferrin receptor (CBL137; Chemicon); mouse anti–γ-tubulin (GTU88; Sigma-Aldrich); mouse anti–acetylated tubulin (6-11B-1; Sigma-Aldrich); and rabbit anti-FLAG (F7425; Sigma-Aldrich). Rabbit anti-IFT20 was raised against full-length recombinant human IFT20 expressed in bacteria and affinity purified using the same protein coupled to Affigel-15 (Bio-Rad Laboratories, Inc.). Donkey secondary antibodies were obtained from Jackson ImmunoResearch Laboratories. The methods used for cloning and construction of EGFP-tagged human Rabs and N-terminally tagged EGFP- and Myc-tagged human RabGAPs have been described previously (; ). EGFP-tagged Arl expression constructs were provided by S. Munro (Medical Research Council, Laboratory of Molecular Biology, Cambridge, UK). Cenexin/ODF2 splice variants were amplified from testis and fetal cDNA (CLONTECH Laboratories, Inc.) and subcloned into pcDNA3.1 to create an N-terminal fusion to the FLAG tag. All constructs were verified by DNA sequencing (DNA sequencing service, Max Planck Institute of Biochemistry, Martinsried, Germany).
The purification of GST-tagged human Rab GTPases, hexahistidine-tagged TBC domain proteins from bacteria, and RabGAP assays were performed as described previously (). Standard assays were performed for 60 min at 37°C, using 100 pmol GST-Rab and 10 pmol hexahistidine-tagged TBC domain protein. All proteins tested corresponded to the full-length open reading frames.
For binding assays, 10 μg GST-Rab protein were bound to 25 μl of packed glutathione–Sepharose (GE Healthcare) in 1 ml total volume of PBS for 60 min at 4°C. The beads were first washed three times in 500 μl with nucleotide exchange buffer (NE100: 20 mM Hepes-NaOH, pH 7.5, 100 mM NaOAc, 10 mM EDTA, and 0.1% [vol/vol] NP-40), followed by two washes in 500 μl nucleotide loading buffer (NL100: 20 mM Hepes-NaOH, pH 7.5, 100 mM NaOAc, 0.1 mM MgCl, and 0.1% [vol/vol] NP-40). The beads were then resuspended in 200 μl NL100, and 20 μl of 100 mM GTP and 10 μg effector protein was added. Binding was then allowed to proceed for 60 min at 4°C, rotating to mix. The beads were then washed three times with 500 μl NL100, and bound proteins were eluted by the addition of elution buffer (NE200: 20 mM Hepes-NaOH, pH 7.5, 200 mM NaCl, 20 mM EDTA, and 0.1% [vol/vol] NP-40), rotating at 4°C for 15 min. Beads were pelleted by centrifugation at 2,000 for 1 min, and the supernatant was transferred to a fresh tube. To remove contaminating Rabs, 50 μl of packed glutathione–Sepharose was added to the eluate and incubated for 10 min at 4°C with mixing. The beads were pelleted by centrifugation at 2,000 for 1 min, and the supernatant was transferred to a fresh tube. This procedure was repeated three times. Eluted proteins were then precipitated using trichloracetic acid and analyzed on 12% minigels stained with Coomassie brilliant blue.
Human telomerase-immortalized retinal-pigmented epithelial cells (hTERT-RPE1; CLONTECH Laboratories, Inc.) were grown at 37°C and 5% CO in a 1:1 mixture of DME and HAMS F12 containing 10% calf serum, 2.5 mM -glutamine, and 1.2 g/liter sodium bicarbonate. For transfection, 0.5–1.0 × 10 cells were plated in 6-well plates. For Rabs and GAP constructs, 0.5 μg plasmid DNA were mixed with 3 μl Fugene-6 according to the manufacturer's protocol (Roche Diagnostics) and then added to 1 well of the 6-well plate. For cenexin constructs, 50 ng plasmid DNA was mixed with 1 μg pBluescript-II as a carrier to avoid aggregation problems seen with high-level expression using the standard protocol. To induce primary cilium formation, the growth medium was replaced with medium lacking serum. After 48–72 h, the cells were placed on ice for 1 h, processed for fluorescence microscopy, and analyzed as described previously (). RNA interference was performed using a published method (). All siRNA duplexes were obtained from Dharmacon, Inc., and the target sequences were as follows: control, CGUACGCGGAAUACUUCGA; Rab8a, CAGGAACGGUUUCGGACGA, GAAUUAAACUGCAGAUAUG, GAACAAGUGUGAUGUGAAU, GAACUGGAUUCGCAACAUU; IFT20, GGAAGAGUGCAAAGACUUU; GMAP-210, GCCAGAGACAAUCUAGCAC.
Cells to be imaged were fixed in −20°C methanol for 5 min and washed three times with PBS. For EEA1 staining, cells were fixed for 20 min in 3% (wt/vol) PFA, quenched for 10 min with 50 mM ammonium chloride, and permeabilized with 0.1% (vol/vol) Triton X-100 for 5 min to allow labeling of internal cell structures. Alternatively, all solutions were made in PBS, and antibody staining was performed for 60 min using a 1,000-fold dilution of antiserum or purified antibody at a final concentration of 1 μg/ml. Secondary antibodies were conjugated to Alexa 488 or Cy3, and DNA was stained with DAPI. Coverslips were mounted in 10% (wt/vol) Moviol 4–88, 1 μg/ml DAPI, 25% (wt/vol) glycerol in PBS. Images were collected using an Axioskop 2 with a 63× Plan Apochromat oil-immersion objective of NA 1.4, standard filter sets (Carl Zeiss MicroImaging, Inc.), a 1300 × 1030 pixel cooled charge-coupled device camera (model CCD-1300-Y; Princeton Instruments) and Metavue software (Visitron Systems). Images were cropped in Photoshop 7.0 or CS2 (Adobe) without contrast or other adjustments and sized and placed using Illustrator 11.0 or CS2 (Adobe).
Fig. S1 shows the effects of RabGAP and dominant-negative Rab expression on primary cilium formation. Fig. S2 shows Rab localization to primary cilia in serum-starved hTERT-RPE1 cells. Fig. S3 gives full details of the Rab8a–cenexin 3 yeast two-hybrid interaction. Online supplemental material is available at . |
To achieve faithful chromosome segregation, a bipolar spindle has to be established. In somatic mammalian cells, the mitotic spindle is nucleated from the two centrosomes, each consisting of two closely associated (engaged) centrioles surrounded by pericentriolar material. During exit from mitosis, the two centrioles separate, and this centriole disengagement is a prerequisite for the centriole duplication in the next cell cycle and, therefore, is tightly regulated (). However, under some circumstances, premature disengagement of the centrioles can occur, leading to multipolar spindles and mitotic defects (; ; ).
Along with the formation of the mitotic spindle, stable attachments of the chromosomes to the microtubules and their alignment at the metaphase plate have to be achieved. Once all of the chromosomes are aligned, the connection between the sister chromatids is severed by the cysteine protease separase (; ). Up to that point, separase activity is held in check both by inhibitory binding of its chaperone securin and Cdk1/cyclin B1 (; ; ). In turn, the levels of securin and cyclin B1 are controlled by the spindle checkpoint () that prevents their degradation as long as kinetochore attachment to microtubules and tension across the kinetochores have not been established. Recently, it has been demonstrated that centriole disengagement at the end of mitosis also requires separase (). Thus, the spindle checkpoint–mediated inhibition of separase protects both sister chromatid cohesion and the connection between engaged centrioles.
In this study, we have analyzed the function of the spindle- and kinetochore-associated protein astrin (; ; ). We find that in the absence of astrin, kinetochore–microtubule attachments are impaired, resulting in a spindle checkpoint arrest. Fixed and live cell analysis of astrin-depleted cells revealed both a high degree of premature centriole disengagement, resulting in multipolar spindles, and a loss of sister chromatid cohesion. The potential involvement of separase in the origin of these phenotypes is investigated.
Astrin has previously been described as a spindle-associated protein involved in mitotic progression (Fig. S1, A and B; available at ; ; ; ). During prophase and prometaphase, a centrosomal pool of astrin is diffusely localized to the pericentriolar material and microtubules emanating from the centrosome, overlapping with but distinct from the areas stained by antibodies against Pololike kinase 1 (Plk1), γ-tubulin, and centrin (). A second chromosome-associated pool of astrin partially overlaps with the outer kinetochore components Hec1 and centromere protein (CENP) E as well as the microtubule tip-binding protein EB1 but is discrete from other kinetochore proteins such as Plk1 and the centromeric markers aurora B and CENP-A (). Therefore, this second pool of astrin is most likely associated with the outer kinetochore. The dual localization of astrin to both centrosomes and kinetochores indicates that it may be required for spindle formation and chromosome segregation.
Supporting the aforementioned conclusion, the depletion of astrin resulted in an increase in the mitotic index and occurrence of cells with multipolar spindles and disorganized DNA (Fig. S1, C–E). Concomitantly, increased cell death by apoptosis was observed (Fig. S1 F; ). These defects were efficiently rescued by the transfection of cells with siRNA- resistant myc-astrin before siRNA-mediated depletion of endogenous astrin () and were not observed upon the depletion of several other outer kinetochore proteins (Fig. S1 G). Further analysis of HeLa-S3 H2B-GFP control cells or astrin-depleted cells by time-lapse imaging showed that astrin-depleted cells also exhibited a profound chromosome alignment defect (). Astrin-depleted cells required substantially more time to assemble a metaphase plate than control cells (63.9 ± 28.5 min in comparison with 27.0 ± 8.0 min; ), and 60% of the cells never accomplished full alignment of all of the chromosomes (, right). Once metaphase chromosome alignment was achieved, chromosomes were lost again from this structure (, t = 104 min and t = 160 min), and cells remained arrested in a metaphase-like state for prolonged periods of time (up to 10 h) and eventually died by apoptosis.
Although most astrin-depleted cells (8/9) initially formed a bipolar spindle, after several hours of mitotic arrest (194.4 ± 46.5 min), bipolarity was lost, and a multipolar spindle was formed (, t = 252 min). In the one remaining case, a multipolar spindle was formed immediately without a preceding period of bipolarity. Together, these data suggest that astrin has functions at both spindle poles and kinetochores and that the lack of astrin leads to a prolonged mitotic arrest.
Consistent with the observed cell cycle arrest, cells lacking astrin displayed Mad2 and strongly BubR1-positive kinetochores, indicating spindle assembly checkpoint activation (). These cells also stained brightly for cyclin B1 and securin, with mean pixel intensities similar to mitotic control cells (), and extracts prepared from them had levels of cyclin B1, securin, and phosphohistone H3 (Ser10) comparable with control nocodazole-arrested cells (). Moreover, the mitotic arrest caused by astrin depletion was relieved by the depletion of Mad2 in addition to astrin or the treatment of astrin-depleted cells with the aurora B inhibitor ZM447493, confirming its dependency on the spindle checkpoint ().
The persistent activation of the spindle checkpoint together with the delay in metaphase plate formation suggested that microtubule–kinetochore interactions require astrin. To test this idea, cells were analyzed after cold treatment () or were preextracted before fixation to differentially preserve kinetochore fibers (). Under both conditions, astrin- depleted cells displayed many unattached kinetochores and fewer stable kinetochore microtubules than control cells, although the phenotype was not as drastic as the one observed in Hec1-depleted cells (; ). The kinetochores that were microtubule associated often appeared to be attached laterally rather than end on in cells lacking astrin (, bottom; insets). Furthermore, although both the core kinetochore protein Hec1 and the spindle checkpoint kinase Bub1 were unaffected (), the kinetochore resident motor protein CENP-E () and its interaction partner CENP-F () were delocalized from the kinetochore in the absence of astrin. These cells remained cyclin B1 positive (unpublished data), confirming that they were still in mitosis. These data suggest that the presence of astrin is required for the kinetochore recruitment or maintenance of CENP-E and CENP-F. In combination with the lack of astrin, this may cause unstable microtubule–kinetochore interactions, unaligned chromosomes, and persistent activation of the spindle checkpoint. However, these finding do not explain why multipolar spindles are formed in cells lacking astrin.
To analyze the molecular basis of the multipolar spindle phenotype in more detail, the localization of centrin, a marker for individual centrioles (), was investigated in astrin-depleted cells. Multipolar spindles can arise by several different routes that can be distinguished by the number of centrioles found at the individual poles. Failure of cytokinesis will lead to multiple spindle poles with two centrin-positive centrioles at each pole. In contrast, aberrant centriole disengagement will cause the formation of multipolar spindles with single centrioles at individual spindle poles (; ). A further possible cause for the loss of spindle bipolarity is the loss of microtubule anchoring at the centrosome, which is observed, for instance, upon tumor overexpressed gene protein (TOGp) depletion (; ).
In contrast to the bipolar spindles of control cells, which displayed two centrin dots at each pole, the multipolar spindles in cells lacking astrin often displayed single centrin dots at each pole (). For a further quantitative comparison, the number of centrin dots per pole in multipolar cells depleted of aurora B that is known to be required for correct chromosome segregation and progression through cytokinesis () or TOGp, a protein important for maintaining intact spindle poles (; ; ), was evaluated (). This approach revealed that 79.2% of multipolar spindles in aurora B–depleted cells displayed two centrin-positive dots per pole ( [top] and C), which is consistent with the idea that these spindles had arisen from a previous cytokinesis failure. Multipolar spindles in TOGp-depleted cells often showed poles with no centrin staining in addition to two normal poles with two centrin dots and therefore contained the highest number of acentriolar poles (43.9%; [bottom] and C). Strikingly, and in contrast to both aurora B and TOGp depletion, in astrin-depleted cells, 55.4% of poles had single centrioles, which is suggestive of aberrant centriole disengagement ().
In line with these data, aurora B and astrin-depleted cells generally displayed pericentrin staining at all poles of multipolar spindles, whereas TOGp-depleted cells often possessed multipolar spindles with only two pericentrin-positive poles, confirming published results (; ). Collectively, these results suggest that the formation of multipolar spindles upon astrin depletion is mainly caused by an untimely loss of the connection between the two centrioles of each centrosome.
In normal cells, the connection between the two centrioles is lost at the end of mitosis or early G1 phase, when the two centrioles are disengaged. It has recently been demonstrated that this disengagement of the two centrioles is dependent on the activity of separase (), a protease that also controls cohesin cleavage between sister chromatids (). One possible explanation for the formation of multipolar spindles in cells depleted of astrin could therefore be an inefficient inhibition of separase during the checkpoint arrest, leading to premature disengagement of the centrioles. In this case, one would predict that cohesion between sister chromatids would also be affected.
Consistent with this idea, immunofluorescence analysis of cells or single chromosomes showed that the chromosomes of mitotically arrested astrin-depleted cells displayed single dots of CREST staining, which is indicative of separated sister chromatids, in comparison with paired dots in metaphase control cells (Fig. S2, A and B; available at ). Furthermore, 65.0% of mitotic chromosomes in chromosome spreads prepared from astrin-depleted cells displayed separated sister chromatids compared with <1.0% of control cell spreads, 97.7% of chromosome spreads of Sgo1-depleted cells, which are known to display a loss of sister chromatid cohesion (), and 9.3% of spreads of CENP-E–depleted cells (; and ; ). Time-lapse analysis of astrin-depleted cells expressing YFP-tagged CENP-A showed that these cells established and maintained cohesion normally in early mitosis through to the formation of (imperfect) metaphase plates but lost cohesion during subsequent mitotic arrest, on average 89.6 ± 42.3 min ( = 10) after forming a metaphase (like) plate (Fig. S2 C). This loss of cohesion was not caused by lack of the centromeric protector Sgo1 (Fig. S2 D). In summary, the data obtained from imaging histone H2B-GFP or CENP-A–YFP expressing astrin-depleted cells indicate that loss of sister chromatid cohesion precedes loss of centrosome integrity in these cells but that both events occur after the formation of an imperfect metaphase plate.
To confirm that loss of centriole and sister chromatid cohesion was a specific effect of astrin depletion rather than an indirect effect of the prolonged mitotic arrest, cells depleted of CENP-E by RNAi or treated after presynchronization with nocodazole for 16 h, both causing extended mitotic arrest, were analyzed (). Although both treatments resulted in a moderate increase in multipolar cells (12.8 ± 3.7% and 23.0 ± 2.8%, respectively), this did not seem to have been caused by centriole disengagement (). Furthermore, neither CENP-E–depleted nor nocodazole-treated cells displayed a substantial loss of sister chromatid cohesion after extended periods of mitotic arrest (; and ). In summary, astrin depletion causes unique defects (67.3 ± 2.1% of multipolar mitotic cells and 65.0 ± 2.6% of separated sister chromatids) that cannot be phenocopied by forcing cells into extended mitotic arrest by other means.
Separation of the sister chromatids in cells lacking astrin would be consistent with the premature activation of separase. This hypothesis was tested by exploiting the fact that active separase undergoes self-cleavage, resulting in a 65-kD C-terminal fragment (). Extracts prepared from mitotically arrested or mitotically arrested and released cells were used to create situations in which separase was either inactive (mitotic arrest) or active (release from mitotic arrest). Immunoblotting revealed the presence of a C-terminal cleavage product only in the control cells that had been released from the mitotic block (, compare lane 1 with lane 2). Importantly, in extracts prepared from astrin-depleted cells, the C-terminal separase cleavage product was present at ∼30% of the level observed in the released control cells (, lane 3). These data suggest that a fraction of separase is active in mitotic astrin- depleted cells.
To directly test the involvement of separase in the astrin phenotype, cells were simultaneously depleted of astrin and separase. Immunofluorescence and Western blotting analysis showed that both proteins could be efficiently depleted either singly or together (). Cells lacking both separase and astrin displayed a similarly elevated mitotic index in comparison with cells depleted only of astrin (15.5 ± 3.7% in the double depletion in comparison with 16.2 ± 3.0% in astrin depletion; ). However, in contrast to astrin depletion alone, cultures depleted of both astrin and separase contained considerably fewer cells with multipolar spindles (16.9 ± 6.0% in the double depletion vs. 68.8 ± 11.5% in the astrin RNAi; ). Importantly, chromosome spreads of these cultures showed that sister chromatid cohesion was restored in cells in which both astrin and separase expression had been repressed (). In contrast, no effect on the TOGp phenotype was observed upon the additional depletion of separase (Fig. S3, available at ). Collectively, these data show that both the aberrant centriole disengagement and the premature loss of sister chromatid cohesion observed in astrin-depleted cells involve separase.
Plasmid transfection and RNAi were performed as described previously (). Sgo1, aurora B, Hec1, TOGp, Ska1, Ska2, CENP-F, and Mad2 siRNA oligonucleotides were described previously (; ; ; ). Astrin, separase, and CENP-E were targeted with 5′-TCCCGACAACTCACAGAGAAA-3′, 5′-AAGCTTGTGATGCCATCCTGA-3′, and 5′-ACTCTTACTGCTCTCCAGTTT-3′ (QIAGEN), respectively. For Western blot analysis of whole cell lysate, cells of one 2-cm plate were harvested and lysed in laemmli buffer. For Western blot analysis of mitotic cells, HeLa S3 cells were collected by mitotic shake off and lysed in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 40 mM β-glycerolphosphate, 10 mM NaF, 0.3 mM Na-vanadate, 1 mM EDTA, 1% [vol/vol] IGEPAL, 0.1% [vol/vol] deoxycholate, 2 mM Pefabloc, 100 nM okadaic acid, and protease inhibitor cocktail without EDTA (Roche Diagnostics). For the analysis of separase activity by Western blotting, -ethylmaleimide (2.5-mM final concentration; Sigma-Aldrich) was added to the lysis buffer.
Astrin aa 1–481 (N-astrin) or astrin aa 1,014–1,193 (C-astrin) were cloned into pQE30 (QIAGEN) or pGEX-5X-1 (GE Healthcare), respectively, and were expressed and purified according to standard protocols. Antibodies against C- and N-astrin were raised in rabbits (Charles River Laboratories) and affinity purified using maltose-binding protein–tagged astrin. Goat polyclonal antibodies against centrin-3 were raised against recombinant full-length His-tagged centrin-3 and were affinity purified using GST-tagged centrin-3. Antibodies against aurora B, CENP-A, CENP-E, CENP-F, BubR1, Bub1, Hec1, Plk1, Mad2, α-tubulin, and myc and CREST serum were described previously (). Other antibodies used in this study were as follows: rabbit anti–pericentrin B, mouse antisecurin, mouse antiseparase (clone XJ11-1B12), and rabbit antiphosphohistone H3 (Ser10) (Abcam); mouse anti–cyclin B1 (Santa Cruz Biotechnology, Inc.); mouse anti-EB1 (BD Biosciences); mouse anti−γ-tubulin (clone GTU88; Sigma-Aldrich); and mouse anti-Sgo1 (Abnova). Rabbit anti-TOGp antibodies were gifts from X. Yan (Third Institute of Oceanography, State Oceanic Administration, Xiamen, China). Secondary antibodies conjugated to HRP, Cy2, Cy3, or Cy5 were obtained from Jackson ImmunoResearch Laboratories. DNA was stained with DAPI (Sigma-Aldrich). Aurora B inhibitor ZM447439 was obtained from Tocris Biosciences.
Image acquisition was performed as previously described (). The fixation for the K-fiber analysis was performed as described previously (). Cold treatment was performed as described previously (). For high resolution images and for time-lapse microscopy of HeLa-S3 H2B-GFP cells, a high resolution imaging system (Deltavision; Applied Precision) on an inverted microscope (IX71; Olympus) equipped with plan Apo 40× NA 0.95, plan Apo 60× NA 1.4, and 100× NA 1.35 oil immersion objectives (Olympus) and a camera (CoolSnap HQ; Photometrics) was used as described previously (). HeLa–CENP-A–YFP cells were filmed on the same system at intervals of 4 min, imaging seven focal planes 2 μm apart, with an exposure of 0.8 s and 100% neutral density.
HeLa S3 cells were treated with Gl2 (control), CENP-E, astrin, or Sgo1 siRNA oligonucleotides for 43 or 35 h (Sgo1), and then 100 ng/ml nocodazole was added for a further 5 h. Alternatively, HeLa S3 cells were arrested in G1/S phase with 1.6 μg/ml aphidicolin for 16 h, released for 6 h, and arrested in G2/M phase for 16 h by adding 100 ng/ml nocodazole. Mitotic cells were collected by mitotic shake off, and chromosome spreads were prepared as described previously ().
Videos 1 and 2 show a control and an astrin-depleted cell progressing through mitosis (E). Fig. S1 shows that the depletion of astrin induces the formation of multipolar spindles. Fig. S2 shows that the absence of astrin causes the loss of sister chromatid cohesion. Fig. S3 shows that the formation of multipolar spindles in TOGp-depleted cells is not dependent on separase. Online supplemental material is available at . |
The internalization and digestion of particulate materials by eukaryotic cells is a complex process that plays important roles in host defense against infection, development, and tissue homeostasis (; ; ; ). In mammals, phagocytosis of microbial pathogens by so-called professional phagocytes such as macrophages and neutrophils is one of the earliest responses in host defense and is critical for controlling inflammation (; ; ; ). Phagocytosis begins with the attachment of particles to cell surface receptors, triggering a reorganization of the plasma membrane and actin cytoskeletal components to facilitate particle internalization and the formation of a phagosome. The phagosome matures over time by sequentially interacting with early, then late endocytic vesicles and finally lysosomes, culminating in the formation of a phagolysosome in which the ingested material is degraded (; ; ). The mechanisms that underlie the early steps of particle internalization and phagosome formation are relatively well understood; however, the molecular machinery regulating phagosome maturation, particularly the process of phagolysosome formation, remains unclear.
Delineating the mechanism of phagosome maturation is critical, as certain intracellular pathogens evade phagolysosomal fusion in macrophages by using a variety of strategies to alter macrophage signaling (). In the case of (; ), a number of mechanisms have been proposed for lysosome evasion, including the effects of ammonia production (), the close apposition between the bacterium and the vacuole membrane (), the ability of surface lipids and cord factor to inhibit vesicular fusion (; ), and the ability of a eukaryotic-like serine/threonine protein kinase G to inhibit phagosome/lysosome fusion in infected macrophages (). Other bacterial pathogens are able to alter phagosomal maturation to survive in macrophages by residing in various organelles, such as late endosomes (), lysosomes (), and rough endoplasmic reticulum () (). The array of interfering strategies that are used by pathogens suggest that the molecular mechanisms regulating phagosome maturation is complex; this point is underscored by recent proteomic analyses of phagosome-associated proteins in mouse macrophages where more than 160 proteins were identified ().
Cellular signaling proteins have been implicated in phagosome maturation. For example, activation of the toll-like receptor (TLR) signaling pathway by bacteria regulates phagocytosis at multiple steps including internalization and phagosome maturation (). In addition, lipid-modulating enzymes such as PI3-kinase and sphingosine kinase, which alter the lipid composition of phagosomal membranes, as well as the lysosome-associated tyrosine kinase Hck (hematopoietic cell kinase) have been shown to regulate phagosome maturation (; ). Extensive studies suggest that newly formed phagosomes carry signaling molecules that direct them to specifically interact with endosomes and then lysosomes. In addition to the cell surface receptors (TLRs), small GTPases such as Rab5 (; ) and Rab7 (), and LAMP1 and LAMP2 (lysosome-associated membrane proteins) () play roles in inter-compartment communication, mediating the fusion process during phagosome maturation. It is likely that the surface of endosomes and lysosomes also carry signaling proteins in order to selectively fuse with incoming phagosomes. However, such proteins have yet to be determined.
The single-cell, soil amoeba is a professional phagocyte that has proven to be an excellent model system for phagocytosis. is a genetically tractable organism with mutants often displaying clear phagocytosis phenotypes that can be easily screened () and has phagocytosis rates severalfold higher than those observed in mammalian macrophages or neutrophils (). and mammalian phagocytes share many common molecular components that regulate engulfment and phagosome maturation. Like mammalian phagocytes, F-actin mediates the formation of the phagocytic cup and the internalization of particles, and the WASP family of actin-regulating proteins also plays important roles in regulating phagocytosis. Also in common is the localization of small Rab GTPases and LAMP proteins in phagosomes and a requirement for Rho and Ras family GTPases for regulating phagocytosis (; ; ; ). Finally, pathogens that evade death in mammalian phagocytes are also able to escape killing by , suggesting the existence of common targets in both host cell types ().
Tyrosine phosphorylation plays a major role in the early signaling events of phagocytosis in mammalian phagocytes (; ) and, as suggested by recent work (), may function in later steps as well. In general, tyrosine kinases constitute a large family of signaling molecules and are key regulators of many cellular processes, functioning to transduce signals within and between eukaryotic cells. It is possible that certain tyrosine kinases also regulate the phagosome maturation process. In the present study, we used as a model system to elucidate the physiological functions of a novel receptor tyrosine kinase (RTK)–like protein we termed vesicle-associated kinase (VSK) 3. VSK3 consists of a signal peptide, a single transmembrane domain, a C-terminal kinase domain, and one N-terminal TIG (immunoglobulin-like fold) domain that is found in the MET (HGF receptor tyrosine kinase) kinase family of higher eukaryotes (). This report is the first to show that an RTK-like protein localizes to the surface of late endosomes/lysosomes and may serve to mediate vesicle fusion and phagosome maturation.
Using a genomic approach to identify receptor-like tyrosine kinases and study their functions in , we searched the genomic database (www.dictybase.org) and found 242 genes that encode proteins containing one of the catalytic domains characteristic of eukaryotic protein kinases. Among them, 46 genes encode the peptide sequence of HRDLXXXN, which is a signature domain in protein tyrosine kinases (). We then analyzed the structure of the 46 putative protein kinase sequences (), and found three previously uncharacterized proteins we termed VSK1, 2, and 3 (see Discussion), which possess a classic receptor kinase domain architecture of a signal peptide, a single trans-membrane domain, and a C-terminal kinase domain. In this study, we focused on the function of VSK3 (). Recently, a comprehensive genomic analysis of the protein kinases in also identified the same three receptor-like kinases, which were named receptor kinases (rk) 1, 2, and 3, respectively (). We propose that VSK is a more suitable name for these proteins to reflect their subcellular localization and potential function (see and Discussion).
displays various biological behaviors during its life cycle. As free-living amoebae, are professional phagocytes capable of internalizing and digesting bacteria and yeast (). Upon starvation, amoebae enter a developmental program during which they aggregate via cAMP-mediated chemotaxis (). To evaluate functions of the VSK3 protein, we determined the expression profile of the gene during differentiation using real-time PCR. The mRNA level of the gene was relatively high in vegetative cells in rich medium, and gradually declined upon starvation (), suggesting a requirement for VSK3 in growing cells. Both cAR1 and Gβ mRNA showed the expected expression profile in control amplification experiments (Fig. S1, available at ).
-null (
) cells were generated through homologous recombination of the genomic sequence flanking the blasticidin resistance cassette into the open reading frame of wild-type (AX2) cells. Disruption mutants were identified by PCR and Southern blot analyses (; Fig. S2, A and B, available at ).
cells were grown on bacterial lawns supported on an agar substrate. The bacteria provide a food source for growing . As bacteria are consumed and cleared, circular plaques form on the bacterial lawns, which reflect growth and the phagocytic capacity of cells. As a response to the loss of nutrients in the center of plaques, cells chemotax into multicellular aggregates and ultimately differentiate into fruiting bodies consisting of dormant spore cells supported by stalk cells.
cells form significantly smaller plaques () relative to wild-type cells, suggesting a defect in growth or phagocytosis.
and wild-type cells in shaking, rich, liquid medium. Under these conditions, cells obtain liquid nutrients through pinocytosis (). We found no significant difference in either the growth rate () or the rate of pinocytosis between strains (Fig. S3 A, available at ), suggesting that the defect in plaque size on bacterial lawns is due to impaired phagocytosis. To test for potential phagocytic defects, phagocytosis was measured as the cell-associated fluorescence intensity of TRITC-labeled yeast after washing (). Cells lacking VSK3 exhibit a clear decrease in yeast uptake in comparison to wild-type cells, indicating that the protein is required for normal phagocytosis.
cells with FITC-dextran and measured the change in FITC fluorescence over time. FITC fluorescence is sensitive to pH change and FITC-dextran has been used as a probe to measure endosomal pH along the fluid-phase endocytosis pathway (). After the cells were loaded with FITC-dextran and then washed with phosphate buffer (pH 6.6) the endosomal pH dropped rapidly, reaching a minimal value of ∼pH 5.0 by 20 min and then returning to the extracellular value of pH 6.6 (), as previously reported ().
clones (); these data indicate that VSK3 does not affect endosomal acidification.
cells did not display considerable developmental phenotypes.
cells form wild-type–like fruiting bodies in plaques on bacterial lawns and on nonnutrient agar.
cells exhibit relatively normal chemotaxis responses to cAMP using a micropipette assay (unpublished data).
To further analyze the phagocytic abilities of mutants, we used a flow cytometry assay to measure uptake of heat-killed yeast fluorescently labeled with TRITC (). In this assay, the fluorescence of ingested yeast particles was determined after quenching the signal from uningested yeast by the addition of Trypan blue. shows representative flow cytometry results of cells incubated with TRITC-labeled yeast over a 2-h time course at 22°C. The first peak shows the background autofluorescence of cells having no internalized TRITC-labeled yeast and the appearance of the second peak over time indicates the rise in the level of ingested fluorescent yeast.
cells display a considerable decrease in yeast uptake during the time course with only a fraction (35.5%) of
cells containing internalized yeast compared with 82.6% for wild-type cells by 2 h ().
To rescue the -null phenotype, we transformed null cells with a nonintegrating, extrachromasomal plasmid that constitutively expresses a VSK3-YFP fusion protein (see ) under the control of the actin 15 promoter. Transformants were selected with various levels of G418 to control the level of VSK3-YFP (). As expected, increasing selective drug pressure increased the level of VSK3-YFP mRNA and VSK3-YFP protein as detected by real-time PCR and Western blot analysis, respectively ().
cells () was substantially rescued in transformants selected with the lower concentration of drug ().
or wild-type cells with a concentration of G418 at 10 μg/ml caused a reduction in phagocytic capacity (). Similar negative effects on phagocytosis have been observed for both loss- and gain-of-function mutants of RacB (), suggesting that a balance of signaling activity is required for optimal phagocytosis.
To measure protein kinase activity and to determine the localization of VSK3 in vivo, we created a VSK3 fusion protein tagged at its C terminus with YFP (VSK3-YFP). For controls, we deleted the kinase domain from VSK3-YFP (VSK3ΔK-YFP), deleted the N terminus and transmembrane domain to fused the kinase domain to YFP (K-YFP) and mutated a conserved lysine to arginine in the presumptive ATP binding site (VSK3K518R-YFP; ). The K518R point mutation was based on previous studies that demonstrated this conversion abrogates kinase activity (; ; ).
cells (), indicating that YFP moiety does not interfere with VSK3 function in vivo.
cells does not rescue the defect (Fig. S3, B and C), showing that the kinase domain is required for VSK3 function.
To assay VSK3 kinase activity, VSK3-YFP, VSK3ΔK-YFP, K-YFP, and VSK3K518R-YFP were partially purified by immunoprecipitation from equal numbers of transformed cells selected with 10 μg/ml of G418. Western blot analyses of immunoprecipitates showed that the level of VSK3-YFP protein was consistently fourfold lower than that of K-YFP and VSK3ΔK-YFP expressed in the same parental cells (). Both VSK3-YFP and K-YFP were able to phosphorylate the tyrosine kinase–specific peptide substrate 60% as well as the control mammalian Src tyrosine kinase (), but did not show any detectable activity in a similar Ser/Thr kinase assay (). We are unable to offer a simple explanation as to why immunoprecipitates from cells expressing K-YFP did not yield more kinase activity relative to the amount of input protein when compared with the full-length VSK3-YFP; synergism from the N-terminal domain or associated proteins may account for the relative higher level of kinase activity from VSK3-YFP immunoprecipitates. As expected, neither VSK3ΔK-YFP nor VSK3K518R-YFP showed detectable protein kinase activity in either assay (, D and E; Fig. S4, available at ). Together, these results demonstrate that VSK3 contains an active tyrosine kinase domain that is required for its normal function.
cells, and examined fluorescence in growing cells using confocal laser scanning microscopy (). VSK3-YFP, VSK3ΔK-YFP, and VSK3-K518R-YFP were observed at the periphery of intracellular vesicles, while K-YFP was seen in the entire cytosol. These findings indicate that sequence determinants outside of the kinase domain dictate VSK3 localization. We incubated these cells with TRITC-dextran, which is taken into cells by pinocytosis and selectively labels late endosomes and lysosomes (; ), and found that VSK3-YFP, VSK3ΔK-YFP, and VSK3-K518R-YFP localize to the surface of dextran-labeled vesicles (). Together, these data suggest that both VSK3 localization and kinase activity are required together for its normal functions ( and ; Fig. S3, B and C; unpublished data).
To further characterize VSK3-containing vesicles, we loaded wild-type cells expressing VSK3-YFP and, as a control, cells expressing VatM-GFP with neutral red, which accumulates in acidic vesicles (). VatM is the 100-kD transmembrane subunit of the vacuolar H(+)-ATPase and localizes to both contractile vacuoles and acidic endosomal vesicles that fuse with phagosomes (). As expected, neutral red colocalizes with VatM-GFP in endosomal vesicles (), with the exception of contractile vacuoles (, arrow). We found that neutral red also accumulates in VSK3-YFP labeled vesicles, indicating that VSK3 resides in acidic compartments ().
It seemed likely that VSK3′s N terminus would be oriented toward the interior of vesicles given the existence of a putative N-terminal signal peptide and a single transmembrane domain, thus enabling the kinase domain to reside in the cytoplasm where ATP is plentiful. To test this prediction, we used a recently established fluorescence protease protection (FPP) assay used to study the topology of transmembrane proteins within live cells (). Cells expressing VSK3-YFP and labeled with TRITC-dextran were exposed to digitonin, which specifically permeablizes the plasma membrane but not membranes of intracellular organelles, and were then treated with trypsin to degrade unprotected proteins and protein moieties within the cytosol (). After digitonin and trypsin treatment, TRITC-dextran fluorescence persisted (), indicating that lysosomes remained intact. If VSK3-YFP were oriented in the membrane such that the YFP moiety was inside the vesicle, it would be protected from trypsin digestion and its fluorescence would persist. However, we observed that the YFP fluorescence was rapidly lost after digitonin and trypsin treatment (), showing that VSK3 is oriented with its N terminus within late endosomes/lysosomes and its C-terminal tyrosine kinase domain facing the cytosol ().
cells expressing VSK3ΔK-YFP by transmission electron microscopy (TEM).
cells as our data indicate it does not rescue the null phenotype (see Fig. S3 B; unpublished data). The YFP epitope was detected with an anti-GFP primary and a gold-conjugated secondary antibody. In control experiments, some nonspecific nucleation of the enhancement reagent occurred on material other than gold during the particle enhancement step (unpublished data) and nonspecific staining was observed throughout wild-type cells not expressing VSK3ΔK-YFP when both primary and secondary antibodies were used (). In cells expressing the VSK3 marker, gold particles were found in high density around intracellular vesicles in bold wild-type and null cells well above background staining (). These data are consistent with VSK3ΔK-YFP localization as determined by epifluorescence microscopy (see ). The morphology of positively staining vesicles varied in shape and size, ranging from 0.1 to 1 μm, but no striking differences were apparent between vesicles observed in wild-type and
cells. Finally, in support of our FPP assay results (see ), high magnification images of individual vesicles revealed that gold particles accumulated on the outside of vesicles (), confirming that the C terminus of VSK3 is positioned toward the cytoplasm.
cells expressing coronin-GFP, which is known to colocalize with F-actin during particle engulfment (). At the point of attachment, the formation of phagocytic cup can be seen as an accumulation of coronin-GFP around the yeast. Once internalized, coronin-GFP rapidly dissociates from the newly formed phagosome. In
cells, as in wild-type cells, coronin-GFP rapidly accumulates at the site of attachment of yeast and remains in association throughout the engulfment process, and then disappears once the yeast is internalized in the newly formed phagosome (). We recorded and quantified the time from the formation of phagocytic cup to the formation of the phagosome in both wild-type (107 s, SD = 14.8 s; = 10) and
(102 s, SD = 10.3 s, = 10) cells, and found that there was no considerable difference between them. Based on these observations, we conclude that VSK3 is not essential for this step of phagocytosis.
To further explore VSK3′s function in phagocytosis, we tracked the dynamic localization of VSK3-containing vesicles as cells were fed heat-killed, Texas red–labeled yeast with confocal microscopy. We used VSK3ΔK-YFP as a marker because the expression of full-length VSK3-YFP to detectable levels for epifluorescent microscopy exerts a dominant-negative effect on phagocytosis (see ), while VSK3ΔK-YFP has no effect on phagocytosis and localizes to the same vesicles (see Fig. S3 A and ). VSK3ΔK-YFP is not detected around yeast particles during the early phase of phagocytic cup formation, but within a minute of full engulfment, VSK3ΔK-associated late endosomes/lysosomes began to surround the yeast-containing phagosome and by five minutes complete fusion occurred (). The kinetics of vesicle fusion we observed is consistent with previously reported rates of phagosome–lysosome fusion in (). These results suggest that VSK3 may be involved in the fusion process between phagosomes and late endosomes/lysosomes. Finally, images captured 20 min after feeding show that VSK3ΔK-YFP still surrounds yeast containing phagosomes, again consistent with late stage localization ().
cells. We monitored the maturation of phagosomes containing Texas red–labeled yeast (red) by their ability to colocalize with lysosomes labeled with FITC-dextran (green) over time (). In wild-type cells, most internalized yeast colocalized with FITC-dextran by 1 h (). In
cells, fewer yeast were internalized, and those that were rarely colocalized with FITC-dextran (). shows a quantitative analysis of colocalization of yeast and FITC-dextran images of wild-type and
cells as a scatter diagram, demonstrating a high degree of colocalization (linearity) in wild-type but not in
cells.
cells using fluorescence microscopy ().
cells had significantly fewer yeast colocalized with lysosomes, demonstrating that without functional VSK3, cells are defective in the fusion of phagosomes and lysosomes.
Although intense study of mammalian phagocytes and has uncovered numerous molecules that regulate the early events of phagocytosis, signaling pathways that mediate the later events of phagosome maturation have yet to be clearly defined. In this study, we have identified and characterized a novel RTK-like kinase, VSK3, that localizes to late endosomes/lysosomes (see ). VSK3 has a characteristic RTK topology, spanning the membrane of intracellular vesicles once with a lumenal N-terminal domain and a cytoplasmic C-terminal kinase domain (see , , and ). The kinase domain actively phosphorylates tyrosine residues (see ), consistent with a localized signaling role for VSK3 on the cytosolic face of vesicles. VSK3 is required for normal phagocytosis but the defect lies in the number of cells in a population that can ingest particles (see and ).
cells cannot efficiently bind to particles, but once attachment occurs, actin-mediated phagocytic cup and phagosome formation are indistinguishable from that observed in wild-type cells (see ). VSK3 does not appear to regulate fluid phase uptake or processing (; Fig. S3 A); however and most significantly, in null cells that do ingest particles, it appears that VSK3 is required for phagosome maturation as phagosome and late-endosome/lysosome fusion is significantly impaired (see ).
In mammalian cells, tyrosine phosphorylation plays an essential role during the earliest stages of phagocytosis in the clustering of Fcγ receptors during particle recognition and uptake and is accomplished by Src and Syk families of tyrosine kinases (; ). In addition to these early functions, another Src-family tyrosine kinase, Hck, was shown to be involved in phagosome maturation in macrophages (). Hck, although not an RTK-like molecule, localizes to a subset of lysosomal vesicles that specifically fuse with phagosomes containing particles whose entry is mediated by Fcγ receptor signaling, implicating a general role for tyrosine kinase signaling throughout the phagocytic pathway.
Metazoans encode a wide variety of single-pass, transmembrane tyrosine kinases. These prototypic RTKs localize to the plasma membrane and function to transduce extracellular signals to intracellular pathways, and only a few have been implicated in phagocytosis. Distinct from mammalian RTKs, VSK3 localizes to intracellular vesicles, but contains an N-terminal, immunoglobulin-like fold TIG domain (), which, according to our work (see ), would face the lumen of vesicles. The presence of TIG domain makes VSK3 most similar to the MET family of mammalian RTKs (), which includes RON kinase, an RTK expressed in macrophages that functions, among other roles, in attenuating the inflammatory response (). In comparison to higher eukaryotes, encodes a limited number of potential RTK-like molecules (). Including this report, functions have been attributed to seven of the nine putative kinases (; ). Although the role of two potential single pass RTKs we term VSK1 and 2 (rk1 and rk2 in ) remain to be determined, it is interesting to note that when expressed as GFP fusion proteins, VSK2 localizes on the membrane of late endosomes/lysosomes like VSK3, while VSK1 is distributed in membranes of other vesicles (unpublished data), suggesting that tyrosine kinase signaling may play additional roles in vesicle trafficking and phagosome maturation.
It has long been recognized that certain pathogenic organisms escape destruction and grow in both and mammalian phagocytes by undermining phagosome maturation (). Death avoidance strategies uncovered thus far indicate that different microorganisms have adapted specialized means to subvert the completion of phagocytosis. For example, disrupts early endosome autoantigen and PI3K function, thus inhibiting trafficking downstream of Rab5 (), whereas inhibits the trafficking of NADPH oxidase and nitric oxide synthase, two antimicrobial enzymes (). Due to its role in phagosome maturation, it is intriguing to speculate that VSK3 may serve as a potential target by an invading pathogen to prevent late endosome/lysosome fusion.
In conclusion, our work shows for the first time that an RTK-like molecule, which resides in late endosomes/lysosomes, functions to mediate phagosome maturation. It has become increasingly clear that receptor signaling occurs not only at the plasma membrane but also along the endocytic pathway (). It may be possible that VSK3 functions like a typical receptor generating specific signals in these cytoplasmic compartments. One of numerous scenarios that can be envisaged is that signals generated in phagosomes by an engulfed particle activate VSK3 in vesicles as they fuse. In turn, activated VSK3 may regulate the downstream fate of the maturing phagosome, similar to the function of TLR9 in the response to bacterial DNA in endosomes of macrophages (). Although such functions remain to be tested, it will be interesting to determine in the future whether localized tyrosine kinase signaling represents a control mechanism for phagosome maturation in other phagocytes besides .
Tyrosine and serine/threonine kinase domains were verified according to in putative protein kinases as identified by the Dictyostelium Sequencing Project (). Protein structural motifs were identified using DNASTAR and sequence analysis tools found at and .
For development, were grown in D3-T nutrient media (KD-Medical) to log phase (1–3 × 10 cells/ml) and harvested by low speed centrifugation (2000 ) for 3 min. Cells were washed in Development Buffer (DB; 7.4 mM NaHPO·HO; 4 mM NaHPO·7HO; 2 mM MgCl; 0.2 mM CaCl; pH 6.5), centrifuged and resuspended in DB to 2 × 10 cells/ml (). For synchronous development in shaking suspension, cells were rotated at 100 rpm on a platform shaker at 22°C in DB and received exogenous 75-nM pulses of cAMP every 6 min. Development and the ability to phagocytose bacteria were assessed on bacterial lawns of grown on SM-agar plates.
Cells were differentiated by exogenous cAMP pulses in DB and 10 cells were harvested at various times. Total RNA was isolated using TRIzol reagent (Invitrogen) according to manufacturer's instructions. 1 μg of DNase-treated RNA was converted to cDNA using the SuperScript first-strand synthesis system (Invitrogen). A 5% volume of the cDNA reaction was used as template for real-time PCR using a Light-Cycle thermal cycler (Roche Applied Science) and PCR products were detected with SYBR Green I. Real-time PCR conditions were according to the QIAGEN protocol except the extension temperature was 60°C. cDNA copy number was determined by using QuantiTect SYBR Green PCR kit (QIAGEN). The primers used to amplify experimental and control cDNAs are as follows: : 5′-TGACCCATATACTGAGAAAG-3′ and 5′-GACGTGTGAATGGAGCGATACT-3′; cAR1: 5′-TGGGCATCTGTCACATTTATCT-3′ and 5′-GGAACTACATTGCACATCATCAC-3′; Gβ subunit: 5′-CAGTGGTGCTTGTGATGCTA-3′ and 5′-ATGTTGTCGTGGGTGTATTG-3′.
The vsk3 disruption construct was generated by inserting the blasticidin resistant (BSR) cassette into the vsk3 cDNA 951 nucleotides from the translation start codon and was cloned into pBluescript KS+. To create vsk3-null strains, the vsk3-BSR construct was cut and purified from the carrier plasmid and transformed into . Transformants were grown and selected in D3-T media containing 10 μg/ml blasticidin in 10-cm cell culture dishes. Individual colonies were picked 5–7 d later and homologous recombinants were identified by PCR and confirmed by Southern blot.
To fuse YFP to the C terminus of VSK3, the vsk3 coding sequence was amplified from cDNA with oligonucleotides that provided 5 BglII and 3 AgeI restriction sites. The amplified product was cloned into the pEYFP-N1 (CLONTECH Laboratories, Inc.) to create pVSK3-YFP. To engineer VSK3ΔK-YFP and K-YFP, phosphorylated primers were designed to anneal at locations flanking the sequence in pVSK3-YFP to be deleted and amplify the remaining plasmid. The PCR product of each reaction was ligated and transformed into TOP10 cells for propagation. For VSK3ΔK-YFP, the following sense and antisense primers were used for the deletion of aa 492–740 from VSK3: aatctctgatatatcaattggttT and gaaatcgttaaaagattgga. For K-YFP, the following primers were used delete the N-terminal domain (aa 1–481), which includes the transmembrane domain, from VSK3: cattttttaataagatctgagtccggtagcgctagc and tttgaaattaaaccaattgatatatcag. PCR fidelity was confirmed by DNA sequencing. The wild-type and mutant VSK3-YFP coding sequences were subsequently cloned into the expression vector pYU20, containing the G418 resistant cassette. Cells were transformed, grown and selected in D3-T media containing 20 μg/ml G418. The predicted molecular weight of the VSK3-YFP fusion protein expressed in cells was confirmed by Western blot with anti-GFP antibodies.
The invariant lysine in the predicted ATP binding site of VSK3 was mutated to arginine by a two-step fusion PCR mutagenesis. Primer 1 anneals to the 5′ end of the VSK3-YFP cDNA and includes a BglII site for subcloning and a ribosome binding site (underscored). Primer 4 anneals to the 3′ end and includes a NotI site for subcloning (underscored). Primers 2 and 3 target the invariant lysine on opposite strands of the cDNA. A single nucleotide was changed from A to G to encode the lysine to arginine conversion and a neighboring alanine (GCA) was mutated to leucine (CTT) to create a HindIII site to screen for positive mutants (underscored). For the first round of PCR, primers 1 and 3 or primers 2 and 4 were mixed with VSK3-YFP cDNA and the resulting amplification products of each reaction were gel purified. For the second, fusion PCR reaction, the purified amplification products were mixed together with the flanking primers 1 and 4. The resulting VSK3-YFP arginine mutant was subcloned into a bacterial shuttle vector and finally placed into the expression vector pCV5 using the BglII and NotI sites. Primer1: ACTATTAAAAAATGATAATAATAAATAAATATATACGGATG; primer 2: TATTGTTGCAATTAAAAGAAATTATTGAATGAAGAC; primer 3: TCAATAATTTCTTTTAATTGCAACAATAATACC; primer 4: CCTTACTTGTACAGCTCGTCC.
Cells were loaded with FITC-dextran for 10 min at 22°C with 2.5 mg/ml FITC-dextran in phosphate buffer, pH 6.6, and washed with ice-cold phosphate buffer. Cells were then incubated at 22°C and the evolution of pH change in endosomes was monitored in aliquots over time by measuring the change in the ratio of FITC fluorescence at 518 nm when excited with 450-nm light to fluorescence at 518 nm when excited with 494-nm light. Endosomal pH values were calculated from a standard curve generated from the resulting fluorescence of FITC-dextran at differing pH values.
Quantitative phagocytosis assays were conducted as described previously (). In brief, cells grown to log-phase (1–3 × 10 cell/ml) were washed and suspended to 2 × 10 cells/ml in phosphate buffer with magnesium (PM; 7.4 mM NaHPO-H0; 4 mM NaHPO-7HO; 2 mM MgCl) and fed TRITC-labeled, heat-killed yeast in shaking suspension (150 rpm) at 22°C. Samples were taken at indicated times and the fluorescence of noninternalized yeast was quenched by the addition of Trypan blue. Fluorescence was measured with a Perkin-Elmer spectrophotometer using 544-nm excitation and 574-nm emission filters. In addition, a flow-cytometry method was used to determine the phagocytosis rates as previously described (). In brief, cells grown to log-phase were fed TRITC-labeled, heat-killed yeast in PM in shaking suspension at 22°C over a time course. After incubation, the cells were quenched and flow cytometry was performed on 10,000 cells for each sample. Fluorescence data were analyzed with FlowJo software. Quantitative pinocytosis assays were performed as described previously () by incubating log-phase cells with TRITC-dextran in suspension culture at 22°C. At various times, aliquots were obtained and extracellular fluorescence was quenched with Trypan blue. Internalized fluorescence was measured with a spectrophotometer as above.
Quantitation of phagosome and lysosome fusion was performed as described previously () with some modification. Log-phase cells were incubated with FITC-dextran (1 mg/ml) in D3-T medium for 3 h to mark lysosomes. Next, heat-killed, Texas red–labeled yeast were mixed 5:1 with in shaking suspension at 160 rpm for 1 h. Cells were then placed in a 4-well chamber for 20 min and allowed to adhere. The uningested yeast were washed away and cells were incubated for an additional hour. Phagosome–lysosome fusion was evaluated by confocal microscopy as a colocalization of yeast and FITC-dextran and was quantified by determining the percentage of phagosomes to phagolysosomes by randomly counting 100 yeast-containing cells.
The membrane topology of VSK3 was determined as previously described () with some modification. Cells expressing VSK3-YFP were incubated with TRITC dextran for 3 h in shaking suspension at 160 rpm in D3-T medium to label lysosomes. Cells were placed in an 8-well Lab-Tek chamber and were imaged by confocal microscopy. A final concentration of 10 μM digitonin, to selectively permeabilize the plasma membrane, and 4 mM trypsin, to digest cytosolic-facing moieties of internal vesicle membrane proteins, were added to the chamber. Changes in the YFP and TRITC signal were recorded over time.
For immunoprecipitation of YFP-tagged proteins, 100 μl of Protein G beads were prepared (Invitrogen) by washing four times with ice-cold PBS and once with ice-cold lysis buffer (1% NP-40, 50 mM Tris, pH 7.5, 150 mM NaCl, protease inhibitors, and 1 mM NaVO4). The beads were resuspended in 1 ml ice-cold lysis buffer and coupled to anti-GFP antibody (CLONTECH Laboratories, Inc.) for 2 h at 4°C and then washed four more times with lysis buffer. To carry out immunoprecipitation, 10 log-phase cells were washed twice with ice-cold PBS containing protease inhibitors and 1 mM NaVO. The cells were resuspended in 5 ml of lysis buffer and vortexed briefly every 5 min for 20 min. Insoluble matter was removed from the lysate by centrifugation (20,000 ) for 15 min. The supernatant was precleared by incubation with 100 μl of Protein G beads (no antibody coupling) for 1 h. The beads were removed by centrifugation and the supernatant was incubated with 100 μl anti-GFP-coupled beads for 3 h at 4°C. The supernatant was removed by centrifugation and the beads were washed four times with ice-cold lysis buffer. Proteins were removed from a 10-μl bed volume of beads and for Western blot analysis. The remaining beads were used to for a tyrosine kinase activity assay. The tyrosine kinase activity assay was conducted by using the SignaTECT Protein Tyrosine Kinase Assay System (Promega) according to the manufacturer's instructions. Ser/Thr kinase activity assay was conducted by using the KinEASE FP Fluorescein Green Assay (Upstate Cell Signaling Solutions) according to the manufacturer's instructions.
Confocal fluorescent observations were made using a laser-scanning microscope (LSM 510 META; Carl Zeiss MicroImaging, Inc.). Either a 40×/1.3 NA or 63×/1.4 NA oil immersion objective lens was used to image cells (). An argon 488-nm laser was used to excite GFP, YFP, and FITC fluorescence and a HeNe 543-nm laser was used for neutral red and TRITC excitation.
cells expressing VSK3ΔK-YFP were washed from growth media and shaken at 100 rpm for 30 min in PBS (pH 7.4). Cells were plated at a density of 75 × 10 cells/cm in PBS on to a circular plastic coverslip (Thermanox) placed in a well of a 24-well plate and allowed to adhere for 10 min. Cells were fixed with 1% formaldehyde, 0.1% glutaraldehyde, 0.01% digitonin, in Pipes/EGTA buffer (15 mM Pipes, pH 7.4, and 1 mM EGTA) for 15 min and were washed 2 × 2 min then 1 × 15 min in 1% formaldehyde in Pipes/EGTA. Next, cells were washed 3 × 5 min with Pipes/EGTA and then incubated in block solution (50 mM NHCl, 0.1% digitonin, and 1% BSA in PBS) for 30 min. Cells were incubated overnight at 4°C with rabbit anti-GFP antibody (ABCAM 290) diluted to 1:500 or 1:1,000 in block solution on a gently gyrating platform. Cells were washed 6 × 2 min then 1 × 5 min in PBS and incubated with goat anti–rabbit Nanogold 1.4-nm particle Fab fragment (Nanoprobes) diluted 1:50 in block solution for 3.5 h at 22°C on a gently gyrating platform. Cells were washed 6 × 2 min in PBS then 3 × 5 min in distilled water. Gold particle enhancer was prepared per manufacturer's protocol (Nanoprobes GEEM GoldEnhance kit) and incubated with cells for 8 min. The reaction was stopped by washing with distilled water 4 × 2 min. Cells were stored in 1% glutaraldehyde in PBS. For TEM analysis, the cells were washed 3× with 0.1 M cacodylate buffer and incubated in 0.5% OsO, 0.8% KFe (CN) in 0.1m cacodylate buffer for 30 min. This was followed by 1 wash with 0.1 M cacodylate buffer and 2 washes with distilled water. The cells were dehydrated in ethanol series of 50, 75, 95, and 100% for 10 min, respectively. The last ethanol concentration was repeated two more times after which the cells were infiltrated with 3:1 ethanol/Spurrs resin for 1 h, 1:1 ethanol/Spurrs for 2 h, and overnight in 1:3 ethanol/Spurrs. The next day the cells were infiltrated with 100% Spurrs resin for 4 h and embedded in beam capsules at 68°C for 24 h. Ultrathin sections were obtained using the MT-7000 ultra microtome and stained with UA. The images were obtained using Philips C10 TEM.
Fig. S1 shows the mRNA levels of the and β genes during development as detected by RT-PCR. Fig. S2. shows the cloning strategy and analysis of disruptant mutants. Fig. S3 A shows that wild-type and mutant strains have similar rates of pinocytosis. Fig. S3 (B and C) shows the results of phagocytosis assays of cells expressing VSK3 kinase-dead mutants. Fig. S4 shows that the VSK3K518R-YFP point mutant does not have detectable kinase activity.
(Video 2) cells. Video 3 shows late endosomes/lysosomes containing VSK3ΔK-YFP fusing with phagosomes containing newly ingested yeast. Online supplemental material is available at . |
Lipopolysaccharide (LPS), the major constituent of the outer membrane of Gram-negative bacteria, is a potent stimulator of the immune system (). Besides its proinflammatory activities, LPS participates in bacterial membrane functions. LPS is essential for bacterial growth and viability since it contributes to low membrane permeability and enhances resistance toward hydrophobic agents (). From a structural point of view, LPS consists of a polysaccharide part attached to a lipid (lipid A). Lipid A consists of a diglucosamine backbone to which two phosphates are linked at positions 1 and 4′, and six or seven ester- and amide-linked acyl chains are bound at positions 2, 3, 2′, and 3′ (). The polysaccharide part, which facilitates the solubility of the molecule in water, consists of two parts: 1), the O-specific chain, an oligosaccharide with a composition varying with bacterial species; and 2), a rather invariable core section, which is located between the oligosaccharide and the lipid A. Wild-type enterobacterial species with O-chains are termed “smooth” and their LPS called “smooth LPS” (S-LPS). Mutants producing LPS lacking O-specific chains are termed “rough” (R) and their LPS designated as Ra, Rb, Rc, Rd, and Re in order of decreasing core length (). The endotoxic molecule with the smallest molecular size and full endotoxic activities is Re-LPS (). Bacteria with Re-LPS phenotypes are more common among pathogens that colonize the upper aerodigestive tract ().
Due to inhalation of airborne particles containing bacteria and LPS, the thin alveolar epithelium is continuously exposed to this potent proinflammatory molecule. When LPS molecules enter the host via airways, they interact with alveolar macrophages in a fluid environment characterized by the presence of pulmonary surfactant, which is involved in reducing the surface tension of the fluid lining the alveoli and in host defense. Several components present in the lipid-rich alveolar fluid, such as surfactant proteins (SP-) A, C, and D (–), are involved in the binding and neutralization of LPS and/or downregulation of LPS responses that promote excessive inflammation and compromise gas exchange.
SP-A is a large oligomeric extracellular protein found primarily in the alveolar fluid of mammals. It belongs to the structurally homologous family of innate immune defense proteins known as collectins for their collagen-like and lectin domains (,). Unlike other collectins, SP-A is mainly associated with surfactant lipids, especially with DPPC. SP-A's ability to bind lipids 1), improves the adsorption and spreading of surfactant membranes onto an air-liquid interface (); 2), protects surfactant biophysical activity from the inhibitory action of serum proteins (); and 3), allows this protein to position and concentrate along with surfactant membranes as an initial defense against inhaled toxins and pathogens. We previously reported that SP-A interacts with DPPC monolayers (,) and with gel-like regions of monolayers of lung surfactant lipid extract (). SP-A is able to bind not only to surfactant membranes but also to pathogen-associated molecular patterns on microorganisms, such as bacterial LPS, which has long been known to bind to SP-A (,). We have recently reported a detailed study of the characteristics of the interaction of human SP-A with bacterial Re-LPS ().
We expect that, given the lipophilic nature of LPS, inhaled LPS might incorporate into the lung surfactant DPPC-rich monolayer and then interact with SP-A at the interface. This would likely make LPS less available for signaling. Thus the first objective of this study was to investigate the miscibility of Re-LPS in DPPC monolayers and the effect of Re-LPS on the lateral lipid organization of these monolayers, determined by epifluorescence microscopy. The second objective was to find out whether SP-A from the hypophase associates with DPPC/Re-LPS-mixed monolayers as well as with DPPC monolayers (,) and whether SP-A modifies lipid lateral organization of such monolayers. The third objective was to visualize the arrangement of fluorescently labeled SP-A (TR-SP-A) in association with DPPC, DPPC/Re-LPS, and lipid A monolayers.
1,2-Dipalmitoylphosphatidylcholine (DPPC) and the fluorescent lipid probe 1-palmitoyl-1-{12-[(7-nitro-2-1,3-benzoxadizole-1-yl)amino]dodecanoyl} phosphatidylcholine (NBD-PC) were obtained from Avanti Polar Lipids (Birmingham, AL). Re-LPS and diphosphoryl lipid A from (serotype Re 595) were purchased from Sigma (St. Louis, MO). The fluorescent probe used to chemically label SP-A, sulforhodamine 101 sulfonyl chloride or Texas red (TR), was obtained from Molecular Probes (Eugene, OR). The organic solvents (methanol and chloroform) used to dissolve DPPC, lipid A, and Re-LPS were high performance liquid chromatography grade.
SP-A was isolated from bronchoalveolar lavage of patients with alveolar proteinosis using a sequential butanol and octylglucoside extraction (). The purity of SP-A was checked by one-dimensional sodium dodecylsulfate-polyacrylamide gel electrophoresis in 12% acrylamide under reducing conditions and mass spectrometry. The oligomerization state of SP-A was assessed by electrophoresis under nondenaturing conditions and electron microscopy as reported elsewhere (,).
Fluorescently labeled SP-A was prepared as described in Ruano et al. (). Briefly, SP-A in 5 mM Tris-HCl buffer, pH 8.3, was incubated with 1 mM TR (SP-A/TR molar ratio of 6:1) for 90 min in darkness at room temperature. To remove unreacted fluorescent reagent, the mixture was exhaustively dialyzed against 5 mM Tris-HCl, pH 7.4. Activity of labeled TR-SP-A compared to that of native SP-A was assayed by testing its ability to self-associate and to induce aggregation of DPPC and Re-LPS in the presence of calcium at 37°C as described elsewhere (–). The effect of TR-SP-A on -A isotherms of DPPC monolayers was essentially identical (within our ability to measure) to that of nonconjugated SP-A.
DPPC, lipid A, Re-LPS, or DPPC mixed with different amounts of Re-LPS were spread from chloroform/methanol 3:1 (v/v) solutions onto a buffer A subphase (150 mM NaCl, 5 mM Tris-HCl, pH 7.4) containing either 150 M EDTA or 150 M CaCl, with or without SP-A, in a thermostated Langmuir-Blodgett trough (302RB Ribbon Barrier Film Balance, NIMA Technologies, Coventry, UK). The concentration of Ca was set at 150 M because at this concentration there is a molar excess of calcium in the subphase with respect to lipid and protein concentrations. The use of low Ca concentrations prevents the extensive protein self-association that occurs at higher Ca concentrations (). Extensive SP-A self-aggregation hampers the interaction of SP-A with lipid films ().
Epifluorescence microscopy measurements were performed on a surface balance whose construction and operation have been described previously (,). DPPC, lipid A, Re-LPS, and DPPC/Re-LPS combinations were mixed in chloroform-methanol solutions with 1 mol % NBD-PC (based on the lipid content). Monolayers were formed by spreading onto a buffered saline subphase (buffer A) containing either 150 M CaCl or 150 M EDTA, with or without 0.08 g/ml of TR-SP-A. After a 1-h period, which would have allowed for solvent evaporation and penetration of the protein into the gas or gas-liquid expanded coexistence phases, the monolayer was compressed at a slow speed (20 mm/s or an initial rate of 0.13 Å/molecule/s) at 25°C. At selected surface pressures, compression was halted, and a video recording was made for a 1-min period of both NBD-PC and TR fluorescence by switching fluorescence filter combinations. As described previously, our apparatus does not allow for instantaneous recording of images at two wavelengths (–). Therefore, images at the two wavelengths are not superimposable because of monolayer movement between the acquisitions of the images. The video images were obtained with a charge-coupled device camera that records in black and white. Images were analyzed with digital image processing using JAVA 1.3 software (Jandel Scientific, San Rafael, CA) as discussed elsewhere (,).
shows the surface pressure area (-A) isotherms obtained at 25°C for monolayers of DPPC and Re-LPS and mixtures of DPPC/Re-LPS. Pure DPPC gave monolayers that exhibited a transition region between liquid-expanded (LE) and liquid-condensed (LC) phases at surface pressures in the range of 7–12 mN/m on a buffered saline subphase containing 150 M CaCl (). The collapse pressure for this phospholipid was ∼70 mN/m. The presence of 150 M CaCl in the subphase did not appreciably modify the -A isotherm of DPPC when compared to that obtained in the absence of calcium (). also shows that the -A isotherms of DPPC/Re-LPS-mixed monolayers were shifted markedly to larger molecular areas relative to the isotherm of pure DPPC (). This shifting depended on the molar fraction of Re-LPS, . The -A compression isotherm for pure Re-LPS monolayers showed an easily distinguishable collapse at ∼56 mN/m and a kink at 36 mN/m, which might indicate a LC/LE transition (, ).
The collapse of all DPPC/Re-LPS-mixed monolayers took place at surface pressures between those of pure DPPC and Re-LPS monolayers (). According to the two-dimensional phase rule, if the monolayer components are immiscible in the condensed and collapsed states, the isotherm will show two distinct collapse pressures corresponding to those for the pure components. However, collapse pressures of mixed monolayers composed of two miscible components will vary with composition (). Thus, the dependence of the collapse pressure on the molar fraction of Re-LPS and the fact that the collapse pressures of the mixed films lie between the collapse pressures of pure components indicate that Re-LPS and DPPC are miscible. To demonstrate more clearly the interaction between these two components of the mixtures, the mean area per molecule was plotted as a function of the mole fraction of Re-LPS at three surface pressures: 11, 32, and 45 mN/m (). The dotted lines are the mean molecular areas calculated by assuming ideal mixing. If an ideal mixed monolayer is formed or the two components are completely immiscible in the two-dimensional state, the plot of mean molecular areas as a function of the mole fraction of one of the components at a given surface pressure would be a straight line. The results show that, for all surface pressures studied, negative deviations from ideal behavior were observed, indicating attractive interaction between DPPC and Re-LPS.
To characterize the possible effects of Re-LPS on DPPC lateral organization, epifluorescence microscopy images of DPPC monolayers containing 1 mol % NBD-PC were recorded in the absence and presence of different amounts of Re-LPS (). The fluorescent dye (NBD-PC) concentrated in the LE phase. which appeared bright in the fluorescence images of the monolayers (–). In the low pressure regime, DPPC monolayers showed black nonfluorescent domains, the LC phase dispersed very homogeneously in a fluorescent environment (LE). As the monolayer was compressed the relative area occupied by the LC domains increased (). Kidney-shaped solid domains, typical of the LE/LC coexistence region of DPPC monolayers (), were observed over the range of 7–12 mN/m. At higher surface pressures solid domains grew to occupy most of the monolayer. Comparison of these images with those obtained in the absence of calcium (,) indicates that calcium lowers the surface pressure corresponding to LE-LC phase transition. Thus dark probe-depleted condensed domains were first seen at ∼5.4 mN/m in the presence of Ca (data showed in ) and at ∼7.4 mN/m in its absence ().
On the other hand, shows that the presence of Re-LPS in DPPC/Re-LPS-mixed films increased the transition pressure from LE to LC phase. Furthermore, the presence of Re-LPS changed the shape of liquid-condensed domains to trefoil-like forms and reduced their size, causing a relative decrease in the LC phase in the DPPC/Re-LPS films. As the amount of Re-LPS in the monolayer increased, the size of the nonfluorescent DPPC-rich LC domains decreased (). For ≥ 0.5 no LC domains were observed and the mixed monolayers are in the LE fluid state. Thus, it is likely that DPPC contributes mainly to the LC domains and that Re-LPS mixed with DPPC are in the brilliant LE phase. In contrast to DPPC monolayers, no LE to LC phase transition could be observed for pure Re-LPS monolayers at relevant lateral pressures (7–30 mN/m) (data not shown). This indicates either that there is a single phase or that LE and LC domains have much smaller dimensions than the optical resolution. In this way, Roes et al. () recently reported that small LE and LC domains can be observed in pure Re-LPS monolayers by atomic force microscopy, the sizes of which are below the optical resolution of a light microscope.
shows the -area per lipid molecule isotherms of monolayers of DPPC alone and in the presence of = 0.2 spread over buffered saline subphases containing 150 M CaCl with or without 0.08 g/ml SP-A. SP-A caused an expansion of either DPPC or DPPC/Re-LPS isotherms, which suggests that the protein was taking up some space in these monolayers or was interacting with the phospholipid monolayer enough to perturb the usual lipid packing. SP-A alone did not adsorb to the interface at the concentration used in the experiments. SP-A needs a lipid monolayer to pull it into the surface. The SP-A-induced displacement of -A isotherms to larger molecular areas was greater for DPPC/Re-LPS than for DPPC monolayers. For instance, at 20 mN/m, the area increase was ∼3 times larger for DPPC/Re-LPS than for DPPC monolayers. SP-A adsorption to the monolayers could affect the interactions between Re-LPS and DPPC, driving their ideal mixing or complete demixing. Either of these would effectively lead to an increase in the mean molecular area per lipid molecule toward the additive line (see ). On the other hand, Re-LPS in DPPC/Re-LPS may promote the adsorption of SP-A, which would cause greater increase in the mean area per lipid molecule in DPPC/Re-LPS compared to DPPC monolayers. The presence of large fluorescent clusters of TR-SP-A in the DPPC/Re-LPS films (see ) seems to support this explanation.
At a surface pressure of ∼50 mN/m the mean area per lipid molecule in the DPPC/Re-LPS films was the same in the absence and presence of SP-A in the subphase (). Therefore, at surface pressures >50 mN/m, SP-A did not occupy space in the monolayers and did not perturb the binary DPPC/Re-LPS monolayers. In contrast, the perturbing influence of SP-A on DPPC monolayers was observed only up to ∼30 mN/m. These observations suggest that specific interactions between Re-LPS and SP-A might facilitate both the adsorption of SP-A to the monolayers and its perturbing effects on the lipid monolayers at higher surface pressures.
shows images obtained from a monolayer of DPPC containing 1 mol % NBD-PC formed on a buffered saline subphase containing 150 M Ca, with or without 0.08 g/ml TR-SP-A. Fluorescence coming from either the lipid probe (NBD-PC) or the fluorescently labeled protein (TR-SP-A) was selectively recorded from the same monolayers by switching the filters. NBD fluorescence showed that at low surface pressures (3.9 mN/m) solid domains were formed in the presence but not in the absence of SP-A. Thus, SP-A increased the formation of solid phases. This SP-A effect on DPPC monolayers in the presence of calcium differs from that observed in its absence, where SP-A did not promote solid phase formation (). also shows that the fluorescence of the protein (TR-SP-A) was not observed in the monolayers at very low pressures (below 3.9 mN/m) and was only seen when LC domains began to appear. At 3.9 mN/m, intense rings of TR fluorescence surrounding DPPC-condensed domains were apparent. As the pressures increased, the protein fluorescence was evenly distributed in the fluid film and the dark LC domains became fuzzy and were penetrated by fluorescent points, indicating TR-SP-A incorporation in the condensed regions. This behavior differs from that observed in the absence of CaCl in the subphase, where SP-A does not directly associate with DPPC in LC phase (,).
shows images obtained from a mixed monolayer of DPPC/Re-LPS ( = 0.2) containing 1 mol % NBD-PC formed over a buffered saline subphase containing 150 M CaCl with or without 0.08 g/ml TR-SP-A. NBD fluorescence showed that at low surface pressure (6–8 mN/m) solid domains were formed in the presence but not the absence of SP-A. At higher surface pressures (8.4 or 11.3 mN/m), the relative area occupied by the LC domains was higher in the presence of SP-A than in its absence. Thus SP-A promoted the formation of DPPC-rich solid domains. SP-A adsorption to mixed DPPC/Re-LPS monolayers could affect the interactions between Re-LPS and DPPC and lead to decreased miscibility between Re-LPS and DPPC. This would favor the formation of DPPC-rich solid domains under compression. The micrographs in the central panel of show that up to ∼11.3 mN/m LE and LC phases coexisted in the monolayers. At 15 mN/m a distinct new phase that dissolves more lipophilic dye appeared (see ). It is difficult to experimentally assess the composition of this new phase. Since it occurs only in the ternary system DPPC/Re-LPS/SP-A but not in the binary systems DPPC/SP-A () and DPPC/Re-LPS (, ), it seems reasonable to infer that SP-A is inducing this new phase as a consequence of its interaction with Re-LPS. It is possible that at surface pressures ∼15 mN/m, DPPC/Re-LPS/SP-A (but not DPPC/Re-LPS) monolayers were in the LC state. This is consistent with the facts that SP-A induced solid domains and that the pressure-area curves for DPPC/Re-LPS films formed on SP-A had an inflection point at ∼16 mN/m, which might indicate the transition pressure to LC phase (). The brilliantly fluorescent LE phase enriched with NBD probe, visualized in epifluorescence images, might be “frozen” within the LC phase. The latter appeared both dark (the typically dye-depleted LC phase) and bright gray in appearance because it had dissolved some of the NBD probe.
also shows that TR fluorescence was only visible when condensed domains began to appear at 6 mN/m. However, the distribution of TR-SP-A in DPPC/Re-LPS ( = 0.2) monolayers was different from that observed in monolayers of pure DPPC. At surface pressures >6mN/m, TR-SP-A accumulated at the fluid-solid boundary regions and formed networks of interconnected filaments in the fluid phase of DPPC/Re-LPS monolayers. These reticular structures were also seen in monolayers containing = 0.3 in which the size of LC domains are reduced with respect to mixed monolayers containing = 0.2 (data not shown). They were also observed in DPPC/lipid A-mixed monolayers ( = 0.2) (data not shown) and in monolayers of pure lipid A formed over a buffered saline subphase containing calcium and TR-SP-A.
() shows the -A compression isotherms for pure lipid A monolayers in the absence and presence of 0.08 g/ml SP-A. The isotherms showed a collapse at ∼60 mN/m and an inflection point at ∼35 mN/m. The presence of SP-A in the subphase led to an increase in the film area of lipid A monolayers, suggesting that SP-A also incorporated into these monolayers. Epifluorescence microscopy images indicated that bright fluorescent reticular structures of TR-SP-A appeared in lipid A monolayers at surface pressures ≥9 mN/m at which LE-LC phases coexist () (, ). These lattice-like structures appear to be specific for the interaction of SP-A with the lipid A moiety of Re-LPS.
Since SP-A self-associates (,) and induces LPS aggregation in the presence of calcium (,,), experiments were performed in the presence of 150 M EDTA to determine whether the effects of SP-A on DPPC/Re-LPS and lipid A films were calcium dependent. () shows typical -A isotherms of DPPC/Re-LPS ( = 0.2) monolayers spread onto a buffered saline subphase containing EDTA, with or without 0.08 g/ml SP-A. SP-A expanded the interfacial DPPC/Re-LPS film in the presence of EDTA (). However, the perturbing influence of SP-A on DPPC/Re-LPS-mixed monolayers was observed over a greater range of surface pressure in the presence of calcium (∼6–50 mN/m) than in its absence (∼10–30 mN/m) ( and ). NBD-PC fluorescence (, ) showed that, in the absence of calcium, SP-A hardly increased the formation of solid domains. With respect to microscopic images obtained with the TR filter (, ), LE-LC phase coexistence was also required for the appearance of fluorescent SP-A around liquid-condensed domains in the presence of EDTA. Once solid domains were formed, TR-SP-A was seen forming a network of interconnected filaments () as had been observed in the presence of calcium (). At a surface pressure of 30 mN/m no protein fluorescence was detected, consistent with protein exclusion from the interface or less SP-A binding to the film.
() shows the -A compression isotherm for pure lipid A monolayers formed over a buffered saline subphase containing 150 M EDTA, with and without 0.08 g/ml SP-A. SP-A also expanded lipid A interfacial films in the presence of EDTA. TR epifluorescence microscopy images (, ) indicated that bright fluorescent reticular structures of TR-SP-A appeared in lipid A monolayers at surface pressures ≥9 mN/m . These results indicate that these lattice-like structures formed by SP-A in either lipid A or DPPC/Re-LPS films are independent of calcium and seem to depend on the specific interaction of SP-A with the lipid A part of bacterial LPS.
In this study, epifluorescence microscopy combined with a surface balance was used to examine the interaction of SP-A with mixed monolayers of DPPC/Re-LPS. The rationale for this study is based on the fact that DPPC/LPS-mixed monolayers might be formed in the lung as a consequence of inhaled airborne particles containing bacterial LPS. Given the lipophilic nature of LPS, these molecules might incorporate into the DPPC-rich monolayer.
Using surface pressure area isotherms, we show that DPPC and Re-LPS were miscible and that there was an attractive interaction between the lipids (). The fact that the collapse pressures of DPPC/Re-LPS-mixed films were between the collapse pressures of pure components (70 mN/m for DPPC and 56 mN/m for Re-LPS) indicates miscibility between both lipids. In addition, binary monolayers of Re-LPS plus DPPC showed negative deviations from ideal behavior of the mean areas of the films, which is consistent with partial miscibility and attractive interaction between the lipids. Epifluorescence microscopy shows that Re-LPS decreased the relative area occupied by the probe-depleted LC domains () and increased the surface pressure corresponding to the LE/LC transition: solid domains began to appear at higher surface pressures in DPPC/Re-LPS films than in DPPC films. Furthermore, the presence of Re-LPS in DPPC films reduced the size of solid domains. Given that the size of LC domains of DPPC-Re-LPS monolayers decreased with increasing Re-LPS mol % in the monolayer, it is likely that LC domains of DPPC/Re-LPS monolayers consisted mainly of DPPC. The dissolution, rearrangement, and decrease in the size of DPPC-rich solid domains induced by Re-LPS are consistent with miscibility and attractive interaction between these two lipids.
The adsorption of SP-A to DPPC/Re-LPS monolayers caused expansion in the lipid molecular areas in both the presence and absence of calcium, which indicates an intercalation of SP-A molecules into the DPPC/Re-LPS monolayer ( and ). The lack of Ca requirement for the interaction of SP-A with DPPC/Re-LPS films is consistent with previous results that indicate that SP-A binds to Re-LPS and DPPC in the absence of calcium (,,,,). However, it would be expected that the presence of Ca influences the interaction of SP-A with DPPC/Re-LPS monolayers since calcium causes conformational changes in the globular domain of SP-A, as detected by fluorescence spectroscopy (,,) and transmission electron microscopy (). This conformational change enhances lipid binding and allows SP-A-SP-A self-association and SP-A-mediated lipid aggregation (,). In addition, Ca binds tightly to the KDO moieties of Re-LPS (with an apparent dissociation constant of 14 M) (), so that Ca could neutralize these negatively charged moieties. Here we show that the presence of calcium influenced the effect of SP-A on the lipid lateral organization of DPPC/Re-LPS monolayers as determined by epifluorescence microscopy. The presence of calcium promoted SP-A-induced formation of solid domains in these monolayers, decreasing the surface pressure corresponding to the LE/LC transition (NBD epifluorescence images of and ). In addition, we found that calcium makes DPPC/Re-LPS monolayers more sensitive to SP-A over a larger surface pressure range ( and ).
The use of fluorescent TR-SP-A allows for the analysis of SP-A interaction with domains or regions of DPPC/Re-LPS monolayers. We found that TR fluorescence was only visible when LC domains began to appear, which occurred at ∼6 and 9.6 mN/m in the presence and absence of calcium, respectively. These results are consistent with the concept that SP-A recognizes the lipid in the gel phase and penetrates the membrane interface through lipid packing defects at liquid-condensed-liquid expanded boundaries (). Once solid domains are formed, TR-SP-A was observed forming a network of interconnected filaments in the fluid phase of DPPC/Re-LPS monolayers, possibly connecting solid domains in which TR-SP-A seems to accumulate. These protein reticular structures were formed when there was a relatively low SP-A concentration in the subphase (0.08 g/ml) in both the presence and absence of calcium ( and ). However, in the absence of this cation no protein fluorescence was detected at surface pressures ≥30 mN/m, which is consistent with protein exclusion from the interface or less protein binding at such surface pressures in the absence of calcium. Given that this protein network was also seen in DPPC/lipid A-mixed monolayers ( = 0.2) and in monolayers of pure lipid A (in both the presence and absence of calcium) ( and ) but not in DPPC monolayers (), we conclude that the formation of extensive lattice-like structures at the interface depends on the specific interaction of SP-A with the lipid A part of bacterial LPS.
These SP-A lattice-like arrays seemed to form through extensive SP-A-SP-A association, stabilized by the binding of the protein to the lipid film. SP-A binds to Re-LPS in solution with high affinity ( = 0.028 M) in a Ca-independent manner (), and the formation of this SP-A network on DPPC/Re-LPS monolayers was calcium independent. However, self-association of SP-A in solution requires Ca, with the calcium activation constant () for human SP-A self-association at physiological ionic strength 12 ± 1.8 M (). Our results suggest that the adsorption of SP-A to DPPC/Re-LPS monolayers facilitated cooperative assembling of SP-A molecules in a lattice-supraquaternary structure even in the absence of calcium. This type of supraquaternary organization of SP-A and cooperative interaction with DPPC/Re-LPS monolayers might play a physiological role. For instance, SP-A could induce aggregation of mixed Re-LPS/phospholipid membranes that are squeezed out from the monolayer on expiration in situ.
Aggregation of membranes containing Re-LPS could be important to reduce LPS toxicity () and facilitate phagocytosis of these aggregates by alveolar macrophages. This would prevent the binding of LPS to its receptor complex on alveolar epithelial and immune cells, which would launch an inflammatory response. On the other hand, the formation of a lattice of protein aggregates on pure lipid A monolayers suggests that this might be the initial step in lesion formation in the outer membranes of Gram-negative bacteria induced by SP-A. It was recently reported that SP-A causes increased Gram-negative bacteria permeability and killing (), presumably by direct effect of SP-A on the properties of the microbial cell membrane ().
It is important to mention that tubular myelin and multilamellar vesicles from native surfactant also contain arrays of SP-A (,). These arrays seem to be formed as a result of SP-A's capability to self-associate and bind to membranes. SP-A arrays remain intact when the lipid is partially removed with acetone (,), and their spacing is comparable to the size of SP-A. Interconnected SP-A molecules seem to form the skeleton of these structures. The formation of this type of SP-A supraquaternary organization adsorbed to surfactant membranes must be of physiological importance. It is thought that these protein arrays may stabilize large surfactant aggregates and decrease surfactant inactivation in the presence of serum protein inhibitors ().
In summary, this study shows that Re-LPS is miscible in the DPPC film. There is an attractive interaction between Re-LPS and DPPC, which results in reduction of the size of DPPC-rich solid domains in DPPC/Re-LPS monolayers. SP-A, the major surfactant protein component involved in host defense, incorporates into DPPC/Re-LPS monolayers and produces expansion in the lipid molecular areas. The perturbing influence of SP-A in DPPC/Re-LPS monolayers is seen over a greater range of surface pressure than in DPPC monolayers in the presence of calcium. With respect to the differential partitioning of fluorescently labeled TR-SP-A into regions of DPPC-Re-LPS monolayers, we found that fluorescent TR-SP-A accumulates at the fluid-solid boundary regions of these monolayers and forms networks of interconnected filaments in the fluid phase in a Ca-independent manner. Such protein reticular structures are also observed when TR-SP-A interacts with mixed DPPC/lipid A and pure lipid A monolayers. These novel results deepen our understanding of the specific interaction of SP-A with the lipid A moiety of LPS and may explain how SP-A behaves as an effective LPS neutralizing agent in the alveolus, making LPS less available to interact with components of the innate immune system in the airways and thus contributing to maintaining the lung in a noninflamed state. |
Interstrand cross-links (ICLs) are highly lethal DNA lesions which block DNA transcription and replication by preventing strand separation. Due to their high cytotoxicity, ICL-inducing agents like mitomycin C (MMC), cisplatin and psoralens are widely used against hyperplasic diseases such as cancer and psoriasis. Furanocoumarins (psoralens) are naturally occurring secondary metabolites in plants, including many fruits and vegetables. Among several ICL-inducing agents, psoralens require UVA-photoactivation following DNA intercalation to chemically react with DNA. 8-methoxypsoralen (8-MOP) is a planar, tricyclic compound which intercalates into DNA duplex at pyrimidines, preferentially at 5′-TpA sites. Upon photoactivation, 8-MOP primarily photoalkylates DNA by cycloaddition to the 5,6-double bond of a thymidine, generating two types of monoadducts (MA) with either the 4′,5′-double bond of the furan (MAf) or the 3,4-double bond of the pyrone (MAp) side of the psoralen (). A unique property of psoralen photochemistry is that the absorption of a second photon by the MAf leads to formation of a pyrone side 5,6-double bond adduct with a flanking thymine in the complementary strand, thus generating an ICL (). Although the yield of psoralen MAs to pyrimidine bases is 3-fold higher than that of ICLs, the latter class of damage appears to have more severe biological effect ().
Genetic and biochemical data indicate that elimination of ICLs is mainly linked to DNA replication and involves several linear repair pathways including structure-specific endonucleases, proteins required for homologous recombination-mediated double-strand break (DSB) repair and error-prone translesion DNA polymerases (). The thirteen FANC proteins, mutated in the Fanconi anaemia cancer-prone syndrome, participate in coordination of excision repair of ICLs in order to reconstitute the genetic material with high fidelity. However, at present, the mechanism of coordination and biochemical activities of the FANC proteins involved in the excision of ICLs remains poorly defined. More importantly, it was believed that bulky DNA lesions such as thymine–psoralen adducts can be eliminated only by the nucleotide excision repair (NER) pathway (). However, with the exception of XPF/ERCC1-deficient cells, cells lacking critical NER components only show a moderate sensitivity to ICL-inducing agents exposure which induces both MA and ICL (). These observations hint us at the existence of an alternative NER-independent repair pathway for ICLs and/or bulky DNA adducts.
Non-bulky endogenous oxidative DNA lesions generated by reactive oxygen species (ROS) are substrates for two overlapping pathways: base excision repair (BER) and nucleotide incision repair (NIR). The BER pathway is initiated by a DNA glycosylase cleaving the -glycosylic bond between the abnormal base and deoxyribose, leaving either an abasic site or a DNA single-strand break which in turn is repaired by an apurinic/apyrimidinic (AP) endonuclease (). Interestingly, it has been shown that mammalian polynucleotide kinase (PNK) can substitute AP endonuclease in removal of 3′-blocking phosphates produced by certain DNA glycosylases (,). In human cells, four DNA glycosylases/AP lyases OGG1, NTH1, NEIL1 and NEIL2, excise the majority of oxidized bases (). Alternatively, in the NIR pathway, the major human AP endonuclease 1, APE1, incises oxidatively damaged DNA in a DNA glycosylase-independent manner (). These two pathways can work in concert to cleanse genomic DNA from 7,8-dihydro-8-oxoguanine (8oxoG) residues, one of the major ROS-induced DNA lesions (). Previously, it was shown that antisense APE1-RNA human cell transfectants exhibited hypersensitivity to MMC, suggesting a potential involvement of APE1 in ICLs repair (). Furthermore, similar to FA cells, Ape1-depleted cells were sensitive to hyperoxia. APE1 is involved in both BER and NIR suggesting that these pathways may be involved in repair of ICLs.
Human homologues of the oxidized base-specific DNA glycosylase endonuclease VIII (Nei) were identified by database search of the genome and named NEIL1, 2 and 3 (human Nei-like proteins 1, 2 and 3) (,). The Nei-like proteins share significant sequence similitude with the Fpg protein, an 8oxoG-DNA glycosylase found in Fpg/Nei/NEIL are structurally and catalytically a distinct family of bi-functional DNA glycosylases endowed with an AP lyase activity that incises DNA at abasic sites by a β,δ-elimination mechanism and leaving single-strand break carrying a phosphate residue at the 3′ and 5′-terminus (,). mutants deficient in both DNA glycosylases, endonuclease III () and show spontaneous mutator phenotype and are hypersensitive to the lethal effects of oxidizing agents (,). Moreover, mouse embryonic stem cell lines deficient in NEIL1 protein were twice more sensitive than control cells to low doses of γ-irradiation (). Indeed, NEIL1 initiates base excision repair of adenine-derived formamidopyrimidines and 5S,6R stereoisomers of thymine glycol (Tg) which are not excised by OGG1 and NTH1 (,). Recently, it has been demonstrated that NEIL1 can excise an oxidized base located in the proximity to single-strand break suggesting its involvement in repair of clustered DNA lesions induced by ionizing radiation (,). Interestingly, S phase specific expression of NEIL1 and excision of base lesions in single-stranded and bubble-structured DNA substrates point to its possible involvement in replication-dependent repair (,).
The purpose of this study was to search for alternative repair pathway(s) involved in removal of ICLs and bulky psoralen–DNA photoadducts. Using a short cross-linked oligonucleotide duplex, we demonstrated that human DNA glycosylase and AP endonuclease initiate alternative repair pathway for bulky psoralen–DNA photoadducts which sought to be substrates for the NER machinery. The potential biological importance of these findings is discussed.
Chemical reagents including 8-MOP were purchased from Sigma-Aldrich. All oligonucleotides were purchased from Eurogentec (Seraing, Belgium), including regular oligonucleotides, siRNAs and those containing Uracil, 5,6-dihydrouracil (DHU), Tg and 3′-terminal phosphate (3′-P). A 21-mer oligonucleotide GFP1, d(GCTCTCGTCTGXACACCGAAG), where X is either T, U or Tg, and complementary ones GFP2, d(GCTCTTCGGTGTACAGACGAG) and GFP3, d(CTTCGGTGTACAGACGAGagc) were hybridized to obtain duplexes referred to as GFP1·2 and GFP1·3, respectively (C). These sequence contexts were previously used to study the repair of psoralen-induced ICLs in human cells (). The following oligonucleotides were used to measure 3′-phosphatase activity: 3′P-Exo10, d(TGACTGCATAp), and complementary 30-mer d(ATGCACATCGTCTACATGCGTATGCAGTCA). Classic AP site substrate was prepared by treating uracil containing oligonucleotide with Uracil-DNA glycosylase (Ung). Oligonucleotides were either 5′ or 3′-end labelled and annealed as previously described (). Oligonucleotides carrying 3′-P residue were 5′-end labelled using 3′-phosphatase-free T4 polynucleotide kinase from Roche Diagnostics GmbH (Meylan, France).
The sequences of siRNA used to decrease APE1 in cells were previously described (). Sequences of control siRNA were 5′ (ACUAUGUAUAGGAGUACGCTT)3′ and 3′ (TTUGAUACAUAUCCUCAUGCG)5′. Human FANCD2 and NEIL1 smart pool siRNAs were obtained from Dharmacon (Perbio Science, France).
strains AB 1157 ( Δ() ) (WT) and its isogenic derivatives BH20 (::kan) and BH110 (:: [Δ()90 ]) were from the laboratory stock. DNA glycosylase deficient strains were constructed by insertional mutagenesis as described (). Briefly, to construct double mutant, the spectinomycine resistance cassette (Spc) was inserted into gene and transferred to the BH20 chromosome by recombination. Cells acquiring the selectable marker were selected and the presence of Spc gene insertion was further confirmed by PCR of the genomic DNA. Single clone containing inactivated ::Spc and ::Kan genes (MS2000) was chosen for further work. Strains AB2480 (isogenic to AB1157 except ) and SW2-8 (isogenic to BW35 (KL16) except ::cam) were gifts from Dr A. K. McCullough (Oregon Health & Science University, Portland, OR) and Dr S. S. Wallace (University of Vermont, Burlington, VT), respectively.
T4 DNA ligase was purchased from Roche Diagnostics GmbH. The purified Ung, Nfo, Fpg, Nth, AlkA and human OGG1, ANPG and NTH1 proteins were from laboratory stock. Human Ape1 protein was expressed and purified from BH110 (DE3) strain to avoid cross-contamination of bacterial AP endonucleases as described (). The expression vectors phNEIL1 and phNEIL2 (), the purified Nei protein () and human DNA polymerase β () were generously provided by Dr Hiroshi Ide (Hiroshima University, Japan), Dr Dmitry Zharkov (ICBFM, Novosibirsk, Russia) and Dr Grigory Dianov (MRC, Harwell, Oxfordshire, UK), respectively. The full-length native NEIL1 and NEIL2 were purified as described previously ().
Five microgram of plasmid DNA pUC18 or 10 pmol of P-labelled GFP1·2 were incubated for 15 min in the dark with 0.1 mM 8-MOP in 50 µl of 100 mM Tris-HCl, pH 7.5, 5 mM EDTA and 50 mM NaCl, then irradiated at 365 nm and 240 kJ/m at room temperature. Plasmid DNA was purified by ethanol precipitation and the presence of ICLs was verified by denaturing agarose gel electrophoresis. GFP1·2 was desalted by spin-down columns filled with water equilibrated Sephadex G-50. Cross-linked and non-cross-linked oligonucleotides were separated by denaturing 20% PAGE. The oligonucleotides were eluted from the gel strips in 2 M LiClO and then acetone precipitated. To obtain MAp residue, the purified cross-linked GFP1·2 was treated with hot alkali ().
The standard reaction mixture (20 µl) for DNA glycosylase activity contained either 0.1 µg of pUC19 or 10 nM of 5′-[P] or 3′-[P]ddAMP-labelled GFP1·2, 25 mM HEPES-KOH, pH 7.6, 100 mM KCl, 1 mM EDTA, 5 mM 2-mercaptoethanol and 6% glycerol, unless otherwise stated. The assay mixture for the Fpg, Nei and OGG1 proteins contained 25 mM HEPES (pH 7.6), 100 mM KCl, 2 mM EDTA, 5 mM 2-mercaptoethanol and 6% glycerol. For Nfo and AlkA, the assay mixture was 20 mM HEPES-KOH, pH 7.6, 50 mM KCl, 0.1 mM EDTA, 0.1 mg/ml BSA, 1 mM DTT, for Nth and NTH1 the same but with 1 mM EDTA, for ANPG 70 mM HEPES-KOH, pH 7.6, 50 mM KCl, 1 mM EDTA, 0.1 mg/ml BSA and 5 mM 2-mercaptoethanol. APE1 assay conditions vary depending on the DNA repair pathway studied as described (). Assays were performed with 5–20 nM of pure protein at 37°C for 10 min, unless otherwise stated. Reaction with pUC18 was stopped by adding 5 µl of 0.25% bromophenol blue, 50% glycerol and 10 mM EDTA and the products were analysed by 0.8% agarose gel electrophoresis (0.5× TBE). Reactions with the oligonucleotides were stopped by adding 10 µl of 0.5% SDS and 5 mM EDTA and analysed as previously described (). Purified reaction products were heated at 65°C for 3 min and separated by electrophoresis in denaturing 20% (w/v) polyacrylamide gels (7 M Urea, 0.5× TBE). Gels were exposed to a Fuji FLA-3000 Phosphor Screen and analysed using Image Gauge V3.12 software.
For the reconstitution of the repair pathway of MAs , 20 nM of GFP1·3 with a single MAp was incubated in the presence of 20 nM NEIL1, 20 nM APE1, 10 nM DNA polymerase β and 2 U T4 DNA ligase in buffer (20 µl) containing 5 µCi of [α-P]dTTP, 20 mM HEPES-KOH, pH 7.6, 50 mM KCl, 0.1 mg/ml BSA, 1 mM DTT, 2 mM ATP and 5 mM MgCl for 10 min at 37°C and then 20 min at 30°C. Reaction products were analysed as described above.
The OGG1, NTH1 and NEIL1 proteins (200 nM) were incubated with 0.2 pmol of 8-MOP+UVA-treated 5-[P]-labelled GFP1•2 at 37°C for 30 min in 20 µl of the standard reaction mixture but in the presence of 50 mM NaBH (a 2 M NaBH stock solution in water was prepared immediately prior to use) or 100 mM NaCl as a control. The reaction was stopped by adding 5 µl of buffer containing 10% SDS, 30% glycerol, 25% 2-mercaptoethanol, 0.1% bromophenol blue and 300 mM Tris-HCl, pH 6.8 and heating 10 min at 60°C. The reaction products were separated by 15% SDS-PAGE and analysed as described above.
All strains of were grown with shaking at 37°C to 3 × 10 cells/ml in LB medium. Cells were centrifuged and resuspended in M9 medium at a density 5 × 10 cells/ml before irradiation. When appropriate, 8-MOP was added to a final concentration of 5 µM, and the suspension was allowed to stand on ice for 10 min in the dark before irradiation. Samples were irradiated at room temperature with HPW125 Philips lamp with a pyrex water filter. The fluence through the sample was 1 mW/cm. All experiments were carried out at least three times.
HeLa cells were routinely grown at 37°C in 5% CO in Dulbecco minimal essential medium supplemented with 10% foetal calf serum, 2 mM glutamine, 100 U/ml of penicillin and 100 mg/ml of streptomycin. To decrease the target protein level, HeLa cells were plated in 6-well plates at the concentration of 180 000 cells/well and 24 h later were transfected with 100 nM of appropriate siRNA using Oligofectamine (Invitrogen) as described ().
Cells were seeded into 96-well plate at the concentration of 10 cells/well in complete medium 72 h after siRNA transfection. ICL induction by photoactivated psoralen was achieved by incubating cells with 10 μM of 8-MOP for 20 min followed by exposure to 10 kJ/m UVA, afterwards cells were cultivated at 37°C in complete medium for increasing periods of time. Cell viability was assessed using Cell Proliferation Kit II (XTT) (Roche Diagnostics GmbH) according to the manufacturer instructions.
APE1−/− cells are sensitive to MMC, however, this cross-linking agent generates a variety of oxidative and non-oxidative DNA lesions and only 10% of them are ICLs (). Consequently, to examine the potential involvement of APE1 in ICLs processing, we used exposure to photoactivated 8-MOP that generates distinct, well-characterized and easily detectable DNA adducts. The sensitivity of HeLa cells with reduced expression of APE1 and FANCD2 towards several genotoxins was examined. In agreement with the previous observations, HeLa cells transfected with Ape1-specific siRNA were highly sensitive to ionizing radiation (IR) and MMC but not to UVC treatments (data not shown). Strikingly, the cells with decreased Ape1 level were also sensitive to 8-MOP+UVA exposure, like cells with reduced expression of FANCD2 protein, suggesting a potential involvement of Ape1 in the processing of psoralen–DNA adducts (A).
The sensitivity of APE1-depleted HeLa cells to cross-linking agents suggest that APE1-catalysed activities may remove psoralen-DNA adducts and thus perform function similar to that of the structure-specific endonuclease XPF-ERCC1 (). In fact, in the NIR pathway APE1 can incise the DNA sugar-phosphate backbone 5′ next to bulky lesions such as 3,-benzetheno-2′-°deoxynucleotides (), thus providing a back-up function to NER pathway. Furthermore, in the BER pathway, APE1 acts downstream of various DNA glycosylases to eliminate genotoxic repair intermediates. Thus depletion of APE1 removes the major AP and NIR endonuclease and 3′-repair diesterase activities in human cells, severely disabling two repair pathways. The first approach was to study whether the purified BER proteins cleave DNA cross-links and for this purpose 8-MOP+UVA treated supercoiled (sc) plasmid DNA was incubated with APE1 and human DNA glycosylases NTH1, NEIL1 and OGG1. Upon cleavage at the site of damage by a DNA repair enzyme, the sc form is converted to an open circular (oc) form and these two forms are identified by electrophoresis in agarose gel (). As shown in B, only NEIL1 converts the sc form to oc one, indicating that it specifically recognizes and incises photoactivated psoralen-induced DNA damage (lane 7).
To further examine whether NEIL1 excises MAs or ICLs or both, the short 21-mer 5′-[P] and 3′-[P]-dCMP-labelled oligonucleotide duplexes, GFP1·2, were treated with 8-MOP+UVA and the cross-linked oligonucleotides were separated from non-cross-linked ones by denaturing PAGE and used as DNA substrates. Interestingly, NEIL1 cleaves only non-cross-linked GFP1·2 (A, lanes 4, 8 and 12). The 5′ and 3′-labelled cleavage fragments migrating at position corresponding to an 11-mer (lanes 4 and 8) and 10-mer (lane 12), respectively, indicate excision of a psoralen–thymine MA at the position 12 of the 21-mer oligonucleotide. Consistent with the established mechanism of action of bi-functional DNA glycosylases/AP lyases, NEIL1 generates a Schiff base intermediate that can be characterized by the covalent trapping of the enzyme–substrate complex by reduction with NaBH (B, lane 4) (). To characterize the substrate specificity of NEIL1, an oligonucleotide containing single MAp was constructed by incubating cross-linked GFP1·2 under hot-alkali conditions (). As expected, base-catalysed reversal of the cross-link yields two closely migrating DNA fragments, the ‘faster migrating’ native and ‘slow migrating’ MAp-containing oligonucleotides (A, lane 5). Appearance of an 11-mer fragment (lane 6) upon incubation with NEIL1 was accompanied by the loss of MAp-containing fragment indicating that NEIL1 excises MAp residues present in duplex DNA. Furthermore, NEIL1 cleaves GFP1·2 irradiated at 405 nm in the presence of 8-MOP suggesting that it also excises MAfs (data not shown).
To further characterize the substrate specificities of the NEIL1 protein, the kinetic constants for the excision of MAp and DHU·G were measured. Comparison of the kinetic constants () shows that MAp is the preferred substrate for NEIL1 ( = 6.3′nM, / = 3.9 min µM) which is recognized with high specificity.
To examine the nature of 3′-termini generated by NEIL1-catalysed excision of MAp, 5′-[P] labelled GFP1·2 duplexes containing an AP site, MAp or Tg residues were used. As shown in B, when excising a Tg, AP site and MAp NEIL1 generates 11-mer DNA fragments migrating the same distance a result that is consistent with β,δ-elimination mechanism of action of this enzyme (lanes 3, 5 and 9). In agreement with this observation, DNA cleavage fragments generated by the NEIL1 and Nei proteins (lanes 8–9) migrate faster than products generated by NTH1 and APE1 containing 3′-α,β-unsaturated aldehyde and 3′-hydroxyl residues, respectively (lanes 2 and 6). Taken together, these results demonstrate that with all DNA substrate tested NEIL1 and Nei generate products carrying a phosphate at the 3′-terminus.
The catalytic rate of human APE1 acting on 3′-phosphoglycolates and 3′-Ps is nearly 200-fold lower than its AP endonuclease function (). Recently, Wiederhold and colleagues () showed that polynucleotide kinase and not APE1 is primarily involved in the NEIL1-initiated BER pathway. To examine whether 3′-Ps are removed by APE1 under reaction conditions used, we incubated 5′-[P] labelled 10-mer DNA fragment carrying 3′-P residue with the Nfo and varying amounts of APE1 proteins at 30 and 37°C. As shown in , at 30°C 5 nM APE1 removes more than 50% of 3′-P (lanes 3 and 7) whereas 20 nM APE1 and 5 nM Nfo eliminate all 3′-P residues generating 10-mer fragment containing 3′-hydroxyl with lower mobility as compared to the substrate DNA (lanes 2, 4, 6 and 8). Addition of 1 mM ATP did not affect activity of enzymes. 3′-Phosphatase activity of APE1 but not that of Nfo is strongly inhibited at 37°C (lanes 10–12 and 14–16). Interestingly, APE1 at higher protein concentration exhibits significant 3′→5′ exonuclease activity generating 9-mer DNA fragment (lanes 4 and 8). These results show that at physiologically relevant protein concentration, APE1 exhibits efficient 3′-phosphatase activity.
As we showed above, NEIL1-catalysed excision of MA generates a single-strand break with 3′-P terminus which has to be eliminated. Hence, we have reconstituted the BER pathway for MAps using purified proteins (). Incubation of a 21-mer GFP1·3 containing a single MAp in the presence of NEIL1, APE1, DNA polymerase β and [α-P]-TTP generated a labelled 12-mer DNA fragment (lane 4). The results indicate that APE1 removes 3′-P residues generated by NEIL1 and allows DNA polymerase β to insert one nucleotide. Addition of DNA ligase completes the restoration of the full-length 21-mer GFP1·3 (lane 5). These data demonstrate that 8-MOP-induced MAs are processed by the short-patch BER pathway. In agreement with biochemical data, the siRNA-transfected cells with decreased NEIL1 level were hypersensitive to 8-MOP+UVA exposure (). It should be stressed that we did not observe significant additive increase in the sensitivity when cells were depleted for both APE1 and NEIL1. This result indicates that both proteins participate to a same pathway to cope with psoralen-induced DNA damage. These and data imply new functions of NEIL1 and APE1 in the removal of DNA adducts generated by cross-linking agents.
To further investigate the excision of psoralen MA in DNA, we examined whether this lesion was a substrate for previously characterized BER enzymes from and human. We challenged a 5′-[P]-labelled oligonucleotide duplex containing single MAp with a variety of highly purified AP endonucleases and DNA glycosylases. Since not all DNA glycosylases possess AP-nicking activity, the assays were made in the presence of AP endonuclease in order to cleave DNA duplex at the potential abasic sites generated by the base excision. When the various and human enzymes were tested on GFP1·2 (A), only incubation with Fpg, Nei and NEIL1 led to the cleavage of the labelled oligonucleotide at the position of the psoralen–thymine MAp. Interestingly, the excision of MAp by the Nei-like enzymes was more efficient than that of Fpg (lanes 5 and 7 versus 4). Despite being used in excess amount, Nfo, AlkA, Nth, OGG1, NEIL2, APE1, ANPG and NTH1 proteins did not act on GFP1·2 (lanes 2–3, 6, 8–11 and data not shown).
Genetic and biochemical studies of have established that both the RecA-dependent recombination and the NER pathways participate in removal of psoralen-induced cross-links (,). Subsequently, it was demonstrated that psoralen MAs are removed by UvrABC nuclease with high efficiency (). To address the physiological relevance of the Nei and Fpg-catalysed excision of psoralen MAs, we assessed the sensitivity of the mutant strains to 8-MOP+UVA exposure. As shown in B, control strain was extremely sensitive to the photoactivated psoralen, whereas and single and double mutants were not particularly sensitive to treatment. These results suggest that in bacteria, the NER system is a major pathway for removal of psoralen MAs present in DNA.
Repair of ICLs and other complex DNA lesions in mammalian cells is linked to replication and required successive involvement of several distinct repair pathways in coordinated manner. At present, the underlying mechanism of this coordination is poorly understood. The results obtained in this study reveal that psoralen–thymine MAs present in duplex DNA are substrate for human oxidative DNA glycosylase, NEIL1. Analysis of the kinetic data demonstrates that NEIL1 can efficiently backup the NER pathway to repair psoralen-induced MAs. It has been shown that NEIL1 catalyses the β,δ-elimination at AP site after oxidized base excision, leaving a 3′-P at the resulting single-strand break (). Here, we demonstrated using size-markers generated by Nei, NTH1 and APE1 that NEIL1 in fact generates 3′-P termini while removing the psoralen adduct. Based on the fact that PNK has a much higher 3′-phosphatase activity than APE1, it has been proposed that in mammalian cells 3′-P generated by NEIL1 and NEIL2 is removed by PNK in the AP endonuclease-independent BER pathway (). Although, human major AP endonuclease is a weak 3′-phosphatase, we demonstrated that at 20 nM protein concentration APE1 efficiently repairs oligonucleotide fragment containing 3′-P residue. In agreement with this result, we have fully reconstituted the repair of MAs using purified APE1, NEIL1, DNA polymerase β and DNA ligase. Following NEIL1 excision APE1 removes 3′-P residue and allows subsequent gap filling and ligation. Consistent with the biochemical data, HeLa cells lacking APE1 and/or NEIL1 become hypersensitive to 8-MOP+UVA exposure. Thus, we added one arm of the BER pathway to the cellular defence used to cope with the genotoxic effects of cross-linking agents. These results demonstrate that NEIL1 is not simply a backup DNA glycosylase for OGG1, NTH1 and NEIL2 but has distinct substrate specificity towards a specific class of bulky DNA adducts.
NEIL1, a 44 kDa protein, was initially characterized as a DNA glycosylase specific for oxidized, saturated and ring-fragmented bases. Several lines of evidence argue for a biological role of NEIL1 in counteracting oxidative DNA damage. Oxidative stress could transiently increase NEIL1 level in human colon carcinoma cells (). Analysis of human polymorphic and mutant variants of NEIL1 revealed that their low DNA glycosylase activities and reduced expression may be involved in the pathogenesis in a subset of gastric cancers and increased risk of metabolic syndrome (,). Cells deficient in the Werner syndrome protein (WRN), a member of the RecQ family of DNA helicases, are hypersensitive to psoralen induced ICLs (,). Lately, it has been shown that WRN interacts with and stimulates NEIL1 in excision of oxidative lesions from bubble DNA substrates suggesting that NEIL1–WRN complex participates in the same repair pathway (). Recently, Vartanian and colleagues () generated knockout (KO) mice with a frameshift deletion and insertion in the gene. The steady-state mitochondrial DNA damage and deletions from liver tissues of −/− KO mice were significantly increased compared to wild-type controls. The mice develop spontaneously severe obesity, dyslipidemia and fatty liver disease, suggesting that accumulation of spontaneous DNA damage in cells lacking NEIL1 may lead to the diseases associated with the metabolic syndrome. Interestingly, dietary consumption or handling of psoralen-containing plants have been shown to cause adverse effects to human health (). In light of a new function of NEIL1, it is tempting to speculate that −/− KO mice may be more sensitive to psoralen-rich diet.
Since human DNA glycosylase can excise bulky MAp, we investigated whether this lesion was a substrate for previously characterized Nfo, AlkA, Nth, Fpg and Nei and human APE1, OGG1, NTH1, NEIL2 and ANPG proteins. The results show that in addition to NEIL1, the bacterial Fpg and Nei proteins can also excise psoralen MAp when present in duplex oligonucleotides. In , psoralen MAs are good substrates for UvrABC nuclease () however both NER and homologous recombination systems are required to remove the psoralen-induced ICLs (). Evolutionary conservation of the new substrate specificity in bacterial Nei-like DNA glycosylases suggests its possible biological role. Therefore, we examined whether in Fpg/Nei-catalysed excision of MAp provides an alternative pathway to classic NER. Surprisingly, DNA glycosylase-deficient mutants, in contrast to mutant, were not sensitive to 8-MOP+UVA exposure indicating that in NER is a major pathway to remove psoralen MAs.
Three-dimensional crystal structures of the Nei–DNA complex and free NEIL1 protein show the same overall fold for both DNA glycosylases (,). Both proteins consist of two domains connected by a linker, this module structure giving rise to a DNA-binding cleft between domains. Based on molecular modelling it was suggested that Nei binds to DNA and flips out damaged base in a shallow hydrophobic pocket (). At present, structural information for Nei and NEIL1 complexed with DNA duplex containing an oxidized base is not yet available. A bulky psoralen MA presents topological constraint for its accommodation in the active sites of Nei-like DNA glycosylases. A 3D crystal structure of the complex of T4 endonuclease V bound to a DNA substrate containing a bulky cyclobutane thymine dimer showed that the enzyme kinks the DNA helix by about 60° and flips out the opposing adenine base complementary to thymine dimer out of the DNA base stack (). Based on these observations we may hypothesize that the mechanism of lesion recognition by Nei-like enzymes is different from that of well-studied DNA glycosylases. To access the C1′ atom, most of DNA glycosylases bent the DNA and flip out an oxidized base into a specific pocket (,), while Nei and NEIL1 similar to T4 endonuclease V may access it by a drastic kink of the helix and flipping out of the complementary base opposite to psoralen MA. Insight into the structural basis for clustered and bulky DNA damage recognition by Nei-family DNA glycosylases will have to await further investigations.
In conclusion, although NEIL1 does not incise DNA containing single ICL, we hypothesized that it may participate in psoralen-induced ICLs removal after XPF-ERCC1-mediated unhooking. Alternatively, but not exclusively, NEIL1/APE1 action could significantly reduce ICLs formation by eliminating MAf residues. |
Hereditary deficiencies of early components of the classical pathway of the complement system are known to predispose to systematic lupus erythematosus (SLE). Among these, C1q deficiency exhibits the strongest association with prevalence greater than 90% suggesting that a physiological activity of the early part of the classical pathway normally protects against the development of SLE (). Mice with targeted deletion of the C1q gene () developed a spontaneous lupus-like disease characterised by the development of anti-nuclear autoimmunity and glomerulonephritis associated with the presence of multiple apoptotic bodies (). Introgression of C1q deficiency onto different genetic backgrounds revealed that in mice C1q operates as a disease modifier. C57BL/6. mice displayed no increase of IgG autoantibodies (autoAbs) or glomerulonephritis (), whilst C1q deficiency backcrossed onto the lupus-prone MRL/Mp background accelerated both the onset and the severity of the autoimmune disease (). Consistent with these observations, C1q reconstitution by bone marrow transplant attenuated the autoimmune disease present in MRL/Mp. mice ().
Currently, there are two main hypotheses to explain the role of complement in the development of SLE, neither of which is mutually exclusive. The first model, defined as the tolerance hypothesis, proposes a role for complement in determining the activation thresholds of lymphocytes, whereby complement enhances presentation of autoantigens to self-reactive immature B cells resulting in their elimination (). The second one, known as the ‘waste disposal’ hypothesis, suggests that in addition to its role in the clearance of immune complexes, complement is involved in the physiological disposal of apoptotic cells (), that have been shown to express lupus autoantigens on their surface (). C1q plays a significant role in the clearance pathway of cellular debris by binding directly or indirectly to apoptotic blebs where it activates complement and mediates phagocytosis by professional and non-professional phagocytes (). Therefore, improper removal of dying cells in the setting of C1q deficiency could result in the stimulation of autoreactive cells leading to autoimmunity.
Immunoglobulin Tg models have been instrumental in understanding B cell regulation revealing several key mechanisms, including receptor editing, deletion, anergy and ignorance. The hen egg lysozyme (HEL)–anti-HEL (Ig) Tg model in particular has been widely used to demonstrate elimination of self-reactive clones by membrane-bound expressed antigen, anergy induction by soluble antigen (sHEL) or ignorance when the amount of antigen is so low that it does not reach the threshold to induce anergy. More recently, it has been shown that intracellular membrane-bound HEL failed to induce tolerance and was instead autoimmunogenic positively selecting Ig B1 cells and inducing large numbers of IgM autoantibody-secreting plasma cells (reviewed in ). Taken together, these findings suggest that in this model the fate of self-reactive B cells is determined not only by the abundance, the avidity of the target self-antigen and the affinity of the B cell receptor but also by the location of the “auto”-antigen. The role of complement has been tested in the Ig Tg (MD4)–sHEL (ML5) model by crossing the double Tg mice with mice deficient in C1q, C4, C3 or CD21/CD35 (). IgG anti-HEL Abs remained undetectable in all complement deficient mice, but C4 and CD21/CD35 deficient B cells displayed reduced surface IgM modulation, indicating a lower degree of anergy induction in these mice (). However, this was not observed in the C1q deficient mice (). This discrepancy could either indicate that C4 operates independently from C1q or reflect differences in the genetic background of the mice used (). Nevertheless sHEL is not the ideal model to study the maintenance of tolerance in SLE as it is neither a natural autoantigen, nor are soluble plasma proteins typically targeted by the SLE autoAbs.
More recently, a model targeting an SLE antigen, DNA, was generated by Weigert and co-workers (). In this model the rearranged variable heavy chain (V) gene derived from a double stranded DNA-binding hybridoma developed in the autoimmune strain MRL/Mp./, was inserted at the Igh locus and was referred to as V3H9R. In contrast to conventional Tg mice, this knock-in model allows the Tg locus to undergo normal editing, isotype switching and somatic mutation. A variety of light chains can combine with the V3H9 to yield anti-DNA Abs () but only few light chains are able to “silence” V3H9R so that it no longer binds to DNA. By virtue of this characteristic the mice expressing only the V3H9 chain (V3H9R mice) can generate anti-DNA specificities. However, when the V3H9R mice were crossed with a Tg knock-in light chain Vκ8 () to generate monospecific Tg mice (V3H9R/Vκ8R mice) (), this combination of heavy and light chain V regions (V3H9/VVκ8) bound only ssDNA and not dsDNA (). Previous studies with the V3H9R mice have shown that the autoreactive Tg B cells accumulated in the splenic marginal zone and were regulated by anergy on non-autoimmune backgrounds such as BALB/c () and C57BL/6 (). However, tolerance could be broken in this model if T cell help was provided in the form of a chronic graft versus host disease (). Consistent with these observations, the double V3H9R/Vκ8R knock-in Tg B cells were regulated by anergy in non-lupus prone mice (BALB/c), whilst in autoimmune prone MRL/Mp./ animals Tg B cells escaped tolerance induction and underwent class-switching and affinity maturation (). These experiments suggested that the mutation in the MRL background allowed the Tg autoreactive B cells to receive T cell help during a germinal center reaction.
To determine whether C1q is involved in selection of self-reactive B cells, we bred the C1q-deficient mice with the V3H9R and the V3H9R/Vκ8R mice and monitored the regulation and activation of anti-DNA Tg B cells over a period of 10 months. The mice in this study were on the autoimmune prone background MRL/Mp expressing the CD95 () gene. The analysis of these mice revealed that the lack of C1q can influence the levels of IgM and IgG Tg-derived antibodies only in the V3H9R model.
MRL/Mp mice were obtained from Harlan Olac, Bichester, UK. MRL/Mp. deficient mice were generated as previously described (). V3H9R.MRL/Mp (), Vκ8R.MRL/Mp () and V3H9R/Vκ8R.MRL/Mp mice were kindly provided by Prof. M. Weigert (Gwen Knapp Center for Lupus and Immunology Research, University of Chicago, Chicago, IL). MRL/Mp. mice were crossed with the V3H9R/Vκ8R.MRL/Mp mice and the resulting V3H9R/Vκ8R.MRL/Mp. were then crossed either with MRL/Mp or MRL/Mp. in order to generate littermate animals. Mice were bled every 3 months starting from 2 months of age and at 10 months they were sacrificed. The mice were genotyped by PCR using specific primers. PCR primers were as follow: mC1qA/5′ (5′-GGGGCCTGTGATCCAGACAGG-3′), mC1qA/In2 (5′-TAACCATTGCCTCCAGGATGG-3′) and neo (5′-GGGGATCGGCAATAAAAAGAC-3′) for the C1q genotyping; MW114 (5′-CTGTCAGGAACTGCAGGTAAGG-3′) and MW162 (5′-CATAACATAGGAATATTTACTCCTCGC-3′) for the V3H9R genotyping (); MW133 (5′-GGTACCTGTGGGGACATTGTG-3′) and MW157 (5′-AGCACCGAACGTGAGAGG-3′) for the Vκ8R genotyping (). Animals were kept under specific pathogen-free conditions. All animal care and procedures were conducted according to institutional guidelines and approved by the local ethical committee.
Flow cytometry was performed using a four-color staining of cells and analyzed with a FACSCalibur™ (Becton Dickinson, Mountain View, CA). The following Abs were used: anti-B220 (RA3-6B2), anti-CD5 (53-7.3), anti-CD11b (M1-70), anti-CD19 (1D3), anti-CD23 (B3B4), anti-CD21/CD35 (7G6), anti-CD90.2 (53-2.1), anti-CD138, anti-IgM (II/41). All Abs were purchased from BD Biosciences Pharmingen (San Diego, CA) with the exception of the anti-V3H9 idiotype (1.209), a kind gift from Prof. M. Weigert (). Biotinylated Abs were detected using an allophycocyanin-conjugated streptavidin Ab (BD Biosciences Pharmingen). Staining was performed in the presence of saturating concentration of 24G2 mAb (anti-FcRII/III). Data were analyzed using WinMDI software (Version 2.8; Scripps Institute).
Serum levels of IgM and IgG were quantified by ELISA, as described previously (). Anti-ssDNA Abs, anti-chromatin Abs, anti-histone Abs and anti-dsDNA Abs were measured by ELISA as described previously (). Microtiter plates were coated with ssDNA prepared from calf thymus DNA (Sigma), chromatin (Lorne Laboratories Ltd., Reading, UK) or histone (Calbiochem, Merck Biosciences, Darmstadt, Germany). For detecting anti-dsDNA Abs, plates were coated with streptavidin (Sigma). ΦX174 double-stranded plasmid DNA (Promega, Southampton, UK) was biotinylated with Photoprobe biotin (Vector Laboratories, Peterborough, UK) and added to the streptavidin. Serum samples were diluted appropriately in PBS 2%BSA, 0.05%Tween-20, 0.02% NaN. Bound Abs were detected with alkaline phosphatase conjugated goat anti-mouse IgG (γ-chain specific) (Sigma–Aldrich, Dorset, UK) and anti-mouse IgM (Southern Biotechnology Associates, Inc., Birmingham, AL). All autoAb results are expressed in arbitrary ELISA units (AEU) in reference to a standard curve derived from serum pools containing high titers of autoAbs. The idiotype Abs were captured with the anti-V3H9 idiotype (1.209) () and detected with alkaline phosphatase conjugated goat anti-mouse IgM or IgG subclasses specific Abs (Southern Biotechnology Associates). The standard curve was derived from an IgM V3H9R/Vκ8R mAb provided by Prof. Weigert or sera containing high IgG subclasses titers of the Tg.
Proteinuria was assessed using Haema-combistix (Bayer Diagnostics, Newbury, UK). Kidneys were fixed in Bouin's solution and paraffin embedded, and sections were stained with periodic acid–Schiff reagent. Glomerular histology was graded in a blind fashion scored on a scale of 0–4, as described before ().
The data are presented as median, with range of values in brackets, unless otherwise stated. The non-parametric Mann–Whitney -test was applied throughout with differences being considered significant for -values < 0.05. Statistics were calculated using GraphPad Prism Version 3.0 (GraphPad Software, San Diego, CA).
C1q deficiency has been shown to accelerate the onset and progression of SLE in MRL/Mp mice (), but the mechanisms underlying these effects are still uncertain. The V3H9R/Vκ8R Tg alleles were transferred onto C1q-deficient mice by crossing V3H9R/Vκ8R.MRL/Mp with MRL/Mp. and experimental cohorts of littermate mice were generated. Whilst in previous studies the expression of the Tg alleles was monitored by measuring the Tg-specific immunoglobulin allotype (), this analysis could not be applied in these mice as the Tg alleles and the endogenous immunoglobulin allotype were indistinguishable. Nevertheless, we were able to detect the combination of the V3H9 with the VVκ8 using an idiotype specific monoclonal Ab named 1.209 (). As the 1.209 mAb can also recognize the V3H9 paired with other endogenous light chains such as Vκ4 (), this anti-idiotype Ab allowed us not only to assess the V3H9R/Vκ8R Tg B cells but also some of the V3H9R Tg B cells. C1q deficiency did not affect the proportion of peripheral blood B cells expressing the Tg in neither of the two models (V3H9R/Vκ8R.MRL/Mp.: 89.1 ± 1.1% versus V3H9R/Vκ8R.MRL/Mp: 90.7 ± 1.5%, = 0.4118; V3H9R.MRL/Mp.: 25.3 ± 1.6% versus V3H9R.MRL/Mp: 22.3 ± 2.3% = 0.1669). The low proportion of idiotype B cells in the V3H9R.MRL/Mp mice indicated that most of the B cells in these mice had either paired the V3H9R Tg allele with a light chain that was not recognised by the anti-idiotype Ab, or edited the V3H9R allele, or used the endogenous immunoglobulin allele. In the V3H9R/Vκ8R.MRL/Mp. mice the B cell receptor was down-modulated to the same extent as in the V3H9R/Vκ8R.MRL/Mp mice (data not shown).
Serological analyses were performed in order to establish whether the Tg autoAbs were expressed. The different cohorts of mice were bled at different time points, and the data presented here are those obtained at 10 month of age when the mice were sacrificed. Substantial levels of idiotype IgM, IgG3, IgG2a and IgG2b Abs were detected in the V3H9R/Vκ8R.MRL/Mp and in V3H9R.MRL/Mp mice indicating that the anti-DNA knock-in Tg alleles on this background were not regulated by anergy unlike when on the C57BL/6 background (). In the V3H9R/Vκ8R.MRL/Mp. mice the levels of idiotype Abs were similar to those detected in the strain-matched C1q-sufficient animals. In contrast, the V3H9R.MRL/Mp. mice had significantly increased levels of idiotype IgM and IgG3 Abs compared to the V3H9R.MRL/Mp mice (A and B), but similar levels of idiotype IgG2a and IgG2b Abs (C and D).
We then measured the levels of IgM and IgG autoAbs. A significant increase in the titre of IgM anti-ssDNA Abs was observed in the V3H9R/Vκ8R.MRL/Mp. mice compared to the V3H9R/Vκ8R.MRL/Mp mice (A). Of note, the V3H9R/Vκ8R.MRL/Mp. mice displayed only markedly increased levels of IgM autoAbs directed against ssDNA (the specificity encoded by the Tg alleles) but not against other lupus autoantigens such as dsDNA, histone or chromatin (B–D). In the V3H9R.MRL/Mp mice the pairing of the V3H9 with endogenous L chains could yield to more specificities. Consistent with this, V3H9R.MRL/Mp. mice had significantly increased levels of IgM anti-ssDNA, anti-dsDNA, anti-histone and anti-chromatin Abs (A–D). We then analysed the levels of IgG autoAbs and found that these were similar between the C1q-sufficient and -deficient cohorts ().
At 10 month of age all the animals were sacrificed. V3H9R/Vκ8R.MRL/Mp and V3H9R.MRL/Mp animals showed histological evidence of a mild glomerulonephritis (median score 1.0, range: 0.0–3.0 and 1.5, range: 1.0–3.0, respectively) reminiscent of the disease observed in MRL/Mp non-Tg mice (). C1q-deficiency in MRL/Mp mice has previously been shown to worsen the kidney pathology (). However, in the Tg mice the absence of C1q did not exacerbate the renal disease compared to the strain-matched C1q-sufficient mice (V3H9R/Vκ8R.MRL/Mp. mice: median score 1.0, range: 0.0–3.0, = 0.4287; V3H9R.MRL/Mp. mice: median range 2.0, range: 1.0–2.0, = 0.4564).
In order to determine if the serological data were accompanied by phenotypic changes in T and B lymphocytes, we performed a comprehensive analysis of the various splenic and peritoneal subpopulations. Cells of at least 7 mice from each cohort were analysed by FACS at the time of the sacrifice using the combinations of markers shown in . In agreement with the findings in other genetic backgrounds (), the V3H9R.MRL/Mp mice had a larger marginal zone (MZ) B cell population compared to V3H9R/Vκ8R.MRL/Mp (A) and an increased proportion of idiotype B cells in the MZ compartment compared to follicular (FO) compartment (5.9 ± 1.0 versus 4.2 ± 0.6, paired -test = 0.0157) (). More interestingly, in the V3H9R.MRL/Mp. mice, which produced higher levels of IgM autoAbs, the MZ B cell population was significantly decreased compared to the V3H9R.MRL/Mp animals (A) and less idiotype B cells were found in the MZ compared to the FO compartment (3.4 ± 0.7 versus 4.8 ± 1.1, paired -test = 0.0042) (). In the V3H9R/Vκ8R.MRL/Mp mice there was no preferential localization of idiotype B cells in the MZ zone (paired -test = 0.056) (). C1q deficiency was associated with a significant reduction in the percentage of MZ B cells (A) and a decrease of idiotype B cells in the MZ compared to the FO area (4.2 ± 0.9 versus 17.5 ± 3.2, paired -test = 0.0013). These findings indicated that MZ B cells might have been activated in the absence of C1q. Activated MZ B cells have been shown to migrate into the T cell zone and differentiate into plasma cells. Indeed, the percentage of plasma cells was increased in the two C1q-deficient cohorts compared to the respective MRL/Mp controls (B).
The peritoneal cavity in mice has been shown to be a site where some self-reactive B cells can escape tolerance () and B-1 cells, one of the major sources of circulating IgM, accumulate. More cells were recovered from the peritoneum of V3H9R.MRL/Mp.-/-mice compared to V3H9R.MRL.Mp mice ( = 0.0051) but not in the V3H9R/Vκ8R.MRL/Mp. versus V3H9R/Vκ8R.MRL/Mp animals ( = 0.1923) (). When the peritoneal B cell subpopulations were analyzed in more detail, all subsets were increased in the V3H9R.MRL/Mp.-/mice compared to the V3H9R.MRL.Mp mice (). On the contrary, in the V3H9R/Vκ8R C1q-deficient animals only B-1a cells were slightly increased compared to their wild type strain-matched controls (), but this did not reach statistical significance ( = 0.0518). We then analysed the V3H9/Vκ8 idiotype expression on the different peritoneal B cells. Surprisingly in the peritoneum of the V3H9R/Vκ8R.MRL.Mp and V3H9R.MRL.Mp mice the majority of the B cells were not expressing the V3H9/Vκ8 idiotype (). Of note the C1q-deficient mice had similar percentage of idiotype B cells in the different peritoneal B cell subpopulations compared to C1q-sufficient mice. The reduced proportion of idiotype B cells in the peritoneum compared to the peripheral blood and the spleen (data not shown) suggests that in this organ the B cells either preferentially used the endogenous alleles or had been activated and undergone editing.
C1q deficiency in humans and in mice has been associated with the development of a lupus-like illness. However, the role of C1q in the regulation of autoreactive B cells remains debatable. Here, we explored the regulation of autoreactive B cells by C1q using anti-DNA knock-in Tg models (V3H9R/Vκ8R and V3H9R) on the lupus prone MRL/Mp genetic background. The analysis of these mice revealed that the MRL/Mp background was in itself sufficient to allow the expression of anti-DNA Tg autoAbs and that the lack of C1q modified this effect only in the single Tg model (V3H9R). However, in the absence of C1q the MZ B cell compartment was significantly reduced in both models and this was accompanied by an increase in plasmocytes.
Recently the V3H9R/Vκ8R anti-DNA knock-in Tg model has been widely used for investigating the mechanisms of regulation of autoAb production in murine models of SLE. There is an increasing evidence that the V3H9R/Vκ8R and V3H9R anti-DNA Tg B cells, which have a relatively weak affinity to ssDNA, are regulated by anergy on normal genetic backgrounds such as BALB/c and C57BL/6, but could be induced to lose tolerance when transferred onto lupus models such as (NZB × NZW)F1 or MRL/Mp./ (). In light of these observations it was important for our study to establish whether the autoimmune prone MRL/Mp background was in itself capable of breaking tolerance. MRL/Mp mice are known to develop a mild autoimmune disease which can be accelerated with different disease-modifying genes such as (), and (). The analysis of the idiotype (V3H9R/Vκ8R) Abs revealed that the Tg MRL/Mp mice indeed had in circulation these idiotype Abs (IgM, IgG2a, IgG2b, IgG3) indicating that a break of tolerance had spontaneously occurred in these mice. One explanation for this is an intrinsic defect in MRL/Mp B cells and there is some evidence in support of this. MRL/Mp mice have been reported to exhibit a defect in maintaining the developmental arrest of V3H9/Vλ anti-dsDNA conventional Tg B cells () and to have a spontaneous B cell hyperactivity in the absence of Ag in the Ig experimental model (). However, the V3H9/Vλ anti-dsDNA autoreactive B cells, despite not being any longer developmentally arrested as in a BALB/c mice, exhibited follicular exclusion and failed to differentiate into plasma cells (). Furthermore Tg MRL/Mp mice expressing Ig have been shown to be able to down-regulate their B cell receptor and to be unable to secrete detectable levels of anti-HEL Abs in the presence of sufficient amount of sHEL (). Similarly anti-laminin Tg MRL/Mp mice were found to be tolerant (). Another potential explanation for the break of the B cell tolerance in the V3H9R/Vκ8R anti-DNA Tg mice, is that in this model the MRL/Mp background was able to provide sufficient T cell help and the presence of idiotype IgG subclasses favour this hypothesis.
We next examined whether C1q could modulate the phenotype of the anti-ssDNA Tg B cells. In the V3H9R/Vκ8R.MRL/Mp mice the absence of C1q increased significantly the circulating levels of IgM against ssDNA but not against other autoantigens including dsDNA. As the pairing of V3H9 with Vκ8 prevents the binding of V3H9 to dsDNA, the elevated amount of IgM anti-ssDNA observed could have been the result of an increased Tg expression. However, the idiotype analysis failed to demonstrate a difference in the levels of IgM idiotype Abs between V3H9R/Vκ8R.MRL/Mp. and V3H9R/Vκ8R.MRL/Mp mice, questioning whether the source of the increased levels of IgM anti-ssDNA was indeed the Tg B cells. On the other hand, the V3H9R.MRL/Mp. mice displayed significantly higher levels of IgM and IgG3 idiotype Abs. As the V3H9R heavy chain can pair with different endogenous light chains generating a wider range of autoAbs, other specificities were tested. Indeed the absence of C1q increased significantly IgM levels against all the lupus autoantigens analysed.
To gain insight into the mechanisms regulating the autoAb production in the C1q-deficient mice we then carried a detailed analysis of the different B cell populations. Several studies have proposed a possible role for MZ B cells in the development of lupus in mouse models. Although studies in immunoglobulin Tg mice have shown that autoreactive B cells can accumulate in the marginal zone under various experimental situations (), it remains to be established whether MZ B cells secrete pathogenic autoAbs in any model of lupus. B cells producing potentially pathogenic autoAbs are thought to home to the MZ () and sequestration to this site is believed to prevent them from entering into the germinal centres and developing the properties of pathogenic B cells. However, recent studies in lupus-prone mice have reported both enlargements () and impaired development of the MZ B cell compartment (). Consistent with previous observations in the V3H9R/Vκ8R model (), the MZ was found to be enlarged in the V3H9R.MRL/Mp mice but no accumulation of MZ B cells was observed in V3H9R/Vκ8R.MRL/Mp mice. Importantly C1q deficiency decreased dramatically the proportion of MZ B cells in these mice and this was accompanied by an increase in the percentage of plasma cells. More interestingly, in the C1q-deficient Tg mice we observed a significant disappearance of idiotype B cells from the MZ suggesting that these cells had been activated. Marginal zone B cells bearing low affinity self-reactive BCR can react to repetitive Ag and produce natural autoAbs of the IgM isotype upon contact with blood-borne pathogens or self Ag (). Moreover, MZ B cells can undergo T cell independent switching to IgG, IgA and IgE in response to pathogen associated molecular patterns (PAMP). As C1q has been shown to be involved in the clearance of apoptotic cells, which are enriched in the typical lupus autoantigens, one could postulate that the increased IgM Tg secretion and the reduction of the MZ B cells in the V3H9R.MRL/Mp. mice might be related to a failure to clear antigens associated with dying cells. In the absence of C1q the ineffectively cleared autoAgs could stimulate the autoreactive MZ Tg B cells to differentiate into plasmocyte resulting into the decrease of MZ B cells and increase of circulating IgM autoAbs. Similarly, in the spleens of V3H9R/Vκ8R.MRL/Mp. mice significantly less idiotype B cells were present in the MZ compartment and more plasmocytes were found. However, these cellular changes were not paralleled by an increase of idiotype IgM Abs in circulation indicating in this model they were not sufficient to induce autoantibody production.
Another potential source of IgM autoAbs are the B1 cells. B1 cells have long been associated with the secretion of natural Abs against self and foreign pathogens, which can occur without obvious inflammatory response. The importance of B1 cells in the pathogenesis of lupus continues to be debated. A major argument for a role of B1a cells in mouse lupus relates to the expansion of this subset in NZB and (NZB × NZW)F1 mice. Reduction of B1 cells for instance via intraperitoneal injection of HO delayed disease onset and reduced disease severity in (NZB × NZW)F1 mice (). However, the expansion of B1a cells was shown not to be critical for the production of IgM or IgG autoAbs in this murine model of SLE (). The Abs secreted by B1 cells tend to be polyreactive with low-affinity cross-reactivity to a variety of self Ags and these characteristics are very different from the pathogenic IgG Abs produced by lupus mice. In our experimental model, we found an increase in the total number of peritoneal and B1 cells in the V3H9R.MRL/Mp. mice compared to their wild type counterparts (). A similar trend was observed for the B1a cells in the V3H9R/Vκ8R.MRL/Mp. mice but this did not reach statistical significance. However, in both models the percentages of idiotype B1 cells between C1q-deficient and -sufficient mice were similar. These findings would indicate the lack of C1q may favour the expansion of peritoneal B1 cells and that these cells might have contributed to the increase of serum IgM autoAb levels observed in the V3H9R.MRL/Mp C1q-deficient mice. Supporting this hypothesis, a recent study showed that C1q deficiency increases the positive selection of B-1 cells and IgM autoAb production by a membrane-bound intracellular auto-Ag ().
In conclusion, using mice expressing site-directed transgenes for anti-DNA autoAbs we have shown that: (i) the MRL/Mp background was in itself capable of inducing the expression of anti-DNA Tg autoAbs, and (ii) V3H9R.MRL/Mp. could influence the production of Tg-derived IgM and IgG3 autoAbs possibly as a result of an impaired disposal of cellular debris. However, we found no evidence of a direct role of C1q in the regulation of self-reactive conventional B cells. Further studies on non-lupus prone genetic backgrounds such as C57BL/6 will be necessary to determine if C1q deficiency can play a more substantial role in shaping the repertoire of the autoreactive B cells. |
The eukaryotic exosome is a complex of 10 core subunits, including a 3′–5′ exonuclease and RNA-binding proteins, that is involved in many aspects of RNA processing and surveillance. The purified exosome alone does not display strong RNA degradation activity () and is targeted to specific substrates by different cofactors (reviewed in ). Two related complexes, TRAMP4 and TRAMP5, target aberrant tRNA (; ; ) and aberrant ribosomal RNA for exosome degradation (; ). These complexes also target a class of cryptic unstable transcripts (CUTs), resulting in their immediate degradation following transcription (; ; ; ). An apparently similar class of RNA has been identified in mouse and human cells; like the CUTs, these superficially resemble mRNAs in being RNA polymerase II (Pol II) transcripts that carry 5′ caps and 3′ poly(A) tails, but lack evident protein coding capacity (reviewed in ). Despite their apparent conservation, little is known about the function of these transcripts.
In many Eukaryotes, transcripts influence chromatin structure via the siRNA/RNAi system. In fission yeast, establishment of centromeric heterochromatin involves the poly(A) polymerases Cid12 () and Cid14 (), which are in the same family as budding yeast Trf4 and Trf5, and Cid14 resides in a TRAMP-like complex. lacks homologs of key components of the machinery that processes siRNAs, but it seemed possible that other mechanisms might transfer information from the transcriptome back to the chromatin structure of the genome.
The TRAMP complexes are composed of a poly(A) polymerase, Trf4 or Trf5, a putative RNA-binding protein, Air1 or Air2, and the putative DEVH-box helicase Mtr4. Trf4 and Trf5 were originally isolated in a synthetic lethal screen with mutations. Conditional double mutants of with showed ribosomal DNA (rDNA) condensation phenotypes (; ), and were reported to display defects in mitotic segregation (; ). In addition, strains lacking both Trf4 and Trf5 (which are synthetic lethal) fail to complete DNA replication (). These reports suggested that Trf4 and Trf5 play some role in DNA metabolism, particularly in the rDNA, although the nature of this role remained unclear.
Wild-type yeast contain a tandem array of approximately 200 rDNA repeats on chromosome XII, the number of which is maintained by regulated recombination (). Each repeat contains the 35S and 5S ribosomal RNA genes separated by intergenic spacer regions IGS1 and IGS2 (see ). The IGS1 region of the rDNA repeat contains a replication fork barrier (RFB) (), where a protein complex including Fob1 imposes a unidirectional block to the progress of DNA replication forks. This region also contains a recombination hotspot that is necessary for double-strand break formation and subsequent recombination between sister chromatids to occur ().
The repeat tracts in two sister chromatids are normally held together by the cohesin complex, which is concentrated over the cohesin-associated region (CAR) in IGS2 (see ) (). This ensures that even if recombination is initiated, it occurs only with the aligned repeat in the sister chromatid, and does not result in a change of repeat number. Two divergent Pol II transcripts generated from the E-pro promoter within IGS1 (see ) () have been suggested to regulate recombination, as replacement of E-pro with a regulated, divergent promoter resulted in cohesin displacement and extensive unequal recombination (). Unequal recombination leads to a net change in repeat number for the chromatid that initiated recombination. However, these analyses did not address potentially distinct roles for the two Pol II transcripts, which we designate here as IGS1-F and IGS1-R. IGS1-F is transcribed through the CAR, consistent with its reported role in cohesin displacement, whereas IGS1-R is transcribed through the RFB.
The rDNA IGS regions, along with the telomeres and inactive mating loci, are silenced for Pol II transcription by the histone deactylase Sir2 (). Loss of Sir2 leads to elevated expression of the IGS1 and IGS2 transcripts () and hyper-recombination within the rDNA array (). This was proposed to be a consequence of cohesin displacement ().
Here we demonstrate that one of the non-coding RNA (ncRNA) transcripts generated from the intergenic spacer regions, IGS1-R, is targeted for degradation mediated by Trf4 and the exosome. Trf4 is recruited to the rDNA spacer region that includes the RFB, via the IGS1-R transcript, and is required for stability of the rDNA repeat copy number. These data provide evidence for novel links between RNA and DNA metabolism in budding yeast.
We speculated that transcripts may be generated from regions of repressed chromatin in the yeast genome but degraded by the TRAMP and exosome complexes. We therefore tested a telomeric region (), the intergenic spacer (IGS) region of the rDNA repeat, a centromeric region () and the silenced mating-type cassettes (/α). In each case, we saw increased levels of a transcript in strains lacking Trf4, but not in single mutant strains lacking Trf5, Air1 or Air2 (, compare lane 2 with lanes 3–5).
At many yeast telomeres, the terminal repeats are flanked by the ‘Y′ region', which is conserved in whole or in part at 16 yeast telomeres. At some telomeres, this region encodes a putative DNA helicase (), designated Yel077c in the case of (). Strand-specific probes demonstrated that the transcript elevated in the Δ strain is an ncRNA expressed antisense to . The ncRNA is ∼6.5 kb in length and was detected by a probe located within () and by a second probe located 2.8 kb further toward the chromosome end (data not shown), showing it to extend across the Y′ region. 5′ RACE generated a product that was enriched in Δ (arrow in ), which was cloned and sequenced. Three 5′ ends were identified, located 106–128 bp beyond the 3′ end of the open reading frame (ORF) of . The ncRNA therefore starts close to the chromosome end and runs antisense through the entire putative helicase ORF. Polymorphisms in sequenced products demonstrate that the ncRNA is transcribed from at least two telomeres (data not shown).
The ncRNA was also elevated in a strain lacking Rrp6 (, lane 6), indicating that it is a target for exosome degradation. Depletion of Trf5 by growth of a Δ strain on glucose medium (, lane 5) increased accumulation of the transcript relative to the Δ single mutant. Overexpression of Trf5, in strains grown in galactose medium, can suppress the phenotype of Δ strains on some TRAMP substrates (), and this was the case for the transcript (, lane 4). We conclude that TRAMP4 and TRAMP5 both participate in degradation, with TRAMP4 probably playing the major role, and function with the exosome to degrade large ncRNAs generated from telomeric regions.
We have not further characterized the ∼1.2 kb transcript, and the structure of the loci prevents unambiguous assignment of the observed ∼1.2 kb transcript to a silent cassette. However, connections between Trf4 and Top1 mutations and rDNA structure (see Introduction) suggested a functional link to IGS transcription, which was therefore further investigated. The ncRNA products of transcription from E-pro, IGS1-F and IGS1-R () are normally present at very low levels () and we speculated that this might reflect rapid degradation involving the TRAMP and exosome complexes. To visualize these ncRNAs, northern blots of RNA from TRAMP and exosome mutants were probed with strand-specific probes to both rDNA intergenic spacer regions (). An Δ strain, which overexpresses IGS transcripts, was used as a positive control. A Δ strain was also analyzed, as double mutants were reported to show synthetic lethality and rDNA condensation phenotypes (; ). Strains carrying Δ and Δ showed elevated levels of three ncRNAs IGS1-R, IGS1-F and IGS2-R (, lanes 6 and 7).
The levels of IGS1-F and IGS2-R were unaltered in TRAMP or exosome mutants, whereas the level of IGS1-R was substantially increased in Δ and to a lesser extent in Δ Δ and Δ strains. No stabilization of IGS1-R was seen in Δ strains, and overexpression of Trf5 under GAL regulation did not suppress the Δ phenotype (, lanes 3 and 4). This suggests that Trf5 does not efficiently target IGS1-R, even in the absence of Trf4. IGS1-R stabilization in the Δ exosome mutant strain was weaker than that in Δ ( (lane 5) and D (lane 3)) and we therefore also examined the core exosome mutant (). Even at permissive temperature (25°C), accumulation of IGS1-R was visible in the strain. We conclude that TRAMP4 functions with both Rrp6 and the core exosome to degrade IGS1-R, whereas degradation of IGS1-F and IGS2-R presumably involves other activities.
Two major forms of IGS2-R of approximately 1.6 kb and 850 nt in length were detected (see also ). The longer IGS2-R species also hybridized to a downstream probe to IGS1, whereas the short species was not detected with this probe, indicating that it is 3′ truncated (data not shown). This would be consistent with termination of some IGS2-R transcripts around the location of the CAR, even in Δ and Δ strains. A shorter probe directed against the 3′ region of IGS1-R (NTS1 short; see ) was also used. This region is extremely AT rich and the probe hybridized poorly; however, only the longer transcripts were detected by NTS1 short probe (, compare lanes 1–4 with 5–8). This indicates that the IGS1-R transcripts show 3′ heterogeneity.
To assess whether increased IGS1-R RNA reflects increased transcription or post-transcriptional stabilization, chromatin immunoprecipitation (ChIP) was performed to determine RNA Pol II occupancy in wild-type, Δ and Δ cells () and Δ (data not shown). This revealed a clear peak of Pol II within the IGS1-R region in the wild-type strain. This peak was elevated in the Δ strain, consistent with constitutive derepression (). In contrast, the Pol II signal in the Δ strain was lower than that in the wild-type strain, despite elevated levels of the transcript.
We next wanted to determine how the TRAMP complex is recruited to the IGS1-R transcripts. The Nrd1–Nab3 heterodimer of RNA-binding proteins was reported to recruit the exosome to substrate RNAs (; ). The longest observed IGS1-R transcript contains 10 potential binding sites for Nab3 (UCUU) including two prominent clusters, and seven binding sites for Nrd1 (GUAA/G). The level of IGS1-R was therefore assessed in ts-lethal and mutant strains (). IGS1-R was strongly stabilized in the mutant, demonstrating that it is targeted by the Nrd1–Nab3 pathway. In contrast, the mutant conferred little or no stabilization (). Allele specificity has, however, previously been reported for mutations in these proteins () and the data do not demonstrate that Nrd1 is primarily responsible for targeting IGS1-R for degradation. On other transcripts, Nrd1–Nab3 are responsible for transcription termination (; , ; ; ). However, we saw no clear differences in the migration of the IGS1-R transcripts in the or mutants relative to the wild type (), suggesting that termination is not strongly impaired.
Primer extension was used to define the 5′ end of the IGS1-R transcript (). The 5′ end was mapped by running the primer extension products alongside a sequencing ladder on a 40 cm denaturing polyacrylamide gel (data not shown). The major 5′ end lies at +599 nt from the end of the 25S rRNA, with some secondary 5′ ends spanning about 20 bp. This position is 26 bp 3′ to that originally described () and around +175 nt from the RFB.
Conventional analyses of polyadenylation using RNase H and oligo(dT) were complicated by the presence of two genome-encoded poly(A) tracts within IGS1-R (). However, the results were consistent with 3′ heterogeneity, as were the northern analyses shown in . Oligo-dT-directed 3′ RACE detected multiple transcripts (), showing that IGS1-R is polyadenylated even in Δ strains, and of 16 3′ RACE clones sequenced, 6 terminated in genomic encoded poly(A) tracts and the remaining 10 clones contained 9 different 3′ ends, demonstrating substantial 3′ heterogeneity ().
Depletion of Trf5 from the Δ strain had little effect on transcript length (, lane 5), whereas transfer of a ts-lethal Δ strain to 37°C resulted in shorter IGS1-R transcripts (, compare lanes 4 and 8). At permissive temperature, the abundance of IGS1-R transcripts was greater in the Δ double mutant than in either single mutant (, compare lanes 2, 3 and 4). These results indicate that IGS1-R is polyadenylated by Pap1 and suggest that efficient polyadenylation promotes IGS1-R degradation.
We tested whether the polyadenylation activity of Trf4 is required for degradation of IGS1-R (). Plasmids expressing either Trf4 or the catalytically inactive Trf4-DADA () fully suppressed the IGS1-R stabilization phenotype in the Δ strain. Recruitment of the exosome and IGS1-R degradation does not therefore require polyadenylation by Trf4 although the presence of the protein is necessary.
TRAMP and exosome mutants were tested for alterations in rDNA copy number using pulsed field gel electrophoresis (PFGE) (). In the wild-type strain, chromosome XII (∼60% of which is composed of rDNA repeats) showed a well-defined length corresponding to an rDNA array of ∼200 repeats. In contrast, Δ and Δ strains showed hyper-recombination phenotypes that caused chromosome XII to appear as smears (compare lane 1 with lanes 6 and 7 in and lanes 1–3 with lanes 7–9 and 13 and 14 ) (; ; ). Strains lacking the TRAMP components, Δ or Δ Δ, showed sporadic deviations from the wild-type rDNA copy number, with the greatest effect in Δ strains. These deviations from wild-type repeat number had only limited penetrance, with three out of eight Δ clones analyzed showing a clear repeat number reduction and two strains showing repeat expansion. We observed no evidence of an rDNA hyper-recombination phenotype, which would be indicated by smearing of the rDNA band, in any Δ single mutant analyzed, in contrast to a previous report ().
A Δ strain was viable on restrictive high methionine medium, and multiple Δ Δ transformants isolated in four independent experiments were all growth-impaired but viable. This is in contrast to their reported synthetic lethality (; ; ), and may reflect differences in alleles or strain background. We also combined Δ with Δ, and growth of the double mutant strain was similar to that of the Δ single mutant. The double mutants of Δ with either Δ or Δ showed large losses in rDNA repeat number (, lanes 10–12 and 15–17), and this phenotype was observed in all clones tested (at least four of each combination). In contrast, the combination of Δ with either Δ or Δ had no apparent effect on rDNA repeat number (). We also tested the exosome mutant Δ (). Loss of Rrp6 alone had no clear effect on repeat number (lane 2), but an Δ Δ double mutant showed reduced heterogeneity relative to Δ single mutant strains (lanes 3–7).
To assess potential links between stability of the rDNA and transcription of IGS1-R, IGS1-F and IGS2-R, the Δ Δ and Δ Δ double and single mutant strains were analyzed by northern hybridization () and Pol II ChIP ().
Levels of IGS1-R appeared to correlate with rDNA copy number instability, being higher in Δ Δ and Δ Δ double-mutant strains than in the Δ and Δ single mutants. In , the quantification of the northern data is normalized to rDNA repeat number. The levels of IGS1-F were highest in the Δ and Δ single mutants, which showed the greatest repeat heterogeneity. In contrast, the levels of IGS2-R appeared principally dependent on the presence or absence of Sir2. The ratio of abundance of the ncRNAs was altered between Δ and Δ strains expressing or lacking Trf4. It is possible that this has an effect on repeat number, although we have no direct evidence for this. Loss of Trf5 from either the Δ or Δ strains had no clear effect on the levels of any of the IGS ncRNAs ().
ChIP for RNA Pol II at the locus () shows that Pol II occupancy is not significantly altered in Δ Δ strains compared to Δ. Hence, the reduced heterogeneity of the rDNA repeats in the Δ Δ double mutant relative to the Δ single mutant is not due to reduced transcription. This also confirms that the high accumulation of transcript is due to increased RNA stability in the absence of Trf4.
We conclude that deletion of , but not , from the Δ or Δ strains leads to greatly increased levels of IGS1-R, due to increased RNA stability, and a drastic loss of rDNA repeats. Deletion of from the Δ strain appeared to reduce the heterogeneity in rDNA repeat number without clearly decreasing average repeat length.
One explanation for the loss of rDNA repeats in Δ mutants would be that Trf4 is a DNA polymerase involved in rDNA replication, as previously proposed (). Were this the case, the polymerase activity of Trf4 would be required to allow the hyper-recombination observed in Δ strains.
To test this, we made use of the suppression of the rDNA repeat heterogeneity seen in Δ strains by loss of Trf4 (). We compared Δ (lanes 1, 4 and 7) and Δ Δ (lanes 2–3, 5–6 and 8–9). These strains also carried plasmids lacking an insert (lanes 1–3), expressing intact Trf4 (lanes 4–6) or expressing the catalytically inactive Trf4-DADA (lanes 7–9). The Δ strains expressing wild-type Trf4 showed a hyper-recombination phenotype (lanes 5 and 6). Hyper-recombination was suppressed in the absence of Trf4 (lanes 2 and 3), but was clearly present when only Trf4-DADA was expressed (lanes 8 and 9). Western blotting confirmed that the mutant and wild-type Trf4 were expressed at similar levels (data not shown). In this experiment, strains were grown on minimal media to select for plasmid maintenance and recombination was less active than on complete YPD media used in the experiments shown in . Thus, the poly(A) polymerase activity of Trf4 is not required for IGS1-R degradation and is also not required for hyper-recombination.
These analyses do not resolve differences in recombination frequency and stability of the rDNA repeat tract. Recombination rates within the rDNA tract can be assessed by integrating a single-copy marker gene and scoring for its loss. The presence or absence of a functional gene can be scored by a colony color test on medium containing Pb, on which Δ colony sectors turn dark brown () (). A construct was integrated into the IGS2 region of a single rDNA repeat in one strain of each genotype used in . Three independent insertion clones from each strain were then scored for colony sectoring phenotypes (). Comparison of sectoring levels in Δ and Δ Δ shows that loss of has little or no effect on recombination frequency. A transgene inserted at this location was previously shown to be repressed by Sir2 (), and we confirmed this for our insert by western blotting (data not shown). Met25-GFP expression was lower in western blots from Δ strains than Δ Δ strains (data not shown), showing that Sir2-dependent silencing was maintained. This indicates that hypoacetylation of H3 and H4 by Sir2 is not lost, since Sir2-dependent silencing requires its deacetylation activity ().
In Δ strains, transcription of the reporter was frequently reduced, resulting in dark pigmentation (). This is consistent with the reduction of IGS1-F expression seen in Δ Δ strains compared to Δ, but the effect was variable between fresh transformants and old cells. The Δ Δ strain proved hypersensitive to Pb ions and could not be assessed in the sectoring assay.
We hypothesized that the altered rDNA stability in Δ strains is due to direct effects of Trf4 on the rDNA or chromatin rather than indirect consequences of defects in RNA processing. In this case, the role of IGS1-R transcription might be to recruit Trf4 to the rDNA in the vicinity of the RFB region. Were this model correct, we would detect an association of Trf4 with the rDNA IGS1-R region that is dependent on RNA Pol II transcription.
ChIP analysis of Trf4-Myc over the rDNA IGS regions () showed clear enrichment over IGS1-R. To confirm that this association was dependent on RNA Pol II transcription, we analyzed an strain, which carries a fast-acting temperature-sensitive mutation in RNA Pol II (). The ChIP signal for Trf4 was substantially reduced across the IGS1-R region at the restrictive temperature, consistent with its recruitment by IGS1-R nascent transcripts, although the ChIP signal for Pol II showed greater reduction than the Trf4 signal (). To ensure that the protein context of the rDNA is not generally altered in this strain, lysates from this experiment were checked by ChIP for the presence of Sir2, which was only mildly reduced at 37°C ().
In order to address whether cohesin is displaced by IGS1-R accumulation in the Δ strains, we performed ChIP analysis using a 13-Myc-tagged cohesin subunit Smc1. The Smc1-Myc fusion is the only form of Smc1 present in the cells, which showed no growth impairment. In addition, rDNA recombination rates were unaltered, as judged by the Pb plate method described above (data not shown), indicating that the fusion protein is fully functional.
In the wild-type background, Smc1-Myc showed the expected distribution with a clear peak over the CAR (). Neither the distribution nor the intensity of the Smc1 ChIP signal was altered by the loss of Trf4. In contrast, the Smc1 ChIP signal was strongly reduced by loss of Top1, either in the presence or absence of Trf4. This makes it unlikely that the effects of Trf4 on rDNA copy number regulation are related to removal of cohesin from the IGS region.
Genome-wide analyses of RNA Pol II density in yeast using ChIP and tiling microarrays showed that almost the entire genome is transcribed (; ). A striking feature of these analyses was that regions of repressed chromatin, notably in the rDNA IGS1 region and telomeres, showed Pol II occupancy that was significantly below the background of ‘non-transcribed' regions. Paradoxically, these repressed regions can be actively transcribed under some conditions, leading to the suggestion that initial rounds of transcription might be needed to establish subsequent silencing (). In strains lacking the poly(A) polymerase Trf4, we detected cryptic transcripts derived from and from regions of repressed chromatin: a telomeric region () and the rDNA intergenic spacer region IGS1-R.
The ncRNA transcript stabilized in TRAMP and exosome mutants is ∼6.5 kb in length and is derived from the Y′ region, which is fully or partially conserved at most yeast telomeres (). The RNA initiates close to the terminal telomeric repeats and runs antisense to an ORF (), which potentially encodes a 143 kDa protein proposed to function as a DNA helicase involved in telomere maintenance (). Sequence analyses showed that ncRNAs are generated from at least two telomeres, and we speculate that they may be transcribed from many or all telomeres. The identification of these ncRNAs might be consistent with the model that cryptic transcripts play a role in establishing silenced chromatin regions (). In , centromeric regions are very small, lack clear heterochromatic regions and were not previously reported to be transcribed. In contrast, centromeric regions in are transcribed, but the RNAs are rapidly degraded by the TRAMP and exosome complexes (). We detected an RNA of ∼1.2 kb apparently derived from , which was elevated in the Δ strain, suggesting that cryptic centromeric transcripts may also be present in .
Previous reports functionally linked Trf4 with Top1 and cohesion and condensation in the rDNA repeats (), and we therefore analyzed the IGS1 transcripts in more detail. Two transcripts, IGS1-R and IGS1-F, are generated by divergent transcription from the E-pro promoter. We showed that IGS1-R, but not IGS1-F, is recognized and degraded by TRAMP4 and the exosome. Expression of a catalytically inactive form of Trf4 showed that its poly(A) polymerase activity is not required for degradation of IGS1-R. The catalytic activity of Trf4 was also dispensable for degradation of mRNA in THO complex mutants (), but was required for degradation of hypomethylated tRNA (). This suggests that polyadenylation aids degradation of structured substrates, but is dispensable on less structured RNAs. The 3′ ends of the IGS1-R transcripts are polyadenylated by the canonical poly(A) polymerase Pap1, and are very heterogeneous. We speculate that this heterogeneity may result in part from termination by collision with oncoming RNA polymerase I molecules.
ChIP analyses showed that Trf4 could be crosslinked to the rDNA over the IGS1-R region, but not over IGS1-F. This association is largely dependent on functional RNA Pol II, since it was substantially reduced in an mutant strain at non-permissive temperature. We therefore propose that Trf4 is recruited co-transcriptionally to the nascent IGS1-R transcripts. The IGS1-R region contains multiple predicted binding sites for the RNA-binding proteins Nab3 and Nrd1, which were previously reported to act as cofactors for the exosome in RNA degradation (). Consistent with this, degradation of IGS1-R was strongly inhibited by a mutation in Nrd1, indicating the involvement of the Nrd1/Nab3 heterodimer in recruiting the TRAMP and/or exosome complexes. Since Nrd1/Nab3 also function in transcription termination by RNA Pol II (; ), they clearly have the potential to bind co-transcriptionally to nascent transcripts. We saw no evidence for effects of mutations in Nrd1 or Nab3 on termination of the IGS1-R transcript, but we predict that they act to recruit the TRAMP complex to nascent IGS1-R transcripts.
Strains carrying Δ displayed sporadic alterations in copy number. In contrast, strains carrying either Δ or Δ showed hyper-recombination, manifested as extensive rDNA repeat length heterogeneity. The combination of Δ with either Δ or Δ resulted in a synergistic phenotype with drastic loss of rDNA repeats. Analysis of an integrated marker indicated that this does not reflect altered recombination rates, indicating that rDNA instability is responsible for repeat loss. Deletion of exosome component Rrp6 in Δ reduced the hyper-recombination phenotype, indicating that Rrp6 is also required for normal rDNA stability.
The cohesin complex is believed to hold sister chromatids together in the rDNA, so that recombination does not lead to alterations in copy number (). However, loss of Trf4 did not detectably affect cohesin binding over IGS1 and IGS2. Changes in rDNA recombination rate without alteration of cohesin association were previously observed in mutants of Lrs4/Csm1 ().
Together the data are consistent with the model in . Transcription of IGS1-F through the CAR located in IGS2, or the balance between transcription of IGS1-F and IGS2-R may play important roles in cohesin displacement (; ). However, we predict that IGS1-R has a distinct function, which is important for rDNA stability, although its exact nature remains unclear. Trf4 and other factors binding to the nascent IGS1-R transcript may enhance rDNA stability at the site of transcription, possibly by promoting the repair of DNA damage. Alternatively, IGS1-R transcripts that escape degradation might exert a dominant negative effect on rDNA repair or recombination at other sites, and these models are not mutually exclusive. The ChIP data showing co-transcriptional recruitment of Trf4 would be consistent with association of the TRAMP complex with the rDNA at the site of transcription via IGS1-R. Rrp6 retains mRNAs with aberrant 3′ ends close to the site of transcription (), and it is conceivable that IGS1-R, which shows high 3′ heterogeneity, could also be linked to the transcription site by Rrp6.
The IGS1-R transcript passes through the RFB, a key region for rDNA copy number regulation. Replication forks emanate from replication origins (termed ARS; ) in each rDNA repeat, but forks moving against the direction of pre-rRNA transcription are stalled at the RFB, presumably to reduce collision with RNA Pol I. Top1 binding sites flank the RFB and are required to relieve DNA supercoiling generated by the high Pol I transcriptional rate. We speculate that the lack of Top1 activity leads to DNA damage over RFB regions carrying stalled DNA replication forks. Strains lacking Sir2 should not have excessive supercoiling at the RFB, but will have more stalled replication forks in this region due to increased activation of rDNA replication origins when Sir2 is absent ().
There are a number of potential links between the exosome and its cofactors and the repair of DNA damage in yeast. Strains lacking Trf4 are hypersensitive to DNA cleavage induced by the Top I inhibitor camptothecin (). This was also the case for Δ Δ double-mutant strains (our unpublished observations), showing this phenotype to be TRAMP-related. Strains lacking Trf4 were also reported to be hypersensitive to DNA damage caused by treatment with the alkylating agent MMS (). In addition, the nuclear exosome components Rrp6 and cofactor Rrp47/Lrp1/C1D are implicated in the repair of UV-induced DNA damage (), and Δ strains are defective in both non-homologous end joining and homologous recombination (). We speculate that the TRAMP and exosome complexes also play a role in the repair of DNA damage at the RFB. Increased histone mRNA levels leading to defects in DNA metabolism have recently been reported (); however, this was not found to be the case for a Δ single mutant, and we have also not observed this in our strains (data not shown).
Recombination events are frequent in Δ mutants but rare in Δ. We predict that recombination-based repair is infrequent in Δ strains due to the action of Top1 in removing supercoils and preventing DNA damage. This may explain the low penetrance of the repeat number change phenotype in Δ, with events altering repeat number occurring only once in many generations. Notably, however, some Δ and Δ Δ samples contained two populations with discrete rDNA lengths ( and data not shown), suggesting that a recombination event had occurred during early growth of the culture. The Δ Δ strains are predicted to undergo frequent rDNA damage repair by recombination. However, the repeat tract collapses as these events are biased toward contraction, being based on strand invasion rather than homologous recombination.
Degradation of the IGS1-R transcript resembles that of the CUTs, which account for a significant proportion of the genome of . A crude estimate based on the microarray analyses of suggests a minimum of 600 CUTs, or just under 10% of the number of annotated genes (see for calculations). Strains lacking Trf4 are reported to show defects in chromosome arm cohesion (; ). Whether this is related to the recruitment of TRAMP to the numerous and widely dispersed CUTs remains to be determined.
Yeast transformation was performed by standard methods. Yeast strains are described in . Cells were grown in YPD (2% peptone, 2% glucose, 1% yeast extract) or synthetic media (0.5% (NH)SO, 1.7% yeast nitrogen base, 2% glucose, amino acids) at 25°C; temperature shifts to 37°C were performed in a shaking water bath. Plasmids are described in and oligonucleotides used for cloning are listed in .
Recombination assays were performed as described by .
Yeast RNA extraction and northern analysis were performed as described (); high molecular weight RNA was separated on 1.2% glyoxal gels. Experimental details and probes are described in and .
PFGE was performed as described (); see for a detailed protocol.
ChIP was carried out as described () with modifications (see ). |
The vascular system is lined with endothelial cells that serve as a barrier separating blood from underlying tissues. Maintenance and modulation of endothelial barrier function is necessary to control the movement of nutrients, fluids, and immune cells between the intravascular and extravascular compartments. Barrier function requires cell–cell interactions, mediated in large part by adherens and tight junctions, which act to restrict the passage of material between cells in the endothelial lining (; ). Loss of cell–cell junctional integrity in the endothelial lining results in increased permeability and edema (; ; ).
Both adherens and tight junctions use transmembrane proteins that interact with homotypic receptors on neighboring cells. Ligation of these receptors stimulates the cytoplasmic recruitment of multiprotein complexes that are involved in signal transduction and interaction with the cytoskeleton. Endothelial adherens junctions are formed by the homophilic interaction of transmembrane VE-cadherin molecules (). The intracellular domain of vascular endothelial–cadherin (VE-cadherin) interacts with β-catenin and recruits molecules such as plakoglobin, α-catenin, zyxin, p120 catenin, and vinculin (; ; ). Tight junctions contain several transmembrane components, including occludin, claudins, and junctional adhesion molecule-A (; ). These interact with intracellular components such as the zonula occludens (ZO) proteins (ZO-1, ZO-2, and ZO-3) and AF-6/Afadin.
Functionally, cell–matrix associations, such as focal adhesions, are similar to cell–cell junctions in that they link the cytoskeleton with external contact points and are important for barrier function (). In focal adhesions, transmembrane integrin receptors bind extracellular matrix and recruit intracellular multiprotein complexes that interact with the cytoskeleton. Many of the proteins found in focal adhesions are also present in cell–cell junctions.
Members of the Enabled/vasodilator-stimulated phosphoprotein (Ena/VASP) family of actin regulatory proteins are found at both focal adhesions and cell–cell junctions, where they interact and colocalize with components of adherens and tight junctions (; ; ). In addition to localization to sites involved in barrier function, Ena/VASP is a well-established substrate for PKA and PKG (; ; ). PKA and PKG signaling pathways are involved in barrier function regulation, which raises the possibility that Ena/VASP could be a downstream effector of these pathways.
At the leading edge of lamellipodia and the tips of filopodia, Ena/VASP localizes to regions with dynamic actin reorganization (; ; ; ; ), where it has been proposed to promote actin polymerization via an anticapping mechanism (; ). Dynamic actin is also found at focal adhesions and cell–cell junctions, and ligation of cell–cell receptors, such as cadherins, results in reorganization of the underlying actin cytoskeleton (; ). Disruption of Ena/VASP has been shown to reduce the formation of perijunctional actin filaments () and perturbs the dynamics of epithelial sheet sealing in (). Although these data suggest that Ena/VASP regulates actin dynamics at cell–cell and cell–matrix junctions, the functional implications of this are unclear.
In mammals, the Ena/VASP family consists of three members, mammalian Ena (Mena), VASP, and Ena/Vasp-like (; ). Because these members have overlapping function and expression patterns, knocking out individual family members results in relatively minor phenotypes in mouse models (; ; ). To determine the biological function of this family of proteins, we have generated an Ena/VASP triple null (mmvvee) mouse. mmvvee mice exhibit neuronal defects, including exencephaly, cobblestone cortex, and lack of cortical fiber tract formation, which have previously been characterized elsewhere (). In addition to these phenotypes, mmvvee mice display severe vascular defects that are described here.
Gross examination of mmvvee embryos revealed profound edema, with 82% of embryos (14 out of 17) at embryonic day (E) 14.5 appearing edematous to the naked eye (). The edema was most pronounced in the torso, where transverse sectioning showed that it was largely restricted to the subdermal region (). Edema did not result from exencephaly because we observed edematous mmvvee embryos that were not exencephalic (unpublished data).
Embryonic edema can result from cardiac defects (; ; ). Congenital heart defects typically manifest with cardiac dilation and congestive heart failure, resulting in increases in the hydrostatic pressure within the vascular system, which drive fluid into the interstitial tissue. To determine if heart defects could explain the edema observed in mmvvee embryos, we analyzed the anatomy of the mmvvee embryonic heart. Histologically, mmvvee hearts appeared structurally normal () with no sign of dilation. Ena/VASP has been implicated in Sema6D signaling events necessary for proper cardiac chamber formation and trabeculation (); however, trabeculation in the mmvvee heart was normal (), and major structural features of the mmvvee heart, including septum, valves, and associated veins and arteries, displayed no defects (not depicted). Closer analysis by electron microscopy showed that mmvvee cardiac myocytes formed Z lines and intercalated discs that appeared similar to those of littermate controls (). Although there is no evidence of structural or functional heart abnormalities that could explain the edema phenotype, we did notice that mmvvee endocardia displayed elongated cell membranes and were less closely associated with the cardiac myocytes than endocardia in the controls (). Because the endocardium is the endothelial lining of the heart, this defect suggested that endothelial dysfunction might be at the root of the edema caused by loss of Ena/VASP.
The loss of junctional proteins VE-cadherin, β-catenin, and FAK in the endothelia of mice results in edema and the regression of blood vessels during development (; ; ; ). To determine the patterning and integrity of blood vessels in mmvvee mice, whole mount embryos were stained for the endothelial-specific marker platelet/endothelial cell adhesion molecule (PECAM) 1, and vessels were found to be grossly intact at E10.5 (). Although major vessels appeared normal in mmvvee embryos, there were differences in small vessels evident at E10.5, including disruption in connectivity between intersomitic vessels in the mid-torso region () and irregular patterning in the vascular plexus of the tail (). These subtle differences in the small vessels of mmvvee embryos are relatively minor in comparison to the vessel disruption observed in mice that were null for endothelial junctional proteins such as β-catenin (). Importantly, these patterning defects are not the beginning of widespread vascular regression because vessels visualized under a dissection microscope were intact in the amniotic sac and body as late as E16.5 (). Vessels in the body and amnion of E16.5 mmvvee mice did not exhibit any differences in patterning or vascular density.
Disruption in endothelial barrier function allows the leakage of plasma proteins into the interstitial space. This alters the oncotic pressure gradient that normally functions to draw fluid back into the vasculature, resulting in the accumulation of interstitial fluid. A more profound disruption in barrier function allows the hemorrhaging of red blood cells. Consistent with a role for Ena/VASP in promoting barrier function, we observed sporadic hemorrhaging in mmvvee embryos (). The earliest observations of hemorrhage were at E14.5 and increased throughout development. Hemorrhaging was also evident by an accumulation of blood in the amnion (). Similar to the pattern seen in the embryo of edema followed by hemorrhage, analysis of embryos at different stages revealed that the amnion first accumulated excess amniotic fluid () followed by hemorrhage. By E18.5, 70% of mmvvee amnions (60 out of 86) exhibited an accumulation of blood. Careful examination of recently harvested mmvvee embryos failed to identify an obvious source of hemorrhage that would account for the accumulation of amniotic blood. Although it is likely that the blood in the amnion is caused by hemorrhaging of the amniotic vasculature, it is possible that it comes from the embryo itself.
Histological analysis indicated a loss of integrity of small blood vessels in mmvvee subdermal tissue (). This was confirmed by transmission electron microscopy, which found that 60% of blood vessels (6 out of 10) >10 μm in diameter were discontinuous, meaning there was an opening in the endothelial lining and escaping red blood cells (). These discontinuous vessels were determined to be venules, based on size and their lack of surrounding adventitia. In addition to discontinuity, mmvvee venules also displayed endothelial cell protrusions into the vascular lumen and subendothelial space (), further suggesting a defect in cell–cell contacts. In contrast, no examples of discontinuous capillaries/small arterioles or larger veins and arteries were evident in mmvvee embryos. Very small vessels (<10 μm), presumed to be capillaries, appeared to be consistently intact in both groups. These findings suggest that hemorrhaging in mmvvee embryos results specifically from open venules.
To measure the effect of Ena/VASP on endothelial barrier function, we used a cell-based assay to test the ability of fluorescently tagged dextran to cross a confluent monolayer of endothelial cells. The dextran molecule had a molecular weight comparable to albumin, the chief protein component of blood plasma. The isolation and culture of sufficient quantities of primary endothelial cells from mmvvee embryonic mice was not possible, so we used human umbilical vein endothelial cells (HUVECs) for the barrier function assays. Consistent with localization in other cell types, Ena/VASP proteins exhibited localization at focal adhesions, stress fibers, and cell–cell contacts in HUVECs (Fig. S1 A, available at ). Previous studies of the role of Ena/VASP on barrier function focused on VASP and relied on the overexpression of VASP fragments to interfere with VASP function (; ; ) or used siRNA to knock down VASP (), which did not eliminate the function of other Ena/VASP proteins expressed in endothelial cells (unpublished data). To overcome these potential limitations, we used a strategy proven to block the activity of all Ena/VASP family members. This approach involves expression of the EGFP-FP4-Mito construct that sequesters all Ena/VASP proteins to the surface of the mitochondria (Fig. S1 B) and mimics loss of function (, ; ). EGFP-FP4-Mito–expressing HUVECs were grown to confluence and used in barrier function assays. When Ena/VASP activity was inhibited, basal barrier function was significantly reduced as compared with control HUVECs expressing EGFP or EGFP-AP4-Mito, a construct that targets mitochondria but does not sequester Ena/VASP (). Inhibition of Ena/VASP activity resulted in a reduction of barrier function similar to that attained through the addition of VEGF, a known permeability enhancing agent, to control monolayers (). The high permeability of Ena/VASP-inactivated monolayers could be further augmented with VEGF treatment, indicating that Ena/VASP is not required for response to VEGF. Rather, Ena/VASP activity appears to be important in the maintenance of basal barrier function. In support of this, overexpression of EGFP-VASP improved barrier function (). Collectively, these data are consistent with a model in which the edema and hemorrhage observed in mmvvee embryos result from defects in endothelial barrier function.
The defective vascular integrity of mmvvee endothelium and reduced barrier function described in the previous sections could result from a failure to recruit the normal complement of cell–cell junction proteins. In fact, overexpression of a fragment of Ena/VASP thought to act as a dominant negative has been shown to block E-cadherin recruitment in keratinocytes (). Furthermore, it has been suggested that phosphorylation of VASP by PKA is important for ZO-1 recruitment to cell–cell junctions (). To determine if the genetic loss of Ena/VASP affects the recruitment of junctional proteins, we isolated primary PECAM-positive endothelial cells from the aortae of control and mmvvee embryos and stained for junctional components. The resulting small clusters of primary endothelial cells showed that localization of PECAM, VE-cadherin, β-catenin, α-catenin, and ZO-1 to cell–cell junctions were unaffected by the absence of Ena/VASP (). Immunohistochemistry on tissue sections also showed a normal distribution of cell junctional markers in the mmvvee vasculature (unpublished data). Similarly, inactivation of Ena/VASP in HUVECs had no effect on the recruitment of the same junctional proteins to established cell–cell junctions () despite the observed barrier defects described in the previous section. The recruitment of junctional proteins was also unperturbed by VASP overexpression ().
To test if the kinetics of cadherin-dependent contact formation was perturbed in the absence of Ena/VASP, we performed calcium switch assays on HUVEC cultures. Chelation of calcium disrupts the ligation and localization of VE-cadherin at cell junctions. After the reintroduction of calcium, the recovery rate of VE-cadherin to junctions of Ena/VASP-inactivated cells was identical to controls, with localization first detected after 30 min and full recovery after 4 h (). These results indicate that neither cadherin-mediated adhesion nor the recruitment of junctional proteins to cell–cell contacts are affected by the absence of Ena/VASP, which suggests that mislocalization or delayed recruitment of junctional proteins does not cause the edema observed in mmvvee embryos.
Loss of Ena/VASP activity in primary endothelial cells () and HUVEC cultures (), either by genetic knockout or protein inactivation, respectively, led to a visible reduction in F-actin content. This reduction in F-actin was measured with microscopy by quantitating the signal intensity of fluorescently tagged phalloidin (). Although overexpression of Ena/VASP in HUVEC monolayers did not cause a substantial increase in F-actin content, it did result in a striking rearrangement of actin into parallel ventral stress fibers that aligned with each other across multiple cell lengths ().
The actin reorganization caused by Ena/VASP overexpression is similar to that induced by shear stress. Under physiological conditions, flowing blood subjects endothelial cells to shear stress, which has been shown to modulate numerous endothelial cell functions, particularly the strengthening of cell–cell junctions and elongation of cell shape in the direction of flow (; ). To determine if the cytoskeletal changes that facilitate response to shear stress are impaired in the absence of Ena/VASP activity, we cultured Ena/VASP-inactivated HUVECs in flow chambers. Confluent monolayers of control HUVECs subjected to shear stress overnight responded as expected by forming parallel stress fibers but, strikingly, Ena/VASP-inactivated HUVECs failed to form stress fibers (). Instead, shear stress appeared to result in a slight decrease in the amount of stress fibers as compared with cells cultured without flow. In contrast, cortical actin staining appeared unaffected by inhibition of Ena/VASP. To determine if shear stress response was also impaired in mmvvee embryos, we looked at actin organization and endothelial cell shape in aortae. Control aortae had robust stress fibers that ran in the direction of blood flow, whereas the stress fibers of mmvvee embryos were less organized. Immunostaining for PECAM in control aortae revealed a strong linear pattern consistent with cells that had elongated in the direction of blood flow, but PECAM staining of endothelial cells in the lining of mmvvee aortae was visibly disorganized (). These data indicate that Ena/VASP is critical for the reorganization of the cytoskeleton and changes in cell morphology in response to shear stress.
Biochemical experiments suggest Ena/VASP binds at, or near, the barbed ends of F-actin and permits continued incorporation of monomeric actin in the presence of capping protein by antagonizing capping activity. We hypothesized that the observed changes in F-actin content were attributable to changes in actin incorporation at barbed ends. To test this possibility, we pulsed permeabilized HUVECs with fluorescently labeled monomeric actin and measured its incorporation at free barbed ends by quantitative microscopy. G-actin incorporation into barbed ends was reduced in cells in which Ena/VASP was inhibited and increased upon the overexpression of VASP (). The sites of the greatest barbed end labeling were at the leading edges of lamellipodia, focal adhesions, and along stress fibers in discrete puncta known as dense bodies, which is consistent with previously published results using nonendothelial cell types (). To determine the effect of Ena/VASP activity on these different structures, we independently measured barbed end labeling at the cell periphery (leading edge), cell–cell junctions, and internal regions of the cell, where the signal was largely caused by incorporation into stress fibers. Barbed end labeling at all of these sites was found to be sensitive to Ena/VASP activity, although the most profound effect was observed along stress fibers, with signal being frequently undetectable in cells expressing FP4-Mito (). This same pattern was found using dissociated primary endothelial cells, with barbed end incorporation reduced in cells from mmvvee aorta as compared with littermate controls (). Ena/VASP has previously been described to decorate stress fibers in discrete puncta (; ), a localization we confirmed in both HUVECs and primary endothelial cells (Fig. S1, A and C). Higher resolution analysis of stress fibers showed that GFP-VASP localizes in puncta that overlap with, or are immediately adjacent to, regions of barbed end labeling (). Puncta were spaced at a mean of 0.852 ± 0.285 μm from each other, a distribution that was unaffected by Ena/VASP activity. The colocalization of Ena/VASP and sites of G-actin incorporation along stress fibers, and the reduction of G-actin incorporation in stress fibers upon the inactivation of Ena/VASP, indicates that Ena/VASP promotes actin monomer incorporation along stress fibers.
Stress fibers are associated with the generation of contractile forces within the cell. Other proteins known to localize in puncta along stress fibers include α-actinin, filamin, tropomyosin, and myosin, all of which are involved in actomyosin contraction (; ; ; ; ; ). Because the actin stress fiber rearrangements that accompany Ena/VASP overexpression are similar to those observed in highly contractile cells, we measured the expression of myosin light chain (MLC) in cells with altered Ena/VASP activity. Ena/VASP overexpression increased levels of MLC over that in controls, whereas inactivation of Ena/VASP reduced MLC expression (). These changes in the level of MLC expression were accompanied by changes in MLC phosphorylation that have been shown to be indicative of enhanced actomyosin contraction ().
To determine if changes in MLC expression and phosphorylation state had the predicted effect on actomyosin contraction, we seeded endothelial cells into three-dimensional collagen matrices and used matrix compaction as an indication of the force exerted upon the collagen by the cells. Consistent with activation of MLC, HUVECs overexpressing Ena/VASP exerted more force on the collagen, whereas Ena/VASP-inactivated cells were less contractile than controls ().
Cell–cell and cell–matrix contacts attach to the actin cytoskeleton and are influenced by force exerted on them through actomyosin contraction. Although the recruitment of adherens and tight junction proteins to cell–cell contacts was not altered by Ena/VASP activity, the gross morphology of the junction was affected. In less contractile Ena/VASP-inactivated HUVECS, staining for junctional marker proteins VE-cadherin, ZO-1, and β-catenin revealed junctions that were slightly smoother than in controls, whereas Ena/VASP overexpression resulted in a striated appearance of junctions, which is indicative of increased force generation ( and ). These results show that the shape of cell–cell junctions is altered by Ena/VASP activity, presumably through an increase in actomyosin contraction. The application of tension upon cell junctions by the cytoskeleton is necessary for the formation and maintenance of junctions in a process known as force strengthening. Thus, reduction of actomyosin contraction in the absence of Ena/VASP may interfere with force strengthening, resulting in the observed barrier function defects.
c
e
l
a
c
k
i
n
g
E
n
a
/
V
A
S
P
p
r
o
t
e
i
n
s
d
e
v
e
l
o
p
e
d
e
m
a
a
n
d
h
e
m
o
r
r
h
a
g
e
,
w
h
i
c
h
l
e
a
d
t
o
e
m
b
r
y
o
n
i
c
l
e
t
h
a
l
i
t
y
.
E
d
e
m
a
a
n
d
h
e
m
o
r
r
h
a
g
e
c
a
n
b
o
t
h
r
e
s
u
l
t
f
r
o
m
d
i
s
r
u
p
t
i
o
n
i
n
e
n
d
o
t
h
e
l
i
a
l
b
a
r
r
i
e
r
f
u
n
c
t
i
o
n
,
w
h
i
c
h
w
e
s
h
o
w
i
s
d
e
c
r
e
a
s
e
d
u
p
o
n
E
n
a
/
V
A
S
P
i
n
a
c
t
i
v
a
t
i
o
n
i
n
a
c
e
l
l
-
b
a
s
e
d
b
a
r
r
i
e
r
f
u
n
c
t
i
o
n
a
s
s
a
y
.
B
a
r
r
i
e
r
f
u
n
c
t
i
o
n
d
e
p
e
n
d
s
o
n
t
h
e
f
o
r
m
a
t
i
o
n
a
n
d
m
a
i
n
t
e
n
a
n
c
e
o
f
c
e
l
l
–
c
e
l
l
c
o
n
t
a
c
t
s
,
w
h
i
c
h
i
n
t
e
r
a
c
t
w
i
t
h
o
n
e
a
n
o
t
h
e
r
a
n
d
t
h
e
u
n
d
e
r
l
y
i
n
g
c
y
t
o
s
k
e
l
e
t
o
n
.
L
o
s
s
o
f
E
n
a
/
V
A
S
P
a
c
t
i
v
i
t
y
r
e
d
u
c
e
s
s
t
r
e
s
s
f
i
b
e
r
f
o
r
m
a
t
i
o
n
,
i
m
p
a
i
r
s
s
h
e
a
r
s
t
r
e
s
s
r
e
s
p
o
n
s
e
,
a
n
d
i
n
h
i
b
i
t
s
a
c
t
o
m
y
o
s
i
n
c
o
n
t
r
a
c
t
i
o
n
.
R
e
l
a
x
a
t
i
o
n
o
f
t
h
e
a
c
t
i
n
c
y
t
o
s
k
e
l
e
t
o
n
i
n
t
h
e
a
b
s
e
n
c
e
o
f
E
n
a
/
V
A
S
P
a
c
t
i
v
i
t
y
m
a
y
r
e
s
u
l
t
i
n
w
e
a
k
c
e
l
l
–
c
e
l
l
j
u
n
c
t
i
o
n
s
t
h
a
t
d
o
n
o
t
s
t
r
e
n
g
t
h
e
n
a
p
p
r
o
p
r
i
a
t
e
l
y
,
c
u
l
m
i
n
a
t
i
n
g
i
n
t
h
e
b
a
r
r
i
e
r
f
u
n
c
t
i
o
n
d
e
f
e
c
t
s
o
b
s
e
r
v
e
d
i
n
m
m
v
v
e
e
m
i
c
e
.
The generation of Ena/VASP-null (mmvvee) mice has been previously described (). All animal work was approved by the Massachussetts Institute of Technology Committee on Animal Care. Because a single allele of Mena was sufficient to prevent the described vascular defects, breeding pairs were selected to generate MMvvee or Mmvvee littermate controls.
Bouin's fixed tissues were embedded in paraffin, sectioned, and stained with hematoxylin and eosin using standard techniques. Sections were imaged with a microscope (Eclipse TE300; Nikon) equipped with a digital camera (Spot Flex; Diagnostic Instruments, Inc.). Whole mount staining was performed using a standard protocol. In brief, E10.5 embryos were fixed in paraformaldehyde, permeabilized in methanol with DMSO, and blocked with hydrogen peroxide. Anti-PECAM1 was used at a 1:20 dilution (BD Biosciences) and visualized using horseradish peroxidase (Vectastain Elite ABC kit; Vector Laboratories). Stained embryos were imaged using a stereoscopic zoom microscope (SMZ-U; Nikon).
Tissues were fixed with 2.5% glutaraldehyde and 2.5% formaldehyde in 0.1 M sodium cacodylate–HCl, pH 7.2, for 60–90 min at 4°C, postfixed with 2% osmium tetroxide for 60 min at 4°C, dehydrated in up to 70% graded alcohol, and en bloc stained with 0.2% uranyl acetate in 70% alcohol for 60 min at 4°C. The samples were then processed for epon embedding. Thin epon sections (400 nm) were cut with a diamond knife, poststained with uranyl acetate and Reynold's lead citrate, and viewed with an electron microscope (G Spirit BioTWIN; Tecnai) operated at 80 kV. Digital images were taken with a camera (2k CCD; Advanced Microscopy Techniques; provided by the Harvard Medical School Electron Microscopy Facility).
Primary endothelial cells were isolated from the aorta dissected from E14.5 embryos. Aorta were rinsed in HBSS without calcium and magnesium and treated with 1 mg/ml collagenase (Sigma-Aldrich) in primary endothelial media for 15 min at 37°C. Collagenase was removed with a PBS rinse and aorta were treated with 0.25% trypsin-EDTA (Invitrogen) for 15 min at 37°C. Aorta were then triterated and passed through a 70-μm cell strainer (BD Biosciences), and dissociated cells were plated directly on collagen I–coated coverslips. Primary endothelial cells were cultured at 37°C in RPMI supplemented with 10% fetal calf serum, penicillin/streptomycin, 1 μg/ml hydrocortisone, 10 U/ml heparin, 1 μg/ml DMSO, and 0.1 mg/ml endothelial cell growth supplement (Sigma-Aldrich) and used for experiments within 48 h. HUVECs (Cascade Biologics) were maintained on collagen I– coated dishes in endothelial basal media–2 (Clonetics). Cells expressing EGFP control, EGFP-Ena/VASP proteins, EGFP-FP4-Mito, and EGFP-AP4-Mito constructs were generated using vesicular stomatitis virus G–pseudotyped retroviruses as previously described (; ). Infected HUVECs were FACS sorted for equivalent levels of expression and used in cell-based assays between passages five and eight.
HUVECs were seeded at confluence in fibronectin-coated Ibidi μ-slide VI flow through chambers (Integrated BioDiagnostics). 10 dynes/cm of laminar shear stress was applied to cells for 16 to 20 h by flowing media through a loop consisting of a media reservoir, peristaltic pump, and compliance chamber. Cells were fixed in warm 4% paraformaldehyde and stained with Alexa-phalloidin. Images were captured with a microscope (Eclipse TE300) using a Plan Fluor objective (10×/0.3; Nikon).
HUVECs and primary mouse endothelial cells were plated on collagen-coated coverslips, and the rate of G-actin incorporation at barbed ends was determined using a previously described technique (). In brief, cells were permeabilized with 0.125 mg/ml saponin in the presence of 0.5 mM Alexa 568–conjugated G-actin (Invitrogen). After a 2-min labeling period, samples were fixed in 0.5% glutaraldehyde, permeabilized with 0.5% Triton X-100, and blocked in the presence of Alexa 350–phalloidin (Invitrogen). Images were acquired with a microscope (Eclipse TE300) using a Plan Flour objective (40×/1.30 DIC; Nikon). Cellular F- and G-actin content was quantitated using Metamorph software.
HUVEC cultures were maintained on collagen-coated 10-cm plates at confluence for 3 d, lysed in SDS sample buffer, sonicated for 10 s, and subjected to SDS-PAGE. Western blot analysis was performed using antibodies to total and Serine19 phosphorylated MLC 2 (Cell Signaling Technology). Blots were probed with anti-RhoA (Cytoskeleton, Inc.) and anti-actin (Chemicon) antibodies to ensure equivalent loading.
10 cells/ml of HUVECs were suspended in 2 mg/ml of type I collagen and cast in 200-μl discs as previously described (). Projected areas of collagen gel discs were measured immediately after release from molds and after 2 d of culture in endothelial basal media supplemented with 50 ng/ml VEGF (R&D Systems), basic fibroblast growth factor (R&D Systems), and phorbol myristate acetate (Sigma-Aldrich) for 2 d.
HUVECs were seeded at confluence on polycarbonate transwell membrane inserts (Corning 3402) and cultured for 3 d. 70 kD of Texas red–dextran (Invitrogen) was added to the top chamber at 2 mg/ml, and its movement into the bottom chamber was monitored over 4 h by spectrophotometer.
Statistical differences between two conditions were determined using test. For multiple conditions, means were compared by analysis of variance. All data found to be significant (P < 0.05) by analysis of variance were compared with Tukey's honestly significant difference post hoc test to reveal statistically different groups.
Fig. S1 shows the localization of endogenous VASP at cell–cell junctions and stress fibers in HUVEC and primary endothelial cell cultures. Online supplemental material is available at . |
Dynamic control of membrane curvature is vital for cells, and protein complexes governing the formation of membrane vesicles use various means of curvature regulation to guide vesicle shape (; ). Apart from coated vesicles formed by highly organized multiprotein complexes (; ; ), mechanisms of geometry creation by other, often much simpler protein ensembles are largely unknown. Budding of enveloped viruses is generally governed by only one dedicated matrix protein, though components of intracellular budding machinery have reportedly been involved (). Matrix proteins generally form the tight lining beneath the viral membrane, indicating their direct interactions with the membrane (). Accordingly, matrix proteins of different viral families have been found to be sufficient to orchestrate membrane budding in cells; their expression and self-assembly on the plasma membrane result in the release of viruslike proteolipid vesicles into the extracellular space (; ). Thus, matrix proteins directly guide membrane curvature by an internal protein lattice, the topological antipode of conventional protein coats that shape intracellular transport vesicles.
The clustering of membrane-associated proteins that are critically involved in budding (e.g., clathrin) generally results in crystalline ordering (; ). Correspondingly, polymerization of a solid protein scaffold that enforces a spherical topology on the vesicle membrane remains the most recognized mechanism of vesicle creation to date (; ). Nevertheless, the mechanisms of curvature creation might be different for vesicles formed by proteins integrated into the vesicle membrane (as opposed to on the membrane, which is common for external protein coats), such as in enveloped viruses or caveolae (; ; ). In this case, interaction between the lipid bilayer and proteins is generally coupled to membrane curvature (), resulting in membrane budding by mere component segregation, as shown in model systems (). Extreme protein crowding on caveolar or viral membranes (; ; ) also suggests involvement of direct protein–protein interactions in establishing the membrane shape. Yet it remains unclear whether such interactions lead to protein polymerization or the weaker fluid-type protein clustering that has been hypothesized to mediate budding by analogy with fluid lipid domains (; ).
To explore the mechanism of shape creation, we reconstituted membrane budding with purified matrix protein, the key structural component of the envelope of Newcastle disease virus (NDV). As for many paramixoviruses, matrix protein of NDV (M protein) plays a key role in virus formation (). M protein is absolutely required for viral egression and expression of this protein results in plasma membrane budding and production of viruslike particles by transfected cells (). The recently reported dependence of NDV formation on lipid rafts (), together with experiments showing direct interaction between M protein and pure lipidic membranes (; ), strongly indicates the synergistic action of M proteins and lipids in the formation of NDV envelopes. We found that the mere interaction of M proteins with the pure lipid bilayer is sufficient to induce self-organization of the proteins into functional budding domains.
The time-resolved admittance measurements technique, traditionally used to resolve detachment or fusion of small vesicles in cells (; ; ; ), was applied to monitor the activity of M protein on the lipid bilayer. We recorded changes in the electrical admittance of a patch, isolated from the planar lipid bilayer (made of a phosphocholine [PC]–phosphoethanolamine [PE]–cholesterol mixture; see Materials and methods) by a small pipette containing 2 μM of M protein. The membrane outside the patch area provided a lipid reservoir to support variations of the patch area. Changes of the imaginary (ΔIm) and real (ΔRe) parts of the admittance were detected ∼1 min after establishing a tight contact between the pipette and the membrane in 7 out of 15 patches (). The ΔIm tracing, which tracks changes in the patch area (), showed periodic variations, with each period consisting of a slow increase followed by a fast decrease of apparent membrane area. Such activity indicates formation of membrane buds (see ); during the initial rising stage the membrane area is retrieved from the lipid reservoir into the bud, whereas the fast area drop indicates its detachment. Excision of the membrane patch from the reservoir membrane led to the impairment of the ΔIm alterations and destabilization of the membrane patch, confirming that variations of ΔIm report changes in the patch area, requiring substantial lipid addition (an isolated patch membrane cannot store enough excess area for multiple bud formation). The periodic increases seen in the ΔIm tracing were not accompanied by any substantial changes of the permeability of any part of the membrane within the patch pipette (measured as membrane conductance at constant holding potential [G]; ; ).
Sharp drops of ΔIm () were often followed by transient rises of ΔRe, illustrating formation of a thin neck connecting the bud and membrane patch, as during the pinching-off of an endosome in a cellular system (, ; ). The amplitude of the ΔRe increase was usually much smaller than the one of the preceding ΔIm drop (), thus the value of the ΔIm jump approached total electrical capacitance of the bud membrane (see Materials and methods) proportional to the bud area. The cumulative distribution function of the values of ΔIm jumps is rather broad and skewed, with a pronounced singularity at ∼1.3 fF (188 jumps in total; ). This singularity breaks the distribution into two parts. Smaller jumps (, left of the yellow line) have normal size distribution, with a mean value of 0.92 ± 0.17 fF (SD, = 40; , left), corresponding to a membrane area of ∼0.1 μm (with specific capacitance of 10 fF/μm). The diameter of a spherical bud of such area is ∼180 nm, close to the typical sizes of an NDV particle (150–300 nm; ). Distribution of the larger jumps is close to log-normal, ranging from 200 to 500 nm consistently with the size heterogeneity of viruslike particules produced by M protein (). Increasing the M protein concentration in the pipette to 5 μM led to an overall increase of the values of ΔIm jumps to 2.7 ± 1.1 fF (SD, = 47; , right).
To directly assay shape transformations of the membrane patch, we visualized the activity of M protein on the membrane of giant unilamellar vesicles (GUVs; PC–PE–cholesterol mixture) containing a fluorescent lipid probe. A small patch of GUV membrane was isolated inside a pipette containing M protein. As in admittance measurement experiments, the membrane outside the patch area provided a lipid reservoir to support budding. demonstrates that shortly after establishing a stable contact between a GUV membrane and a pipette containing 2 μM of M protein, the fluorescence of the membrane patch inside the pipette increased sharply as the proteins adsorbed on the membrane (). The subsequent membrane rearrangements resulted in formation of round vesicles of different diameters visible near the patch, confirming the assumption on the spherical topology of the buds. The vesicles' sizes are more broadly distributed and generally larger than those observed on the planar lipid bilayer, likely because of differences in lateral tension for each lipid system (). Unidirectional budding of multiple vesicles demonstrates that the adsorbed proteins impose negative curvature on the membrane (here defined as the mean curvature of the membrane monolayer covered by proteins). With retrieval of the membrane area into the vesicles, the GUV diameter was progressively decreasing; thus, contact with the pipette did not interfere with lipid exchange between the external reservoir and the patch membrane. Finally, the GUV membrane detached from the pipette and multiple vesicles were seen moving inside the GUV ( and Video 1, available at ).
The moments of vesicle detachment were also resolved by admittance measurements. The scheme in illustrates the complete sequence of membrane budding and fission. First, the membrane bud closed (, red arrow), reflecting the abrupt narrowing of the membrane neck connecting the bud and membrane patch (; ). Afterward, the neck conductance (proportional to ΔRe; see Materials and methods) dropped below the level of resolution, indicating membrane fission (, blue arrow). This final drop was detected in a small fraction of trials (∼2%). Generally, ΔRe steadily decreased below the level of resolution (), likely because of the gradual elongation and/or thinning of the neck. Nevertheless, appearance of freely moving intralumenal vesicles () corroborates the ultimate fission of the vesicle necks.
Intralumenal vesicles were also efficiently formed when 4 μM of M protein was applied from a thin pipette and placed near a GUV by a weak pulse of positive hydrostatic pressure. Shortly after the protein application, changes in membrane fluorescence as well as membrane deformations were detected. They initially appeared as bright domains and invaginations associated with the GUV membrane ( and see ), and then transformed into intralumenal vesicles moving inside the original GUV (; and Video 2, available at ).
Vesicle formation was stimulated by cholesterol and membrane charge (). We compared the efficiency of M protein binding to large unilamellar vesicles (LUVs) containing different amounts of cholesterol and charge lipids. For all lipid compositions tested, a fraction of M proteins bound tightly to the LUV membrane (). The binding efficiency was not affected by the membrane charge (), corroborating earlier findings that M protein adsorption on the lipid bilayer is predominantly nonelectrostatic (). Thus, charge lipids enhance the budding activity of already bound M proteins. Cholesterol, however, stimulates both adsorption and budding activity of M protein. Notably, with addition of 30 mole fraction × 100 (mol%) of cholesterol, which doubles the bending rigidity of the GUV membrane (), the budding efficiency of M protein was not diminished but rather augmented (). This finding demonstrates that cholesterol, an abundant component in the NDV membrane, can actively participate in the virus budding (). Interestingly, the presence of PE also augments protein adsorption () and supports effective membrane budding (), likely through its intrinsic negative curvature.
Overall, vesicle formation and changes of membrane fluorescence were detected with four different batches of M protein (18 experiments total). No comparable changes were observed on GUVs perfused with the buffer containing no protein or 4 μM BSA (, right; and Video 3, available at ). We conclude that through interactions with the lipid bilayer, M protein implements the genetically encoded information required to create virus geometry.
To gain insight into the mechanism of curvature creation, we analyzed structural alterations in the lipid bilayer induced by M protein. Such alterations, correlated with membrane deformations, were first evident from the increase of fluorescence of membrane patches during vesicle budding (). A similar increase of membrane fluorescence is induced when M protein binds to LUV (), whereas BSA caused no effect at comparable concentrations (). Adsorption of M protein to LUV induced comparable dequenching of two different fluorescent probes, rhodamine (Rh)–dioleoyl-PE (DOPE) and boron dipyrromethane difluoride (BODIPY)–G, but did not alter the fluorescence of LUV containing nonquenched dyes (). The similar behavior of two chemically different fluorophores and the lack of influence of proteins on nonquenched dyes preclude specific interactions between the fluorophores and the protein. Furthermore, the increase of steady-state anisotropy of the BODIPY-G fluorescence upon M protein addition was detected for both quenched and nonquenched dye (Fig. S1, available at ), suggesting that membrane-associated proteins impose general constraints on lipid mobility ().
Changes in membrane fluorescence of LUV were detected only at relatively high protein concentrations sufficient to produce membrane deformation (see ). At those concentrations, we detected leakage of contents from LUV loaded with aqueous fluorescent markers, either small (8-aminonaphthalene–1,3,6–trisulfonic acid/p-xylene-bis-pyridium bromide [ANTS/DPX]) or large (70 kD FITC-dextran). Proteolytic treatment of M protein greatly impaired the release efficiency ( C and S2, available at ). Release efficiency was comparable for both markers for the same amount of the protein added (), in agreement with vesicle bursting. Previously, we established that M proteins did not form any conductive pathways in the lipid bilayer, such as proteolipid pores (, G tracings). Rather, the vesicles' rupture indicated membrane deformations induced by the protein. As liposome volume can be considered fixed at short time scales, substantial membrane deformations (e.g., membrane invaginations) tend to increase the surface/volume ratio of a liposome, thus stretching and ultimately rupturing the liposome membrane as in experiments on the osmotic rupturing of LUVs (). Thus, the bending of the lipid bilayer by M protein is generally correlated with an increase in both intensity and anisotropy of membrane fluorescence.
The relatively high protein concentration required to reconstitute M protein activity suggests that protein condensation in budding areas is the likely cause of fluorescence intensity. Indeed, viral M proteins assemble into a tight layer under the viral envelope and also can aggregate in vitro (). Here, the experiments on GUV containing Rh-DOPE in a self-quenched concentration directly demonstrate formation of distinct membrane domains. Shortly after protein application to GUV, bright spots formed within the original GUV contour (). The spots enlarged and merged as the GUV quickly deformed away from its initially spherical shape ( and Video 4, available at ). Some bright spots appeared as budlike membrane invaginations, similar to those observed with the M protein of vesicular stomatitis virus (). On deflated GUVs flattened on the coverslip, the bright spot either budded away as small vesicles or continued growing and merging in large circles ( and Video 5), which is behavior that has been previously described for fluidlike lipid domains (; Laradji and Sunil Kumar, 2005; ). The likely cause for the fluorescence dequenching in the domain areas is limitation of lipid mobility by membrane-associating M proteins (), which would impede energy exchange between the fluorophores.
Self-assembly of M proteins into circular domains on the lipid surface was further confirmed by EM observations. Circular patterns were detected after M protein adsorption on a lipid monolayer preformed on the air–water interface (; ). No such objects were detected in control experiments when only M protein or lipids were applied (not depicted). Though a circular shape () is a characteristic of fluids, similar patterns have also been detected for polymerized protein coats whose shape is defined by the polymerization pattern (). However, growing via merger that gives rise to a wide difference in domain sizes (from submicrometer clusters to micrometer-sized domains; ), in striking difference to the well-defined size of protein lattices (), requires internal fluidity of the domains. Thus, the generic tendency of M protein to self-aggregate (; ) is moderated on the membrane so that fluidlike proteolipid domains form.
The dynamics of vesicle formation observed by admittance measurements are consistent with the domain-driven mechanism of budding, originally proposed for fluid lipid domains (). Growing lipid domains destabilize and collapse into a closed vesicle, which might still remain attached via a thin neck, when the energies of both become comparable, closely resembling vesicle formation by M proteins (, arrow). The subsequent vesicle separation is triggered through instabilities in the domain boundary (; ) and doesn't require the participation of specialized fission proteins. A domain merger could account for large deviations in the size of vesicles produced by M protein; although the smaller vesicles would represent domains budding independently (, left), the larger vesicles result from a domain merger.
The formation of vesicles from fluid domains in a planar bilayer with high lateral tension σ (typically σ is ∼10 N/m; ) requires substantial energy to pull lipid material from the reservoir and bend it into a sphere. For a 100-nm vesicle, such energy would reach several thousand kT, where k is Boltzmann's constant and T is the temperature in degrees Kelvin (ΔF is ∼8π+ σS, where S is the vesicle area and the bending modulus is ∼20 kT). However, if M proteins are as tightly packed on the vesicle membrane as inside the virus, the number of proteins per 100-nm vesicle is >1,000 (e.g., at 0.05 protein/lipid ratio on the membrane surface; ). At such densities, the energy cost to pull material and bend it into a sphere per protein is low (approaching 1 kT). Thus weak interactions between proteins and lipids in the domains can combine to provide enough energy for curvature creation. This estimation corroborates the notion that the weak association of M proteins on the membrane can energetically support membrane deformations.
Besides providing the required energy, the same association of M proteins controls membrane geometry, producing membrane vesicles of the desired shape. Long-range coordination of membrane deformations required for vesicle formation is based not on the intrinsic topology of the protein lattice but on proteolipid interactions within the fluidlike budding domain. These interactions are manifested as intrinsic curvature of the domain, which is evident for unidirectional vesicle budding ( and ) and the line tension of the domain boundary. Both factors drive membrane curvature, creating viruslike membrane vesicles from the pure lipid bilayer. Although fluid domain–driven budding is generally sensitive to various membrane parameters (, , ), we demonstrated that vesicle populations with a narrow size distribution indeed can be obtained (; ; ). Thus, despite its intrinsic simplicity, weak protein condensation on a membrane surface provides a powerful tool to regulate membrane shape and topology.
M protein was purified from the “clone 30” strain of NDV as described previously (), with 5 mM Ca added to all buffers used during purification. The obtained M protein pellet was dissolved in 1 M KCl, 20 mM Hepes, and 0.2 EDTA, pH 7.4. Concentration of the protein was measured by BCA Protein Assay kit (Thermo Fisher Scientific). All of the experiments were conducted in 100 mM KCl, 20 mM Hepes, and 0.2 mM EDTA, pH 7.4 (buffer A). G ganglioside conjugated with BODIPY-FL (Invitrogen) in the polar head region (BODIPY-G) was synthesized as described previously (). Dioleoyl-PE (DOPC), 1-palmitoyl-2-oleoyl-PE (POPC), DOPE, 1,2-dioleoyl-phosphoglycerol (DOPG), and DOPE–lissamine Rh B sulfonyl (Rh-DOPE) were obtained from Avanti Polar Lipids, Inc. The following lipid compositions were used (mole ratio is indicated): DOPC/DOPE/cholesterol, 58:28:10 (PC–PE–cholesterol); POPC (PC); POPC/cholesterol, 66:30 (PC–cholesterol); POPC/DOPG, 81:15 (PC + charge). All were supplemented with 4 mol% of Rh-DOPE or BODIPY-G. For experiments with nonquenched fluorophores in PC–PE–cholesterol, the amount of Rh-DOPE or BODIPY-G was decreased to 0.2 mol% and the amount of PC and PE was increased proportionally.
100-nm LUVs were prepared by extrusion in buffer A or buffer (osmotically balanced with A) containing ANTS/DPX or 70 kD FITC-dextran in self-quenched concentration, as described previously (). GUVs were prepared by electroformation using platinum wire electrodes (Goodfellow Metals; ). The electroformation was performed in sucrose buffer, osmotically equilibrated with buffer A. The resulting GUVs were either detached from the electrode and put in buffer A or left on the electrode and perfused with buffer A.
5 μM of M protein was incubated for 5 min with LUVs of different lipid compositions at different protein/lipid ratios. The amount of LUV was normalized for the total fluorescence of Rh-DOPE incorporated. The LUV fraction was separated from unbound protein using the Ficoll gradient flotation method () and analyzed by SDS-PAGE using SYPRO Ruby protein gel stain (Invitrogen).
Leakage of ANTS or FITC-dextran and changes of fluorescence intensity of Rh-DOPE or BODIPY-G after addition of the M protein to LUV was determined at ambient temperature by spectrofluorimetric measurements using a luminescence spectrometer (Aminco-Bowman SLM-2; Spectronic Instruments, Inc.). The normalized fluorescence intensity F was recalculated from integral fluorescence intensity of LUV as follows: F = (F−F)/(F−F), where F corresponds to F before the protein addition and F – to F after complete disruption of LUV (infinite dilution of the fluorophores) by detergent (0.1% of Triton X-100; Sigma-Aldrich). 380/520-nm excitation/emission wavelengths were used for ANTS/DPX signal detection, 550/590 nm for Rh-DOPE, 505/525 nm for BODIPY-G, and 490/520 nm for FITC-dextran.
The visualization of GUVs attached to the electrode was performed on an inverted microscope (Axiovert 200; Carl Zeiss, Inc.) using a 40×, 0.75 NA objective (ACHROPLAN; Carl Zeiss, Inc.). GUVs detached from the electrode were settled on the bottom of a 170-μm-thin glass 35-mm dish. The dishes were preincubated with 1 g/liter BSA for 1 min and thoroughly washed with buffer A to reduce GUV binding to the glass. The interaction of M protein with GUVs detached from the electrode was recorded using Axiovert 200 or Olympus IX-70 inverted microscopes both equipped with 150×, 1.45 NA objectives (Olympus). The images were digitized by CoolSNAP EZ (Photometrics) or an intensified charge-coupled device camera (VE1000SIT; Dage-MTI) connected to IPLab (BioVision) or Metamorph Flashbus (MDS Analytical Technologies), respectively.
The analysis technique was adapted from . In brief, PC–cholesterol lipid solution in methanol/chloroform (9:1) was deposited on a buffer droplet. After 1-h equilibration, a carbon-coated gold EM grid (Electron Microscopy Sciences) was placed on top of the buffer droplet where the lipid monolayer has been formed. M protein was applied to the buffer and, after 1-h incubation, the grid was removed and stained with uranyl acetate (2% solution) for further observations with a transmission EM (Tecnai G2; FEI Company).
Planar lipid bilayers were prepared by the Mueller-Rudin technique from the PC–PE–cholesterol mixture in squalane and patch clamped as decribed previously (). Admittance measurements were performed using a patch clamp amplifier (Extracellular Patch Clamp 8; HEKA) and a PC-44 acquisition board (Signalogic) with on-board software lock-in () using a 5,000-Hz, 100-mV sinewave superimposed with −20 mV of holding potential.
= (Δ + (ωΔC − Δ
))/Δ ≈ Δ.
Online supplemental material describes measurements of the M protein purity and details of the protein enzymatic treatment (Fig. S1), as well as measurements of the steady-state anisotropy of BOPIPY-G fluorescence (Fig. S2). Video 1 shows M protein–driven vesicle formation from a membrane patch isolated from GUVs by a patch pipette. Video 2 shows formation of such vesicles by transient protein application to a GUV, whereas Video 3 shows no effect of BSA application. Videos 4 and 5 show temporal and spatial changes in membrane fluorescence induced by M proteins on GUV. Online supplemental material is available at . |
Intracellular transport of membrane organelles is critical for various processes such as endocytosis (; ), secretion (), neuronal signaling (), and organization of endomembranes (). The driving force for intracellular transport is provided by organelle-bound molecular motors, which move cargo organelles along microtubules (MTs; motors of kinesin and dynein families) or actin filaments (AFs; myosin family motors; ). Experimental evidence suggests that MTs and AFs play distinct transport roles (; ). MTs generally serve as tracks for long-range transport, whereas AFs support the local movement of organelles (; ).
It has been shown that membrane organelles use both types of cytoskeletal tracks for transport. In a pioneering study, showed that membrane organelles in the cytoplasm extruded from squid axon could switch from moving along an MT to moving along an AF. Later studies demonstrated that mitochondria (), synaptic vesicles (), and pigment granules (; ) use both AFs and MTs for various aspects of transport. Although multiple approaches have been developed to study the regulation of transport along individual cytoskeletal tracks (MTs or AFs), the question of how the switching between the two major transport systems is regulated remains unknown. Unlike organelle movement along individual tracks, these events are impossible to reliably detect on the light microscopy level because of the high densities of MTs and AFs in the cytoplasm.
A classic model system for studies of the transport of membrane organelles along the two types of cytoskeletal tracks is melanophores, pigment cells whose major function is the redistribution of membrane-bounded pigment granules to ensure color changes in the animal (). Pigment granules are induced by intracellular signals to either aggregate at the cell center or redisperse uniformly throughout the cytoplasm. During these movements, pigment granules use both MT and AF tracks. It is believed that pigment aggregation occurs predominantly along MTs, whereas pigment dispersion involves a combination of MT- and AF-based transport, suggesting that the switching between the two types of cytoskeletal tracks has to be tightly regulated by signaling events. Because these types of movement occur uniformly in response to cell-wide stimuli, observation of pigment movements in these cells allows us to distinguish the contribution of each type of cytoskeletal tracks and to develop computational approaches to detect the events of switching between the two types of transport.
In this study, we used melanophores as a model system to develop a new approach to directly measure switching between AF- and MT-based transport using a combination of experimental measurements and computational modeling. This approach allowed us, for the first time, to measure the parameters that determine how fast pigment granules switch back and forth between the MTs and AFs (the transferring rate constants) and to determine how intracellular signals modify these parameters to control the predomination of one cytoskeletal transport system over the other.
To measure switching rate constants between two types of cytoskeletal tracks, we developed a two-step computational approach for modeling pigment transport in melanophores. As the first step, we used experimental particle-tracking measurements of pigment granule movement separately along MTs and AFs in response to pigment aggregation and dispersion signals. For the modeling of MT-based transport, we measured pigment granule trajectories in the presence of dispersion and aggregation stimuli and calculated granule displacement along the cell radius by projecting the granule tracks onto the radial lines drawn from the cell center to the periphery. Measurements were performed in cells with intact AFs or with AFs disrupted with latrunculin ().
).
; ). These constants define how many granules transfer from one state into another over a unit of time (1 s; see Materials and methods for a description of calculation methods). For the modeling of AF-based transport, we measured pigment granule trajectories in the presence of dispersion and aggregation stimuli in cells treated with nocodazole to disrupt MTs. Granule displacement was calculated as the linear displacement from the starting to the ending point of each granule trajectory. For modeling purposes, we assumed that AF-dependent transport could be quantitatively described as two-dimensional diffusion. The major parameters used for modeling of MT- and AF-based transport of pigment granules are summarized in Table S1 (available at ).
As the second step of modeling, we combined the computational models of the aforementioned individual MT and AF movements.
), which define how fast pigment granules transfer from MTs onto AFs and from AFs onto MTs.
and
values.
and
in computational simulations were systematically changed (see Materials and methods) to fit the reference plots obtained in the experiment ().
and
values for pigment aggregation and dispersion, which reproduced the redistribution of pigment density observed in an experiment.
and
were found to be in the range of 5–7 min and 0–0.005 min for dispersion and 3–5 min and 8.5–12 min for aggregation, respectively.
and
values during dispersion and aggregation:
(disp) = 6.5 min,
(disp) = 0.0025 min,
(aggr) = 4.5 min, and
(aggr) = 10.7 min. As seen in , the curves representing the dynamics of pigment redistribution in experiments () and simulations () closely match each other.
Comparison of the switching rate constants during aggregation and dispersion reveals several remarkable aspects of regulation of the switching between organelle transport along the two types of cytoskeletal tracks.
value that reflects the switching from MTs to AFs is similar during pigment aggregation and dispersion, suggesting that this type of switching is not affected by the pigment aggregation and dispersion stimuli.
(disp), is extremely low, suggesting that during dispersion, the transfer of pigment granules from AFs to MTs is negligible, and, thus, the transfer onto AFs is essentially irreversible.
undergoes a dramatic (∼10,000 fold) increase during aggregation, suggesting that the transfer from AFs onto the MTs is the major regulating factor that determines which cytoskeletal track will be used by each organelle for motility. Therefore, we conclude that transferring of pigment granules between MTs and AFs is regulated through a change in a single parameter, the rate for transfer onto MTs ().
We hypothesize that such an overwhelming increase or decrease in the transferring onto MTs is achieved by the cooperation of two independent mechanisms. The first mechanism involves control over the activities of pigment granule–bound molecular motors (), which move pigment granules along MTs (plus end–directed kinesin; ; ) and minus end–directed dynein () or AFs (myosin V; ). An increase in the activity of each motor type should drag pigment granules onto a specific cytoskeletal track and, therefore, lead to an increase of a corresponding switching rate constant. Because our data show that pigment granules transfer onto AFs with similar rates during aggregation and dispersion, we suggest that myosin V activity does not play a substantial role in the transferring regulation. We further suggest that a primary role in this process is played by the regulation of cytoplasmic dynein, whose activity sharply increases during aggregation and decreases during dispersion () and, therefore, correlates with the MT switching rate constant changes.
The second hypothetical mechanism implicates regulation of the properties of the cytoskeletal tracks. For the pigment aggregation to happen, pigment granules that move along AFs must be captured by MTs to begin the centripetal movement. Because pigment aggregation involves a striking increase in the switching onto MTs, it is possible that the properties of MT tracks themselves are changed to enhance their ability to bind pigment granules. These changes may involve changes in the MT dynamics, an increase in the MT density, or both. Another possibility involves regulation of the binding to MTs of the dynactin subunit p150, whose presence at the growing (plus) ends of MTs has been shown to play an important role in the interaction of MTs with membrane organelles such as pigment granules (). Multiple changes in the MT tracks may be combined to enhance the capturing of pigment granules.
The results of our study have global implications in the regulation of a wide variety of intracellular transport events that involve switching between MTs and AFs, including secretion, endocytosis, axonal transport, and positioning of membrane organelles in the cytoplasm (; ; ; ). This suggests that during these transport events, regulation of switching onto MTs may also play a critical part in determining the type of cytoskeletal track that will be used by each particular organelle. Although at present it is impossible to directly observe these events in living cells, future development of live cell imaging and improved resolution of in vivo microscopy will enable the detailed studies that will shed light on this important problem.
Fish melanophores were cultured from the scales of Black tetra () onto carbon-coated glass coverslips as described previously (). Pigment aggregation was induced with 5 × 10 M adrenalin. Pigment dispersion was induced by washing out adrenalin via five to six changes of fish tissue culture medium. In some experiments, 5 mM caffeine was introduced into the last washing solution to facilitate dispersion.
MT- and AF-dependent components of the motion were examined in separate sets of experiments. To examine the MT component, we tracked pigment granules in cells lacking AFs, which were obtained by treatment with the actin-disrupting drug latrunculin (), or we obtained MT-based movement parameters by tracking pigment granules in intact AF-containing cells by fitting a granule movement trajectory by a straight line to determine the MT axis (). Movement of a pigment granule along the MT axis was then analyzed by breaking the displacement into periods of uninterrupted runs to the MT minus end (to the cell center), to the plus end (to the cell periphery), and pauses using multiscale trend analysis algorithm, which was previously described (). The AF-dependent component of the motion was examined by measuring the distance between the initial and the final position on the pigment granule trajectory in cells with disrupted MTs ().
Changes in pigment levels over time were quantified from time sequences of bright-field images of melanophores stimulated to aggregate or redisperse pigment granules by measuring gray levels at five points distributed along the cell radius. Points were chosen by dividing the distance between the margin of the pigment aggregate and the cell margin into five equal intervals. Changes in pigment densities over time were calculated from percentages of gray levels between 0 (pigment level at the cell periphery in the dispersed state) and 1 (pigment level at the cell periphery in the aggregated state) averaged over 10 cells ().
The model is formulated in a continuous approximation in terms of a two-dimensional pigment density, (, ) (here and below, bold font is used to indicate vector quantities). If normalized to unity,it can also be regarded as a probability density function that determines the probability, (, )
, to find a granule at location at time . This full granule density, which was compared with the experimental data, is the sum of densities of granules bound to MTs and AFs: (, ) =
(, ) +
(, ).
(, ), in turn, consists of densities of states with a plus or minus MT motor activity or pauses:
(, ) =
(, ) +
(, ) +
(, ).
Overall pigment dynamics is governed by granule transport along MTs and AFs and, therefore, by the dynamics of individual states. The corresponding governing equations can be derived from granule mass conservation (; ) or as a probability master equation (). The formulation is based on assumptions outlined in the following paragraphs.
#text
xref
italic
sub
#text
bold
sub
italic
xref
disp-formula
#text
xref
italic
sub
disp-formula
#text
Given randomness in filament directionality and frequent switching from one filament to another, transport along AFs is described as diffusion, with the effective diffusion coefficient
().
), whereas other terms represent rates of transitions of the granule between the states shown schematically in (the term
describes the rate with which the granules in state
fall off the MT ends).
and
, the values of which are determined by the spatial organization of MTs. The time dependence of rate constants is limited to short (∼1 min) time intervals, during which the system switches from aggregation to dispersion or vice versa. The rate constants are assumed to have changed their values from initial to final after these short transients. Note that the transients occurring with the onset of dispersion are somewhat longer than those associated with the beginning of aggregation (Fig. S1 B, available at ).
The system of governing equations () is subject to boundary conditions at the inner and outer boundaries (the computational domain is depicted in Fig. S2 A). At the inner boundary, there is a net influx (outflow) of pigment granules into (out of) the lamella during dispersion (aggregation), whereas at the outer boundary, the net flux is zero.
and
(see Determination of k and k section).
(, ) and
(, ). In the case of dispersion, all variables have zero initial values.
(, 0) =
(, 0) =
(, 0) = 0 and
(, 0) = . In the experiments, the initial pigment distribution averaged over 15 scans in five cells was not completely uniform but rather well approximated as ∝ 1 + 0.2exp(−( −)/), with = 8 μm.
(, 0).
and
from the observed pigment dynamics, other model parameters must be tightly constrained. Constraining parameters simplifies because transitions between the MT-bound states occur much faster (seconds; ) than between MTs and AFs (minutes; ). The separation of time scales allows one to uncouple the MT-bound states and treat them as a quasi-closed subset that attains equilibrium on a fast time scale.
–
, can then be estimated separately from
and
using the data from single-particle tracking (). For this purpose, two types of cells were used: cells with intact AFs and cells with disrupted AFs.
,
, and
(Fig.
–
.
–
can be determined. The results shown in Fig. S1 B were obtained for the data presented in Fig. S1 A and were smoothed by fitting either to a constant (aggregation) or to a single exponential function (dispersion).
(i = 1–6) during dispersion is therefore approximated bywhere t = 1 min, and the values of
(the aggregation
value),
, and
(the steady-state dispersion value) as well as the time constants
are shown in Table S1 for cells with intact AFs; similar results were obtained for cells with disrupted AFs.
–
can be evaluated by conditional analysis of the particle-tracking data, without additional assumptions. In this method, the rate constants are calculated directly from conditional statistics on durations of a given state terminated by another given state and from frequencies of such events, but the available conditional data were less extensive. It is interesting that two very different methods gave similar results, as evident from Fig. S1 C.
at a certain time obtained by the two methods. If these values were equal, the corresponding point would lie exactly on the line = . The trend lines are close to the bisector, which indicates agreement.
and
, and for the effective diffusion coefficient of migration along the actin network,
(Table S1). The latter can be estimated from the average displacements, <Δ>, over a certain time interval, Δ, in fish melanophores with disrupted MTs (). For dispersion, <Δ> ≈ 1.1 μm over Δ = 14 s, and a rough estimate,
∼ <Δ>/Δ, yields the value of ≈0.09 μm/s, which is comparable with that measured in frog melanophores (). For aggregation, <Δ> ≈ 0.33 μm over the same time period, and
≈ 0.008 μm/s.
Other model parameters were either obtained from direct measurements (average radius of the cell, ≈ 57 μm, and radii of the aggregate ) or taken from previously published papers (; ). The latter applies to γ, the fraction of MTs that reach the cell periphery (Table S1). Note that γ is set to zero for dispersion to reflect the increased disassembly of MTs at the periphery during dispersion.
Finally, we determine a function that describes the granule influx at the inner boundary during dispersion. Because of pseudoequilibrium among MT-bound states, () =
() −
() ∝
().
at the inner boundary are determined by transitions to and from
and by the diffusion of
:where the effect of diffusion along AFs at = is described in a single-exponential approximation with
∼
/
and
It then follows from the above equations () that
∝ exp(−(
)( −
)) + δexp(−(
)( −
)), where −1/
and −1/
are the eigenvalues of the linear system.
= 1 min as before.
,
, and δ in Table S1 correspond to the set (
,
) that provides the best fit to the experimentally observed pigment dispersion (see Determination of k and k section).
and
(
=
()/(1 + γ); see ) to determine the parameter set that minimizes the difference between model predictions and the experimental measurements. Because of radial symmetry, the problem can be reduced to one spatial dimension.
(, ) =
(, ) (σ = 0, +, −, AF), both in the above equations () and in boundary and initial conditions.
The above equations () are linear and therefore can be treated analytically. However, because the unknown parameters enter both the equations and boundary conditions and are to be found from fitting the experimental data, numerical solution of the problem becomes the only practical option. The equations () have been solved numerically using a newly developed capability of the Virtual Cell computational framework to solve advection-diffusion-reaction equations (). The algorithm utilizes a hybrid method that switches between central difference and upwind discretization schemes for the advection term depending on the local Peclet number (). The method was validated extensively against exact solutions and through regression testing. The one-dimensional computational domain was sampled evenly using 81 nodes. This resulted in a mesh size of 0.5 μm in the case of dispersion and one of 0.55 μm in the case of aggregation. Integration was performed with 1-ms time steps. The error of the numerical solution is estimated to be <1.5%.
sub
sup
xref
#text
italic
sub
sup
xref
#text
Fig. S1 shows parameter constraints, and Fig.
and
. Table S1 contains parameters values. Online supplemental material is available at . |
italic
xref
#text
italic
xref
ext-link
#text
Kc cells were maintained in Schneider's medium with 10% FBS in the presence of penicillin and streptomycin (Invitrogen). HeLa cells were maintained in DME with 10% newborn calf serum in the presence of penicillin and streptomycin (Invitrogen) and were passaged for no more than 6 wk.
Kc cells (10/well in 384-well plates) were transfected with prealiquoted dsRNAs (0.5 μg/well) followed by a 48-h incubation to allow for protein turnover. Subsequently, cells were treated with 5 μg/ml dox (Sigma-Aldrich) for an additional 48-h incubation. Cell viability was determined by measuring cellular ATP levels (CellTiter-Glo luminescent cell viability assay; Promega). Positive hits for this screen were defined as dsRNAs that increased cellular ATP levels by at least two units of SD above the mean level of the plate, a z score of two, in duplicate.
Confirmation of these hits was performed using newly synthesized dsRNAs. cDNA was amplified from fly genomic DNA using primers optimally designed for RNAi (sequences provided by N. Ramadan, RNAi Screening Center, Harvard Medical School, Boston, MA). dsRNA synthesis was conducted according to the manufacturer's recommended protocol (MEGAscript; Ambion). Kc cells (2 × 10/well in 96-well plates) were transfected with dsRNAs (1 μg/well) followed by treatment with dox. Each dsRNA was tested in triplicate and in at least two independent experiments. Cell viability was determined as a ratio between ATP levels of cells treated with a given dsRNA in the presence of dox and control cells treated with the dsRNA only. Statistically significant protection (P < 0.05) against dox-induced cell death was determined by comparing the cell viability of cells treated with RNAi in the presence of dox with the cell viability of cells treated with dox alone. Fold protection against dox was determined as the ratio between the mean cell viability of cells transfected with dsRNA in the presence of dox and cells treated with only dox.
Kc cells were transfected in 96-well plates with the 62 dsRNAs that were confirmed in the primary screen followed by treatment with dox for 8 h to induce caspase activation. Each dsRNA was tested in triplicate and in at least two independent experiments. Caspase-3/7–like activity was quantified using a luciferin-labeled DEVD peptide substrate (Caspase-Glo 3/7 assay; Promega). Fold caspase-3/7–like activity was determined as the ratio between the mean caspase-3/7–like activity of cells transfected with dsRNA in the presence of dox and cells treated with dox only.
Kc cells were transfected in 96-well plates with individual dsRNAs targeting the 62 genes confirmed in the initial screen. After a 24-h incubation, these cells were transfected with 1 μg dsRNA for an additional 48-h incubation when cell viability was quantified (as described in the Rescreen section). Fold protection against RNAi was determined as the ratio between the mean cell viability of cells transfected with the indicated dsRNA and RNAi and the mean cell viability of cells transfected with RNAi only. For the caspase activation assay, Kc cells were transfected with dsRNA as indicated followed by transfection with dsRNA against and an additional 24-h incubation.
Low-passage HeLa cells were transiently transfected (5 × 10/well in 96-well plates) with a pool of four siRNAs unless otherwise indicated (50 nM; predesigned ON-TARGETplus siRNAs; Dharmacon) using HiPerFect transfection reagent (QIAGEN). After a 48-h incubation, cells were treated with dox (concentrations and incubation times are noted in figures). siRNAs were tested in triplicate for each independent experiment. Cell viability and caspase-3/7 activity was quantified as described for KC cells. Images were obtained using bright-field microscopy.
For detection of caspase cleavage, HeLa cells were transfected with siRNAs (1.5 × 10/well in six-well plates) followed by treatment with dox. Cells were lysed directly in SDS sample buffer and subjected to SDS-PAGE analysis. Caspases were detected by immunoblotting using the following antibodies: caspase-9 (R&D Systems), caspase-2 (Alexis), caspase-3 (Cell Signaling Technology), and cleaved caspase-3 (Cell Signaling Technology). hARD1 was detected using a rabbit polyclonal antibody (provided by M.C. Brahimi-Horn, University of Nice, Centre National de la Recherche Scientifique, Nice, France).
For hARD1 reconstitution experiments, HeLa cells were transfected with a pool of two siRNAs targeting the 5′ untranslated region of hARD1 (sequences CUGACUGCGCCUUCACGAUUU and GCUGACUGCGCCUUCACGAUU; Dharmacon) in six-well plates followed by a 24-h incubation. This was followed by transient transfection of C-terminal–tagged hARD1-/his using TransIT-LT1 transfection reagent (Mirus) and an additional 24-h incubation. After dox treatment, cells were lysed and analyzed by Western blotting.
Images of HeLa cells were obtained using a microscope (Eclipse TE300; Nikon) with a 10× Ph L objective lens and a camera (ORCA-ER; Hamamatsu) at room temperature. Openlab 3.1.7 acquisition software (Improvision) was used to analyze the images.
Fig. S1 shows a characterization of dox-induced cell death and controls for RNAi screen. Fig. S2 presents semiquantitative analysis of dsRNAs to validate the RNAi library. Table S1 provides a comprehensive list of genes identified in the primary screen. Online supplemental material is available at . |
xref
italic
#text
Our previous work on kinetochore assembly suggests BubR1 as a potential 3F3/2 antigen (). To test this possibility, we immunoprecipitated BubR1 from meiotic metaphase (cytostatic factor [CSF] arrested) and spindle checkpoint extracts (CSF extracts with addition of sperm chromosomes and nocodazole), followed by Western blotting with the 3F3/2 antibody. BubR1 purified from both extracts reacted with the 3F3/2 antibody to a similar extent, whereas treatment with λ-phosphatase removed the 3F3/2 signals (). Thus, BubR1 is a 3F3/2 antigen whose phosphorylation occurs before checkpoint activation. As Plk1/Plx1, which usually interacts with its substrates (), is the 3F3/2 kinase (; ), we tested whether BubR1 associates with Plx1. Indeed, endogenous Plx1 coimmunoprecipitated with BubR1 (), and recombinant polo box domain (PBD) from Plx1 also binds to endogenous BubR1 in extracts (unpublished data).
Next, we determined the kinase requirement for the formation of the kinetochore 3F3/2 phosphoepitope on sperm chromosomes. Endogenous kinetochore 3F3/2 epitopes on sperm chromosomes purified from checkpoint extracts were first dephosphorylated with λ-phosphatase and endogenous kinases inactivated by -ethylmaleimide (NEM; ). Chromosomes were then incubated with active recombinant kinases and formation of the 3F3/2 epitope was monitored by immunofluorescence staining. Even though Plx1 alone at high concentrations was sufficient to generate the 3F3/2 epitope in this assay (), Plx1 near its physiological concentrations (10 ng/μl Plx1; ) failed to reconstitute the 3F3/2 phosphoepitope (). This is not surprising, as recognition and phosphorylation of substrates by Plx1 frequently require a priming phosphorylation by another kinase, such as Cdk1/cyclin B, which generates a binding site for PBD (). Thus, we investigated the priming requirement for the 3F3/2 epitope. Although active Plx1 or Cdk1/cyclin B alone was not sufficient to generate the kinetochore 3F3/2 signals, a combination of the two kinases generated robust 3F3/2 signals (), suggesting that Cdk1 functions as a priming kinase to promote the 3F3/2 phosphoepitope.
We then reconstituted the 3F3/2 epitope in BubR1 in a purified system. First, we showed that both Cdk1 and Plx1 phosphorylated recombinant BubR1 (Fig. S1 A, available at ). To investigate the effect of priming phosphorylation on BubR1, we incubated recombinant BubR1 with unlabeled ATP in the presence or absence of Cdk1/cyclin B, followed by incubation with or without Plx1 in the presence of γ-[P]ATP and purvalanol A, a Cdk1 inhibitor (). Although Plx1 alone phosphorylated BubR1, BubR1 that had been incubated with Cdk1 was phosphorylated fourfold more efficiently by Plx1 (). Thus, priming phosphorylation of BubR1 by Cdk1 enhances its phosphorylation by Plx1.
Second, we identified the amino acid residue important for priming phosphorylation. PBD recognizes the S-S/T-P sequence in mitotic Plx1 substrates (; ), in which the Ser/Thr residue preceding Pro is phosphorylated. Sequence analysis indicated that there is only one S-T-P (aa 604–606) conserved between human and BubR1 (Fig. S1 B). We found that priming phosphorylation of BubR1-T605A by Cdk1/cyclin B had no effect on the level of its subsequent phosphorylation by Plx1, in contrast to the wild-type BubR1 (). Thus, Thr 605 is a critical site for priming phosphorylation of BubR1 by Cdk1.
Third, we reconstituted the 3F3/2 epitope in BubR1 in vitro. Recombinant BubR1 and BubR1-T605A were phosphorylated with Cdk1/cyclin B and/or Plx1, followed by Western blotting with the 3F3/2 antibody (). Either kinase alone was not sufficient to form the 3F3/2 epitope. The presence of both kinases robustly generated the 3F3/2 epitope on wild-type BubR1, but not on BubR1-T605A. Furthermore, generation of the 3F3/2 epitope does not require the kinase activity of BubR1 (; ). Thus, Cdk1 and Plx1 act synergistically to generate the 3F3/2 epitope in a Thr 605–dependent manner. Consistent with this, recombinant GST-BubR1, but not GST-BubR1-T605A or GST-BubR1-T605E, coprecipitated with the endogenous Plx1 in CSF extracts (). Thus, Thr 605, likely through its phosphorylation, facilitates the BubR1–Plx1 interaction.
Lastly, we determined whether BubR1 is the 3F3/2 antigen in checkpoint extracts. CSF extracts were first immunodepleted of endogenous BubR1 and then incubated with recombinant BubR1, BubR1-T605A, or BubR1-T605E together with sperm chromosomes and nocodazole. Chromosomes were purified onto coverslips and underwent the assay of rephosphorylation by Cdk1/cyclin B and Plx1, as described in . Depletion of BubR1 reduced the kinetochore 3F3/2 signals 30-fold, whereas addback of recombinant BubR1 rescued the signals (). However, the 3F3/2 signals failed to recover to a substantial extent in the BubR1-T605A and BubR1-T605E addback samples, even though both proteins were properly targeted to kinetochores. Thus, BubR1 is a physiological kinetochore 3F3/2 antigen, and Thr 605 is required for the generation of the 3F3/2 epitope. In contrast, the presence of residual 3F3/2 signals on BubR1-T605A/E– rescued kinetochores suggests that another kinetochore 3F3/2 antigen may exist whose phosphorylation or kinetochore localization is under the control of BubR1.
We next determined whether the Thr 605 phosphorylation is required for checkpoint arrest. BubR1 was depleted from CSF extracts to >95% (, lane 1), and recombinant BubR1, BubR1-T605A, or BubR1-T605E was then added back to endogenous levels (, lanes 2–4). Subsequently, sperm chromosomes and nocodazole were added to activate spindle checkpoint, and samples were then split and incubated with or without calcium. Aliquots were then taken at various times and assayed for Cdk1 kinase activity. In the absence of calcium, Cdk1 kinase activity in all extracts remained high, indicating a stable meiotic metaphase arrest (, bottom). Upon addition of calcium, which triggers the transition from meiotic metaphase into interphase, BubR1-depleted extracts entered interphase with a low Cdk1 activity, whereas addback of BubR1 maintained the high Cdk1 activity caused by the activation of the spindle checkpoint (; ; ; ). However, addition of BubR1-T605A or BubR1-T605E failed to prevent meiotic exit, indicating that Thr 605 in BubR1 is required for the checkpoint arrest. Mutations on Thr 605 did not nonspecifically inactivate BubR1 because of misfolding of the mutant proteins, as BubR1-T605A was as active as BubR1 in inhibition of APC-Cdc20 ().
We determined the cellular basis for the lack of checkpoint arrest. BubR1 controls the kinetochore localization of Plx1 and the checkpoint protein Mad2, as demonstrated in BubR1-depleted extracts (; ; ). Localization of Plx1 and Mad2 to kinetochores in BubR1-depleted extracts was recovered upon addition of recombinant BubR1, but not BubR1-T605A or BubR1-T605E, even though mutant BubR1 proteins were efficiently targeted to kinetochores (; and ). This lack of recruitment of Plx1 and Mad2 was not because of a global change in the outer kinetochore structure, as the checkpoint protein Mps1 was efficiently targeted to kinetochores in all analyzed extracts (; ). Thus, Thr 605 is specifically required for the checkpoint arrest and for the recruitment of Mad2 and Plx1 to kinetochores.
BubR1 is a kinase whose activity is required for checkpoint arrest (; ). Thus, we determined whether phosphorylation of BubR1 by Cdk1 and Plx1 affects its kinase activity as assayed by its autophosphorylation. Without prephosphorylation by Cdk1 and Plx1, the BubR1 kinase activity was undetectable (; ). However, prephosphorylation of BubR1, but not the kinase-dead BubR1, by Cdk1 and Plx1 increased its autophosphorylation activity at least 100-fold (). Thus, the Cdk1- and Plx1-mediated phosphorylation activates BubR1.
Next, we determined the relative contribution of Cdk1 and Plx1 and the role of Thr 605 in activation of BubR1 in a similar assay. Although prephosphorylation of BubR1 by Cdk1 or Plx1 alone enhanced its kinase activity to some extent, phosphorylation of BubR1 by both Cdk1 and Plx1 synergistically activated BubR1 by 10 or 30 times as compared with BubR1 prephosphorylated by Cdk1 or Plx1 alone, respectively (). This synergistic activation of BubR1 requires Thr 605 in BubR1, as the BubR1-T605A mutant only showed an additive activation by Cdk1 and Plx1 (). Thus, Cdk1 and Plx1 synergistically activate the kinase activity of BubR1 in a Thr 605–dependent manner.
Responses to checkpoint activation usually consist of a cell cycle arrest and repair of the cellular defects, which activates the checkpoint in the first place. In the case of the spindle checkpoint, lack of attachment or tension arrests cells in mitosis and alters the kinetochore structure to promote attachment and tension. It has been shown previously that the checkpoint protein BubR1 is not only essential for inhibition of APC/C (; ; ) but also regulates the kinetochore–microtubule attachment and chromosome congression (; ; ; ). We report here that both functions of BubR1 are coordinately regulated by Cdk1 through its phosphorylation on BubR1 Thr 605. First, this phosphorylation is required for checkpoint-mediated mitotic arrest. Mechanistically, phosphorylation on Thr 605 does not affect the in vitro inhibitory efficiency of BubR1 toward Cdc20-APC/C. Instead, this phosphorylation controls the targeting of Mad2 to kinetochores, which is essential for mitotic arrest.
Second, Cdk1-mediated phosphorylation promotes the formation of the 3F3/2 epitope. The 3F3/2 antigen is a mysterious kinetochore protein that has been hunted by cell biologists for over a decade (). The kinetochore 3F3/2 phosphoepitope is generated by Plk1/Plx1 in response to the lack of tension across sister kinetochores (; , ; ; ). We demonstrate here that the checkpoint protein BubR1 is the kinetochore 3F3/2 antigen, whose phosphorylation requires the synergistic action of Cdk1 and Plx1, as priming phosphorylation on BubR1 Thr 605 by Cdk1 is essential for the formation of the 3F3/2 signals. Surprisingly, the 3F3/2 phosphoepitope in BubR1 is observed in both CSF extracts and checkpoint extracts () but not in interphase extracts (unpublished data), indicating that formation of this epitope is tension independent. Thus, the mitotic/meiotic state provides a permissive environment for the formation of the 3F3/2 biochemical epitope, but its functional specificity to the lack of tension is likely determined by its kinetochore localization.
As Plk1 is not required for checkpoint arrest in human cells (; ; ), the 3F3/2 epitope is probably not involved in mitotic arrest but is likely to act in regulating kinetochore–microtubule interactions (; ; , ). Indeed, Plk1 promotes the assembly of the bipolar spindle and the generation of tension (; ; ). As the formation of the 3F3/2 epitope by Cdk1 and Plx1 drastically activates the kinase activity of BubR1, we speculate that the active BubR1 kinase controls the kinetochore structures and/or promotes microtubule attachment to kinetochores. Although our extract system precludes us from analyzing the exact physiological function of the 3F3/2 phosphoepitope in tension signaling, the molecular information on the 3F3/2 antigen and its kinase presented in this study nevertheless opens the door for future characterizations of the tension responses in culture cells.
Antibodies against Mad2, BubR1, Mps1, and Plx1, as well as 3F3/2 ascite, have been described previously (). Baculoviruses for Plx1 and Cdk1/cyclin B were provided by J. Maller (University of Colorado, Denver, CO) and H. Piwnica-Worms (Washington University, St. Louis, MO), respectively. Active recombinant Plx1 and Cdk1/cyclin B were expressed in Sf9 cells for 44 h and then treated with 250 nM okadaic acid for 4 h before harvesting. BubR1-T605A and BubR1-T605E mutants were generated by site-directed mutagenesis. Recombinant BubR1, BubR1-T605A, BubR1-T605E, and BubR1 kinase dead (K788R) were expressed in as GST fusion proteins and purified using glutathione agarose (GE Healthcare).
Meiotic metaphase extracts (CSF extracts) and checkpoint extracts from eggs and demembranated sperm nuclei were prepared as described previously (). CSF extracts that were either mock depleted or depleted of BubR1 were incubated with demembranated sperm nuclei and nocodazole. In rescue experiments, recombinant BubR1 proteins were added to depleted extracts before the addition of sperm nuclei and nocodazole.
Immunofluorescence images were captured at 23°C on a microscope (Axiovert 200M; Carl Zeiss, Inc.) using an oil-immersion objective lens (100× 1.4 NA; Plan-Apochromat), a digital charge-coupled device camera (Orca-ER; Hamamatsu Photonics), and Openlab 5.0.1 (Improvision). For quantitative comparison of fluorescence intensities, images were acquired and processed identically. In , the intensity of each kinetochore 3F3/2 signal was determined relative to that of BubR1 obtained from the same kinetochore.
Chromosomes were purified onto coverslips from spindle checkpoint extracts (). To remove the 3F3/2 phosphoepitope and inactivate endogenous kinases, coverslips were incubated with λ-phosphatase (New England Biolabs, Inc.) and subsequently treated with NEM (). Coverslips were then incubated with 5 ng/μl His-Plx1 and/or 3 ng/μl Cdk1/cyclin B in the kinase reaction buffer (KRB) (20 mM Hepes, pH 7.8, 15 mM KCl, 10 mM MgCl, 1 mM EGTA, 0.5 μM microcystin LR, and 0.1 mg/ml BSA) supplemented with 2 mM ATP for 1.5 h at room temperature. Coverslips were next stained with the 3F3/2 ascite at a 1:8,000 dilution at 4°C overnight, with the Alexa Fluor 594 and 488 secondary antibodies (Invitrogen) for 1 h at room temperature, and with DAPI.
Phosphorylation of BubR1 by Cdk1 or Plx1 was performed in a total volume of 10 μl KRB for 30 min at room temperature, using 1 μg BubR1 and 0.2 mM ATP, with either 250 ng Cdk1/cyclin B or 30 ng His-Plx1. The radiolabeling assay was performed similarly, except in the presence of 2 μCi γ-[P]ATP.
To assay BubR1 autophosphorylation in , 550 ng of recombinant BubR1 was first phosphorylated by Cdk1/cyclin B in KRB supplemented with 0.2 mM of unlabeled ATP for 75 min at 25°C, and then by Plx1 for 45 min in the presence of 65 nM purvalanol A, a potent inhibitor of Cdk1/cyclin B (). The Plx1 kinase was then removed from phosphorylated BubR1 by incubating with anti-Plx1 antibody/protein A beads for 1 h at room temperature. Subsequently, phosphorylated BubR1 was assayed for autophosphorylation by incubating in KRB plus 0.2 mM ATP, 2 μCi γ-[P]ATP, 65 nM purvalanol A, and 0.1 mg/ml ovalbumin for 40 min at room temperature. Control experiments were done in parallel using GST protein as a substrate to undergo prephosphorylation by Cdk1/cyclin B and Plx1. After subsequent depletion of Plx1, the control samples were mixed with 550 ng of recombinant BubR1, which had not been phosphorylated by either kinase, and assayed for BubR1 autophosphorylation.
Cdk1 and Plx1 phosphorylate BubR1 in vitro. Online supplemental material is available at . |
B cell lymphoma 2 (Bcl-2) family members are critical regulators of programmed cell death (). Proteins of this family share homology in four conserved regions termed Bcl-2 homology (BH) domains and can be divided into anti- and proapoptotic proteins. The antiapoptotic proteins Bcl-2, Bcl-x, Bcl-w, myeloid cell leukemia (Mcl) 1, and Bfl-1/A1 are characterized by all four BH domains. Proapoptotic homologues can be further subdivided into two subfamilies. The multi–BH domain Bax homologues, including Bax, Bak, and Bok/Mtd, contain BH1–3, whereas the proteins of the BH3-only subfamily, which comprise Bad, Bid, Bim, Bmf, Puma, Noxa, Nbk/Bik, and Hrk, only share the BH3 domain. BH3-only proteins are essential initiators of apoptosis, and once activated, they regulate the ability of the multi–BH domain members Bax and Bak to undergo a conformational switch and to oligomerize in the outer mitochondrial membrane (; ). Activated Bax and Bak then induce mitochondrial membrane permeabilization and subsequent release of proapoptotic factors, e.g., cytochrome , from the intermembrane space into the cytosol. Cytosolic cytochrome induces formation of the apoptosome and ultimately triggers execution of the intrinsic apoptosis signaling cascade.
Deregulation of Bcl-2 family proteins has been implicated in the development of many malignancies (). In addition to deregulated expression of antiapoptotic Bcl-2 (), disruption of multidomain proapoptotic Bcl-2 homologues is also critically involved in tumorigenesis. Loss of Bax is a frequent event in human cancer and is related to tumor progression, poor prognosis, and clinical resistance to anticancer therapy (; ). Furthermore, loss of the gene contributes to oncogenic transformation and tumor development in mice (). Recently, it has been shown that inactivation of BH3-only proteins is implicated in human cancer (). In line with a tissue-specific expression of the BH3-only protein Nbk/Bik (; ) with strong expression in the kidney (), loss of Nbk and the more broadly expressed Bim is a common feature of clear cell renal cell carcinoma (; ).
Bcl-2 family members can form homo- and heterodimers with other members of the protein family. It is considered that the BH3 domain constitutes an amphipathic α helix, which binds to a hydrophobic groove formed by the BH1, BH2, and BH3 domain of antiapoptotic Bcl-2 family members. Binding to and inactivation of antiapoptotic Bcl-2 family members is crucial for BH3-only proteins to initiate apoptosis (). The ability of proapoptotic Bcl-2 family proteins to interact with their antiapoptotic siblings led to the so-called rheostat model: the ratio of pro- to antiapoptotic Bcl-2 members determines the apoptotic fate of the cell (). A recent study revealed, however, that BH3-only proteins selectively bind to specific sets of prosurvival proteins and that only certain pairs associate with each other under physiological conditions (). These complementary binding profiles implicate two classes of antiapoptotic Bcl-2 proteins, one comprising Mcl-1 and Bfl-1/A1, the other Bcl-2, Bcl-x, and Bcl-w. It has been suggested that efficient induction of apoptosis depends on neutralization of both classes of prosurvival proteins.
Moreover, numerous studies show a central role of Bax rather than Bak in Bcl-2–regulated cell death. This is especially true for Puma and Nbk, which both have been shown to act via a Bax-dependent/Bak-independent pathway (; ; ). There is also accumulating evidence that distinct BH3-only proteins act specifically through the activation of Bak (; ). We have shown that disruption of Bax is sufficient to confer resistance to Nbk under conditions of abiding Bak expression. This indicates that induction of cell death by Nbk is mediated specifically via a Bax-dependent and apparently Bak-independent pathway (). In this light, the role of Bak and putative Bak inhibitors in Nbk-induced apoptosis remains enigmatic. Thus, we cannot rule out the possibility that loss of Bax protects cells just by decreasing the amount of Bax/Bak-like molecules under a critical threshold, which would be necessary for Nbk to induce apoptosis.
Here, we show that Bak is fully functional and sufficient to mediate apoptosis by specific activators, e.g., Bcl-x. Moreover, evidence is provided for the mechanism underlying Nbk resistance in Bax-deficient cells. Although Nbk triggers Bax activation, it nevertheless fails to relieve Mcl-1–mediated inhibition of Bak. Moreover, Mcl-1 protein levels are increased upon Nbk expression. By this mechanism, Nbk enforces inhibition of Bak by Mcl-1. Collectively, these results provide a molecular rationale for the Bax dependency of the BH3-only protein Nbk.
We have shown previously that loss of Bax protects cancer cells from apoptosis induced by the BH3-only protein Nbk (). To further investigate the mechanism of Nbk- induced cell death and to study the regulation of the Bax homologous protein Bak in detail, we induced expression of a myc-tagged Nbk protein in the parental wild-type (wt) HCT116 cell line and in isogeneic cell lines devoid of Bax (Bax) or Bak (Bak) expression alone or of both (Bax/Bak; ). Loss of protein expression was achieved by knockout of the Bax gene () and Bak knockdown by short hairpin RNA (), respectively.
Transduction of these cell lines with an adenoviral vector, Ad-mycNbk-tTA, for the regulated expression of Nbk leads to high levels of Nbk protein in the absence of doxycycline, i.e., under on conditions, whereas expression of the transgene was almost completely repressed by the addition of doxycycline to the culture medium, i.e., under off conditions (). Western blot analyses showed that forced expression of Nbk is accompanied by cleavage of the initiator caspase-9 and the effector caspase-3 in HCT116 wt and in HCT116 Bak (knockdown) cells. In contrast, neither caspase-9 nor caspase-3 cleavage could be detected in HCT116 cells devoid of Bax expression, regardless of whether or not they expressed Bak. This is not caused by lower Nbk expression because Western blot analyses affirmed equal expression levels of Nbk under on conditions in all four cell lines.
To confirm that loss of Bak does not affect regulation of apoptosis upon Nbk expression, we performed flow cytometric analyses to quantify fragmentation of genomic DNA. Apoptotic cells were identified as cells with hypodiploid, i.e., sub-G1, DNA content. After 24 h of transduction with Ad-mycNbk-tTA, a mean of ∼40% of the HCT116 wt and 45% of the HCT116 knockdown cells became apoptotic under on conditions (). In sharp contrast, identical transduction with Ad-mycNbk-tTA failed to significantly induce apoptosis in Bax knockout and Bax/Bak double deficient HCT116 cells. These results establish that, at least in Nbk-induced apoptosis, Bax and Bak do not exert redundant functions. Whereas Bak knockdown does not affect Nbk-induced apoptosis, loss of Bax efficiently protects HCT116 cells from Nbk-induced apoptosis. Furthermore, the presence of Bak does not influence the proportion of cells undergoing Nbk-induced apoptosis, regardless of whether or not they express Bax.
To corroborate that all death occurs by apoptosis and to control viability, we analyzed cells induced to express Nbk by propidium iodide (PI)/Annexin V–FITC staining. As in the case of the DNA fragmentation analysis, Nbk induced apoptosis in the HCT116 wt cells and in Bak-deficient HCT116 cells, as analyzed by detection of phosphatidylserine exposure and PI negativity (Fig. S1, available at ). Under on conditions, 41% of wt cells were detected as Annexin V positive/PI negative (early apoptotic) and an additional 13% showed Annexin V/PI positivity (late apoptotic) at 24 h after transduction with Ad-mycNbk-tTA. In contrast to Bak knockdown cells, Bax-deficient cell lines displayed <6% of Annexin V–positive/PI-negative cells and only 6% of double positive cells.
To further address the role of Bax and Bak and of mitochondria in Nbk-induced apoptosis, we determined the release of cytochrome and dissipation of the mitochondrial membrane potential upon exposure to Ad-mycNbk-tTA. Western blot analysis of cytosolic extracts obtained at 24 h after infection with Ad-mycNbk-tTA showed that Nbk expression induces the release of cytochrome in both HCT116 wt and HCT116 Bak cells but not in Bax-deficient cell lines ().
Using the potential-sensitive dye JC-1, we used flow cytometry to analyze whether release of cytochrome is accompanied by mitochondrial permeability transition and loss of mitochondrial membrane potential (ΔΨ). This revealed loss of ΔΨ in Bax-proficient cells upon mycNbk expression, regardless of the presence or absence of cellular Bak. This loss of ΔΨ occurred in a mean of 42 and 62% of the HCT116 wt and HCT116 Bak cells, respectively. Conversely, only 16% of the Bax knockout and 10% of Bak/Bax double deficient cells exhibited decreased mitochondrial membrane potential upon Ad-mycNbk-tTA transduction under on conditions (). Compared with HCT116 wt cells, induction of cell death by Nbk is slightly increased in Bax cells (, S1, and S2, available at ). This is not caused by a decrease in the level of antiapoptotic proteins such as Bcl-2 or Bcl-x (not depicted), but it seems to reflect the increased Bax expression level in these cells ().
These data demonstrate that loss of Bax, but not of Bak, expression protects HCT116 cells from Nbk-induced apoptosis. Nevertheless, regarding the simple rheostat model of a balance between pro- and antiapoptotic proteins, we could not rule out that loss of Bax protects cells just by decreasing the amount of proapoptotic multi–BH domain proteins under a critical threshold necessary for Nbk to induce apoptosis. Thus it might be that, in Bax-deficient cells, the amount of Bak protein is not sufficient to mediate apoptotic signals by Nbk. To test if increased expression levels of proapoptotic multi–BH domain proteins can sensitize cells for Nbk-induced apoptosis, we established cell lines in which Bax or Bak is constitutively overexpressed. The parental cell line DU145 does not express Bax because of a frameshift mutation in the gene, whereas is not mutated and endogenous Bak is expressed to a moderate extent. By use of a retroviral vector, HyTK-Bax, we established DU145 cells stably reexpressing the Bax-α cDNA under the control of a cytomegalovirus (CMV) promoter () and a Bax-negative vector control cell line (). Infection of these cell lines with Ad-mycNbk-tTA resulted in comparable levels of Nbk expression in both cell lines under on conditions ().
Flow cytometric analysis of DNA fragmentation revealed that Nbk expression failed to induce apoptosis in the Bax-negative control cells. Despite strong Nbk expression, only 7% of the cells showed hypodiploid DNA content, compared with 5 and 6% under control and off conditions, respectively. In contrast, Bax-expressing DU145 cells readily underwent apoptotic DNA fragmentation with 39% of Bax-reexpressing DU145 cells showing sub-G1 DNA content (). This confirms the Bax dependency of Nbk-induced cell death observed in the HCT116 system.
In parallel, we generated DU145 cells stably overexpressing Bak by the use of a CMV promoter–driven pcDNA3-Bak vector as previously described (). DU145 control cells, transfected with an empty vector, show moderate Bak expression, whereas transfection of DU145 cells with pcDNA3-Bak resulted in high Bak protein expression (). Transduction with Ad-mycNbk-tTA resulted in high and comparable expression of Nbk under on conditions. However, in sharp contrast to Bax, Bak failed to sensitize the DU145 cells for apoptosis induction in response to Nbk expression (). We have shown previously that Bak overexpression in Bax-negative DU145 cells sensitizes for DNA damage–induced apoptosis by the topoisomerase II poison epirubicin (). Thus, the resistance of Bax-deficient DU145 cells to Nbk- induced apoptosis is not caused by an insufficient function of the Bak protein. This underlines the idea that, in contrast to Bax, functional Bak is not sufficient to transduce the cell death signal triggered by Nbk. This is in contrast to other BH3-only proteins, which trigger apoptosis via Bak- or Bax/Bak-dependent pathways. Notably, Bcl-x–induced apoptosis is entirely dependent on Bak, but not on Bax, in mouse embryonic fibroblasts (). This is in line with our results obtained in the DU145 system. Bax-overexpressing DU145 cells were resistant to Bcl-x, whereas Bak-overexpressing DU145 cells were highly sensitized for induction of apoptosis by Bcl-x (). These results confirm that Bak is functional. Thus, the failure of Bak to become activated by Nbk is not caused by an impaired Bak protein.
To further investigate the specific role of Bax and Bak in Nbk-induced apoptosis, we generated DU145 cells stably expressing EGFP fusion proteins of Bax or Bak. To this end, the cDNA for Bax or Bak was inserted into the pEGFP-C1 vector, and resulting plasmids were transfected into DU145 cells. In agreement with an expected cytosolic localization of Bax, EGFP-Bax showed a homogenous, mostly cytoplasmic staining pattern (), with some Bax being localized constitutively at the mitochondria (). EGFP-Bak showed a different reticular localization, which corresponds to constitutive association of Bak with, e.g., mitochondrial membranes and the ER (; and see ). Nbk expression induced a strong clustering of EGFP-Bax after 24 h, which indicates quantitative redistribution of Bax from the cytosol to the mitochondria and oligomerization of activated Bax. In contrast to EGFP-Bax, and despite the fact that translocation is not necessary in the case of Bak, negligible redistribution of EGFP-Bak was detected under on conditions (). This effect was far less than the effect observed for EGFP-Bax. Exposure to the anticancer drug 5-FU or Bcl-x did, however, trigger massive clustering of Bak (; unpublished data). Upon Nbk expression, EGFP-Bax coalesces into clusters that remain closely associated with mitochondria (). There, Bax localizes to mitochondrial tips and constriction sites, associates with Drp1 and Mfn2, and participates in apoptotic fragmentation of mitochondria (; ). Eventually, EGFP-Bax is located in small punctate structures. During this process, mitochondria cluster around the nucleus and cells undergoing apoptosis shrink and show rounded shape (Fig. S2).
These data establish that Nbk is unable to induce apoptosis via a Bak-dependent pathway. We therefore aimed to test whether this is caused by the presence of an endogenous Bak inhibitor. Because of its constitutive membrane association, Bak is kept tightly in check by its binding to inhibitory proteins, such as VDAC2, and prosurvival members of the Bcl-2 family. As activation of Bak is reported to be specifically inhibited by two prosurvival members of the Bcl-2 family, Bcl-x and Mcl-1 (), we investigated interaction of Nbk with these proteins by coimmunoprecipitation studies. To this end, we used Bax-proficient DU145 cells expressing myc-tagged Nbk. MycNbk was precipitated and the resulting samples were examined for the presence of Mcl-1 and Bcl-x by Western blot analysis. Mcl-1 could not be detected in anti-Nbk immunoprecipitates (, left, on), indicating that Nbk does not interact with Mcl-1. However, the opposite was true for Bcl-x, which was readily detected in Nbk immunoprecipitates, thereby confirming that Nbk specifically binds to Bcl-x. Additionally, as neither Bax nor Bak interact with Nbk, Nbk appears to act as a derepressor, which inhibits Bcl-x function, rather than a direct activator of proapoptotic Bax-like molecules.
On the basis of a recent publication by , which demonstrated high affinity of the Puma BH3 peptide to all antiapoptotic Bcl-2 proteins, we expressed Puma in DU145 Bax cells as a control. Both Mcl-1 and Bcl-x were detected in anti-Puma immunoprecipitates (, right, on). This indicates that, in contrast to Nbk, Puma targets both antiapoptotic proteins, Bcl-x and Mcl-1, in vivo. In line with the assumption that Puma, like Nbk, functions as a derepressor, we did not observe previously reported interactions of Bax with a Puma BH3 domain in a cell-free system () or of Bax with Puma in vivo (). Neither did we observe an interaction of Puma with Bak. A reciprocal immunoprecipitation was also performed to confirm the aforementioned data. This time, Mcl-1 was precipitated from DU145 Bax cells overexpressing Nbk or Puma. In accordance with the immunoprecipitations described in , Nbk was not detectable in anti–Mcl-1 immunoprecipitates (, left, on), whereas Puma was readily detected in Mcl-1 immunoprecipitates (, right, on). These data demonstrate that different BH3-only proteins target different antiapoptotic Bcl-2 family members in a cellular context. This is well supported by Biacore data showing specific interactions of BH3 domain peptides with antiapoptotic Bcl-2 family proteins ().
We next asked if Nbk or Puma might differentially disrupt the Mcl-1–Bak interaction. Under off conditions, Bak can be detected in anti–Mcl-1 immunoprecipitates (). Despite high levels of Nbk expression under on condition, this Mcl-1–Bak interaction is not disturbed by Nbk (, left). In contrast to Nbk, Puma binds to Mcl-1 and displaces Bak, as Bak is no longer detectable in anti–Mcl-1 immunoprecipitates after Puma expression (, right). Mcl-1 interacts with Bak (but not with Bax), as Bak (but not Bax) was detectable in anti–Mcl-1 immunoprecipitates (). This was observed under conditions that promote association of prosurvival proteins with Bax, i.e., the use of nonionic detergent Triton X-100. In contrast, no such findings have been reported for Bak, and Bak has also been shown to associate with Mcl-1 in the presence of CHAPS. Collectively, these interaction studies indicate that Nbk, in contrast to Puma, is insufficient to displace Bak from Mcl-1, as Nbk specifically interacts with Bcl-x but not with Mcl-1.
Interestingly, we found that Mcl-1 expression is increased in cells expressing Nbk. Inhibition of Bak by Mcl-1 might be enforced by this stabilization of Mcl-1. In this regard, a recent study showed that the BH3-only protein Puma also stabilizes Mcl-1 levels (). To investigate Mcl-1 stabilization, we blocked proteasomal degradation of Mcl-1 by the use of the ubiquitin proteasome inhibitor MG132 or induced expression of Puma and compared Mcl-1 levels to those upon Nbk expression in DU145 cells. Western blot analysis revealed that inhibition of the proteasome by MG132 and expression of Puma lead to increased levels of Mcl-1, and expression of Nbk gave comparable results (). However, in contrast to Puma, which may stabilize Mcl-1 through direct interactions (), Nbk does not bind to Mcl-1. Thus, other and/or indirect mechanisms must be involved in the up-regulation of Mcl-1 in response to Nbk expression. As shown by coimmunoprecipitation experiments, Puma binds to Bcl-x, and additional overexpression of Nbk strongly reduces this interaction (Fig. S3, available at ). It might be that Puma is displaced from Bcl-x after Nbk expression and then is available to bind and stabilize Mcl-1. However, expression levels of Puma are moderate in DU145 cells, and expression of Nbk leads to Mcl-1 stabilization, even after down-regulation of Puma by siRNA (). Thus, stabilization of Mcl-1 by Nbk is independent of Puma. Furthermore, stabilization of Mcl-1 by Nbk or Puma is specific, as expression levels of Bcl-x and Bcl-2 (which is barely detectable in DU145) do not change after Nbk or Puma expression ().
To functionally address the contribution of Mcl-1 in Bak regulation during Nbk-induced cell death, we down-regulated Mcl-1 by RNA interference in Bax-deficient and Bax-reexpressing DU145 cells (). Mcl-1 expression was specifically decreased in both cell lines by Mcl-1 siRNA but not by control siRNA. As expected, Bax-deficient DU145 cells were resistant to Nbk-induced apoptosis, and reconstitution of Bax sensitized these cells for Nbk-induced apoptosis. Lipofection of Mcl-1 siRNA sensitized both Bax-deficient and -proficient DU145 cells for Nbk-induced apoptosis, whereas an irrelevant control siRNA had no effect (). Consistent with these data, Nbk-triggered release of cytochrome is increased after down-regulation of Mcl-1. DU145 cells treated with control siRNA did not show cytochrome release, whereas Mcl-1 siRNA causes release of cytochrome after Nbk expression in these Bax-deficient cells (). In Bax-proficient DU145 cells, cytochrome release was increased by Mcl-1 siRNA but not by control siRNA. Measurement of phosphatidylserine exposure and PI uptake confirmed the increased numbers of apoptotic cells upon Mcl-1 down-regulation. Although Bax-deficient DU145 cells were resistant to Nbk-induced phosphatidylserine exposure, reconstitution of Bax sensitized these cells for Nbk-induced apoptosis. Furthermore, down-regulation of Mcl-1 sensitized both cell lines for Nbk-induced phosphatidylserine exposure. About 26% of DU145 mock cells were detected as Annexin V positive (13.5% late apoptotic [PI positive] and 12.5% early apoptotic [PI negative]). Overall, 40% of Bax-reexpressing DU145 cells were stained with Annexin V–FITC (16.5% late and 24% early apoptotic; Fig. S4, available at ). Thus, knockdown of Mcl-1 renders Bax- deficient/Bak-expressing DU145 cells susceptible to Nbk-induced cytochrome release and apoptosis.
During apoptosis, Bak undergoes a conformational change leading to the exposure of an N-terminal epitope. To study if Nbk-induced apoptosis results in Bak activation after siRNA repression of Mcl-1 in Bax-deficient DU145 cells, we performed immunofluorescent stainings with a conformation-specific antibody directed against the Bak N terminus. siRNA-mediated down-regulation of Mcl-1 enabled Nbk to induce Bak activation in DU145 cells, whereas an irrelevant control siRNA had no effect (). In agreement with the conformational change of Bak, Nbk expression induced a strong clustering of EGFP-Bak after Mcl-1 knockdown (). In contrast, cells treated with control siRNA showed no EGFP-Bak clustering after induction of Nbk expression.
In Bax-deficient DU145 cells treated with Mcl-1 siRNA, EGFP-Bak, unlike EGFP-Bax, constitutively colocalized with mitochondria in healthy cells and was not detected in the cytosol (, left). Upon activation of Bak by induced expression of Nbk, EGFP-Bak, like EGFP-Bax, localized to mitochondrial tips and coalesced into large clusters that remained associated with mitochondria (, right). Eventually the phenotype is comparable to EGFP-Bax, with small punctate EGFP-Bak structures, clustering of mitochondria, and a round-up apoptotic shape off the cells (Fig. S5, available at ).
According to a recent paper, Bak is not only sequestered by Mcl-1 but also by Bcl-x (). To study if Nbk can inactivate Bcl-x to release and activate Bak, we performed coimmunoprecipitation experiments using a Bcl-x–specific antibody (). Both Bax and Bak can be detected in anti–Bcl-x immunoprecipitates under off conditions, indicating binding and inactivation of these proteins by Bcl-x. Upon induction of Nbk expression in the absence of doxycycline, Nbk was coprecipitated with Bcl-x, whereas coprecipitation of Bak and, even more pronounced, of Bax with Bcl-x was reduced. This indicates that Nbk binds to Bcl-x, thereby partially displacing Bak from Bcl-x and almost completely freeing Bax from its binding to Bcl-x. Identical results were obtained after down-regulation of Mcl-1. However, as shown in , only after down-regulation of Mcl-1 did Nbk expression induce activation of Bak.
The functional studies established that Nbk-induced cell death is increased after down-regulation of Mcl-1 (). Regarding a displacement model for activation of Bax/Bak by BH3-only proteins, the knockdown of Mcl-1 may cause the release of BH3-only proteins bound to Mcl-1, such as Puma and Bim, which could render the system more sensitive to apoptosis. To study a putative role of these BH3-only proteins in Nbk- induced apoptosis, we analyzed the effect of siRNA-mediated down-regulation of Bim, Puma, or both in Bax-deficient DU145 mock cells after knockdown of Mcl-1. Knockdown of the prespective proteins was confirmed by Western blot analysis (). Down-regulation of Mcl-1 sensitized these Bax-deficient/Bak-expressing DU145 cells for Nbk-induced apoptosis. Nevertheless, the additional knockdown of Bim, Puma, or both failed to affect Nbk-induced apoptosis. Upon induction of Nbk, ∼26% of cells showed sub-G1 DNA content regardless of Puma or Bim expression (). Finally, the Mcl-1 inhibitor Noxa is barely detectable in DU145 cells (unpublished data). These results demonstrate that sensitization to Nbk-induced apoptosis after down-regulation of Mcl-1 is a direct effect of Nbk that is independent from these additional BH3-only proteins.
Nbk is a BH3-only protein that is expressed in a restricted subset of human epithelial tissues with strongest expression in the kidney (). In line with a potential role of Nbk in tumor suppression, loss of Nbk is a common feature of renal cell carcinoma (), and the chromatin locus 22p13.3, which contains , is frequently deleted in human colorectal and breast cancers (). Expression of Nbk triggers apoptosis in breast, lung, prostate, and colon carcinoma, as well as glioma- and melanoma-derived cell lines (; ; ; ).
Because Nbk is a constitutively active protein, it is mainly regulated on the transcriptional level, and induction of Nbk in response to genotoxic stress is mediated in a p53-dependent manner (, ). Based on previous analyses that Bax deficiency protects cancer cells from Nbk-induced apoptosis, even though these cells retain expression of the Bax homologous protein Bak, it appears that the proapoptotic function of Nbk depends on the presence of Bax but not of Bak (; ).
Interestingly, it has been suggested most recently that BH3-only proteins can function via two mutual, yet not exclusive, modes of action (; ). For the BH3-only proteins tBid, Bim, and, more controversially, Puma (), a physical interaction and direct activation of both Bax and Bak has been proposed. These BH3-only proteins were therefore described as direct activators.
The other mode of BH3-only proteins is derepression, which means the release of proapoptotic potential of Bax or Bak by counteracting the inhibitory effect of antiapoptotic Bcl-2 proteins. Derepressors can release Bax and/or Bak from their inhibitory antiapoptotic counterparts through competitive binding of their BH3 domain to antiapoptotic Bcl-2 proteins. Strong evidence for the derepressor model comes from the observation that Bax and Bak can mediate apoptosis without discernable association with the putative BH3-only activators (Bim, Bid, and Puma), even in cells with no Bim or Bid and reduced Puma ().
Nbk interacts with antiapoptotic Bcl-2 family members, but not with the proapoptotic multidomain proteins Bax (; ; ; ) and Bak, indicating that Nbk functions as a derepressor to activate apoptosis. However, regarding the proposed rheostat model where the relative concentration of antiapoptotic, Bcl-2–like, and proapoptotic family members determines the fate of a cell, the role of Bak in Nbk-induced apoptosis is poorly investigated.
In this study we demonstrate that specific loss of Bax, but not of Bak, confers resistance to Nbk-induced apoptosis. Furthermore, we show that, in contrast to Bax reexpression, expression of Bak does not sensitize Bax-deficient DU145 cells for Nbk-induced cell death. Nbk fails to induce clustering and oligomerization of EGFP-Bak fusion protein but induces strong EGFP-Bax translocation. This demonstrates that Nbk-mediated apoptosis ultimately depends on Bax, whereas Bak is dispensable.
Interestingly, Bak is sequestered by the antiapoptotic proteins Bcl-x and Mcl-1 (but not by Bcl-2), and activation of Bak requires neutralization of both proteins Bcl-x and Mcl-1 by BH3-only family members (). In turn, we showed that Nbk interacts with Bcl-x, but interaction with Mcl-1 could not be detected. Moreover, Nbk fails to disrupt Mcl-1–Bak interaction. Notably, Bak is sequestered and inactivated by Mcl-1, even in the presence of strong Nbk expression. This is in contrast to Puma, which binds to Bcl-x and Mcl-1 and releases Bak from Mcl-1 binding. In this context, it is interesting that peptides derived from BH3-only domains exhibit differences in their binding selectivity to antiapoptotic Bcl-2 proteins. Determination of the binding constants by Biacore experiments revealed that Puma potently engaged all the prosurvival proteins, including Mcl-1, whereas Nbk showed very low affinity to Mcl-1 (). Thus, it appears to be reasonable that Nbk blocks the inhibitory function of Bcl-x on Bak, but the antiapoptotic Mcl-1 is capable of backing up and still keeping Bak in check as a second restraint. Inhibition of Bak by Mcl-1, which in turn cannot be inactivated by Nbk, is therefore responsible for the failure of Nbk to induce Bak activation. Indeed, down-regulation of Mcl-1 by siRNA sensitizes Bax- deficient DU145 cells for Nbk-induced apoptosis and enables Nbk to activate Bak. This hints at a staggered barrier formed by antiapoptotic proteins, which is counteracted by the likewise graduated proapoptotic potential of BH3-only proteins.
Mcl-1 is readily down-regulated in response to certain death stimuli. The short half-life of Mcl-1 is attributed to constitutive polyubiquitination and subsequent degradation of Mcl-1 by the proteasome, which is a prerequisite for induction of apoptosis after UV irradiation (). Polyubiquitination of Mcl-1 is catalyzed, e.g., by Mule/ARF-BP1, a novel E3 ubiquitin ligase that binds to Mcl-1 via a BH3 domain and marks Mcl-1 for proteasomal degradation (). The BH3-only protein Noxa also plays a crucial role in Mcl-1 degradation. By displacing Bak from Mcl-1, Noxa enhances polyubiquitination and subsequent proteasomal degradation of Mcl-1 (). In contrast to Noxa-induced degradation of Mcl-1, we found that the cellular Mcl-1 protein level was even increased after Nbk expression. This effect of Nbk expression was specific for Mcl-1 and was not observed for Bcl-2 or Bcl-x. A similar phenomenon has been previously reported in the context of Puma overexpression. In these experiments, Mcl-1 was stabilized by binding to Puma, and it has been suggested that the interaction between Mcl-1 and Puma and the resulting stabilization of Mcl-1 may represent a novel mechanism to regulate and prevent apoptosis (). However, Mcl-1 is stabilized by Nbk even when Puma is knocked down. Thus it is unlikely that Puma, after being displaced from Bcl-x by Nbk, mediates Mcl-1 stabilization in this setting. In our experimental system, persistent binding of Bak to Mcl-1 might therefore be responsible for enhanced Mcl-1 expression via stabilization, as Nbk, unlike Puma, does not bind to Mcl-1 to displace Bak. Consistent with these data, we found that adenoviral overexpression of Puma, despite stabilizing Mcl-1, can induce apoptosis in a Bak-dependent manner (in Bax-deficient DU145 Bak cells), although to a lesser extent compared with DU145 Bax cells (unpublished data). Therefore, stabilization of Mcl-1 may exert distinct regulatory functions in the fine-tuning of apoptosis and attenuating Puma-induced apoptosis. Moreover, Mcl-1 stabilization by Nbk interferes with activation of Bak because Nbk, in contrast to Puma, fails to bind and inactivate Mcl-1. Thus, in Nbk-induced apoptosis, Bak cannot compensate for Bax because of the low affinity of Nbk to Mcl-1 and the up-regulation of Mcl-1 after expression of Nbk, which may further interfere with activation of Bak. Support for this model comes from the observation that the BH3-only protein Noxa, which targets Mcl-1, cooperates with Nbk to activate mitochondrial cytochrome release; however, the impact on Bak activation was not addressed in this study and all experiments were performed in a Bax-proficient background ().
Collectively, our data support a model in which the proapoptotic activity of BH3-only proteins is governed by their propensity to interfere with an at least dual layer of protection of Bak (). In our system, binding of Nbk inactivates Bcl-x and releases Bax and Bak (). Although the interaction between Nbk and Bcl-x seems to be stronger, Nbk also binds to Bcl-2 (; ; ), indicating that Bcl-2 is at least partially inhibited by Nbk. Once the capacity of Bcl-2 to bind Bax is exceeded, the unbound Bax protein becomes activated. Notably, the Bax inhibitors Bcl-2 and Bcl-x are not stabilized by Nbk. In contrast, Bak, released from Bcl-x upon Nbk expression, is still kept in check by Mcl-1, and this is even enforced by the up-regulation of the Bak inhibitor Mcl-1 by Nbk. This Bak inhibition may be further supported by other factors that do not necessarily belong to the Bcl-2 family, such as the VDAC2 protein (; ). Mcl-1–mediated repression of Bak thereby provides a molecular rationale for the strict Bax dependency of apoptosis induced by Nbk in the present setting. Thus, BH3-only proteins that, like Nbk, act as derepressors bind to distinct antiapoptotic proteins. Nevertheless, the consequently derepressed proapoptotic multidomain proteins can then be counteracted by a second row of antiapoptotic proteins. This adds a higher degree of specificity and opportunities for the fine-tuning of death signals. The outcome of triggering this tightly balanced interaction network would be dictated by both the mix of BH3-only proteins induced by a given death stimulus and the abundance of their antiapoptotic counterparts forming this second barrier to apoptosis. Support for the biological relevance of such a Bak inhibitor model comes indirectly from cancer biology. There, disruption of the gene impairs myc-induced islet cell depeletion upon myc expression and facilitates tumorigenesis in a myc-driven pancreatic β-cell tumor model, whereas genetic inactivation of has no impact (). In fact, loss of Bax is a frequent event in human cancer, whereas Bak expression persists in most cancers. Thus, targeting Bak inhibitors should be a rewarding strategy to overcome restraints in apoptosis signaling in cancer and resulting resistance mechanisms.
HEK293 and DU145 cells were obtained from the American Type Culture Collection or the Deutsche Sammlung von Mikroorganismen und Zellkulturen. HCT116 wt cells and their isogeneic knockout sublines HCT116 Bax/Bak were provided by B. Vogelstein (Johns Hopkins Cancer Center, Baltimore, MD). The stable knockdown of Bak was performed in either HCT116 wt or HCT116 Bax knockout cells yielding Bax/Bak and Bax/Bak cells. Transfectants were generated and cells were cultured as previously described (; ; ). Expression of BH3-only proteins was suppressed by addition of 1 μg/ml doxycycline to the culture medium (Tet-off condition).
Monoclonal mouse anti-Bax antibody (clone YTH-2D2; raised against a peptide corresponding to aa 3–16) was purchased from Trevigen, Inc., and goat anti-Nbk antibody (N-19; raised against an epitope mapping to the 19-aa N terminus of human Nbk) and rabbit anti–Mcl-1 (H-260; epitope corresponding to aa 1–260 mapping at the N terminus of Mcl-1 of human origin) were purchased from Santa Cruz Biotechnology, Inc. Polyclonal rabbit anti–Bcl-x antibody (raised against aa 18–233 of rat Bcl-x) and anti-Bim (against aa 22–40 of human Bim) were purchased from BD Biosciences, monoclonal mouse anti–human cytochrome (clone 7H8.2C12) was purchased from BD Biosciences, and goat anti–caspase-9 antibody (raised against human caspase-9 aa 139–330) was purchased from R&D Systems. The polyclonal rabbit anti-Bak antibody (raised against a peptide corresponding to aa 14–36) was purchased from Dako. Polyclonal anti–human actin antibody and anti-Puma, produced in rabbits, were obtained from Sigma-Aldrich. Polyclonal anti–human caspase-3, produced in goat, was obtained from R&D Systems. Secondary anti–rabbit, anti–goat, and anti–mouse HRP-conjugated antibodies were obtained from Promega or SouthernBiotech. After cell trypsination and washing, protein expression was detected by ECL-based Western blot analysis as described previously (). For analysis of cytochrome release, cells were lysed in a hypotonic digitonin buffer as described previously ().
The recombinant adenovirus for Puma or Bcl-x expression was constructed for mycNbk as described previously (). In brief, the E3 region of Ad5 was replaced by tTA under the control of a CMV promoter and an SV40 poly(A) tail. A Puma or Bcl-x cDNA was introduced into the E1 region under the control of the Tet-off system to achieve conditional expression in the absence of doxycycline. The resulting DNA construct was transfected in HEK293 packaging cells to produce vector stocks. Adenoviral stocks were propagated according to standard procedures described in .
Apoptosis was determined on a single-cell level by measuring the DNA content of individual cells with a FACScan (BD Biosciences) as described previously (). Cellular DNA content was measured by flow cytometry with logarithmic amplification in the FL-3 channel. Data are given in percentage of hypoploidy (sub-G1), which reflects the percentage of apoptotic cells. Alternatively, apoptotic cell death was determined by PI staining and measuring binding of Annexin V–FITC upon exposure of phosphatidylserine to the cell surface ().
After infection with recombinant adenovirus at the MOI of 25 in the presence or absence of doxycycline, cells were collected by centrifugation at 300 at 4°C for 5 min. Mitochondrial permeability transition was determined by staining the cells with 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-benzimidazolylcarbocyanin iodide (JC-1; Invitrogen) as described previously (). Mitochondrial permeability transition was then quantified by flow-cytometric determination of cells with decreased red fluorescence, i.e., with mitochondria displaying a lower ΔΨm.
DU145 EGFP-Bax or EGFP-Bak cells were seeded on coverslips, infected with Ad-mycNbk-tTA, and cultured for 24 h under on or off conditions. Cells were washed with PBS and fixed for 30 min with ice-cold 1% paraformaldehyde. For immunofluorescence staining of mitochondria, cells were permeabilized with ice-cold 100% methanol for 1 min. Cells were washed again, and nonspecific binding of antibodies was blocked by incubation with 8% BSA for 30 min at room temperature. Primary antibodies (mouse anti–Tom 20 antibody [BD Biosciences]) were diluted in 1% BSA in PBS and added to the cells overnight at 4°C. Incubation with secondary antibodies (Alexa Fluor 594–conjugated chicken anti–mouse IgG [Invitrogen]) was performed for 1 h at room temperature. For staining of DNA, cells were incubated in PBS + DAPI for 5 min. Cells were washed and mounted in fluorescence mounting medium (Dako). For overview images, cells were inspected at room temperature with a fluorescent microscope (BX50; Olympus) equipped with a 40×/0.75 objective lens (UPlanFL; Olympus) and a camera (micropublisher 5.0 RTV; QImaging). For subcellular localization pattern analysis, cells were inspected at room temperature with a microscope (Axiovert 200; Carl Zeiss, Inc.) equipped with a 63×/1.4 objective lens (Plan-Apochromat; Carl Zeiss, Inc.) and a digital camera (ORCA ER; Hamamatsu Photonics). Images were acquired by Openlab software (Improvision) and vertical slices (0.2-μm separation) were deconvoluted with Openlab 5.0.2 nearest–neighbor deconvolution algorithm on Mac OSX 10.4 (Apple Inc.).
Coimmunoprecipitations were performed using the ExactaCruz IP/Western blot reagents (Santa Cruz Biotechnology, Inc.) according to the manufacturer's instructions. In brief, 1.5 × 10 cells/75-cm flask were infected with Ad-mycNbk-tTA or Ad-Puma-tTA and cultured for 24 h with or without doxycycline. Cells were harvested and lysed as described previously (). Lysates were precleared with preclearing matrix (Santa Cruz Biotechnology, Inc.). 150 μl of the precleared cellular extract was shaken in the presence of the IP antibody–IP matrix complex at 4°C for 4 h. Immunoprecipitations were done and examined by Western blot analysis as described previously ().
Mcl-1 siRNA and control siRNA were purchased from Santa Cruz Biotechnology, Inc. Lipofection of the cells was performed by use of transfection reagent (Santa Cruz Biotechnology, Inc.) according to the manufacturer's instructions. On-target plus siRNA against Puma, Bim, Mcl-1, and control siRNA for experiments shown in and was purchased from Dharmacon. Transfection of the cells was performed by use of transfection reagent (DharmaFECT; Dharmacon) according to the manufacturer's instructions. After 24 h, cells were transduced with Ad-mycNbk-tTA at an MOI of 25.
Fig. S1 shows Nbk-induced phosphatidyl serine exposure of HCT116 cell lines. Fig. S2 shows changes of the subcellular localization pattern of EGFP-Bax and mitochondria after Nbk expression. Fig. S3 shows coimmunoprecipitation studies of Puma and Bcl-x. Fig. S4 shows Nbk-induced phosphatidyl serine exposure of DU145 cell lines after Mcl-1 knockdown. Fig. S5 shows subcellular localization pattern of EGFP-Bak and mitochondria upon Nbk expression. Online supplemental material is available at . |
The immunological synapse (IS) is a specialized junction between a T cell and an antigen-presenting cell (APC) that forms within seconds after contact between the T cell receptor (TCR) and the appropriate antigen on the APC (). The mature IS is characterized by the clustering of TCR with key signaling (e.g., PKCθ and ZAP-70) and adhesion molecules at the contact site and the exclusion of other membrane proteins such as CD43, which has a bulky extracellular domain, from the synapse region. Thus, the IS provides an accessible cellular system for studying the general question of how cells organize their cell cortex and plasma membrane into distinct functional domains. In this paper, we show the differential and essential roles of ezrin/radixin/moesin (ERM) proteins during this process.
IS formation depends on reorganization of the actin cytoskeleton in T cells induced by initial TCR signaling (; ). The normal accumulation of filamentous actin (F-actin) at the T cell–APC interface is thought to stabilize a continuous contact between the two cells, and inhibition of actin polymerization blocks immunological synapse formation (; ; ). The significance of cytoskeletal changes is further illustrated by the rapid activation-induced transition from a spherical T cell with abundant microvilli to a polarized cell with few microvilli but with ruffles at the contact site ().
The ERM proteins provide a regulated linkage between the cytoskeleton and plasma membrane, especially in actin-containing cell surface structures such as microvilli and membrane ruffles (; ). Human ERM proteins, which are ∼580 residues long and share ∼75% sequence identity, are classified as members of the band 4.1 superfamily, as they have an ∼300-residue N-terminal 4.1 ERM (FERM) domain (; ). The FERM domain binds to membrane-associated proteins, whereas an F-actin–binding site lies near the C terminus (; ). ERM proteins are negatively regulated by an intramolecular interaction between the FERM and C-terminal domains that masks binding sites for F-actin and at least some membrane proteins (; ; ; ). The intramolecular interaction can be relieved by phosphatidylinositol 4,5-bisphosphate in conjunction with the phosphorylation of a conserved threonine in the C-terminal domain (T567 in ezrin, T564 in radixin, and T558 in moesin) to open up the molecule and unmask the F-actin and membrane-binding sites (; ; ). Recently, phosphorylation of an additional conserved threonine in the FERM domain, T235, was also suggested to reduce the intramolecular association in ezrin ().
The presence of a single essential ERM protein in and suggests a single basic function for ERM proteins (). In vertebrates, the three closely related ERM members emerged by gene duplication, suggesting they may perform overlapping functions (). Antisense phosphorothioate oligonucleotide treatment to reduce all ERM expression in cultured cells indicated that they are at least partially redundant (). In addition, no substantial differences have been reported in the ability of ERM proteins to bind multiple ligands (e.g., CD43, CD44, and intercellular adhesion molecules; ). Thus, it is currently believed that ERM proteins perform similar, largely redundant functions. However, their very distinct tissue distributions hint at individual cell type–specific functions (; ; ).
In primary human lymphocytes, moesin and ezrin but not radixin are expressed, with moesin being the quantitatively dominant ERM protein (). In resting T cells, phosphorylated ERM proteins are localized in microvilli (). Stimulation of T cells by chemokine or antigen induces rapid ERM dephosphorylation (). This presumably facilitates the loss of microvilli to create a region of the plasma membrane for T cell–APC conjugate formation. In addition, exclusion of CD43 from the contact site is mediated by moesin, ezrin, or both (; ; ). CD43 is an abundant, highly glycosylated, and sialylated transmembrane protein () that is proposed to impede synapse formation by its negative charge and large size. Thus, the interaction between CD43 and ERM proteins seems to be required for the removal of CD43 for contact formation (; ). Existing work refers to moesin and ezrin function in T cells as redundant ().
In this study, we report that ezrin and moesin perform distinct and critical roles during immunological synapse formation requiring their regulation by phosphorylation and differential association with specific ligands.
To study the potentially distinct roles of ezrin and moesin in T cells, we first examined their localization in Jurkat T cells. Using specific antibodies to the two ERM proteins expressed in Jurkat T cells (Fig. S1, available at ), we found that in resting T cells, ezrin was mostly cytoplasmic, whereas moesin was enriched at the cell cortex (). However, when the cells were activated for 30 min with a soluble TCR antibody, OKT3, ezrin was redistributed to the membrane cortex (). The differential distribution of ezrin and moesin in resting T cells and the redistribution of ezrin from the cytoplasm to the cell cortex upon activation was also found in primary human T cells (). Next, we determined the localization of ezrin and moesin in the mature IS using the Raji B cell line as the APC, a system that has been well studied (; ; ; ). Jurkat T cells were incubated for 30 min with superantigen-loaded Raji B cells, and synapse formation was verified by clustering of the TCR and by formation of F-actin–rich protrusions that establish contact with the APC (; ; ). We found that ezrin was redistributed to the IS, whereas moesin was excluded from the contact site (). As Raji B cells also express ezrin, its enrichment close to the contact site could be contributed from the B cell. To circumvent this, anti-CD3/CD28–coated polystyrene beads, which mimic activated APCs (; ), were mixed with Jurkat T cells for 15 min, which led to conjugate formation with TCR and PKCθ at the contact site (unpublished data). Ezrin and F-actin were concentrated at the contact site, whereas moesin was excluded from it (). These data demonstrate that ezrin and moesin are differentially localized in T cells and show largely complementary distribution in the mature IS.
Within 30 s after T cell/APC contact formation, TCRs cluster at the IS along with PKCθ and the tyrosine kinases Lck and ZAP-70 (; ). Therefore, we examined whether ezrin and moesin redistribution occurs within this time frame. Jurkat T cells were stimulated with anti-TCR for 1 min before fixation and immunostaining. Whereas ezrin is mostly cytoplasmic in resting T cells, it was recruited to regions enriched with TCR and PKCθ after 1 min of activation. Moesin, on the other hand, was excluded from these clusters (). F-actin undergoes rapid depolymerization at the contact site after cell–cell contact and polymerizes again beneath the synapse and is enriched in the T cell protrusions (). When we looked at T cells 1 min after activation, F-actin was excluded from TCR clusters in the same way as moesin (). Quantification of clusters formed within 1 min of activation showed that ezrin and PKCθ enriched with TCR in 86 ± 6.5% and 82.3 ± 2% of the cells, respectively. In contrast, moesin was found in clusters with TCR in only 19.6 ± 2.7% of the cells (). Similar results were found when the c ells were activated by incubation with anti-CD3/CD28–coated polystyrene beads for 1 min (). These data demonstrate that the spatial changes in ezrin and moesin distribution upon T cell activation are rapid and occur along with the clustering of TCR and other signaling molecules.
As ERM protein activity can be regulated by phosphorylation of a conserved C-terminal threonine, we examined the changes in phosphorylation of the relevant threonine, T567, in ezrin and T558 in moesin during T cell activation. In resting cells, moesin is much more highly phosphorylated than ezrin (). Upon 1–3 min of activation by soluble anti-TCR antibody, moesin is substantially but not completely dephosphorylated, as reported previously (). Moesin dephosphorylation is followed by phosphorylation of both ezrin and moesin after ∼20 min of activation ().
When we examined the distribution of phosphorylated ERM (pERM) in the mature IS using antibody that recognizes both phosphorylated T567 ezrin and T558 moesin, it was localized both at the periphery of the synapse and at the membrane cortex outside of the contact site in conjugates between T cells and either superantigen-loaded Raji B cells or anti-CD3/CD28–coated beads (). It is known that upon chemokine stimulation of T cells, moesin is quickly and mostly dephosphorylated on T558, enabling microvilli collapse (). When we looked at pERM at 1 min of activation, we found that similar to moesin, it was excluded from TCR clusters (). In resting T cells and after 1 min of activation, moesin is the major phosphorylated ERM protein (). Thus, pERM staining at early activation time points represents mostly the remaining phosphorylated moesin, whereas at later time points, it represents both phosphorylated ezrin and moesin that are phosphorylated on the conserved C-terminal threonine.
As we found changes in the phosphorylation of ezrin and moesin during T cell activation (), we next wanted to explore their possible involvement in IS formation. To inhibit dephosphorylation, we used the phosphatase inhibitor calyculin A, which is known to inhibit ERM dephosphorylation in lymphocytes (). 100 nM calyculin A blocked moesin dephosphorylation induced by T cell activation (Fig. S2, available at ). When Jurkat T cells were preincubated for 10 min with 100 nM calyculin A followed by 1-min activation with either anti-TCR () or anti-CD3/CD28–coated polystyrene beads (), TCR clustering was inhibited in >90% of the cells, and, in 80% of them, no ezrin clustering was detected (). In these cells, moesin was not excluded from any cortical membrane areas, and continuous staining for pERM across the cortex membrane was seen (). In addition, F-actin did not depolymerize beneath any membrane segment in these cells (). This result suggests that moesin dephosphorylation in stimulated T cells may be required for moesin and F-actin exclusion from the contact site and for proper TCR cluster formation.
We next examined the need for the specific phosphoregulation of ezrin and moesin during IS formation. Two phosphomutants were prepared for both ezrin and moesin in which the two threonines (T235/T567 and T235/T558, respectively), which are known to be regulatory phosphorylation sites in other systems (; ), were replaced by either alanine (ezrin-AA and moesin-AA) to preclude phosphorylation or glutamic acid residues (ezrin-EE and moesin-EE) to mimic phosphorylation. We analyzed the effects of expression of these proteins on clusters and synapse formation in Jurkat T cells. As both N- and C-terminal tagging of wild-type moesin and ezrin led to mislocalization in resting T cells (unpublished data), we used untagged proteins and verified enhanced protein expression by Western blotting (Fig. S3 A, available at ). When T cells expressing ezrin-AA were activated for 15 min with anti-CD3/CD28–coated polystyrene beads, only 7% of the cells formed a mature contact with T cell protrusions around the bead (). In contrast, 83% of wild-type ezrin-expressing cells formed a mature contact with activating beads after 15 min of activation (). Ezrin-EE was localized to the membrane cortex in resting T cells (Fig. S3 B), as it presumably represents a constitutively active form, thereby precluding an analysis of its redistribution upon stimulation. When moesin-AA was overexpressed in T cells followed by 1-min activation with anti-TCR, only 50% of the cells had TCR clusters, and it was difficult to determine whether moesin is excluded from those clusters, as overexpressed moesin-AA was highly enriched in the cytoplasm (). However, overexpressed moesin-EE was localized at the T cell cortex, and when the cells were activated for 1 min with anti-TCR, it was not removed from any portion of the membrane, and TCR clusters were not detected in 95% of the cells (). Together, these results suggest that the rapid dephosphorylation of moesin and, later, the phosphorylation of ezrin are necessary for IS formation. The differential distribution of ezrin and moesin and their phosphoregulation during T cell stimulation and IS formation were reproduced in primary human T cells (Fig. S4).
After the finding that ezrin-AA prevents IS formation, we set out to explore where the signaling pathway is compromised. Elevation in cytoplasmic calcium is one of the earliest events after T cells activation (; ; ), so we examined changes in cytoplasmic calcium in T cells expressing ezrin and moesin phosphomutants. Calcium levels were monitored continuously for 270 s after activation using the cytoplasmic dye Fluo-LOJO. When T cells expressing wild-type ezrin or moesin were activated with anti-TCR, elevated calcium levels were recorded in 89% and 82% of the cells, respectively (). However, in cells expressing ezrin-AA, cytoplasmic calcium elevation was detected in only 18% of the cells. In contrast, calcium increase was detected in 72% of moesin-AA–expressing cells (). These data demonstrate that although ezrin is phosphorylated at later stages of synapse formation, ezrin-AA inhibits cytoplasmic calcium elevation, an early event during synapse formation. Ezrin-EE and moesin-EE, both of which localized at the cell cortex, inhibited calcium elevation to a similar extant (), agreeing with our findings that these two proteins inhibit initial TCR clusters and mature synapse formation.
Several studies have suggested that CD43 is removed from the IS by binding to ERM proteins, with some discrepancy as to which ERM member is responsible (; ; ). As we found differential roles and distribution for ezrin and moesin in IS formation, we examined which ERM protein binds CD43. CD43 was found to coimmunoprecipitate with moesin from both resting and activated Jurkat T cells () but not with ezrin from either resting or activated cells (). Immunofluorescence revealed colocalization of moesin with CD43 in resting and activated T cells (). This result shows that CD43 binds selectively to moesin rather than ezrin and that CD43 associated with moesin is excluded from the synapse.
As we found that CD43 binds moesin but not ezrin in T cells (), we sought to identify specific binding partners for ezrin. Immunoprecipitation of ezrin followed by mass spectrometry analysis was used to identify potential ezrin-binding proteins. ZAP-70, a T cell–specific Syk tyrosine kinase critical for IS formation (; ; ), specifically coimmunoprecipitated with ezrin but not with moesin in both resting and 30-min–activated Jurkat T cells () and, in fact, coimmunoprecipitated throughout the time course of IS formation (not depicted). ZAP-70 binds ezrin directly, as recombinant His6–ZAP-70 purified from insect cells and immobilized on nickel beads binds pure ezrin (). The binding selectivity of ZAP-70 for ezrin over moesin is potentially determined by their FERM domains, as recombinant His6–ZAP-70 preferentially binds the ezrin FERM domain (Fig. S5 C, available at ). This domain was previously shown to include the binding site for several ligands ().
In resting Jurkat T cells, ZAP-70 is localized at the cell cortex (), whereas the bulk of ezrin is cytoplasmic (). However, after 1 min of activation by soluble TCR antibody, ezrin and ZAP-70 precisely colocalize in clusters (). To decipher the apparent paradox of ezrin and ZAP-70 distributions in resting T cells, we analyzed the properties of the ZAP-70–bound fraction of ezrin in T cells. For this, ZAP-70 was quantitatively immunoprecipitated from Jurkat T cells (Fig. S5 A) together with ∼15% of total ezrin (Fig. S5 B). Thus, a small fraction of ezrin may be bound to ZAP-70 at the cortex and not readily detectable by light microscopy. By characterizing the ezrin coprecipitating with ZAP-70 with an equivalent amount of ezrin that was directly immunoprecipitated, we found that ZAP-70 preferentially binds the threonine T567-phosphorylated form of ezrin, which is known to be the activated form and, thus, is more likely to be associated with membrane-bound proteins ().
To explore the role of ezrin phosphorylation on ZAP-70 localization, its localization in Jurkat T cells expressing ezrin mutants was examined. In 88% of the cells expressing ezrin-AA, both ezrin and ZAP-70 were cytoplasmic and remained largely cytoplasmic after activation with anti-TCR (). These results suggest that ezrin-AA perturbs ZAP-70 localization at the cell cortex and prevents its clustering upon T cell stimulation. The cortical distribution of ezrin-EE did not affect ZAP-70 cortical localization but inhibited their coclustering at the cortex altogether (). Binding of ZAP-70 to the different ezrin mutants was verified by Western blot analysis (Fig. S5 D). To determine whether phosphorylation of one or both threonines is required for ZAP-70's cortical recruitment, two single-A ezrin mutants were generated and introduced into Jurkat T cells (ezrin-T235A and ezrin-T567A). Although 85% of resting cells expressing ezrin-T235A showed a wild-type distribution of ezrin and ZAP-70, 91.4% of resting cells expressing ezrin-T567A had a cytoplasmic distribution of both ezrin and ZAP-70 (). Similar results were obtained when cells were stimulated with anti-TCR; namely, ezrin- T235A expression allowed proper cluster formation, whereas both ezrin and ZAP-70 remained cytoplasmic in ezrin-T567A–expressing cells (). Collectively, these results show that threonine T567 in ezrin is critical to allow the proper localization of ZAP-70 in resting cells and for the rapid recruitment of ZAP-70 during IS formation.
Our findings that proper ZAP-70 localization in T cells and recruitment to the IS requires phosphoregulated ezrin most likely indicates the need for direct interaction between these two proteins. As an independent method to assess whether ezrin is necessary for ZAP-70 localization, Jurkat T cells were specifically depleted of ezrin using siRNA (). 73% of resting T cells depleted of ezrin had a cytoplasmic distribution of ZAP-70 (), which failed to cluster and remained cytoplasmic after 1 min of activation with anti-TCR (). When ezrin-depleted cells were activated for 15 min with anti-CD3/CD28–coated polystyrene beads, conjugate formation was inhibited in 84% of the cells, and ZAP-70 remained cytoplasmic (). These data further demonstrate that ZAP-70 requires ezrin for proper localization in T cells and for clustering at the IS and that moesin cannot provide this function.
Formation of the IS involves segregation of the TCR with key signaling molecules to the T cell–APC interface and the exclusion of other specific membrane proteins. These dynamic changes are accompanied by TCR-mediated structural changes and cellular organization of the cytoskeleton. ERM proteins have been implicated in morphological changes and membrane–cytoskeletal rearrangements during synapse formation, and they are often considered to be functionally redundant. We examined the distribution and roles of the two ERM proteins expressed in T cells, ezrin and moesin, during synapse formation. Our results show that ezrin and moesin are differentially localized and, through phosphorylation-regulated processes, perform functionally distinct roles during IS formation as well as interact with different binding partners.
Functional redundancy for vertebrate ERM proteins is consistent with the findings that loss of the single ERM protein in the insect () or nematode () is lethal, whereas gene knockouts of moesin () or radixin in mice are not, although radixin-deficient mice do have defects associated with liver function and hearing (; ). However, ezrin knockout mice die within 2 wk of birth, possibly as a result of defects in intestinal epithelial cell morphology (). These differential phenotypes do not disprove functional redundancy, as ERM proteins show distinct tissue distributions, with ezrin being found highly enriched in polarized epithelial cells (), radixin being the dominant member in liver (), and moesin being enriched in endothelial cells and lymphocytes (; ). In support of redundancy, when each ERM was individually suppressed in cultured thymoma cells, their microvilli were not affected, but loss of all three eliminated microvilli (). However, none of these studies explored the question of whether ERM members might play distinct roles within the context of a single cell.
What functions might ezrin and moesin perform in synapse formation? During IS formation, changes in T cell shape occur, including the disappearance of microvilli and flattening of the membrane followed by actin polymerization and ruffle formation at the T cell–APC interface. We show that in resting T cells, moesin and phosphomoesin are enriched at the cell cortex. Upon stimulation, moesin is very rapidly dephosphorylated and lost along with its ligand CD43 specifically from the contact site, and both are then absent from this region until at least 30 min after IS formation. The relevance of moesin dephosphorylation in this process is implied by the finding that treatment with the phosphatase inhibitor calyculin A inhibited both moesin dephosphorylation and its local loss during T cell activation. More specifically, expressing a mutant that mimics constitutive phosphorylation (moesin-EE) abolished the localized loss of moesin at the contact site as well as IS formation. Thus, rapid moesin dephosphorylation plays a critical role in preparing a region of the cell cortex for synapse formation.
Ezrin plays a different role. In resting T cells, the bulk of ezrin is localized to the cytoplasm in a dephosphorylated dormant state. Within 1 min of activation, ezrin becomes enriched at the contact site with TCR and key signaling molecules. At later times, ezrin is phosphorylated and colocalizes with F-actin at the immunological synapse, a region devoid of moesin. The importance of ezrin phosphorylation in this process was demonstrated by examining cells in which a nonphosphorylatable mutant (ezrin-AA) was expressed. Initial TCR clusters were formed in 50% of these cells, but mature synapse formation was almost completely abolished. This implies that delayed phosphorylation of ezrin contributes to stabilizing the spatial segregation of molecules at the contact site. Surprisingly, ezrin-AA expression also blocked the rapid calcium elevation that is seen within 30–60 s after T cell stimulation. As the inhibition of calcium mobilization was specific to ezrin-AA (and was not detected with moesin-AA), it suggested that ezrin may be involved upstream in the signaling cascade of T cells rather than merely stabilizing the mature synapse.
We have found that ezrin binds directly to ZAP-70, a T cell–specific nonreceptor tyrosine kinase that plays an integral role in TCR signaling, in resting T cells as well as throughout synapse formation. ZAP-70 is localized to the cell cortex in resting Jurkat T cells and is rapidly phosphorylated by Lck and recruited to TCR clusters upon T cell activation (; ; ; ). A previous study showed that ezrin and/or moesin interacts with Syk tyrosine kinase, a protein closely related to ZAP-70 (). The significance of proper recruitment of ZAP-70 to the TCR signaling complex is demonstrated in patients with common variable immunodeficiency in which ZAP-70 fails to properly localize, leading to defective T cell differentiation (), and by impaired T cell activation in various forms of the severe combined immunodeficiency disease, in which ZAP-70 was affected (). Biochemical characterization of ZAP-70–ezrin interaction revealed that ZAP-70 preferentially binds the threonine-phosphorylated form of ezrin. This fraction of ezrin most likely represents an activated form of ezrin that can be attached to membrane-bound proteins rather than the dormant, nonphosphorylated ezrin that we found to be mostly cytoplasmic. We have also examined whether ZAP-70 could be the tyrosine kinase that phosphorylates ezrin, but our in vitro experiments did not support this possibility (unpublished data).
We found that ezrin-AA expression perturbs ZAP-70 distribution and prevents its recruitment to clusters upon activation. Expression of single-alanine mutants of ezrin (T235A or T567A) showed that phosphoregulation of only one of these two threonines, T567, is required for ZAP-70 localization in T cells and recruitment to the IS. Depletion of ezrin by specific siRNA treatment from Jurkat T cells leads to the same phenotype as expression of the ezrin-AA mutant: namely, cytoplasmic distribution of ZAP-70 and lack of clustering at the site of activation. These results demonstrate the central role of ezrin regulation in the correct targeting of ZAP-70 in the T cell.
As far as we are aware, this is the first study to show functionally distinct and phosphoregulated contributions for ezrin and moesin in a defined cellular process. Moreover, we were able to show that mutants altered in the ability to be dephosphorylated or phosphorylated revealed distinct contributions of ezrin and moesin in IS formation. In addition, we have uncovered ligands that bind selectively to ezrin and moesin. The molecular basis for the functional distinction and binding preferences between such closely related proteins is not yet resolved and suggests that fruitful approaches will be able to determine which sequences define their distinct contributions and what biochemical differences distinguish ezrin and moesin function in T cells.
All cells were purchased from American Type Culture Collection. Antibodies against human ezrin, moesin, and pERM have been described previously (; ). Rabbit anti–human ERM and mouse anti–human ZAP-70 were obtained from Cell Signaling Technology. R-phycoerythrin–labeled mouse anti–human CD3 and mouse anti–human CD43 were obtained from BD Biosciences. Mouse anti–human PKCθ was obtained from Santa Cruz Biotechnology, Inc. OKT3 mouse anti–human CD3 was purified from the OKT3 hybridoma cell line. Rhodamine phalloidin, AlexaFlour568 phalloidin, donkey anti–rabbit and donkey anti–mouse AlexaFlour488, and goat anti–rabbit and goat anti–mouse AlexaFlour568 were obtained from Invitrogen. HRP-conjugated goat anti–rabbit antibodies were obtained from MP Biomedicals.
Human moesin-wt entry vector was generated by PCR from pQE16-moesin using the primers 5′-CCAGTGTGGTGGAATTCATGCCCAAAACGATC and 3′-AGATATCTCGAGGCTAATTAAGCTTGTTACATAGACTCAAATTCG. PCR product was ligated into TOPO TA cloning vector (Invitrogen) for sequence verification, digested using EcoRI and XhoI restriction enzymes, and ligated into pENTR 3C entry vector (Invitrogen). Human moesin-AA and moesin-EE entry vectors were generated from moesin-wt entry vector in two steps. First, T235A/E mutants were generated by PCR using the following primers: T235A (5′-GAATGACAGACTAGCACCCAAGATAGGC), T235A (3′- GCCTATCTTGGGTGCTAGTCTGTCATTC), T235E (5′-GAATGACAGACTAGAACCCAAGATAGGC), and T235E (3′-GCCTATCTTGGGTTCTAGTCTGTCATTC). PCR products and moesin-wt entry vector were digested using BglI and BglII restriction enzymes, ligated, and sequence verified. Second, AA/EE mutants were generated by PCR using the following primers: T558A (5′-GACAAATACAAGGCACTGCGCCAGATC), T558A (3′-GATCTGGCGCAGTGCCTTGTATTTGTC), T558E (5′-GACAAATACAAGGAACTGCGCCAGATC), and T558E (3′-GATCTGGCGCAGTTCCTTGTATTTGTC). PCR products and entry vectors from first step were digested using BglI and XhoI restriction enzymes, ligated, and sequence verified. All entry vectors were then exchanged into pDEST 12.2 expression vector (Invitrogen) by LR clonase reaction. Human ezrin entry and expression vectors were generated in a similar way by J.I.A. Thoms (Cornell University, Ithaca, NY). GST–ZAP-70 expression vector was provided by A.C. Chan (Genentech, Inc., South San Francisco, CA).
T cells or cell conjugates were plated onto poly--lysine–coated glass slides, fixed for 30 min at RT with 3.7% formaldehyde followed by permeabilization in 0.1% Triton X-100 in PBS for 2 min, and rinsed three times in PBS. Cells were then incubated for 1 h with 5% BSA in PBS followed by incubation with primary antibody in 5% BSA in PBS for 1 h, washed in PBS, and incubated with appropriate secondary antibody (or phalloidin) in 5% BSA in PBS for 1 h. After additional washes, 5 μl Vectashield (Vector Laboratories) was added to the cells, and slides were covered with coverslips. Cells were observed on a microscope (Eclipse TE-2000U; Nikon) with a 100× 1.4 NA lens using a spinning disk confocal imaging system (UltraView LCI; PerkinElmer) and a 12-bit digital output charge-coupled device camera (C4742-95-12ERG; Hamamatsu).
Jurkat and primary T cells were lysed and resolved by SDS-PAGE followed by transfer to polyvinylidene difluoride membranes (Immobilon-P; Millipore) using a semidry transfer system (Integrated Separation Systems). After 1-h blocking in 5% dry milk, TBS–Triton X-100 membranes were incubated with primary antibody for 1 h, washed, and incubated for 1 h with appropriate HRP-conjugated secondary antibody. Blots were developed using ECL (GE Healthcare).
Jurkat and primary human T cells were activated with 10 μg/ml OKT3 antibody for the indicated times. For stimulation with beads, 10 T cells were mixed with 2 × 10 anti-CD3/CD28–coated polystyrene beads (Invitrogen) for the indicated times. For stimulation with B cells, Raji B cells were fluorescently labeled with 5-chloromethylfluorescein diacetate (CellTracker; Invitrogen) and loaded with 2 μg/ml SEE superantigen (Toxin Technology). B cells were then mixed with an equal number of T cells for the indicated times.
Transient transfection of Jurkat T cells was performed using GenePORTER (Gene Therapy Systems) according to the manufacturer's instructions. Transfection efficiency was between 60 and 70% as determined by cotransfection with red fluorescent protein. Cells overexpressing untagged ERM mutants were easily identified by their higher fluorescent intensity compared with nontransfected cells.
Human peripheral blood lymphocytes were isolated from citrate-anticoagulated whole blood by dextran sedimentation (Blood Centers of America) followed by density separation over Ficoll-Hypaque (Sigma-Aldrich). The resulting mononuclear cells were washed in PBS and further purified by nylon wool and plastic adherence as described previously (). The resulting lymphocytes were >90% CD3 T lymphocytes as verified by FACS analysis.
Jurkat T cells were lysed in cold radioimmunoprecipitation assay buffer (0.1% SDS, 1% Triton X-100, 1% deoxycholate, 150 mM NaCl, 1 mM EDTA, and 25 mM Tris, pH 7.4) with protease inhibitors, and lysates were centrifuged for 10 min at 55,000 rpm at 4°C. Supernatants were mixed with either protein A–Sepharose beads (Sigma-Aldrich) precoated with antimoesin polyclonal antibodies or cyanogen bromide–activated Sepharose 4B beads (Sigma-Aldrich) precoated with antiezrin polyclonal antibodies and incubated for 2 h with constant rotating. Beads were washed six times with radioimmunoprecipitation assay buffer, and pellets were resuspended in SDS sample buffer, boiled, and resolved on 7.5% SDS-PAGE.
Jurkat T cells were cotransfected with the relevant ERM construct and red fluorescent protein. 48 h after transfection, cells were plated on a poly--lysine–coated glass-bottom culture dish and labeled with 1 μM Fluo-LOJO (TefLabs). Cells were observed before and during activation with 10 μg/ml OKT3 antibody on an Eclipse TE-2000U microscope using the UltraView LCI spinning disk confocal imaging system.
Protein bands excised from the SDS-PAGE gel were destained before digestion with trypsin overnight at 37°C. Peptides were extracted with a solution of 70% acetonitrile and 0.1% trifluoroacetic acid and were desalted using C-18 ZipTip (Millipore). The samples were then lyophilized and resuspended in 0.1% trifluoroacetic acid before mass spectrometry (MS) analysis. Protein identification was performed by nanoflow reversed phase liquid chromatography tandem MS using a nanoflow liquid chromatography system (Agilent 1100; Agilent Technologies) coupled online with a linear ion trap MS (LTQ; ThermoElectron). Nano-reversed phase liquid chromatography columns were slurry packed in-house with 5 μm of 300-Å pore size C-18 phase (Jupiter) in a 75-μm i.d. × 10-cm fused silica capillary (Polymicro Technologies) with a flame pulled tip. After sample injection, the column was washed for 20 min with 98% mobile phase A (0.1% formic acid/water) at 0.5 μl/min, and peptides were eluted using a linear gradient of 2–42% mobile phase B (0.1% formic acid/acetonitrile) for 40 min at 0.25 μl/min and then 98% mobile phase B for 10 min. The linear ion trap MS was operated in a data-dependent mode in which each full MS scan was followed by five MS/MS scans in which the five most abundant molecular ions were dynamically selected for collision-induced dissociation using normalized collision energy of 35%. Tandem mass spectra were searched against the UniProt human proteomic database from the European Bioinformatics Institute () using SEQUEST (ThermoElectron). The fully tryptic peptides that have SEQUEST cross correlation scores >2.0 (+1), 2.5 (+2), or 3.0 (+3) and ΔC values >0.1 were considered legitimate identifications.
ZAP-70 was subcloned into pFastBac HTb vector and transformed into Sf9 cells. Protein expression was performed according to the manufacturer's instructions (Invitrogen). Ezrin was expressed in the strain M15 (). Recombinant ZAP-70 was purified and immobilized on nickel beads as previously described (), and recombinant ezrin was purified as previously described (). For in vitro binding assay, similar amounts of ZAP-70 and ezrin recombinant proteins were mixed in cold buffer (50 mM Tris, pH 7.4, 100 mM NaCl, and 1 mM DTT) and incubated for 2 h with constant rotating. Beads were washed six times with the aforementioned buffer, and pellets were resuspended in SDS sample buffer, boiled, and resolved on 7.5% SDS-PAGE.
2.5 × 10 Jurkat T cells were electroporated once at 300 V and 700 μF using the Xcell Gene Pulser (Bio-Rad Laboratories) in the presence of 100 nM ezrin-specific siRNA or luciferase GL2–specific siRNA. Cells were cultured for 48 h and analyzed by immunoblotting or immunofluorescence.
Fig. S1 shows ERM expression in Jurkat T cells. Fig. S2 shows ERM dephosphorylation inhibition by calyculin A. Fig. S3 shows expression of transfected proteins in Jurkat T cells by Western blotting and immunofluorescence. Fig. S4 shows ERM protein localization in primary human T cells. Fig. S5 shows immunoprecipitation of ZAP-70 from Jurkat T cells. Online supplemental material is available at . |
Cdk1 and its associated protein cyclin B1 are required for entry into and maintenance of the mitotic state in mammalian cells (; ; ). Exit from mitosis in mammalian cells requires the inactivation of Cdk1, the protein kinase that drives the mitotic state (). Inactivation follows the destruction of the Cdk1-activating subunit cyclin B1 by proteolysis (; ; ), a process normally activated at metaphase by anaphase-promoting complex/cyclosome (APC/C)–driven ubiquitination (; ). Failure to degrade cyclin B1 results in constitutively active Cdk1 and indefinite arrest in mitosis (; ; ; ). As Cdk1 inactivation is not required for progression past metaphase, vertebrate cells and in vitro cell model systems can arrest either in metaphase or in later stages of mitosis in the presence of constitutively active Cdk1 (; ; ).
Inactivation of Cdk1 itself has been considered to be necessary and sufficient to induce a rapid exit from mitosis. Exposure of cells to specific inhibitors of Cdk1 causes rapid mitotic exit (). The APC/C E3 ubiquitin protein ligase processively ubiquitinates specific sequence tags (), principally D-box () or KEN-box () motifs, in multiple target proteins in the course of mitotic exit () and targets them for proteasome destruction. The degradation of two proteins, cyclin B1 and securin, is linked to proper mitotic exit. Destruction of cyclin B1 is absolutely necessary for mitotic exit (; ). Although the destruction of securin is required for proper chromosome segregation, failure to destroy securin does not block mitotic exit ().
In this study, we analyze the state of cells exposed to Cdk1 inhibitors in combination with the suppression of proteolysis and present evidence that the mitotic state (defined as the continuous presence of condensed chromosomes) of a mitotic spindle and of mitotic phosphoprotein antigens is sustained for a long period in the absence of Cdk1 activity, but only when APC/C-dependent protein degradation is simultaneously suppressed.
We find that the capacity to sustain mitotic status correlates with the persistence of phosphorylated Cdk1 substrates in the absence of Cdk1 activity. Thus, our results demonstrate that Cdk1 inactivation alone is not sufficient to induce mitotic exit. Instead, key serine/threonine protein phosphatases, which are required for mitotic exit, are largely inactive during mitosis and must be reactivated by a proteolytic event so that they, in turn, can dephosphorylate Cdk1 substrates and enable mitotic exit. Our results show an unexpected convergence of the mammalian system with yeast in which phosphatase activity is required for mitotic exit ().
HeLa cells were collected in mitosis by exposure to S-trityl--cysteine (STLC), a potent and specific inhibitor of the microtubule motor protein Eg5 (), or to nocodazole, an inhibitor of microtubule assembly (). We then tested the effect of cell exposure to the specific Cdk1 inhibitor roscovitine or to the protease inhibitor MG132. The mitotic state was determined by flow cytometric assay of the presence of MPM-2, a well-established mitosis-specific phosphoprotein substrate and mitotic marker (; ). As previously demonstrated (; ), exposure to Cdk inhibitors such as roscovitine for 2 h induced rapid mitotic exit (). On the other hand, exposure to MG132 sustained the mitotic state ().
We then tested the effect of exposing HeLa cells to the combination of roscovitine and MG132. Importantly, the presence of MG132 substantially prevented roscovitine-treated cells from losing MPM-2 phosphoproteins despite the absence of Cdk1 activity (). The retention of MPM-2 phosphoproteins provides evidence for continued mitotic status in the combined presence of roscovitine and MG132. Quantitation of flow cytometric data over a time course demonstrated that the percentage of mitotic cells exposed to both drugs initially diminished but then remained stable at ∼60% of the time 0 value (). The difference from the fate of cells exposed to roscovitine alone was striking. Essentially, no cells remained mitotic after 60 or 90 min of exposure to roscovitine in the continued presence of STLC or of nocodazole ().
Continued mitotic status, which is represented by the retention of MPM-2 phosphoantigen, was supported by the retention of phosphorylated histone H3. Histone H3 is phosphorylated on S10 during mitosis by Aurora kinase (; ; ), and phosphorylated H3 has been used as a specific mitotic marker (). Western blots showed that phosphorylated H3 was absent after 2 h of roscovitine exposure but was retained in the combined presence of roscovitine and MG132 (, (PS10)H3). Aurora A, securin, and cyclin B1, proteins normally degraded on mitotic exit (), were largely absent after 2 h of roscovitine treatment but were retained in cells exposed to the combination of roscovitine and MG132 ().
We obtained similar results after the exposure of cells to two other Cdk inhibitors, CGP74514A and purvalanol A. First, we conducted dose-response experiments to determine the minimum concentration of inhibitors capable of inducing the complete mitotic exit of nocodazole-arrested cells (unpublished data). Then, mitotic cells were exposed to the Cdk inhibitors in the presence of MG132. Results were comparable with those with roscovitine. Neither 7.5 μM CGP74514A nor 25 μM purvalanol A were able to drive mitotic exit in the presence of MG132 (). Furthermore, we also assayed the effect of substituting MG132 with another proteasome inhibitor, AM114, and obtained similar retention of the mitotic status in the absence of Cdk activity (). Western blot analysis of the extracts from treated cells showed similar patterns of the phosphorylation of Cdk substrates and of phosphorylated H3, Aurora A, and cdc27 (). Chromosome spreads confirmed the mitotic status of cells treated either with CGP74514A + MG132 or with CGP74514A + AM114 (). The different combinations of inhibitors of Cdk and proteasome argue against the possibility that our results are artifactually caused by MG132 specifically dampening the inhibitory activity of roscovitine. Additionally, we extended our 2D FACScan analysis to another human cell line, HCT116, and the results were qualitatively and quantitatively similar: compared with nocodazole-blocked cells, 85% of cells incubated with AM114 + roscovitine were MPM2 positive (unpublished data). Importantly, a nontransformed human cell line, IMR90, also remained substantially mitotic on exposure to combinations of either MG132 + roscovitine in the presence of nocodazole, AM114 + roscovitine, or AM114 + CGP74514A (Fig. S1, available at ). We conclude that nontransformed cells remain fully mitotic in different conditions of Cdk1 inhibition, a result comparable to that with transformed cells.
Immunofluorescence microscopy confirmed mitotic status in HeLa lacking Cdk1 activity. shows images with roscovitine + MG132. As expected, MG132 prevented the destruction of proteins normally degraded during mitotic exit. On continuous exposure to the Eg5 inhibitor STLC, control cells contained condensed chromosomes and a single aster spindle (tubulin label). The chromosomes were positive for the mitotic markers phosphohistone H3 and MPM-2. Lamin B, a marker for interphase nuclear envelopes, was dispersed, as is normal during mitosis (). In contrast, cells treated with roscovitine for 2 h in the presence of STLC exited mitosis and, thus, lost condensed chromosomes, the monoastral spindle, histone H3 phosphorylation, and the MPM-2 phosphoantigen but had gained a lamin nuclear border (). These controls showed the status of markers in continued mitotic arrest or in mitotic exit. On exposure to the combination of roscovitine + MG132, cells appeared like the mitotic cells blocked in STLC and showed no sign of mitotic exit by any criterion despite the absence of Cdk1 activity. All Cdk and proteasome inhibitor combinations yielded similar images with respect to mitotic exit (unpublished data).
We quantitated the percentage of the mitotic population that retained two mitotic markers, monoastral spindles or S10 phosphorylated histone H3, during 2 h of exposure to either roscovitine alone or to the combination of roscovitine and MG132 (). For this purpose, cells were first collected by mitotic shake-off after block in STLC, which yielded ∼80% mitotic cells that were positive for each of the two markers at time 0 (). By 1 h, roscovitine exposure resulted in almost a complete loss of monoastral spindles or phosphorylated H3. In contrast, cells exposed to the combination of roscovitine and MG132 almost completely retained these markers compared with time 0. Similar results were obtained with antiphospho-S10 histone H3 (). Other Cdk and proteasome inhibitors yielded similar data (unpublished data).
To exclude the possibility that our results were caused by artifactual interference of the proteasome inhibitor with suppression of Cdk1 activity by Cdk1 inhibitors, we conducted an in vitro assay of Cdk1 histone H1 phosphorylation activity in the presence of purvalanol A or roscovitine alone, or of Cdk1 histone H1 phosphorylation activity in the combined presence of purvalanol A + MG132 or roscovitine + MG132. The result () clearly showed that both purvalanol A and roscovitine substantially suppressed Cdk1 activity and that the further addition of MG132 did not interfere with Cdk1 inhibition. All drugs were assayed in vitro at the same concentrations that were effective in cells.
The prevention of mitotic exit using the combination of roscovitine, purvalanol, or CGP74514A with either MG132 or AM114 suggests that the suppression of degradation of a key protein protects the cell from mitotic exit in the absence of Cdk1 activity. To confirm that retention of mitotic status was not dictated by the continued presence of cyclin B1 or securin, we assayed the effect of retention of these two proteins, whose degradation is known to be essential for proper mitotic exit. For this assay, we expressed nondegradable mutants (cyclin B1 R4A2 and securin KEN DM [destruction box mutant]) of the two proteins using either single- or double-transfection protocols. Transfection with cyclin B R4A2 showed that the presence of nondegradable cyclin B was not sufficient to maintain mitosis because in the presence of roscovitine, expressing cells were negative for MPM-2, as was the case in nontransfected controls (). In contrast, transfected (GFP-positive cells) and nontransfected cells both remained positive for MPM-2 in the presence of roscovitine + MG132 after 2 h ().
After double transfection, cells retained their mitotic status in the presence of MG132 + roscovitine, as indicated by Western blot analysis of the mitotic-specific phosphorylation of Cdk substrates as well as by the presence of mitosis-specific markers, phosphorylated H3 and cdc27 (). We found that neither cyclin B1 R4A2 nor securin KEN DM was degraded on exposure of mitotic cells to roscovitine (). Nonetheless, cells rapidly exited mitosis, as assayed by the loss of S10 phosphohistone H3, PT244cdc27, and Cdk phosphosubstrates. As before, the combination of roscovitine and MG132 prevented mitotic exit. Endogenous cyclin B1 and securin acted as internal controls in these experiments and degraded during mitotic exit in the presence of roscovitine (). Similar Western blot results were obtained after single transfection with either cyclin B1 R4A2 or securin KEN DM alone (unpublished data).
The combined presence of a Cdk1 inhibitor and a protease inhibitor induced mitotic cells to proceed to cell cleavage (). We have carefully observed the status of cells blocked by a combination of roscovitine and MG132, and we have found the initiation of monoastral furrowing in a substantial subpopulation of cells. This furrowing event was essentially the same as that previously observed in cells treated with monastrol (another inhibitor of Eg5) and forced to exit mitosis by suppression of the spindle assembly checkpoint (). Indeed, cells frequently formed what we interpreted as a bud, creating a small daughter cell that contained no chromatin (). Such bud formation occurred with approximately equal frequency in cells treated with roscovitine alone or with the combination of roscovitine and MG132. The difference in outcome was that treatment with roscovitine alone caused the loss of mitotic chromosomes, whereas budded cells treated with both roscovitine and MG132 retained mitotic chromosomes.
The budding was a true furrowing event, as the passenger proteins survivin (), Aurora B (), and TD60 () relocalized to the neck of the bud both in cells treated with roscovitine alone or with the combination of roscovitine and MG132 (). Furthermore, PRC1 and anillin, two proteins that localize to the furrow and are critical to cell cleavage (; ; , ), were also present at the bud necks (). The percentage of cells that exhibited budding is summarized (). We interpret these results as indicating that furrowing occurs in the absence of Cdk1 activity and in the absence of proteolysis. However, such furrowing occurs in the absence of mitotic exit.
Because monoastral cells proceeded to furrow in the presence of roscovitine and MG132, we next addressed the fate of cells released from STLC block in the presence of roscovitine and MG132. The question was whether cells without Cdk1 activity would proceed through the normal stages of mitosis or would proceed directly to furrowing in the absence of Cdk1 activity. For this experiment, HeLa mitotic cells were collected by shake-off after 18 h of block in STLC and were released from STLC in the presence of MG132 alone or roscovitine + MG132. When cells were released without further treatment, <17% remained mitotic at 4 h as determined by FACScan analysis (). In contrast, release into MG132 for 2 h yielded 68% in mitosis compared with 73.5% in the initial mitotic population. The same result was obtained when cells were released into MG132 for 2 h and treated with a combination of roscovitine and MG132 for a further 2 h. Despite the absence of Cdk1 activity, there was no further loss of mitotic population (66.2% at 4 h). Observed by immunofluorescence microscopy, the population of cells treated with the combination of roscovitine and MG132 was largely prometaphase (bipolar spindles with most chromosomes aligned to the metaphase plate but with a few chromosomes misaligned) or metaphase (bipolar spindles with all chromosomes aligned to the metaphase plate), as shown in . Thus, the absence of Cdk1 activity did not prevent cells from maintaining a bipolar spindle and a metaphase chromosome array. Furthermore, no cells treated in this manner proceeded to furrow in this time frame, but all remained in early mitosis in the absence of Cdk1 activity (Fig. S2 A, available at ). It is of interest to note that coincubation with roscovitine and MG132 caused the reproducible loss of metaphase cells by reversion to prometaphase (Fig. S2 A).
Maintenance of the mitotic state in the absence of Cdk1 activity might depend on the retention of phosphorylated Cdk1 substrates during combined Cdk1 suppression + MG132 treatment. In , using an antibody specific for Cdk1 substrate consensus motif phosphopeptides, we showed that the phosphorylation status of Cdk1 substrates () was indeed retained in cells exposed to the combination of roscovitine + MG132 or CGP74514A + MG132. Western blots showed that mitotic cells were highly positive for a large number of Cdk1 phosphoepitopes and that these substrates were absent after 2 h of Cdk1 inhibitor exposure ().
These results suggested that protein phosphatases were not acting on Cdk1 substrates in the absence of Cdk1 activity. As this offered a possible mechanism for sustained mitotic status without Cdk1, we directly assayed for the effect of phosphatase inhibition in cells treated with roscovitine and found that the dephosphorylation of Cdk1 substrates was indeed suppressed in mitotic cells exposed to the combination of roscovitine plus either okadaic acid or calyculin A (), two specific inhibitors of PP1 and PP2A protein phosphatases (; ; ). These results correlated with FACScan data showing retention of the mitotic marker MPM-2 in STLC-treated cells (). Results quantitated for cells collected in mitosis either with STLC or with nocodazole () show substantial mitotic retention with roscovitine + either okadaic acid or calyculin A. In comparison, the inhibition of cdc25, a mitotic protein phosphatase (for review see ), was without effect on mitotic exit of roscovitine-treated cells, nor did its suppression protect cells from the dephosphorylation of Cdk1 substrates (Fig. S3, available at ).
Chromosome spreads of cells treated with Rosc + either okadaic acid or calyculin A showed that condensed mitotic chromosomes were maintained at high levels with the combined suppression of Cdk1 and phosphatase activity (Fig. S4, available at ). Representative chromosome spreads are shown in .
Condensed chromosomes were equivalently maintained with either MG132 alone or Rosc + MG132 but unlike roscovitine alone, in which the majority of cells had reformed nuclei by 2 h (Fig. S4). Interestingly, mitotic arrest with Rosc + either okadaic acid or calyculin A occurred despite the loss of cyclin B1 and securin by 2 h (). Thus, phosphatase activity did not appear to be required for activation of cyclin B1 or securin degradation by APC/C-linked proteolysis.
Maintenance of the mitotic state depends on Cdk1 activity. Loss of Cdk1 activity normally occurs at the metaphase to anaphase transition once the spindle assembly checkpoint and other mitotic checkpoints have been satisfied (). Mitotic exit normally occurs through the cdc20-activated APC/C-dependent degradation of two key substrates, cyclin B1 and securin.
We have demonstrated that mitotic cells will remain mitotic for several hours in the absence of Cdk1 activity provided that APC/C-dependent protease activity is suppressed. Continued mitotic status in the absence of Cdk1 activity has been verified by several independent criteria: the continued presence of condensed chromosomes, the presence of a mitotic spindle, the presence of the mitosis-specific phosphoantigen markers MPM-2 and S10 of histone H3, and the presence on chromosomes of mitotic passenger proteins. Finally, and most importantly, we show that Cdk1 substrates remain phosphorylated in the absence of Cdk1 activity.
Thus, our results require a revision of the prevailing paradigm, which holds that destruction of cyclin B1, which inactivates Cdk1, is itself necessary and sufficient to induce mitotic exit. This paradigm requires that suppression of Cdk1 activity should therefore induce mitotic exit even in the absence of cyclin B1 destruction. Instead, our results show there is a pathway downstream of Cdk1 inactivation that requires both further proteolysis and phosphatase activation to complete the mitotic exit pathway.
Our results contrast with a recent study that proposed that cells exposed to both a Cdk1 inhibitor and an inhibitor of proteolysis undergo cell cleavage but remain competent to revert to mitosis when the Cdk1 inhibitor is removed (). The interpretation of these results as evidence for a reversible exit from mitosis dependent on the continued presence of cyclin B1 rested on the reversibility of cell cleavage and the reappearance of a rounded mitotic cell. Other markers that would confirm that mitotic exit had indeed occurred in the absence of Cdk1 activity combined with the suppression of proteolysis were not examined except phosphorylated nucleolin.
suggested that mitotic exit occurred with some Cdk1 inhibitors such as purvalanol A but not with roscovitine. Importantly, our results have been obtained with three different Cdk inhibitors and with two different proteasome inhibitors, eliminating the possibility that the capacity to remain in mitosis in the absence of Cdk1 activity is dependent on a particular drug combination.
In accord with , we find that a portion of blocked monoastral cells undergo transient furrowing. This furrowing is accompanied by the relocation of proteins important to cytokinesis, such as the passenger proteins Aurora B, survivin, and TD60, and of PRC1 to the cell cortex along a bundle of microtubules. As a result, these cells form a bud that contains no chromatin. Such furrowing is quite similar to that previously shown to occur in cells with monoastral spindles in which the metaphase checkpoint had been suppressed by the introduction of mutant Mad2 (). These results and previous work (; ; ; ) indicate that cytokinesis is an event that is independent of, and neither synchronous nor synonymous with, mitotic exit. Furrowing can occur well after mitotic exit has occurred or, as in the case presented here, can occur in the absence of mitotic exit. However, induction to undergo furrowing may depend on the absence of Cdk1 activity.
Cells released from STLC arrest in the presence of MG132 proceed to metaphase and maintain a bipolar metaphase spindle for at least an additional 2 h after the suppression of Cdk1 activity by roscovitine. The absence of Cdk1 activity clearly does not, of itself, drive cells forward from metaphase to anaphase. It is of interest that a substantial percentage of cells revert from metaphase to prometaphase when Cdk1 activity is suppressed in MG132-treated cells (Fig. S2), whereas none are driven forward to anaphase. When chromosomes are lost from the metaphase plate, they appear to have a merotelic (both kinetochores associated with one spindle pole) kinetochore alignment (unpublished data). Therefore, it appears that continuous Cdk1 activity is required to maintain proper metaphase chromosome alignment. This apparently unique role for Cdk1 activity in maintaining bipolar kinetochore attachment has not been noted before and deserves attention.
Retention of cells in mitosis is not caused by the continued presence of either cyclin B1 or securin in the absence of Cdk1 activity, as nondegradable mutants of these proteins do not prevent mitotic exit in the absence of Cdk1 activity. As MG132 nonetheless prevents mitotic exit, the reasonable conclusion is that a protease substrate other than cyclin B1 or securin must be degraded to permit mitotic exit in the absence of Cdk1 activity. Two proteins other than cyclin B1 and securin must be degraded for mitosis to progress, but these proteins, cyclin A (; ) and Emi1 (; ), are both eliminated very early in mitosis and are unlikely to play a role in mitotic exit.
The putative protease substrate must be involved in phosphatase activation. Cdk1 substrates remain substantially phosphorylated for hours in the combined presence of different Cdk and proteasome inhibitors, and we obtain parallel results on exposing cells to the combination of roscovitine with protein phosphatase inhibitors (okadaic acid or calyculin A).
Suppression of the PP1 and PP2A protein phosphatase families is required for entry into mitosis (; ; ; ; ), and both okadaic acid and calyculin A specifically inhibit the PP1 and PP2A protein phosphatase families (). Reasonable candidates for control of mitotic exit are members of the PP1 family, as they remain suppressed in mammalian cells by phosphorylation until metaphase and can be prolonged in this suppressed state by exposure of mitotic cells to either okadaic acid or calyculin A (). Furthermore, microinjection of anti-PP1 antibody arrests mammalian cells at metaphase (). Similarly, the two PP1 proteins in are suppressed by cdc2 phosphorylation in mitosis, and their reactivation is required to proceed past metaphase (). PP1 activity is also required for mitotic exit in () and (; ).
Our results with the combination of roscovitine and the phosphatase inhibitors okadaic acid or calyculin A indicate that phosphatase activity is required for exit from the mitotic state, presumably by the dephosphorylation of mitotic Cdk1 substrates, and, importantly, that there is little phosphatase activity evident on Cdk1 substrates in the presence of roscovitine and MG132. The continuing mitotic state, which is characterized by condensed chromosomes and by the presence of MPM-2 and S10-phosphorylated histone H3 markers, indicates that the phosphatase activity required for mitotic exit is minimal when APC/C-dependent proteolysis has been suppressed by MG132. The potential role of protease inhibition in the suppression of phosphatase-dependent mitotic exit is of substantial interest.
The key phosphatase activity must be downstream of the cdc20-driven APC/C-dependent protease activity, as cells treated with phosphatase inhibitors in the presence of roscovitine have lost both cyclin B1 and securin () but remain metabolically mitotic through the absence of the phosphatase-dependent destruction of Cdk1 substrates.
In light of a potential key role for protein phosphatases in mitotic exit, it is important to note that both budding and fission yeast contain mitotic exit networks that are dependent on unique protein phosphatases of the cdc14 family (; ; ). Although mammalian cdc14A and B are functional homologues of cdc14 phosphatases in the yeast system (), suppression of cdc14A in mammalian cells does not prevent mitotic exit (). We do not believe that the cdc14 mechanism is likely to play an equivalent role in the pathway we describe here, as cdc14 (Clp1/Flp1) directly inactivates cdc2 (the homologue of Cdk1) by regulating its phosphorylation status. In this scenario, phosphatase inhibitors would not be expected to retain mitotic status in the absence of Cdk1 activity. In light of our results with combined Cdk and phosphatase inhibitors, it is interesting that in , Cdk activity is required to activate Net1 and, thus, cdc14 in the mitotic exit pathway and that this overrides PP2A-dependent metaphase arrest (). It will be of substantial interest if a parallel exists between our results and the yeast pathway that uses Cdk activity and phosphatase activation for mitotic exit.
Our results strongly implicate phosphatase activity in key events in mitotic exit and protease in activating this pathway. Although it remains to be seen whether these effects directly or indirectly involve cdc14, it is nonetheless of great interest that there appears to be a convergence between the yeast systems and the mammalian system in the functional requirement for activation of protein phosphatase activity to enable mitotic exit.
In summary, we find that cells without Cdk1 activity remain mitotic by several criteria as long as there is no APC/C-dependent protease activity and that this effect is dependent on mitotic suppression of a key protein phosphatase activity. It will now be of great interest to elucidate this important pathway, to determine the protease substrate that is retaining these cells in mitosis, and to identify the protein phosphatase whose activation is apparently required for mitotic exit.
HeLa cells were grown in DME (Invitrogen) supplemented with 10% FCS and were maintained in a humid incubator in a 5% CO environment at 37°C. Mitotic cells were collected by shake-off after 16-h incubation with either 0.1 μg/ml nocodazole (Sigma-Aldrich) or with 7.5 μM STLC (Novabiochem). After centrifugation at 1,000 rpm for 5 min at 37°C, cells were resuspended in media in the continuous presence of the mitotic inhibitors with either 100 μM roscovitine (Calbiochem) or a combination of roscovitine with 20 μM MG132 (Sigma-Aldrich), or cells were resuspended with 0.5 μM okadaic acid (Calbiochem) or 30 nM calyculin A (Calbiochem) for up to 2 h. Alternatively, mitotic cells were resuspended in media in the continuous presence of the mitotic inhibitors with either 7.5 μM CGP74514A (Calbiochem) or 25 μM purvalanol A (Calbiochem) or with the combination of CGP74514A or purvalanol A with either 20 μM MG132 (Sigma-Aldrich) or AM114 (Calbiochem) for up to 2 h. In experiments using CGP74514A or purvalanol A, proteasome inhibitors were added 1 h before addition of the Cdk inhibitors.
Cells were analyzed by 2D flow cytometry using MPM-2 monoclonal antibody recognizing mitosis-specific phosphoepitopes () and propidium iodide, a marker of DNA content. Cells were fixed, incubated with MPM-2 antibodies, and labeled with FITC-conjugated anti–mouse IgG secondary antibodies (Jackson ImmunoResearch Laboratories) and propidium iodide as described previously (). Data were collected with a flow cytometer (FACScan; Becton Dickinson) using propidium iodide in the first dimension and MPM-2 in the second dimension (presented as a dot plot) and were analyzed with CellQuest software (Becton Dickinson). For each sample, 10,000 events were collected, and aggregated cells were gated out.
Mitotic cells were fixed with 2% PFA in PBS for 20 min at 37°C, permeabilized with 0.2% Triton X-100 in PBS for 3 min, stained in suspension, and spun onto coverslips at 215 for 3 min. The following antibodies were used for indirect immunofluorescence microscopy. Monoclonal antibody to β tubulin (Sigma-Aldrich) was used at a 300-fold dilution, MPM-2 mouse monoclonal antibody (Upstate Biotechnology) was used at 1:100, and Aurora B was detected with a rabbit polyclonal antibody (Abcam) at a 500-fold dilution. PRC1 affinity-purified rabbit antibody (gift from W. Jiang, Burnham Institute of Medial Research, La Jolla, CA; ), JH human autoimmune serum recognizing human TD60 (), and rabbit polyclonal antisurvivin (Novus Biological) were all used at 500-fold dilutions. Goat anti–lamin B (Santa Cruz Biotechnology, Inc.) was used at a 300-fold dilution. Rabbit polyclonal antibody to anillin (gift from C. Field, Harvard University, Cambridge, MA) was used at a 600-fold dilution. Secondary antibodies (Jackson ImmunoResearch Laboratories), including FITC-conjugated affinity-purified goat anti–mouse, anti–rabbit, and rabbit anti–goat IgG antibodies, were used at 1:250; Cy3-conjugated affinity-purified goat anti–mouse was used at 1:400. DNA was detected by incubation with 0.2 μg/ml propidium iodide in PBS for 5 min after incubation with secondary antibodies. Samples were observed using a microscope (Optiphot; Nikon) coupled to a laser-scanning confocal apparatus (MRC-600; Bio-Rad Laboratories) using the image acquisition software COMOS (Bio-Rad Laboratories); scans were obtained using a 40× NA 1.0 oil objective (Nikon). Treated cells, which were transfected with the nondegradable (R42A)-cyclin B1-GFP, were fixed and stained in the same manner as the above samples and were finally stained with DAPI with Vectashield mounting medium (Vector Laboratories). Images were acquired with an epifluorescence microscope (BX61; Olympus) equipped with a CCD camera (Retiga-SRV; QImaging) driven by Volocity software (Improvision) with a binning of 2 using a planApo 60× NA 1.42 objective (Olympus). Figures were processed in Photoshop version 7.0 (Adobe) and assembled in CANVAS version 8.0 (Denaba Systems).
Kinase assays were performed as described previously (). In brief, HeLa cells were treated with 0.2 μg/ml nocodazole for 16 h, and mitotic cells were collected by shake-off, washed once with PBS, and lysed in ice-cold buffer C as described previously () containing fresh protease inhibitors (2 μg/ml leupeptin, 20 μg/ml aprotinin, 2 μg/ml pepstatin, and 1 mM PMSF), phosphatase inhibitor (1.3 mM -nitrophenyl phosphate), and 2 mM DTT. Mouse monoclonal anti-Cdk1 antibody (Abcam) was incubated on ice with mitotic lysate (2 μl/100 μg protein) for 30 min before the addition of protein A–Sepharose beads (Zymed Laboratories). After the addition of beads, the lysates were incubated for 5 h at 4°C, rotating end over end. Immune complexes were washed once with 600 μl buffer C and twice with kinase assay buffer (20 mM Tris, pH 7.6, and 20 mM MgCl). A master kinase cocktail mix was prepared containing kinase assay buffer, 1 μg/ml histone H1, 0.2 μCi/μl γ-[P]ATP, 30 μM ATP (VWR), and 1 μM DTT. Master kinase cocktail was added to separate tubes containing water (vehicle), purvalanol A (25 μM final), roscovitine (100 μM final), MG132 (25 μM final), purvalanol A and MG132 (25 μM final each), or roscovitine and MG132 (100 μM final and 25 μM final, respectively). For immunoblots, immunoprecipitated Cdk1 complexes were resolved by SDS PAGE, transferred to polyvinylidene difluoride membranes, blocked with 5% nonfat milk, and probed with anti-Cdk1 antibody (1:500 dilution; Abcam).
To prepare chromosome spreads, cells in suspension were pelleted by centrifugation at 215 for 2 min and were resuspended in 75 mM KCl at 37°C for 30 min. Cells were then centrifuged again (2 min at 215 ) and fixed overnight at −20°C in 75% methanol and 25% acetic acid. Spreads were then obtained by resuspension of fixed cells and centrifugation onto coverslips (Cytospin centrifuge; Thermo Scientific) at 2,800 rpm for 3 min. The chromosome spread was then washed with PBS, stained for 5 min with 0.5 μg/ml propidium iodide in PBS, washed again twice, and mounted for observation.
Cells were lysed in 50 mM Tris-HCl, pH 7.4, 250 mM NaCl, 5 mM EGTA, and 0.1% NP-40 and were supplemented with protease and phosphatase inhibitors for 30 min on ice. 20 μg of lysates was resolved on polyacrylamide gels and transferred to nitrocellulose sheets. Materials are listed with their dilutions as follows: β tubulin was detected with a mouse anti–β tubulin monoclonal antibody (Sigma-Aldrich), 1:1,000; mouse anti–cyclin B1 monoclonal antibody (clone GNS1; Santa Cruz Biotechnology, Inc.), 2,000-fold dilution; rabbit polyclonal anti–Aurora A (Cell Signaling), 1,000-fold dilution; polyclonal antisecurin antibody (gift from H. Zou, University of Texas Southwestern Medical Center at Dallas, Dallas, TX; ), 500-fold dilution; polyclonal antiactin, 10,000-fold dilution (Sigma-Aldrich); rabbit antiphospho-S10 H3 (Upstate Biotechnology), 5,000-fold dilution; mouse monoclonal anti-myc (clone 9E10; Santa Cruz Biotechnology, Inc.), 1,000-fold dilution; rabbit antiphospho-(serine)-CDK substrate antibody (Cell Signaling), 1,000-fold dilution; and rabbit antiphospho-(T244)-cdc27 antibody (Abcam), 1,000-fold dilution. Nitrocellulose sheets were then incubated with HRP-conjugated goat anti–mouse and anti–rabbit IgG secondary antibodies. Protein–antibody complex was detected by enhanced chemiluminescence (Pierce Chemical Co.).
Fig. S1 shows that nontransformed cells are retained in mitosis after loss of Cdk activity when proteasome activity is also suppressed. Fig. S2 shows that the absence of Cdk1 activity permits bipolar spindle formation and metaphase chromosome alignment on release from STLC, but many cells revert to prometaphase. Fig. S3 shows that cdc25 phosphatase activity has no effect on mitotic exit. Fig. S4 shows quantitation of treated cells for chromosomes versus interphase nuclei. Online supplemental material is available at . |
Members of the Myc family of nuclear protooncogenes are known to stimulate cell division and transformation. However, the roles of Myc in cell differentiation have been difficult to assess, in part because of the early embryonic lethality of N- () and c- ()–null mice and the difficulties in experimentally distinguishing proliferation from differentiation using in vitro approaches. In this study, we used B cell–specific deletion or overexpression of c- and/or N- to explore the roles of Myc in B lymphocyte development and lymphoma formation. B cell development in the bone marrow (BM) and fetal liver are marked by a series of cell fate decisions, which are controlled by checkpoints that ensure that all mature B lymphocytes are capable of producing functional antibodies (for review see ). Pluripotent hematopoietic progenitors initially become committed to the B lineage in response to growth factors and stromal cell interactions. The least mature committed B cell progenitors, called pro–B cells, have their Ig heavy chain (HC) and light chain (LC) genes in germline (GL) configuration. The Ig HC genes rearrange before the Ig LCs, and with successful in-frame D-J and then V-D-J juxtaposition, the Ig HCμ appears at the cell surface along with the signal-transducing proteins Igα and Igβ and the surrogate LC λ5 and Vpre-B1 and 2 as a complex called the pre–B cell receptor (BCR; termed large pre–B cells). The pre-BCR then signals large pre–B cells to proliferate, mature to the small pre–B cell stage, extinguish rearrangement of the other allele of the Ig HC (a process called allelic exclusion), and initiate Ig LC transcription and rearrangement (Vκ and Vλ). Successful in-frame Ig V-J rearrangement allows the Ig HC and LC proteins to pair and form IgM molecules on the surface of immature B cells, at which point they migrate from the BM to the periphery, where development continues.
Formation of the pre-BCR represents a critical checkpoint for ensuring that maturing B cells have in-frame V-D-J gene rearrangements. Signaling is induced by pre-BCR aggregation in lipid rafts, resulting in activation of the Src family protein tyrosine kinases (PTKs) Blk, Lyn, and Fyn, which then phosphorylate tyrosine residues within Igα and Igβ. This results in recruitment and activation of the Syk PTK (and to a lesser extent ZAP70) and the adaptor SLP-65 (also known as BASH or BLNK). Phosphorylation of SLP-65 by Syk further activates multiple signaling pathways, including the Ras–Raf and phosphatidylinositol 3-kinase (PI3K) pathways, as well as the Tec family PTKs Btk and Tec, which, together with Syk, activate PLCγ, resulting in Ca mobilization (see ; for review see ). Altogether, these events result in expression and/or activation of a set of transcription factors including extracellular regulated kinase, nuclear factor of activated T cells (NFAT), nuclear factor κB (NF-κB), and Myc.
Targeted deletion studies in mice have revealed essential roles for the majority of the described surface and cytoplasmic molecules in the development of pre–B cells (for review see ). However, the roles of nuclear factors such as Myc in B cell development remain unclear, in part because of genetic redundancy by related transcription family members or lethality after tissue-wide deletion. The Myc family of basic helix-loop-helix transcription factors bind DNA sequences called E boxes (CACGTG) as a heterodimer with the related basic helix-loop-helix factor Max (for review see ). This results in initiation of transcription, although transcriptional repression by Myc has been noted, probably through inhibition of the Miz-1 transcription factor (; ). Of the three mammalian family members, only c- and N- are expressed in B lymphocytes. c- is initially expressed during the pro–B cell stage in response to cytokines, including IL-7 (; ). After pre-BCR expression in large pre–B cells, both c- and N- are expressed during the maturation and expansion to small pre–B cells (). Thereafter, only c- is expressed in immature and mature B cells after B cell activation. In this study, we took a unique genetic approach in mice to show that Myc stimulates B cell differentiation and expansion downstream of the pre-BCR and Tec PTKs. Moreover, we uncover a novel feature of Myc in amplifying [Ca] as a mechanism to concurrently stimulate the expression of both Myc and Ca-regulated genes, which are essential for both cell division and differentiation.
To examine the roles of Myc in early B cell development and to overcome potential functional redundancy between c- and N-, we generated mice deficient in both N- and c- genes selectively in B cells using the Cre-loxP system (). c- mice () were bred to N- mice (), which also carried the CD19 transgene (Tg), to generate N- c- CD19cre mice. The CD19 promoter drives expression of the Cre recombinase only in B lineage cells throughout their development (). Semi-quantitative PCR (qPCR) analysis of sorted B220CD43 and B220CD43 B cell progenitors in the BM revealed ∼75/90% deletion of c- and ∼70/95% deletion of N- genes, respectively, in the presence of CD19cre (Fig. S1 A, available at ).
Total BM cells were isolated from mice deficient in c- and/or N- selectively in B cells, and B cell development was analyzed using flow cytometry (see ; ). Decreasing c- expression alone (c- CD19cre) impairs the generation of B220CD43 pre–B cells, whereas decreasing N- expression alone (N- CD19cre) does not consistently have any effect (unpublished data). However, depletion of total expression in N- c- CD19cre mice results in maximal decreases in the percentage () and number () of B220CD43, B220IA pre–B cells, immature B cells (B220IgM), and mature B cells (B220IgM) relative to CD19cre mice. These results suggest that Myc is required for pro–B to pre–B cell maturation.
In mice, pre–B cells proliferate in response to cytokines such as IL-7 produced by BM stromal cells (). Because IL-7 signaling is known to stimulate the expression of c-, we examined the ability of B cell progenitors to divide in response to IL-7. Total BM cells from N- c- CD19cre or CD19cre mice were labeled with 5-carboxyfluorescein diacetate succinimidyl ester (CFSE) and cultured in the presence or absence of IL-7 for 36 h, and the cell division profile of B cell progenitors was examined. As shown in , the majority of B cell progenitors from either N- c- CD19cre or CD19cre mice do not divide in the absence of IL-7. However, in response to IL-7, pre–B cells from CD19cre mice undergo an average of four divisions, whereas cells from CD19cre mice either do not divide at all or undergo several divisions, likely as a result of the loss of cells that contain the deleted c- alleles (Fig. S1 B; ).
To determine whether the reduction in levels affects the development of peripheral B cells, we purified splenocytes from N- c- CD19cre and CD19cre mice and characterized B cell subsets by flow cytometry. Splenic B cells are subdivided into additional stages of maturation and function, including transitional 1 and 2 (T1 and T2), follicular mature, and marginal zone (MZ; ). As shown in , CD19cre mice exhibit a significant reduction in the number of all B cell subsets. This correlates with reduced serum IgM, IgG2a, and IgG1 in CD19cre compared with CD19cre mice after immunization with keyhole limpet hemocyanin (KLH; ). These results suggest that Myc regulates final B cell numbers in part by regulating pre–B cell development and, potentially, peripheral B cell differentiation.
Peripheral B cells can be further classified as B2 conventional B cells, which reside in lymphoid tissues such as the spleen, and as B1 B cells, which represent the majority of B cells found in peritoneal and pleural cavities and are IgM, CD5, and CD23. As shown in , deletion of c- and N- together results in a decrease in both CD5IgM peritoneal B1 B cells and CD5IgM B2 B cells. These results suggest that Myc is essential for the development of both B2 and B1 B cells and for establishing final B cell numbers.
It is unclear from the gene-targeting strategy whether the effects of Myc are exclusively in modulating the expansion of a population of cells that have already differentiated to the pre–B cell stage or whether Myc may also stimulate the differentiation of pre-BCR cells. To address this question, we took a classical genetic approach to determine whether Myc could rescue B cell differentiation in the absence of upstream signaling molecules known to be required for both expression and pre–B cell differentiation. In pre–B cells, genes are expressed, in part, by a pathway involving pre-BCR formation; the activation of Tec kinases, PLCγ, and PKC; and NF-κB translocation to the promoter (see ; ). First, we determined whether Myc could rescue B cell differentiation in the absence of pre-BCR formation by breeding Eμ- transgenic mice (), which express c- exclusively in B lineage cells throughout development, to mice deficient in recombinase-activating gene 2 (RAG2; ), a gene required for the initiation of V(D)J recombination. RAG2 mice lack pre-BCR expression, resulting in a complete block in B cell development at the B220CD43CD25 pro–B cell stage (, Hardy fraction C). We find that c- Tg is sufficient to stimulate the differentiation of RAG2-deficient pro–B cells based on the acquisition of pre–B cell characteristics, including the down-regulation of CD43 and the up-regulation of CD25, IA, CD22, and heat-stable antigen (HSA; ). This translates to a rescue in the relative number of pre–B-like cells to a level equivalent to that of wild-type (Wt) mice based on an increase in the number and ratio of B220CD43 cells to B220CD43 cells in BM from Eμ-/RAG2 as compared with RAG2 mice (unpublished data). c-Myc also stimulates GL transcription of the Igκ LC in purified B cell progenitors from Eμ-/RAG2 as compared with RAG2 mice (). Myc does not appear to induce allelic exclusion of the Ig HC (Fig. S2 C, available at ), suggesting that Myc Tg is not sufficient to mediate all of the signals from the pre-BCR.
To further examine the proliferative capacity of Eμ-/RAG2 versus RAG2 B cell progenitors, we labeled BM cells with BrdU, which is incorporated into DNA during cell division, and with CFSE. The percentage of B220BrdU cells is higher in Eμ-/RAG2 compared with RAG2 mice (Fig. S2 A), and B cells from Eμ-/RAG2 mice undergo considerably more cell divisions than B cells from RAG2 mice in response to IL-7 stimulation (). In contrast, we do not see any difference in the number of apoptotic B220 cells between Eμ-/RAG2 compared with RAG2 mice based on the acquisition of the apoptotic marker annexin V (Fig. S2 B).
To further characterize the upstream signaling pathway involved in Myc-dependent differentiation, we determined whether Myc could rescue B cell differentiation in the absence of the Tec family PTKs, which mediate proliferation and differentiation downstream of the pre-BCR and BCR (see ). We bred Eμ- mice to mice doubly deficient in expression of the Tec family PTKs Btk and Tec (). In humans, loss of function mutations in the gene lead to X-linked agammaglobulinemia, an immunological disease whereby Ca signaling is impaired (; for review see ) and the generation of pre–B cells is blocked. In mice, pre–B cell development is only partially inhibited in -deficient mice and proceeds normally in -deficient mice but is essentially completely inhibited in double-null mice. Thus, we asked whether Myc could rescue B cell development when Tec family signaling is ablated. As shown in ,
mice are deficient in their ability to generate pre–B cells based on a reduction in B220CD43 B cell progenitors compared with Wt mice.
B cell progenitors, both the percentage () and total number (not depicted) of B220CD43, B220CD22, and B220IA pre–B cells and B220IgM immature B cells are completely rescued. These results provide additional genetic evidence that Myc stimulates B cell development downstream of the pre-BCR and Tec family PTKs.
Because Tec kinases have been reported to both induce and suppress tumor formation (), we investigated how the transformation capacity of Eμ- Tg is altered by the loss of Btk and Tec. We find that the heterozygosity of and is sufficient to increase the tumor frequency in Eμ- mice, with the 50% tumor incidence being reduced from 100 d for Eμ- mice to 60 d for Eμ-/
(). The percent tumor-free incidence was decreased further as additional alleles of or were deleted. These results suggest that deregulated c- synergizes with the loss of Tec signaling during B cell lymphoma formation. Interestingly, ∼75% of the tumors that develop in Eμ-
or Eμ-
mice express surface IgM (unpublished data), indicating that they have an immature B cell phenotype relative to the majority of tumors derived from Eμ- Tg mice, of which >80% are IgM pre–B cell tumors ().
Ca signaling is induced by the pre-BCR and is modulated by Btk (for review see ), and, in T lymphocytes () as well as in many other cell types, is required for the induction of differentiation. Because Myc rescues B cell differentiation in the absence of the pre-BCR and Btk, we investigated whether Myc can bypass requirements for normal Ca signaling or whether Myc amplifies Ca signaling in the absence of the pre-BCR or Btk. In B cells, the sustained Ca flux after BCR cross-linking is thought to be caused by the concurrent activation of PI3K, which leads to the production of membrane-associated PI-3,4,5-P3 and to the recruitment and phosphorylation of Btk. Btk activation results in full and sustained PLCγ activation, peak IP production, and maximal release of [Ca] (see ; for reviews see ; ).
To investigate whether Ca signaling is affected during Myc-dependent differentiation, we labeled total BM cells from Eμ- Tg and Wt mice as well as Eμ-/RAG2 and RAG2 mice with the Ca-binding dye indo-1 () and fluorescent-conjugated antibodies against B cell markers and measured the ability of B cell progenitors to flux Ca after stimulation with the Ca ionophore ionomycin or anti-IgM. Surprisingly, we find that B cell progenitors from both Eμ- Tg and Eμ-/RAG2 mice exhibit elevated basal [Ca] and prolonged duration of Ca flux compared with Wt and RAG2 mice, respectively ( and not depicted). Remarkably, Eμ- Tg also rescues peak and sustained [Ca] in B cells (before and after anti-μ stimulation) in the absence of Btk and Tec kinases (Eμ- Tg/
), which are the major mediators of sustained Ca signaling downstream of the pre-BCR (). These results suggest that Myc amplifies [Ca] signaling in B cells downstream of the pre-BCR and Tec kinases.
We next determined whether the elevated Ca flux we observe in Myc-expressing B cells results in changes in known nuclear mediators of Ca signaling. In normal T and B cells, Ca influx activates the phosphatase calcineurin (Cn), which dephosphorylates the NFAT family of transcription factors (NFATc1–4), allowing them to translocate from the cytoplasm to the nucleus and activate transcription of target genes (for reviews see ; ). We examined the translocation of NFAT in purified splenic B cells from Eμ- Tg and normal Wt littermates both before and after BCR cross-linking. Purified B cells from Wt littermates require 15 min of anti-IgM stimulation to achieve the maximal translocation of NFATc1 ( and Fig. S5 C, available at ) and NFATc2 ( and S5 D) to the nucleus. In contrast, B cells from Eμ- Tg mice have the majority of both NFATc1 and c2 in the nucleus, even before stimulation with anti-IgM. These effects are not the result of increased NFAT expression in Eμ- B cells, as mRNA (Fig. S4 A) and protein levels (Fig. S5, A and B) are reduced or unchanged relative to Wt B cells.
mice ( and S5 F). Transfection of Myc-null fibroblasts (HO.15 cells) with c-Myc () results in greatly enhanced NFAT activity based on increased activation of an NFAT-luciferase reporter construct (Fig. S3 E), demonstrating that Myc also increases NFAT activity. These results suggest that the sustained increase in [Ca] signaling in Myc-expressing B cells has a major influence on key Ca-regulated nuclear events.
To assess whether the elevated Ca flux in Myc-expressing B cell progenitors may contribute to the abilities of Myc to stimulate maturation and proliferation, we first determined how [Ca] levels in these mice correlate with differentiation, proliferation, and survival status. We find that Ca levels positively correlate with maturation and proliferation in Eμ-/RAG2 mice based on the down-regulation of CD43 on B220 cells and dilution of CFSE. B cell progenitors with the most mature phenotype (B220CD43) and highest proliferative capacity have the highest levels of Ca flux, whereas the cells with the least mature phenotype (B220CD43) and lowest proliferative capacity have the lowest levels of Ca flux (Fig. S3, A–C). The observed increase in [Ca] is not the result of differences in cell cycle status because gated G0/G1 cells from Eμ- Tg have consistently elevated [Ca] relative to G0/G1 cells from normal littermates (Fig. S3 D). These results suggest that there is a correlation between amplified Ca signaling and the ability of Myc-expressing B cells to undergo differentiation and division.
We next examined whether normal Ca signaling is required for maturation and division of Wt and Myc Tg pre–B cells. First, we determined whether impairing endogenous Myc expression results in impaired Ca flux. Total BM cells from N- c- CD19cre or CD19cre mice were stimulated in vitro with anti-μ for 2 h to induce expression followed by ionomycin stimulation to induce Ca flux. As shown in , both pre–B/immature B and mature B cell populations from N- c- CD19cre mice exhibit reduced peak and sustained Ca levels relative to CD19cre mice. To determine whether a reduction of Ca levels affects the ability of both Wt and Eμ- cells to proliferate and mature, we cultured total BM cells from Wt and Myc Tg mice in EGTA to limit the amount of available extracellular Ca and assessed maturation and proliferative capacity after the addition of increasing doses of extracellular Ca. As shown in Fig. S4 (B and C), chelation of Ca with EGTA results in the impairment in Ca flux and proliferative capacity after ionomycin stimulation. EGTA also blocks the ability of Wt B cell progenitors to mature from the B220CD43 stage to the B220CD43 stage (, first and second panels), which is rescued by the addition of extracellular Ca (, third to fifth panels). B cell progenitors were unable to divide in low Ca conditions based on limited CFSE dilution of live (TO-PRO-3) cells (, first and second panels), which is also rescued by the addition of extracellular Ca (, third to fifth panels). Progenitors from Eμ-/RAG2 mice were also unable to mature (Fig. S4 D) or divide efficiently (not depicted) under low Ca conditions.
As Eμ- B cells exhibit increased NFAT nuclear translocation, we also determined whether the Cn–NFAT pathway is required for the division of Wt and Eμ- B cells. Purified B cells from Wt and Eμ- mice were stimulated with anti-μ for 72 h in the presence or absence of cyclosporine A (CsA), a specific inhibitor of Cn. As shown in , Wt B cells undergo an average of four divisions (, left), and Eμ- B cells undergo an average of five divisions (, right) in response to anti-μ, whereas cell division is completely inhibited by CsA. Collectively, these results suggest that normal Ca signaling is required for pre-BCR–mediated pre–B cell proliferation and maturation and for Myc to stimulate the differentiation of B cell progenitors.
The increase in [Ca] that occurs in response to stimuli such as BCR ligation is transient, in part because Ca is resequestered into the endoplasmic reticulum by the sarcoplasmic ER Ca ATPase (SERCA) or is extruded from the cell by plasma membrane Ca-ATPase (PMCA) pumps. Because Myc amplifies [Ca] in B cells in the absence of Btk and Tec (which regulate influx), we hypothesized that Myc could be altering the extent of Ca efflux. First, we measured the relative expression of , , and , the respective pumps expressed in B lineage cells (). Although we did not find consistent differences in the expression of and in mature B cells from Eμ- mice relative to Wt mice (unpublished data), levels of mRNA are significantly decreased in Eμ- mature B and Eμ-/RAG2 pro–B cells (), whereas mRNA is increased in sorted pro–B cells from N- c- CD19cre versus CD19cre mice (Fig. S4 E). The decrease in expression correlates with a decrease in total PMCA protein () and export of Ca across the plasma membrane in purified B cells from Eμ- mice relative to Wt mice in response to anti-μ (), as measured by fluorometric analysis of extracellular media containing a membrane-impermeable version of the Ca-binding Indo-1 dye. These results suggest that Myc negatively regulates Ca extrusion in B cells and are consistent with a recent report that Myc interacts with the human promoter, resulting in decreased expression (). Indeed, we also find using chromatin immunoprecipitation (ChIP) that Myc interacts within the homologous region in the mouse PMCA4 promoter (Fig. S4 F).
To determine whether reduced expression in Myc-expressing cells is important for Myc-induced proliferation, we infected BM cells from Wt or Eμ- mice with murine stem cell virus (MSCV) retroviruses containing internal ribosomal entry site (IRES)–GFP and human cDNA (MSCV--GFP) or vector alone (MSCV-GFP). Analysis of the Phoenix retroviral packaging cell line indicates that MSCV--GFP cells exhibit a substantially decreased baseline and peak [Ca] after ionomycin stimulation relative to GFP cells, whereas MSCV-GFP cells flux normally (). Infection of Wt and Eμ- BM with MSCV--GFP virus results in increased expression over MSCV-GFP–infected cells by real-time PCR (). MSCV-GFP–infected Wt and Eμ- pre–B cells (B220GFP) divide normally in response to IL-7 in vitro based on considerable dilution of the CFSE-like SNARF-1 dye. In contrast, B cell proliferation is inhibited in MSCV--GFP–infected Wt and Eμ- B cells in a dose-dependent manner ( and not depicted).
The roles of Myc proteins in cell differentiation remains one of the most controversial issues in Myc biology. Early studies performed in cell lines and more recent studies in primary keratinocytes resulted in the prevailing view that Myc stimulates cell cycle entry while inhibiting differentiation (for review see ). However, in primary chicken and murine B cells engineered to overexpress and in humans with Burkitt's lymphoma, B lymphocytes mature to the Ig B cell stage relatively unperturbed despite the constitutive overexpression of throughout B cell development (; ). These results suggest that B cells are able to mature relatively normally despite the presence of deregulated Myc.
In this study, we took a genetic approach to address the roles of Myc in the differentiation of mammalian cells using primary B cells. First, we found that conditional deletion significantly inhibits B cell development at the pro–B to pre–B cell transition (), which is the stage at which the combined signaling from the pre-BCR and IL-7 normally induce c- and N- expression.
) whereby B cell development is blocked just before the stages at which c- and N- are normally expressed. We found that the Eμ- Tg efficiently stimulated the differentiation and expansion of pre–B-like cells from pro–B cells in the absence of pre-BCR formation (in RAG2 mice).
background, whereby B cell development is nearly completely blocked at the large pre–B cell stage. These results collectively provide genetic evidence that Myc acts downstream of the pre-BCR and Btk/Tec to stimulate pre–B cell development and are consistent with previous studies suggesting that the pre-BCR is required for induction and pre–B cell development (; for review see ).
The notion that Myc may initiate differentiation was first proposed by in keratinocytes in vitro. Recent studies have suggested roles for Myc in inducing the differentiation of hematopoietic and epidermal stem cells, in part by regulating the expression of adhesion molecules, thus releasing them from a differentiation-inhibiting niche (). In this study, we show that constitutive Myc can act in a cell-autonomous manner to rescue differentiation in lineage-committed progenitor cells. These results collectively suggest that the phenotypic consequences of Myc during development may depend on the differentiation status of the cell: Myc expression in stem cells and lineage-committed progenitors results in differentiation, whereas Myc expression in mature postmitotic cells results in reentry into the cell cycle and inhibition of terminal differentiation.
Although -null or double-null mice do not spontaneously develop pre–B cell tumors, the absence of both and significantly enhances the incidence of pre–B cell leukemia as compared with –null mice (). Although the exact mechanism of tumor suppression by Tec family PTKs is not known, the loss of Btk and/or SLP-65 in B cells results in sustained IL-7 receptor (IL-7R) expression and an increased proliferative and/or survival response to IL-7. Here, we show that loss of Btk/Tec significantly accelerates B cell tumor formation in Eμ- mice (). In addition, we find that Eμ- Tg substantially increases the proliferative potential of B cell progenitors in response to IL-7, whereas decreasing expression inhibits IL-7–mediated proliferation. Thus, sustained IL-7R expression in -null mice, in cooperation with constitutive Myc activity as occurs in Burkitt's lymphoma, substantially increases the pool of dividing cells that is capable of acquiring epigenetic alterations that promote tumorigenesis. We conclude that Tec PTKs act as tumor suppressors that attenuate deregulated c- and that IL-7 synergizes with Myc to increase the pool of dividing B cell progenitors. The ability of Myc to drive differentiation to the IgM-positive stage in Btk/Tec mice while having accelerated transformation further suggests that it is not the inhibition of maturation itself but rather the degree of growth factor responsiveness that determines tumor susceptibility. Consistent with the latter notion, loss of Jak3 kinase, an essential IL-7 signaling molecule, reduces the transforming potential of the Eμ- Tg ().
Because Ca signaling is important for virtually all aspects of embryogenesis (for review see ) and pre-BCR and Btk stimulate Ca signaling pathways, we were prompted to investigate how Ca signals were affected during Myc-dependent B cell development. Surprisingly, we found that both basal [Ca] levels and the duration of Ca flux are elevated in B cells from Eμ- Tg and Eμ-/RAG2 mice. Furthermore, an increase in the relative level of [Ca] in Eμ-/RAG2 Tg mice positively correlates with increased proliferation and differentiation and is required for Myc to stimulate B cell development and proliferation. The increase in [Ca] level may occur by down-regulation of the Ca efflux pump PMCA4b because Myc interacts with the promoter, and both expression and Ca efflux are decreased in Eμ- B cells. Moreover, enforced expression inhibits Myc-induced cell proliferation. These results are consistent with a study in developing T lymphocytes in the thymus, which depend on Ca signaling during the maturation and expansion of CD4CD8 pre–T cells after rearrangement of the T cell receptor β chain and formation of the pre–T cell receptor (). In addition, gene-targeted mutations in other genes involved in Ca signaling, including the Src family PTKs Lyn/Fyn/Blk, Syk, SLP-65, LAT (linker for activation of T cells), Btk/Tec, the p85 subunit of PI3K (), and PLCγ1/2 (), also result in impaired B cell development at the pre–B cell stage. Altogether, these results strongly support a role for Myc and Ca signaling in the maturation of pre–B cells.
A study in primary B lymphocytes indicates that the amplitude and duration of Ca signals have profound influences on the type of transcriptional response (). For example, NF-κB and JNK are selectively activated by a large transient [Ca] rise, whereas NFAT is activated by lower but sustained [Ca] levels. We also find that B cells from Eμ- Tg mice, which exhibit elevated, sustained [Ca] levels, also exhibit increased NFAT translocation even in the absence of upstream signals from the pre-BCR and BCR. In addition, B cell nuclear fractions from Myc Tg mice appear to contain mostly the short isoform of NFATc1 (NFATc1A; ), which, in contrast to the other NFATc1 isoforms (NFATc1B and c1C), does not promote apoptosis in lymphocytes (). These results suggest that the elevated [Ca] in Myc-expressing B cells has profound influences on transcriptional and biological responses as well. Indeed, we find that anti-IgM–induced B cell proliferation of both Wt and Myc Tg B cells is completely abrogated by treatment of cells with CsA, and a recent study indicates that the B cell–specific deletion of Cn results in defects in B cell proliferation and function ().
One common feature of many types of cancers is their reduced dependence on external growth factors, and Myc-induced tumors share this property. Indeed, Myc activates at least three genetic programs that are normally growth factor dependent, including cyclin E/Cdk2 kinase, E2F-dependent transcription, and protein synthesis pathways (for review see ). How can one transcription factor with limited target gene specificity induce so many biochemical and biological changes that are normally dependent on multiple signaling pathways? Our results suggest that although Myc directly stimulates the expression of many E box–containing genes, Myc also increases sustained [Ca] and enhances NFAT translocation (). This allows for the concurrent activation of Ca-dependent target genes even in the apparent absence of upstream mediators of Ca signaling, as might occur under conditions of poor growth factor availability or limited extracellular Ca. The synergy between Myc and [Ca] to stimulate B cell development and proliferation are strikingly similar to the signaling requirements underlying immunogenic versus tolerogenic responses to antigen in B lymphocytes. Although activation of Myc alone or a low level Ca signal alone provokes apoptosis or anergy (tolerance; ), Myc expression in conjunction with a sustained Ca/NFAT signal results in a mitogenic response and immunity. Here, we find that Myc amplifies [Ca], thus tuning levels into the range required for optimal NFAT activation and translocation. The unique function of Myc to amplify Ca signaling may help explain why transformed cells are relatively resistant to low growth factor and Ca conditions () despite known requirements for Ca signaling during G1→S transition.
C57BL/6 Eμ- transgenic mice were obtained from Jackson ImmunoResearch Laboratories and were genotyped by PCR according to instructions from the supplier. To generate mice carrying conditionally inactivated c- and N- genes, mice carrying a floxed c- allele (c-
; ) were bred to N- mice () expressing the recombinase Tg under the control of the CD19 promoter (). -deficient mice have been described previously (). Mice were housed under specific pathogen-free conditions. All mouse procedures were approved by the University of Washington Institutional Animal Care and Use Committee.
For cell division experiments, cells were stained with CFSE according to the manufacturers' instructions (Invitrogen). CFSE-labeled cells were cultured in complete RPMI + 10% FBS (Hyclone) in the presence or absence of 10 ng/ml murine recombinant IL-7 (R&D Systems) or 5 μg/ml anti-μ (F(ab′)2 fragment; Thermo Fisher Scientific) for the indicated times at 37°C and 5% CO. In some experiments, EGTA was added at a final concentration of 0.5, 0.75, or 1 mM to IL-7–driven cultures in the presence or absence of 0.2, 0.4, or 0.6 mM CaCl, and cyclosporine was added at a final concentration of 100 ng/ml to anti-μ–driven cultures. Thereafter, the cells were stained with the indicated antibodies for analysis by FACS or harvested for measurement of [Ca] (see section Measurement of intracellular calcium and DNA content). The vital dye TO-PRO-3 (Invitrogen) was added to the indicated samples (1-nM final concentration) before acquisition to distinguish live and apoptotic cells. The number of discrete CFSE peaks was determined using MODFIT software (Verity Software House).
The Myc-null and Myc-expressing fibroblast cell lines HO.15 and HO.15 + c-Myc, respectively, and the Phoenix packaging cell line were grown in DME supplemented with 10% FCS, glutamine, penicillin, streptomycin, and essential amino acids. For retroviral gene transfer into primary BM cells, human cDNA was subcloned into the HpaI site of the modified MSCV MIG vector upstream of the EGFP gene and IRES. The MSCV vector carrying only the IRES-EGFP cassette was used as a control. The cDNA of human was provided by M. Husain (Department of Physiology, University of Toronto, Toronto, Canada; ). Ecotropic viral stocks were generated by Ca phosphate–mediated transfection of subconfluent Phoenix cells in the presence of chloroquine. Viral supernatant was collected and filtered after 3 d. For infection of Wt or Myc Tg BM, ∼15 × 10 cells were plated per well of a six-well plate and infected 48 h later with 4 ml of fresh viral supernatant and 4 μg/ml polybrene by spinoculation. Two rounds of infection were performed, and cells were then labeled with 1 μM SNARF-1 (S-22801; Invitrogen) for 15 min at room temperature. Cells were washed twice with complete RPMI + 10% FCS and cultured in fresh medium containing 10 ng/ml IL-7 for an additional 48 h. Thereafter, the cells were stained with αB220-phycoerythrin (PE) for analysis of B220 GFP cell division by FACS.
Single-cell suspensions were prepared and analyzed by flow cytometry as previously described (). The majority of antibodies used in this study are referenced in . Additional antibody conjugates used in this study include B220-PE-Cy-5.5 (eBioscience), FITC anti–mouse CD21, biotin anti–mouse CD22.2, biotin anti–mouse IA (BD Biosciences), FITC–annexin V (Invitrogen), IgM-PE-Cy7 (SouthernBiotech), and IgD-FITC (BD Biosciences). Cytometry was performed on FACScan and LSR flow cytometers (BD Biosciences). Data were analyzed using CellQuest (Becton Dickinson) and FlowJo (version 6.3.2; Tree Star, Inc.) softwares. To determine relative numbers of B lineage cells in each population (), percentages from flow cytometry were multiplied by the total BM cellularity to obtain absolute numbers of pro–B cells (B220CD43), pre–B cells (B220CD43), immature B cells (B220IgM), and mature B cells (B220IgM). To determine relative numbers of peripheral B2 B cells in each population (), percentages from flow cytometry were multiplied by the total spleen cellularity to obtain absolute numbers of total B cells (B220IgM), follicular mature (IgMIgD), T2 (IgMIgD), and T1/MZ (IgMIgD) B cells.
For measurement of [Ca], cells were washed and resuspended in 0.5 ml HBSS containing 3% FBS at 10–10 cells/ml (for harvested BM cultures) or 5 × 10–8 × 10 cells/ml (for freshly isolated BM). EGTA-treated cell cultures were resuspended in Ca-free HBSS/3% FBS. Indo-1–acetoxymethyl (pentaacetoxymethyl ester; Sigma-Aldrich) was added at a final concentration of 10–20 μM, and incubation was performed for 30 min at 37°C. The indo-1 fluorescence ratio (400:530 nm) of the cells was acquired as a function of time using a flow cytometer (BD-LSR I; Becton Dickinson). For each experiment, collection of a 30-s baseline measurement was followed by stimulation with either 25 μg/ml anti-μ or 1 μg/ml ionomycin (EMD) as indicated. For pre-BCR cross-linking of c- and N-–deficient or Wt BM, indo-1–loaded cells were first preincubated with 25 μg/ml anti-μ for 2–3 h at 37°C.
For simultaneous analyses of Ca flux and DNA content, fluo-4 Ca-binding dye (Invitrogen) was added at a final concentration of 1 μM in place of indo-1–acetoxymethyl. The cells were washed twice with HBSS/3% FBS and stained with the indicated antibody conjugates at room temperature for 45 min followed by Hoescht (Invitrogen) for 45 min at 37°C. Thereafter, the cells were washed twice, resuspended in a 1–2-ml final volume, and maintained at 37°C for 5 min before and during analysis of [Ca].
Ca efflux was measured with a cell-impermeable form of indo-1 (5-μM final concentration; Invitrogen). 10 purified splenic B cells were washed twice with Na-free/Ca-free efflux buffer (modified from ) and resuspended in 2 ml efflux buffer (10 mM Hepes, pH 7.4, 1 mM MgCl, 5 mM KCl, 135 mM choline chloride, 10 mM glucose, and 0.1% BSA). Fluorescence of cell suspensions was detected with a Fluorescence Master Series fluorometer (Photon Technology International) at an excitation wavelength of 350 nm and emission wavelengths of 400 and 485 nm. Collection of a baseline measurement was followed by stimulation with 10 μg/ml anti-μ (F(ab′)2 fragment), and ratiometric Ca values (400:485 nm) were plotted as a function of time.
For NFAT-luciferase assays, HO.15 (Myc null) or HO.15 + c-Myc fibroblasts were plated in 24-well tissue culture dishes at 90% confluence and, 24 h later, were transfected in triplicate with pNFAT-luciferase plasmid or pCIS-CK negative control plasmid containing the luciferase reporter gene without any cis-acting elements (Stratagene) and the pRL Renilla luciferase control reporter vector (Promega) using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. Treatment with 60 ng/ml PMA + 1 μg/ml ionomycin or PMA/ionomycin + 100 ng/ml CsA was performed 18 h after transfection. After 6 h, luciferase assays were performed with a luminometer (Monolight 1500; Analytical Luminescence Laboratories) and the Dual Luciferase Reporter assay system (Promega). Firefly luciferase values for each transfection were normalized to Renilla luciferase activity (pRL), and data were expressed as relative luciferase activity versus that obtained with the pCIS-CK negative control plasmid. Mean values of a representative experiment of three performed are displayed ± SEM.
Mice were given water containing 0.8 mg/ml BrdU (Sigma-Aldrich) for 2 d. BM was harvested, surface stained for B220, and fixed, permeabilized, and treated with 30 μg DNase I (Sigma-Aldrich). Incorporation of BrdU into DNA was measured by flow cytometry using anti–BrdU-FITC (BD Biosciences) according to the manufacturer's protocol.
Single-cell suspensions of total splenocytes and BM cells were isolated as previously described (). B lymphocytes were enriched by negative selection on magnetic microbeads coupled to rat anti–mouse CD43 (Miltenyi Biotec) or using the mouse B cell Negative Isolation kit (Invitrogen) according to protocols provided by the manufacturers. The recovered splenic B cells were rested in RPMI 1640 medium before stimulation for biochemical analyses or were washed into efflux buffer for Ca efflux assays. B lineage cells were enriched from total BM by positive selection on magnetic beads coupled to CD45R/B220 (Invitrogen) or CD19 (Miltenyi Biotec) according to the manufacturers' instructions or by cell sorting on a FACSaria (BD Biosciences).
Purified splenic B cells were stimulated with 25 μg/ml anti-μ (F(ab′) 2 fragment) in serum-free RPMI medium. At the indicated time points, the cells were washed twice in ice-cold PBS and resuspended in hypotonic buffer containing protease inhibitors. Cytoplasmic and nuclear extracts were prepared by the method of . SDS-PAGE and Western blotting were performed using standard techniques. Anti-NFATc1, anti-NFATc2, anti–β-actin, anti-Max, and donkey anti–goat IgG-HRP were purchased from Santa Cruz Biotechnology, Inc. Rabbit anti–mouse IgG HRP secondary antibody was purchased from Invitrogen. Goat anti–rabbit IgG HRP secondary antibody was purchased from Bio-Rad Laboratories. Anti-PMCA monoclonal antibody (clone 5F10) was purchased from Millipore. Anti–lamin A/C antibody was purchased from Cell Signaling Technology. Anti–α-tubulin monoclonal antibody was purchased from Sigma-Aldrich.
Purified splenic B cells from Myc Tg mice were cross-linked with 1% formaldehyde for 10 min at 37°C. ChIP was performed with the ChIP assay kit (Millipore) according to the manufacturer's instructions using anti–c-Myc (N-262; Santa Cruz Biotechnology, Inc.) or control rabbit IgG (rIgG; Santa Cruz Biotechnology, Inc.) antibodies. After reversal of cross-links, DNA was precipitated and detected by qPCR with pairs of primers (underlined) specific to the mouse PMCA4b promoter (189-bp region beginning at −2,872 relative to the putative transcription start site and −4,285 relative to the translation start site: 5′-AGCGCTCACGTTCTAGAAACTTGTGTGTCCCTCAGTGCAGCAGGACTTTAGTGGATTCCTGAAACTGGAGGTCTCCATCACACGCTGTTACTTGAACAGGTATATGTCTCTGATTCTCCCGGAGCAGTTCTGTAGCGCTCT
) and the APEX1 promoter as a positive E-box control ( forward, 5′-TACCACGAACAACCCAGAACC-3′; reverse, 5′-GTACCTGACCTCCCAACGAAG-3′). Real-time qPCR was used to quantify the fold enrichment relative to background detected with rabbit IgG for each primer set. The irrelevant gene 45S was amplified to normalize samples ( forward, 5′-TGAATTGTGGCCCTGAGTGATAGG-3′; reverse, 5′-GAGTGGTGTTTGTGTGTGTGTTGG-3′).
For PCR analysis of c-and N- alleles, single-cell suspensions of total BM from control or mutant mice were stained with anti-B220 and anti-CD43 antibodies, and cell sorting was performed on a FACSaria; alternatively, cells were magnetically labeled with CD19 microbeads and positively selected with an autoMACS Separator (Miltenyi Biotec). The FACS-sorted B220CD43 and B220CD43 populations were directly lysed in PCR lysis buffer (). Purified CD19 B lineage cells were cultured with 10 ng/ml IL-7 for 0, 15, or 22 h before lysis with PCR lysis buffer. The resulting DNA was used directly for PCR at 10,000 genomes/μl. Primers used to amplify the floxed undeleted and deleted c-and N- alleles have been described previously (; ). Deletion of was quantified by real-time PCR measurements of the c- PCR product in the presence or absence of CD19cre. PCR analysis for D-J and V-D-J rearrangement of the Ig HC was performed as previously described ().
RNA from purified Wt and Eμ-c- splenic B cells, Wt and Eμ-c- total BM, or FACS-sorted B220 Eμ-c-/RAG2, RAG2, and CD19 c- N- B cell progenitors was extracted using the RNAqueous-4PCR kit (Ambion). cDNA was generated using Superscript II Reverse Transcriptase (Invitrogen). Samples were normalized using β (β forward [5′-TCCTTCGTTGCCGGTCCAC-3′] and β reverse [5′-ACCAGCGCAGCGATATCGTC-3′]) or α (α forward [5′-AGTGAGCACTGTTAAGAGACTGCC-3′] and α reverse [5′-CGCAGCCAGATGACTAGAGTACAA-3′]). c- levels were determined using the following primers: c- forward (5′-ACCAACAGGAACTATGACCTC-3′) and c- reverse (5′-AAGGCAGTAGCGACCGCAAC-3′). The following primers were used for murine PMCA4: forward (5′-TCGTGACAGCCTTCAATGACTGGA-3′), reverse (5′-AGGTCACCGTATTTGATCTGGGCA-3′), and human PMCA4 ( forward [5′-ATGACCCACCCTGAATTCGCCATA-3′] and reverse [5′-TGGTTGCAATCCACCGCATTGT-3′]).
Primers specific for the various isoforms of murine NFATc1 and NFATc2 have been described previously (): forward (5′-GGTAACTCTGTCTTTCTAACCTTAAGCTC-3′), reverse (5′-GTGATGACCCCAGCATGCACCAGTCACAG-3′), forward (5′-CCCATCCGCCAGGCTACAGCCGCAGTAA-3′), reverse (5′-TTCGGTAAGTTGGGATTTCTGAGTGGTACC-3′), forward (5′-CCCATCCGCCAGGCTACAGCCGCAGTAA-3′), reverse (5′-TGAGTGGTACCAGATGTGGGTCCAGTTTAT-3′), forward (5′-CACGCCTTCTACCAAGTACACAGGAT-3′), and reverse (5′-ACAGTCGATGGTGGCTCTCATGTT-3′). Primers used to amplify hypoxanthine-guanine phosphoribosyl transferase () were as follows: forward (5′-GTTGGATACAGGCCAGACTTTGTTG-3′) and reverse (5′-GAGGGTAGGCTGGCCTATAGGCT-3′).
Primers used to amplify GL Igκ transcripts have been described previously (). Quantitative deletion analysis of floxed c- alleles was performed using genomic DNA from FACS-sorted CD19cre or cre c-
N-
pro– and pre–B cells and primers specific for the floxed c- allele () using 45S measurements to normalize. Experiments were performed using a real-time PCR system sequence detector (model 7300; Applied Biosystems) and a PCR system (Mx4000; Stratagene).
Mice were immunized with KLH protein (Calbiochem) emulsified in complete Freund's adjuvant (CFA; 1:1 vol/vol mixture of 1 mg/ml of sterile protein solution–CFA) by subcutaneous injection at the base of the tail. Two injection sites were administered with 50 μl of the mixture. Mice were killed after 7 d. KLH-specific antibody production was measured with a KLH-coated ELISA system. For IgM, IgG1, and IgG2a measurements, serums were diluted 1:200, 1:135, and 1:135, respectively.
One-tailed test was used for all analyses except the Kaplan-Meier analyses, in which we used Prism software (version 4; Graph Pad) to generate two-tailed p-values.
Developed films were scanned using Photoshop (version 8.0; Adobe), and adjustments of contrast and brightness were performed with Photoshop software. Scanned images were imported into Canvas (version 9.0.2; ACD Systems of America) for figure preparation. In reference to , the nuclear α-lamin control Western blots shown in (C and E) and the β-actin control Western blot shown in were obtained by probing a separate gel. For evaluation of c- and N- deletion by densitometry (Fig. S1), band intensities representing the ratio of deleted to floxed alleles (with background subtracted) are shown below each lane and were quantified by densitometry on an imaging system (AlphaImager 3400; Alpha Innotech).
Fig. S1 shows the specific deletion of c- and N- in B lineage cells. Fig. S2 shows that the Eμ- Tg stimulates proliferation but not apoptosis of RAG2 B cell progenitors or Ig HC exclusion in Wt B cell progenitors. Fig. S3 shows that the elevated Ca flux of Eμ- B cell progenitors correlates with increased maturation and proliferation and increased NFAT activity. Fig. S4 shows that chelation of Ca impairs Ca flux, maturation, and proliferation of Wt and Eμ- B cell progenitors in vitro, that Myc interacts with the mouse promoter in Eμ- B cells, and real-time PCR analyses of message from purified splenic B cells and message from c- and N-–deleted pro–B cells. Fig. S5 shows total levels of NFAT protein in whole cell lysates from Wt and Eμ- B cells and fractionation controls for the separation of cytoplasmic and nuclear fractions depicted in (C–F). Online supplemental material is available at . |
Chromosome segregation is a complex and dynamic process during which paired sister chromatids attached to the microtubules of the mitotic spindle are equally segregated in the two daughter cells. To avoid defects in chromosome segregation, mitotic exit is delayed until all chromosomes are properly attached to the mitotic spindle by a control mechanism named the spindle assembly checkpoint (SAC; ; ).
SAC activation requires numerous proteins that localize at the kinetochores in early mitotic stages. The main components of this surveillance mechanism have first been identified in budding yeast and include MAD1, MAD2, MAD3 (), BUB 1, BUB3 (), and MPS1 (). In higher organisms, all of these mitotic checkpoint proteins localize at unattached kinetochores during prometaphase. In addition, other kinetochore and centromeric proteins such as centromere protein A (CENP-A), C, I, F, E, hMIS12, the Ncd80 complex, the aurora B complex, and the RZZ (Rod, ZW10, Zwilch) complex are necessary for recruitment of the SAC and are components of the kinetochore structure (; ). Localization of SAC proteins at kinetochores seems to be hierarchical, implying that the recruitment of some depends on the prior recruitment of others (). In response to unattached kinetochores, the SAC is activated, resulting in the inhibition of CDC20, an activator of the multisubunit E3 ubiquitin ligase anaphase-promoting complex/cyclosome (APC/C) that is responsible for the metaphase→anaphase transition (). CDC20 is sequestered together with checkpoint proteins in complexes containing MAD2, BUB3, and BUBR1, thus preventing APC/C activation and premature mitotic exit (). When kinetochores are properly attached to spindle microtubules, the spindle checkpoint is turned off. This allows the APC/C to ubiquitinate mitotic proteins that will then be degraded by the proteasome machinery, triggering mitotic exit (). In higher eukaryotes, SAC is required in normal conditions to check whether chromosomes are correctly attached to microtubules before anaphase onset (; ). Inactivation of SAC genes always results in severe chromosome missegregations in mammalian cells. In addition, the depletion of SAC proteins inhibits the mitotic arrest induced by microtubule-depolymerizing drugs ().
The PRP4 gene encodes a 150-kD serine-threonine protein kinase that has been implicated in the regulation of mRNA splicing in , and mutations in lead to the accumulation of pre-mRNAs (). However, the homologue protein has also turned up in screens for genes involved in mitosis (), and, in , the expression of a dominant truncated PRP4 protein reportedly induced mitotic aberrations, suggesting a dual function in RNA splicing and mitosis in this organism (). Therefore, we decided to further explore the possible mitotic role of PRP4.
We raised a polyclonal antibody against the PRP4 protein. Western blotting revealed the expected ∼150-kD protein (, left; − lane). By immunofluorescence experiments, the PRP4 protein appeared on fixed interphase cells as nuclear punctae (), which is in agreement with the localization previously described (; ). This nuclear signal disappeared in cells transfected with siRNA (, middle; compare top with bottom) as well as the protein band by Western blotting (, + lane; and Fig. S1, A and B, available at ).
The PRP4 protein kinase was previously shown to associate with isolated mitotic chromosome arms (). To further reexamine this subcellular localization, dividing HeLa cells were methanol fixed and stained for CENP-A (as an inner kinetochore marker) and affinity-purified anti-PRP4 antibodies (). During prometaphase, most kinetochores (59.7 ± 19.1%; Table S1, available at ) stained positive for PRP4. During metaphase and anaphase, 27.0 ± 12.8% and 10.9 ± 7.1% of all kinetochores, respectively, displayed positive PRP4 labeling, suggesting that PRP4 leaves the kinetochore during metaphase. A strong granular signal was also detected in the cytoplasm, indicating that only a small fraction of the protein was kinetochore associated.
To confirm this kinetochore localization and eliminate the cytoplasmic PRP4 pool, isolated chromosomes prepared from mitotic cells were spread, fixed onto glass coverslips, and processed for immunofluorescence using anti-PRP4 antibodies. We found that PRP4 colocalized with BUBR1 at the outer kinetochore region (). Moreover, isolated chromosomes prepared from HeLa cells expressing a GFP-tagged PRP4 protein also showed a clear colocalization with MAD2 at the outer kinetochore region (Fig. S1 C).
This kinetochore localization prompted us to investigate whether PRP4 was playing a role during mitosis. To do so, we tested three siRNA oligonucleotides directed against different parts of the mRNA. After 2 d of RNAi treatment with any of the oligo sets, PRP4 protein levels were reduced by at least 80% (Fig. S1 B) compared with control-transfected cells.
Cultures subjected to siRNA treatment displayed reduced mitotic indexes of 3.1 (±0.8) versus 7.2% (±1.1) for control cells after 24 h and 2.8 (±0.4) versus 7.1% (±0.3) after 48 h ( = 3; ∼1,000 cells). Among the few mitotic PRP4-depleted cells, most were apparently in prometaphase and anaphase. Well-defined metaphase cells with aligned chromosomes were rare and represented only 11.9% of mitotic cells versus 46.4% in control cells (). In addition, most of the anaphase and telophase cells exhibited abnormal figures (22.2 vs. 1.6% for control cells) and often showed lagging chromatids/chromosomes in the spindle midzone ( and ). 21.0% of the PRP4-depleted cells showed residual DNA bridges during cytokinesis (). As a consequence of these segregation defects, during interphase and very late cytokinesis, we observed a sevenfold increase in the frequency of cells with micronuclei ( and ). We decided to check whether a GFP-tagged PRP4 RNAi-resistant construct (GFP-PRP4rr) could save the phenotype caused by RNAi. GFP or GFP-PRP4rr were cotransfected with control or PRP4 siRNAs (). Western blot analyses showed that the endogenous PRP4 protein level was knocked down after RNAi, but the tagged protein remained stable (). In control RNAi conditions, GFP- or GFP-PRP4rr–positive cells contained ∼13% of micronuclei. In GFP- or GFP-PRP4rr–positive cells after RNAi, this percentage was ∼35% and ∼20%, respectively, suggesting that the PRP4 RNA–resistant construct was rescuing the RNAi phenotype (). Together, these data suggest a role for during mitosis.
To further characterize chromosome dynamics and the timing of mitosis in control and PRP4-depleted cells, we filmed HeLa cells expressing an H2B-GFP transgene (). In control HeLa H2B-GFP cells ( = 11), chromosomes correctly aligned to form a metaphase plate (, top; panel at 21 s) before anaphase onset (, top; panel at 37 s [control]; and Video 1, available at ). After RNAi ( = 12), anaphase always started before chromosome alignment (, bottom; panel at 23 s; and Video 2). also shows that all recorded PRP4-depleted cells do not congress their chromosomes to the metaphase plate and that mitotic exit occurs earlier than control cells. Together, these data show that PRP4 is required for chromosome congression and segregation but also for mitotic timing.
To confirm this acceleration, the time between nuclear envelope breakdown (NEBD) and anaphase onset was analyzed by live cell imaging of HeLa cells. This time was 56.2 ± 23.1 min ( = 36) in control cells (Fig. S2 B, available at ). In PRP4-depleted cells, this time was only 43.1 ± 12.9 ( = 43), 46.4 ± 7.8 ( = 10), and 33.6 ± 8.0 min ( = 36) after transfection with the different siRNAs (Fig. S2 B). The mitosis peak time () was 49 min for the control cells and 38, 41, and 31 min for PRP4 knockdown cells (siRNAs 1–3), confirming that mitosis duration was reduced after RNAi (Fig. S2 A).
The gene was previously shown to be required for pre-mRNA splicing (; ). We then checked whether this process was compromised after RNAi in HeLa cells (see Materials and methods and Fig. S3, available at ). Pre-mRNA splicing was identical to control cells. Thus, the mitotic effect of PRP4 knockdown is unlikely to be a secondary consequence of general splicing defects.
These defects were similar to those previously observed after the depletion of other SAC components such as MAD2 and BUBR1: shorter mitosis duration and chromosome segregation defects (). In addition, MAD2- and BUBR1-depleted cells do not arrest in mitosis after treatment with nocodazole, a drug that depolymerizes mitotic spindle microtubules.
To test for a possible role of in SAC signaling, we monitored mitotic progression by time-lapse video microscopy in the presence of 50 ng/ml nocodazole, conditions in which SAC is activated. When control cells entered mitosis, they rounded and remained blocked in that stage for the duration of the experiment (∼5 h; > 25), indicating a mitotic arrest caused by the presence of a functional SAC (, top). In contrast, when MAD2- or PRP4-depleted cells were challenged in the same conditions, the cells rounded and became adherent after 102.5 ± 14.9 min ( = 9) and 97.0 ± 12.3 min ( = 10), respectively, indicating their return into interphase without chromosome segregation (, middle and bottom). In addition, the mitotic index was analyzed in a time course experiment in control, MAD2-, or PRP4-depleted cells in the presence of nocodazole (). Although the mitotic index increases during the experiment in control cells, indicating checkpoint activation and mitotic arrest, the mitotic index did not increase in MAD2- and PRP4-depleted cells, indicating a SAC failure. In parallel, a rescue experiment was performed by coexpression of the PRP4 siRNA with GFP or GFP-PRP4rr constructs described previously (). The mitotic index of PRP4-depleted cells was low in the presence of GFP. In contrast, a full rescue was observed in the presence of the GFP-PRP4rr construct, indicating that the GFP-tagged PRP4 protein was able to complement PRP4 depletion.
To investigate how the absence of the PRP4 protein affected the SAC, we examined whether the RNAi treatment disturbed the kinetochore localization of several proteins involved in the SAC. The CENP-A, aurora B, BUBR1, and Hec1 proteins were all localized at the kinetochores of prometaphase PRP4-depleted cells (; ). On the other hand, MPS1, MAD1, and MAD2 were not detected at these kinetochores, although their protein levels were normal by Western blotting ( and S1 A). Therefore, our data indicate that PRP4 is specifically required for the ordered recruitment/maintenance of MPS1, MAD1, and MAD2 at the kinetochore.
The fission yeast homologue gene seems to be necessary for pre-mRNA splicing (), but our experiments show that this function is not conserved or required for cell viability in the human (Fig. S3). Moreover, all of the checkpoint proteins analyzed so far by Western blotting proved to be expressed at control levels. This indicates that the defects observed in this study are direct and are not caused by the disappearance of known checkpoint proteins. In agreement with a direct function for PRP4 during the checkpoint, a pool of the PRP4 protein localizes at the kinetochores of isolated chromosomes, the right place to perform this mitotic function. Moreover, PRP4 has recently been identified as a component of the human mitotic spindle phosphoproteome, which contains kinetochore-associated proteins ().
In this study, we found that PRP4 is important for chromosome alignment. In addition, several lines of evidence support our conclusion that belongs to the family of SAC regulatory genes. First, after PRP4 depletion, severe chromosome segregation defects are observed. Second, the cells do not arrest in mitosis in the presence of nocodazole. By themselves, these two criteria indicate a requirement of PRP4 for SAC function. This effect can be easily explained by the lack of MPS1 recruitment that is itself needed for the recruitment of MAD1 and MAD2.
Interestingly, PRP4 also reduces the mitotic duration. This places PRP4 in the checkpoint category, including MAD2 and BUBR1 (components of the soluble mitotic checkpoint complex), which are required to maintain a minimal mitosis duration (). How does PRP4 mediate this additional function? The simplest hypothesis could be that PRP4 regulates the cytoplasmic pool of the mitotic checkpoint complex. Indeed, even if a fraction of the PRP4 protein is localized at the kinetochore region, most of the protein is cytoplasmic with a granular and punctiform distribution during mitosis. This localization is intriguing compared with those of previously characterized checkpoint proteins. Regardless, appears to be an important gene regulating SAC, and further studies will be required to provide additional mechanistic insights into how PRP4 controls checkpoint function and mitotic duration.
HeLa and HeLa stably expressing GFP-tagged H2B (provided by Dr. H. Kimura, University Of Kyoto, Kyoto, Japan) cell lines were maintained in DME supplemented with 10% FCS and antibiotics (penicillin and streptomycin).
siRNA oligonucleotides were purchased from Eurogentec. For human RNAi, three different sequences were used: 5′-UGCAAGAGCCAACCAAGAA-3′, 5′-GCAGGAAUCUUCGUCUGAU-3′, and 5′-UGAUAUGUUUGCUGCGUAU-3′. For the rescue experiment, a siRNA with the sequence 5′-AGACCAACGUAAGAAAGUA-3′ was used. For RNAi, the sequence used was 5′-UACGGACUCACCUUGCUUG-3′ (). A random siRNA was used as a control. siRNAs were transfected into cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Each experiment was repeated at least three times.
The siRNA-resistant construct (GFP-PRP4rr) was generated by PCR using the vector pBlueScriptR-PRP4 (IMAGE clone 4821304; GenBank/EMBL/DDBJ accession no. ) and two primer pairs (mutations and restriction sites are underlined) with the following sequences: (A) 5′-TCACCGCCGCGGAGACCCAG-3′, (B) 5′-CCTTTAGCTGGTGTACTTTTCGTGTCTTCAGG-3′, (C) 5′-CCTGAAGACACGAAAAGTACACCAGCTAAAGG-3′, and (D) 5′-CTTGTTTAAATTTTTTCCTGGATG-3′. Two PCRs were performed with the oligonucleotides A/B and C/D. Both PCR products were mixed and amplified with primers A and D. The resulting fragment was cloned into pEGPC3 (BD Biosciences) using EcoRI and BamHI restriction sites to generate the GFP-PRP4rr construct. The mutations were confirmed by DNA sequencing. To generate a recombinant histidine-tagged PRP4(aa 504–688), a partial sequence of human was amplified by PCR from the vector pBlueScriptR-PRP4 using the primers 5′-GAATTTAAAGGAAGTCTTTCTGAA-3′ and 5′-TTATTGTAACGTTTATCTAGGA-3′ (restriction sites are underlined). The PCR product was cloned into the pET-21a expression vector (EMD) using EcoRI and NotI restriction sites.
Competent BL21 was transformed with pET-21a-PRP4(aa 508–688) (EMD). The recombinant protein expression was induced with 1 mM IPTG in cells during 4 h at 25°C. The protein was purified according to the manufacturer's instructions (QIAGEN), dialyzed overnight in PBS at 4°C, and used for rabbit immunization.
To investigate the duration of mitosis, HeLa cells transfected with control or with each siRNA oligonucleotide were filmed on an inverted video microscope (time exposure of 10 ms) for 48 h after RNAi treatment. Images were acquired every 1 min. Videos were analyzed to determine the delay between the NEBD and anaphase onset.
For kinetochore and centromeric protein localization, cells were fixed for 30 min in PHEM buffer (60 mM Pipes, 25 mM Hepes, 10 mM EGTA, and 8 mM MgCl) containing 4% PFA and 0.5% Triton X-100 and were washed three times during 5 min in PBS. To monitor PRP4 protein localization at the kinetochores, the cells were prepermeabilized for 60 s in PBS and 0.1% Triton X-100 and fixed for 10 min at −20°C in methanol.
500 ng/ml nocodazole was added to cultured cells for 2 h. Mitotic control cells or cells expressing the GFP-PRP4rr construct were harvested by mitotic shake. Culture medium with mitotic cells was recovered and supplemented with 75 mM KCl for 30 min at 37°C. Cells were spun onto coverslips at 2,000 rpm for 10 min. Coverslips were fixed for 15 min in PBS containing 4% PFA and were rinsed in PBS and processed for immunofluorescence. Staining was viewed with an inverted confocal microscope (DM IRE2 SP2; Leica) equipped with a 63× NA 1.4 objective lens using the LCS 3D software (Leica).
Cells were transfected with the plasmid pAdCMV-glob () and/or siRNA using Lipofectamine 2000. After 48 h, cells were collected, and total RNA was isolated with the Nucleospin RNA II kit (Macherey Nagel) and treated with 4 U Turbo DNase (Ambion). Reverse transcription reactions were performed in the presence of 1 μg of total RNA, 0.5 μg oligo dT, and 200 U SuperScript II Reverse Transcriptase (Invitrogen) for 50 min at 42°C. PCR reactions were performed with the Expand high-fidelity PCR system (Roche) for 40 cycles. Primers were complementary to exons 1 and 2 of the β-globin plasmid but not for the endogenous β-globin gene. PCR products were separated in agarose gels. If transgenic mRNA splicing is effective, a 334-bp PCR product is generated. In case of defective mRNA processing, a longer 464-bp product is generated.
Fig. S1 shows protein levels in control and PRP4-depleted cells and GFP-PRP4 kinetochore localization. Fig. S2 presents quantification of mitosis duration in control and PRP4-depleted cells. Fig. S3 shows that human PRP4 knockdown does not inhibit RNA splicing. Video 1 shows a control mitotic HeLa cell expressing GFP-tagged H2B 48 h after control RNAi. Video 2 shows a mitotic HeLa cell expressing GFP-tagged H2B 48 h after PRP4 RNAi. Table S1 shows the percentages of PRP4-labeled kinetochores during prometaphase, metaphase, and anaphase. Online supplemental material is available at . |
When I started as an undergraduate at the University of Melbourne, I was mostly interested in psychology and mathematics. My dad's a mathematician, so I've always been comfortable with math, and it came easily to me. But then I had some fantastic biology professors in my first year at university. They really got me interested in biology, so I shifted my emphasis to zoology and botany.
In particular, I was interested in ecology and invertebrate zoology, but I felt like I didn't have the patience for the fieldwork that was required, especially for ecology research. So I became much more enamored with cell biology, and in particular in plant cell biology.
In Australia, the undergraduate degree is three years long, and then you have what's called an honors year, when you work in a research lab on a small, independent research project. That was my first real introduction to lab research. Again, I had some great professors who really made the internal workings of plant cells interesting to me.
That year, I was working on how algae assemble their cell wall components, so I got to thinking a lot about how proteins move around in cells, and how cells move proteins around. That really appealed to me as a great biological problem. I decided when I was looking for a Ph.D. lab that I wanted to study protein trafficking. I looked around at various labs in Australia and settled on Marilyn Anderson, who was actually in the botany department at Melbourne, where I had been an undergraduate. She was just moving to take a position in the biochemistry department at La Trobe University.
I moved out there with her, and she had a great project that looked at the trafficking of defense proteins to the plant vacuole, which is like the yeast vacuole and the mammalian lysosome. It's a storage compartment, and these defense proteins are stored there in extremely high quantities.
My thesis work demonstrated that for a particular family of protease inhibitors (one kind of defense protein), there is a COOH-terminal domain that's required for delivery of their precursor protein to the vacuole. Once the precursor protein has been delivered to the vacuole, proteolysis occurs to release the fully active protease inhibitors.
A mentor who's been a great influence to me is Trevor Lithgow, who was also at La Trobe in the biochemistry department and is now at the University of Melbourne. He had been a postdoc with Jeff Schatz, who was one of the luminaries in the field of protein trafficking to mitochondria. Trevor was a local boy from the suburbs of Melbourne who went to the Schatz lab and had a fantastic time.
He really encouraged me to find the best lab that you can and do an amazing postdoc and have a great time, because it's a really fantastic time of life, and it's an opportunity that you don't necessarily appreciate at the time. I looked at labs in Europe and the US, and I settled on Randy Schekman's lab in Berkeley for a couple of reasons.
First, I thought the yeast system was fantastic because of its great genetics and the ability to do both biochemistry and genome-wide analyses. Then, obviously, Randy is at the top of his field and is also a great guy. It was a great environment to work in. An added bonus was that Randy also offered a postdoc position to my husband (who is in a similar field), which solved our two-body problem of two scientists trying to get jobs together.
We had reason to believe that a coat protein called Sec24 might be involved in selective capture of newly synthesized proteins into vesicles that leave the ER and deliver proteins to downstream compartments. I first studied isoforms of Sec24 involved in cargo capture. Later, Randy came up with an idea that makes a postdoc want to run and hide. He wanted to try alanine-scanning of the surface of Sec24. It's a 105-kD protein, meaning there are a not-insignificant number of residues to target on the surface that might be cargo-binding sites! He convinced me to do it, so we picked charged residues to target because we knew that some proteins used ER export motifs that are charged. I generated a library of about 20 alleles of Sec24 that had been mutagenized in these charged residues. We've now published two of them that correspond to specific cargo-binding sites. That work has dovetailed really nicely with ongoing structural work from Jonathan Goldberg's lab that also defined these two sites.
Growing up, actually, I was very anti-Vegemite. But then you can't work with yeast and be anti-Vegemite, so now I'm incredibly pro-Vegemite. I'm pro-beer also. All yeast products are good to me [].
Well actually, brewing beer is something my husband and I got into when we were in Randy's lab. A friend from Australia set us up with a little home brewing system. I should point out, we don't use our lab strains. We really do use proper brewing yeast. We had done it in Berkeley quite a bit, and when I moved to New York, I decided I really wanted to brew a Christmas ale for the department Christmas party. I thought I'd do it in the lab so everybody could come down and see fermentation in action, and everybody in the department could have a connection to this beer.
The result of having this brewing yeast in the lab, which is more robust than our lab strains, was that we ended up with a terrible contamination of the diploid brewing yeast that took over some of our plates. For about a month, things weren't making sense anymore; strains were growing really well when they shouldn't be.
So we bleached everything in sight, threw out all the suspect stuff, and my students and postdocs made me promise to never brew beer in the lab again! I still brew for the department Christmas party, but I do it at home now.
The project that I've taken with me from Randy's lab to start on my own here at Columbia is expanding our understanding of how that cargo/coat recognition occurs. In particular, the question that we're interested in now is whether the folding of the cargo influences its interaction with the coat proteins, and if so, how. This question stems from an observation—although one that's probably not universally true—that misfolded proteins are not packaged into ER-derived transport vesicles.
One possible reason is that the ER contains a whole host of chaperones that help these proteins fold. Once you've removed secreted proteins from the ER environment, they can't fold properly because they no longer have access to those chaperones. If misfolded proteins escaped the cell, they could be toxic, so there's a real necessity to retaining them.
The question is, What is the mechanism by which misfolded proteins are recognized and prevented from gaining access to a vesicle? We just published our first paper on this topic in September, so that was a real landmark and really exciting for all of us. Things are moving along. And it really is a lot of fun. |
italic
#text
The meeting began, appropriately, with a dedication to Bjorn Afzelius, presented by Romano Dallai, a close friend and colleague. In the 1930's, a Swiss physician, Manes Kartagener, observed the characteristic triad phenotype that came to be known as Kartagener's Syndrome (, sinusitis, and bronchiectasis) (). Afzelius was the first to observe that a number of male patients with Kartagener's Syndrome also were sterile. He determined that the sperm and airway cilia from these patients had abnormal axonemal ultrastructure; some patients were missing, for example, the dynein arms required for flagellar motility (). The relationship between these ultrastructural observations on cilia and flagella and the pathology of , however, remained enigmatic for over 20 years after Afzelius' study.
Afzelius, the slated keynote speaker at the meeting, was unable to attend. Instead, George Witman presented a comprehensive review of the role of cilia in disease, starting with Afzelius' investigations on Kartagener's syndrome patients. He reviewed recent breakthroughs on flagellar assembly and the discovery of IFT, first described in the Rosenbaum laboratory (), and the seminal observations of Nobutaka Hirokawa and colleagues from the University of Tokyo. The latter group showed that the “9+0” cilia of the embryonic node have a unique motility and are required for the development of normal left/right organ asymmetry (). Thus, a definitive link to ciliary motility was finally made, explaining the developmental and cell biological basis for in Kartagener's patients. Note also that some nodal cilia have recently been reported to be 9+2 (); for more discussion of this point, see the section below on cilia as signal receivers.
Witman then described how studies on IFT in led to the discovery that defects in primary cilia caused PKD in mammals (). The connection between cilia and PKD was soon clarified by the discovery that the polycystins—products of the genes that are defective in PKD in humans—are displayed on the primary cilia of kidney tubule cells (; ). Thus, mutations in ciliary proteins (the polycystins) or a complete loss of cilia (due to defects in IFT) result in the pathology of PKD. This initial correlation between primary cilia and a specific disease was then followed by a multitude of research papers linking a variety of cystic and developmental diseases to normal cilia function.
italic
#text
The technique of cryo-electron microscopy/tomography coupled to computer enhancement has contributed new insights into the high resolution structure of the axoneme and the complex inner and outer dynein arms which are composed of multiple polypeptides. Reports by both Daniela Nicastro (Brandeis University) on the structure of ciliary axonemes and dynein, and Takashi Ishikawa (ETH, Zurich) on the dynein complex of wild-type and mutant , provided new three-dimensional information on the dynein motor complex and how it may be functioning to cause microtubule outer doublet sliding. However, these two groups differed in their reports with respect to the angle of orientation the outer arm dynein heads have relative to the doublet microtubules, which has important implications for the mechanism of force generation by the multi-headed dynein. Both reports were founded on the elegant negative stain images of the dynein motor by Stan Burgess (University of Leeds) who presented single-particle analysis of the circular motor head of the dynein heavy chain and how it might be moving. Using replicas obtained by quick freeze-deep etching, Pietro Lupetti (University of Siena) reported that the two dynein heads are positioned at different distances from the A-tubule. Furthermore, the stalk and the stem domains do not reside in the same plane as the head, which is contrary to what had been previously reported for isolated dynein adsorbed to a flat surface. In the first use of tomography to study the structure of IFT particles, Gaia Pigino (University of Siena) showed that these are complex structures, but that periodicities of 8.5 nm or 25 nm could be observed in a given group of IFT particles depending on location within the particles.
Reports on dynein structure were complemented by biochemical analyses of the dynein motor complex by Stephen King (University of Connecticut Health Center) and continued in reports on the control of dynein activity by the dynein regulatory complex (DRC) by Mary Porter (University of Minnesota), the role of calmodulin and the radial spokes in the control of flagellar beating by Elizabeth Smith (Dartmouth College), and the function of kinases and phosphatases, some associated with the flagellar radial spokes, in the control of dynein activity by Winfield Sale (Emory University) and by Avanti Gokhale (Emory University). Additional important analyses of the diverse, functional capabilities of dynein subforms and domains were described by Ritsu Kamiya (University of Tokyo), and new advances on dynein regulation were presented by Chikako Shingyoji (University of Tokyo). Part of the dynein control mechanism involves the orientation of the central pair microtubules, and David Mitchell (SUNY Upstate Medical University) provided an update on how this complex central pair apparatus rotates, modulating the radial-spoke based activity of the dynein arms in different halves of the doublet circle. This central pair complex also contains kinesin (), whose function is required for motility but in a manner yet to be determined.
One of the highlights of the meeting was a group of talks on the control of ciliary length and stability, and ciliary resorption, in particular as a prelude to cell division. Although IFT is involved in length control because it delivers the required axonemal and membrane precursors to the flagellum (), length is, in turn, controlled by a kinase–phosphatase signaling system. Paul Lefebvre (University of Minnesota) has defined this system in an elegant fashion using the long-flagella (lf) mutants of that grow flagella to two or three times their normal length. Two of these mutants encode protein kinases: LF4, a MAP kinase, and LF2, a protein kinase of the CDK family. The LF4 and LF2 kinases, which localize to the flagella and the cytoplasm, respectively, may act either as length sensors, perhaps by monitoring “time of flight” on moving IFT particles, or as enforcers of length control by activating disassembly of flagella that have grown in length beyond a predetermined set point. In addition, William Dentler (University of Kansas) reported that disruption by Brefeldin A of membrane delivery to the flagellum from the Golgi induced flagellar disassembly and inhibited flagellar assembly without affecting IFT.
A related series of presentations addressed the relationship between resorption of the primary cilium and the ability of the cell to exit G and reenter the cell cycle. This cilia–cell cycle control hypothesis has been discussed for decades, but other than an observed correlation between cilia resorption and cell cycle progression no direct cause-and-effect data have been presented. This has been remedied by several sets of data addressing cell cycle control. First, Lynne Quarmby (Simon Fraser University) showed how mutations in the Cnk and Nek kinases (of which there are many different ones in ciliated organisms) of affected flagellar resorption and the cell cycle. Second, Erica Golemis (Fox Chase Cancer Center) reported an interaction between the cell cycle kinase Aurora A and the pro-metastatic scaffolding protein HEF1 that occurs at the basal body in response to external cues. This interaction activates Aurora A kinase activity, the target of which is HDAC6, a deacetylase that removes the acetate group from axonemal tubulin, thus destabilizing the axoneme as a prelude to disassembly of the primary cilium. Tubulin deacetylation by HDAC6 is necessary and sufficient for cilia destabilization and resorption, and this is required for entry into the cell cycle from Go (). Interestingly, previously identified an Aurora-like kinase (CALK) as a key component in flagellar resorption in . Third, the coregulation of the cilia cycle and the cell cycle was nicely documented by Zhaohui Wang (Yale University), who presented evidence that IFT27, a small G protein and a component of IFT Complex B, also plays a role in cell division. The amount of IFT27 is constant per cell, and thus as the cell grows, the concentration of IFT27 falls. When the concentration of IFT27 decreases below a certain level, resorption of the flagella is induced, and the cell enters S-phase. A rise in IFT27, due to new transcription, is then required for the completion of cell division. If IFT27 does not rise (i.e., in a knocked-down cell), the cell cycle is blocked at cytokinesis.
italic
#text
xref
italic
#text
An exciting new development in ciliary assembly and function in relation to disease () concerns the Bardet-Biedl syndrome (BBS) proteins. BBS is a disease syndrome resulting in a large array of pathologies, including obesity, defective bone development, diabetes, and other developmental disorders; there are at least 12 different BBS genes and the BBS proteins are located primarily in the centrosome–ciliary complex (). Max Nachury (from the Peter Jackson group, Genentech and Stanford) reported that many of the BBS proteins sediment together on gradients as a complex, called the BBSome, providing some of the first evidence that the diverse pathologies resulting from mutations in the different BBS genes are related. Alexander Loktev (Genentech/Stanford), also of this group, reported that the BBSome might be membrane associated, both in the cytoplasm and in the cilium itself. Unfortunately, primary cilia cannot (yet) be isolated in sufficient quantities from tissue culture cells, or from , so the ciliary membrane association cannot be directly tested biochemically in these cells. The Genentech group, collaborating with Val Sheffield (University of Iowa), hypothesized that the BBSome is involved in the movement of ciliary membrane vesicles from the cytoplasm to the cilia surface. Membrane trafficking to and within the cilium occurs through the interaction of the BBSome with Rabin8, a GEF for the small GTPase Rab8, as inhibition of the production of active (i.e., GTP-bound) Rab8 blocks ciliation and induces BBS phenotypes in zebrafish. Related data from Gregory Pazour (University of Massachusetts Medical School) showed that RAb8 is important for opsin transport to the OS; a mutation in Rab8 causes opsin to remain in the inner segment in photoreceptor cells. Pazour also showed that the T22N mutation in Rab8 blocks ciliary targeting of PKHD1-GFP (i.e., fibrocystin-GFP).
Bradley Yoder (University of Alabama) used conditional alleles of two genes (Kif3A and Tg737/IFT88) required for ciliogenesis to demonstrate the connection between primary cilia and obesity (one of the phenotypes in Bardet-Biedl syndrome). Systemic loss of cilia in adult mice results in animals that eat too much (hyperphagia) and are thus obese and have related defects, such as problems with glucose homeostasis. By disrupting cilia only in the arcuate nucleus of the hypothalamus, Yoder could show that this mimicked the hyperphagic effect caused by a systemic loss of cilia, thus strongly suggesting that primary cilia in the CNS mediate the reception and/or processing of satiety signals ().
It is now relatively clear that the primary cilia of cells in developing tissues such as mammary gland and bone are absolutely essential for normal tissue morphogenesis. Investigations of bone development by Courtney Haycraft (University of South Carolina) and mammary gland development by Kimberly McDermott (University of California, San Francisco) from the Tlsty lab showed that these tissues do not develop properly in the Tg737/IFT88 mouse in which the primary cilia on most cells are either not present or greatly reduced in size (). Tubule branching in the mammary gland is greatly inhibited, and bone development likewise is highly abnormal, including polydactyly and improper bone length. The exact role of the primary cilia in the morphogenesis of these tissues is not yet known, but it almost certainly is going to involve a role in cell polarity during embryogenesis and subsequent postnatal development. With respect to cell polarity, Chonnettia Jones (Emory University) reported use of a conditional allele for Tg737/IFT88 in the cochlea. This resulted in a loss of the primary cilium (the kinocilium), and a defect in the organization of the stereocilia (which are actin-based microvilli). The alignment of the basal bodies is also disrupted, indicating that Tg737/IFT88 interacts with the planar cell polarity pathway (a noncanonical Wnt signaling pathway) and has a distinct role in placement of the centrioles. Using genetics combined with image analysis, Wallace Marshall (University of California, San Francisco) has begun to identify the basal body components required for positioning, while Susan Dutcher (Washington University) reviewed the basal body mutants in that assemble morphologically abnormal basal bodies. These mutants all have consistent phenotypes: flagellar assembly defects, errors in cleavage furrow placement, and supersensitivity to taxol.
In interesting work on the role of primary cilia in wound healing reported by Christensen (University of Copenhagen), it is clear that the cells which are moving to close the wound all have their primary cilia pointing in the direction of cell movement, whereas the nonmotile cells behind the wound do not. The possibility here is that the primary cilia on the wound edge first point in the proper direction because of external signals, thus reorienting the ciliary basal body/centriole, which in turn orients the associated cytoplasmic microtubules whose repositioning finally activates the dendritic assembly of actin to protrude the leading membrane edge, resulting in directed cell motility.
There are now many reported examples of how primary cilia sense the environment either by mechano-, chemo-, or photoreceptors on the membrane of the primary cilium. William Snell (University of Texas Southwest Medical Center) reported a role for several different kinases in the mating reaction in ; this event depends on interaction of the flagella of plus and minus mating types and requires a functional IFT system. Søren Christensen (University of Copenhagen) reported the presence of a variety of receptors on the ciliary membrane including PDGFαα (), which could coordinate both growth control and directional cell migration in development and wound healing. Work reported by Kathryn Anderson (Memorial Sloan-Kettering Cancer Center), Bradley Yoder (University of Alabama), and Søren Christensen showed that components of the Hedgehog (Hh) signaling pathway, including the Hh receptor (patched) and its interacting protein (smoothened), are present in the primary cilium as well. Two years ago showed that the transcription factor Gli 3, which is inhibited by smoothened upon Hh activation of patched, is localized to the tips of primary cilia. Perhaps the most detailed work on receptors associated with cilia has been performed on polycystins 1 and 2 (PC1 and 2, the gene products of PKD1 and 2, respectively), which are found on the ciliary membrane, in addition to other areas in the cell such as the ER and the apical cell surface. The primary cilium has been shown by Helle Praetorius (University of Aarhus) to be able to sense flow; Jing Zhou (Harvard University), Michael Caplan (Yale University), and others have proposed that movement of fluid in the kidney tubules causes ciliary bending and the consequent opening of Ca channels encoded by PKD2. The resultant calcium flow into the cell keeps the cell in the nondividing state. Lack of fluid flow and/or inhibition of calcium influx due to the loss/malfunction of PC2 allow cleavage of the cytoplasmic C terminus of PC1. Michael Caplan (Yale University) showed that the C terminus then finds its way into the nucleus, activating signaling pathways that modulate the uncontrolled cell division characteristic of PKD.
This hypothesis of PKD pathogenesis via regulated intramembrane proteolysis () controlled by fluid flow () may, however, be oversimplified because Gregory Germino (Johns Hopkins University) and colleagues identified a two-day postnatal interval in mice that dramatically determines the kidney's response to PKD1 inactivation. Loss of PC1 at any point before this time resulted in severely cystic kidneys within three weeks, while loss of PC1 after this critical two-day period did not result in a cystic pathology for as long as five months. Bradley Yoder (University of Alabama) noted a similar pattern of late onset cystic disease in his systemic model of Tg737/IFT88 inactivation. So, although cilia are indeed important in normal kidney development and maintenance (; ; ), it would appear that they are doing more than merely sensing fluid flow via polycystin-mediated calcium channel activation. For example, Stefan Somlo (Yale University) also presented data on the flow hypothesis of PKD, suggesting that calcium flux could occur in the kidney tubule cells when PC2 is not present in the cilia, by use of a mutant form of PC2 that retains its ability to act as a calcium channel but that does not localize to the primary cilium. Thus, PC2 is required for the calcium response, but it appears that ciliary PC2 is not required. However, the most direct positive evidence for the mechanosensory activity of kidney cilia continues to be the work of , who showed that moving a single cilium caused calcium entry into that cell. Continuing this work using freshly isolated renal tubules, Helle Praetorius (University of Aarhus) reported that renal flow induces activation of purinergic P2 receptors, resulting in a calcium increase that could be inhibited by externally applied ATP scavengers such as apyrase. Thus, flow may be inducing an ATP release that in turn activates purinergic P2 receptors leading to the calcium increase. If specific (P2Y2) purinergic receptors are knocked out, the flow response is significantly lower (P < 0.05). Note that the matter is far from settled, however, because one class of P2 receptors (P2YR) has the ability to release calcium from internal stores ().
The activity of receptors and calcium in the cilium requires now a return to the discussion of introduced at the beginning of this review. Two hypotheses were proposed to explain how leftward flow of extraembryonic fluid in the node, caused by the motile nodal cilia, produces left/right body asymmetry. Nobutaka Hirokawa (University of Tokyo) proposed that the flow moves vesicles filled with morphogens (sonic hedgehog and retinoic acid) from right to left (), activating signaling pathways in the receiving cells on the left side of the node, resulting in asymmetric folding. Alternatively, Martina Brueckner (Yale University) proposed that the two populations of cilia present in the node (recall ) work in concert to set up the body plan. Those in the center of the node are motile and are responsible for the leftward fluid flow, which in turn bends the nonmotile cilia on the left side of the node, inducing calcium flow through ciliary PKD channels (). Brueckner reported that inversin, which has an essential function in left/right development, relocalizes in the cytoplasm of cultured IMCD3 cells in response to elevated calcium. Asymmetry of inversin localization in cells on the left side of the node also occurs, and this is thought to lead to activation of signaling pathways that ultimately results in asymmetric morphogenesis of the body organs.
A very important series of results were reported linking cilia abnormalities to a range of human diseases beyond those (PKD, BBS) already mentioned. Karl Lechtreck (University of Massachusetts Medical School) showed that the protein hydin—first identified as a component of one of the central pair structures in that were so elegantly detailed by David Mitchell (SUNY Upstate Medical University)—is involved in the development of hydrocephalus (cerebrospinal fluid accumulation in brain ventricles). In stunningly beautiful videos of ciliary motility on cells lining the brain ventricles, hydin-deficient ependymal cilia were observed to be stiff and unable to form the beat pattern characteristic of normal cilia. Hydrocephalus develops presumably because a lack of proper motility results in decreased fluid transport. In a related presentation, Heymut Omran (University Children's Hospital, Freiburg) reported for the first time that a mutation in the DNAI2 gene encoding outer arm dynein intermediate chain 2 results in primary ciliary dyskinesia (characterized by defects in motile cilia of the respiratory tract, embryonic node, and sperm) in some patients.
Meckel-Gruber syndrome is a rare recessive disorder characterized by cystic kidneys, polydactyly, and incomplete skull closure. Helen Dawe (University of Oxford) showed that Meckel-Gruber syndrome proteins Mks1 (localized to the centrosome) and MKS3 (meckelin, localized to the primary cilium) are required for centrosome migration and subsequent ciliogenesis, perhaps through interactions with Rho kinase and myosin II. Veena Singla (University of California, San Francisco) reported that the OFD1 (orofaciodigital syndrome type I) gene product localizes to the basal body, and cells lacking the OFD1 gene product cannot generate cilia. Thus, in the absence of cilia due to the loss of a key basal body component, malformations of the face, oral cavity, and digits, which together with PKD are characteristics of OFD syndrome, occur. Surprisingly, loss of OFD1 does not affect the cell cycle in embryonic stem cells in culture. Cholangiocytes (epithelial cells lining the bile duct) in the PCK rat, a model for autosomal recessive polycystic disease (ARPKD), do not express fibrocystin due to a mutation in the PKHD1 gene. Anatoliy Masyuk (Mayo Clinic) showed that the basal bodies in this rat model are heterogeneous in size, contain extra appendages, and are positioned incorrectly in the cell. These structural and functional abnormalities cause hepatic cystogenesis.
The protein cystin, ∼5% the size of fibrocystin, is disrupted in the cpk mouse model of ARPKD. Cystin has an N-terminal myristoylation signal, a ciliary localization signal, and two nuclear localization signals. Often, but not always, cystin resides at the ciliary tip. Lisa Guay-Woodford (University of Alabama) showed that a myristoyl-electrostatic switch () controls the cycling of cystin on and off the membrane and hence the localization of cystin to the cilium or the nucleus. The S17A mutation enhances the association of cystin with the ciliary membrane, preventing its nuclear localization; when in the nucleus, however, cystin interacts with necdin and affects the expression of renal cell genes. Among these are cdc2 and c-myc, the former explaining why inhibitors of cyclin-dependent kinases are effective in treating cystic disease in the mouse kidney.
In a report clearly showing that tubulin polyglutamylation is required for cilia development in vertebrates, Iain Drummond (Massachusetts General Hospital) showed that in zebrafish, the fleer (flr) genes have a pleiotropic affect; flr mutants have kidney cysts, cannot form rod outer segments, have hydrocephalus, and left/right asymmetry defects. In addition, flr mutant cilia are short, and lack the outer side of the B-tubule of the axoneme, which is where polyglutamylated tubulin is located (). Thus, flr protein is involved, perhaps as a cofactor, in tubulin glutamylation. The gene dyf-1 is a homologue of flr and is required for tubulin glutamylation in sensory neuron cilia as well ().
Finally, Uwe Wolfrum (Johannes Gutenberg University of Mainz) presented clear electron microscopic evidence showing that IFT proteins can be found on vesicles near the synapse in rod inner segments before vesicle docking and fusion with the periciliary membrane. These data are consistent with the hypothesis of , who have proposed that the cilium originally derived from a specialized membrane patch to which coated vesicles were transported by primordial IFT machinery. Their hypothesis is supported by the homology of certain IFT proteins to current day COPI and clathrin-coated vesicle proteins. Furthermore, the work of Wolfrum points out the importance of this transport network for cilia assembly and should caution investigators interpreting, for example, knock-down experiments with cilia or IFT-related proteins. A knock-down phenotype may not have anything to do directly with the ciliary apparatus itself, but perhaps to some component of the steps of vesicle budding, transport, and fusion that are required in the cytoplasm before the actual formation of a cilium or flagellum.
#text |
Cell fusion is a fundamental process in the biology of eukaryotic cells; it is essential for fertilization and occurs at key times during somatic development (; ; ). Yeast conjugation, or mating, can be divided into five broad steps: cell to cell signaling, cell polarization, cell fusion, nuclear congression, and nuclear fusion (; ; ). Mating begins when two cells of the opposite mating type respond to the pheromones released by a nearby prospective mating partner. The cells repolarize to form shmoos; that is, each cell forms a projection with its apical end directed toward the prospective mating partner. The two cells adhere at the shmoo tips to form a prezygote. Degradation of the cell wall and fusion of the plasma membranes allow the cells to fuse, providing continuity of the two formerly separate cytoplasms. Finally, the nuclei congress in a microtubule-dependent manner (), and the nuclear envelopes (NEs) fuse to produce a single diploid nucleus (, ).
Like other fungi, undergoes cell division and mating without NE breakdown (). Therefore, NE fusion is essential for the production of a diploid nucleus. However, the pathway controlling NE fusion is not well understood. Yeast NE fusion is a homotypic reaction in that both membranes correspond to the same subcellular compartment. However, nuclear fusion is inherently more complex than viral membrane fusion because it entails the fusion of two pairs of membrane bilayers. Therefore, it is analogous to the fusion of other organelles (; ; ). Like mitochondrial fusion, NE fusion requires the alignment and fusion of both inner and outer membranes.
Yeast karyogamy or mutants are defective in nuclear fusion and can be divided into two major categories. Type I mutants display defects in nuclear congression in that the nuclei fail to move together and typically exhibit defects in the cytoplasmic microtubules (; ; ). In type II mutant zygotes, the nuclei become closely apposed yet remain unfused (). In certain class II karyogamy mutants, the apposed nuclei appear to be connected by membrane bridges that run between the outer nuclear membranes (; ). Both of these classes of karyogamy mutants typically progress into mitosis, yielding haploid progeny.
All of the proteins known to be required for nuclear membrane fusion (e.g., Kar2p, Kar5p, Kar7p/Sec71p, Kar8p/Jem1p, and Prm3p) are localized to the NE (; ; ; ; ). However, one of the mysteries of NE fusion is that many of the requisite proteins reside within the NE lumen, where it is unlikely that they could play a direct role in the initial stages of membrane fusion. Nevertheless, the genetic data are supported by in vitro fusion experiments that indicate a role for the luminal proteins Kar2p, Kar5p, and Kar8p/Jem1p in the overall fusion process (; ). To date, it has been difficult to define the role of these proteins, although their roles may be similar to functions that they are already known to perform in the ER (; ; ; ; ).
The yeast microtubule-organizing center, the spindle pole body (SPB), is critical to both nuclear congression and nuclear membrane fusion (for review see ). The cytoplasmic microtubules that bring the two nuclei together during congression are nucleated from the half-bridge of the SPB (). Several type I karyogamy mutants have defects in SPB-associated proteins, including Kar1p and Mps3p (; ; for review see ). During NE fusion, the SPBs are thought to dictate the initial site of membrane fusion (). However, the steps of membrane fusion have remained unclear. Two models have been proposed to describe these steps (; ). In the one-step model, membrane fusion occurs at the SPB, at whose margins the inner and outer membranes are continuous. Such fusion could allow the inner and outer nuclear membranes to fuse simultaneously. In the alternative model, nuclear fusion occurs in three separate steps. First, the outer membranes fuse, allowing the contents of the NE lumens to become continuous. Next, the inner membranes fuse, allowing continuity of the nuclear contents. Finally, the SPBs fuse within the plane of the membrane. Early electron microscopy studies supported the one-step model, with nuclear membrane and SPB fusion occurring simultaneously (, ; ; for review see ). However, the luminal localization of proteins required for karyogamy, such as Kar2p, was difficult to reconcile with the one-step model.
We have used two experimental approaches to determine whether nuclear fusion during mating occurs via one or three steps. First, electron tomography (ET) was conducted on fixed wild-type zygotes at different stages of nuclear fusion. ET is superior to serial section electron microscopy for the reconstruction of spatial details in three dimensions in several ways (). First, it allows the examination of thick slices of a cellular sample, provides images with essentially isotropic resolution at 4–6 nm, and allows a flexible analysis of 3D structures. Second, live cell time-lapse microscopy was conducted on zygotes tagged with fluorescent protein markers for each relevant compartment to map the stages of nuclear fusion. Live cell fluorescence microscopy lacks the spatial resolution to visualize SPB fusion but, in contrast to static electron microscopy, allows direct determination of the temporal order of events. Thus, the two approaches complement one another. Collectively, our data show that nuclear fusion occurs by three distinct steps: outer membrane fusion, inner membrane fusion, and SPB fusion.
Although the general stages of cell fusion and karyogamy have previously been described by electron microscopy (; ), specific intermediates of nuclear fusion were not well characterized. Therefore, we elected to use ET to provide a more detailed examination of the outer and inner NE, the SPBs, and cytoplasmic microtubules during nuclear fusion.
Nuclear congression and membrane fusion occur within 10–15 min after cell fusion, making it difficult to capture intermediates in nuclear fusion (; and our unpublished data). To identify zygotes at the appropriate stage of mating just before and during membrane fusion, we scanned populations of mating cells by light microscopy to identify times at which a large fraction of the cells was at an appropriate stage of conjugation. These populations were cryoimmobilized with a high pressure freezer, fixed by freeze substitution, embedded in plastic, and serially sectioned for examination by ET (see Materials and methods). Individual sections were scanned at low magnification to identify zygotes that were parallel to the plane of sectioning and that had a narrow zone of cell fusion (measured orthogonal to the long axis of the zygote), which indicated that cell fusion had occurred just before freezing. Such cells were found with a frequency of roughly one in 1,000. In this population, we identified zygotes in which the NEs were unfused, in which NEs were partially fused, in which NE fusion had been completed, and at several stages of SPB fusion (see –
). Tomograms of 22 wild-type zygotes were analyzed (Table S1; available at ). In some cases, it was possible to reconstruct two or three serial semithick sections, which provided 0.5–1 μm of sample thickness and an area of ∼2 × 4 μm in the section plane. From these 3D image data, models highlighting the relevant cellular structures (outer and inner NE, SPBs, and microtubules) were generated.
A representative model of a zygote, which had completed cell and NE fusion, is shown in , with all structures indicated (see the tomographic slice in A and the model in D; also see Video 1, available at ). Microtubules (, yellow) were traced from their point of nucleation near the SPB either to their true end or until they left the reconstructed volume (). The margins of the SPBs were traced in pink. Outer and inner nuclear membranes were traced on every fourth tomographic slice along the z axis, an interval of 4–5 nm (, D and E; green and blue, respectively). After the major cellular structures had been traced, IMOD software was used to mesh the outlines, providing a 3D rendering of the modeled objects (). In the example shown in , the cell walls and plasma membrane (purple) were continuous, indicating that cell fusion had been completed. The nuclei were joined by continuous inner and outer nuclear membranes (, B and E; blue and green, respectively), indicating that NE fusion had been completed. Cell fusion had occurred recently in this zygote, as indicated by the presence of vesicles near the zone of cell fusion (, red). In addition, the zone of cell fusion and region of nuclear fusion were both very narrow. Both regions expand as the zygotes mature (). Often the NEs appeared to be stretched out in zygotes in which nuclear fusion had not advanced far beyond the initial stages. The stretched appearance may reflect random movement of the nuclei and tension on the NEs. In , the stretching appeared asymmetric; however, in other zygotes, stretching was more symmetric ().
The proximity of the nuclei, the morphology of the membranes, the position of the SPBs, and the width of the region of nuclear fusion (if present) were all used to order the zygotes along a presumed pathway of nuclear fusion. Tomograms were carefully examined to determine whether outer membrane fusion, inner membrane fusion, and SPB fusion had occurred. In a representative tomogram of a mating pair that had completed cell fusion but not yet initiated nuclear fusion (), one can see two nuclei with both outer and inner membranes still separate (, green and blue, respectively). Note the elongated shapes of the nuclei, which narrow to a rounded point in the region closest to one another. Electron-dense material is visible at the tips of both nuclei. On either side of the forward edges of the nuclei are the SPBs embedded in the NE (, pink disks). The dark-stained layers extending from the margins of the SPBs over the forward edges of the nuclei are the half-bridges. Cytoplasmic microtubules were observed connecting the two SPBs but have been omitted from the models for clarity. Vesicles clustered near the remnant cell walls at the zone of cell fusion indicate that cell fusion had occurred recently in this zygote.
In 21 of our 22 tomograms, NE fusion had been initiated or completed. One tomogram appeared to capture a particularly interesting intermediate in NE fusion (). Although the SPBs were unfused (, left), a different tomographic slice showed that the outer nuclear membranes were continuous (, middle; green in model). Close examination of slices through the region of fusion showed that the inner membranes were closely apposed but still distinct (, middle; arrowhead; light blue in model panel), indicating that outer NE fusion must precede inner envelope fusion. It is likely that this zygote represents a true intermediate caught just after the initiation of NE fusion, but this state has been observed only once. Because of the rarity of this class, it is formally possible that this image represents an aberrant event in which nuclear fusion had not proceeded normally. Therefore, this issue was examined with an alternative method (see below).
Both the outer and inner membranes were continuous between the two nuclei in 20 tomograms, indicating that NE fusion had been completed at the time of rapid freezing ( and ). In seven of these zygotes, two completely separate and distinct SPBs were observed ( and , A and B), demonstrating that SPB fusion had not yet occurred. In zygotes with separate, unfused SPBs, the median width of the nucleus at the region of fusion was only 110 nm (measured across the narrowest part of the nucleus between the outer edges of the NE; SD = 30 nm, with one outlier of 302 nm; ). The narrowness of the isthmus suggests that these zygotes had only just completed nuclear fusion.
The last group of 13 tomograms included zygotes with SPBs in the process of fusion or already fused SPBs (). In all such zygotes, both inner and outer NE fusion was complete. In eight zygotes in the initial or intermediate stages of SPB fusion, the nuclear fusion zone had widened to a median width of 153 nm (SD = 92 nm; one outlier of 1,415 nm; ). In five zygotes, SPB fusion was either complete or almost complete. In the four zygotes in which the membranes could be measured, the nuclear fusion zone had expanded to a median width of 567 nm (SD = 93 nm). Collectively, these data demonstrate that SPB fusion occurs considerably after the completion of nuclear membrane fusion.
Six layers of the SPB can be detected by ET (). The central plaque, which is embedded in the NE, is one of the most electron-dense layers, and this structure was modeled to mark the location of the SPB. Extending from the margin of each SPB is the half-bridge, an electron-dense region of the NE (for review see ). By ET, the half-bridge is made up of five layers ().
Several distinct morphologies of SPBs were seen in cells frozen during the course of nuclear fusion, suggesting that SPB fusion may occur in several stages. In 8 of the 13 zygotes, the central plaques of the SPBs were clearly separate but joined by their half-bridges. shows a representative tomogram in which two central plaques are joined by the multilayered half-bridge. In one zygote, the half-bridges appear to have become joined along their lateral margins; although the central plaques were close together, they had not yet fused (). This situation is particularly evident when the SPBs were viewed en face rather than in cross section (, middle and inset). In two zygotes, the SPB central plaques appeared to be partially fused (). Finally, in two zygotes, we found only a single SPB central plaque, implying that fusion was complete (). Based on these observations, a reasonable order for the pathway of SPB fusion would entail interactions first between the half-bridges, most likely via lateral interactions along the half-bridge margin, followed by fusion of the central plaques. Regardless of the detailed pathway of SPB fusion, these results clearly show that SPB fusion occurs as a secondary event and does not initiate NE fusion.
ET is not well suited to distinguish closely spaced temporal events. To complement ET, we developed a method to observe nuclear fusion in live cells using fluorescently tagged proteins as markers for different cellular compartments. To observe SPB congression relative to NE fusion, we initially used Spc42p-RFP to mark the SPB and SS-3XGFP-HDEL to label the lumen of the NE ( and Video 3, available at ). Spc42p is a key component of the SPB central plaque (). SS-3XGFP-HDEL is translocated into the lumen of both the NE and ER because of the secretory signal sequence (SS) at its N terminus, and it is retained in the NE/ER by virtue of the C-terminal ER retention signal (HDEL). The signal sequence is cleaved after translocation, so the fluorescing protein will hereafter be referred to as 3XGFP-HDEL. The NE and ER are continuous in yeast; however, the 3XGFP-HDEL signal appears substantially brighter in the NE, as previously observed for lumenal proteins (). Data collection was initiated at or soon after cell fusion, as judged by the appearance of the zygotes by differential interference contrast microscopy.
In the example shown in , the SPBs were already in close proximity at the beginning of the experiment (t = 1), indicating that nuclear congression had already occurred. Initially, the donor NE was bright, and the acceptor NE was dim. As the experiment progressed, the acceptor NE gradually increased in fluorescence. The initial gradual increase in fluorescence will be referred to as the slow phase of 3XGFP-HDEL transfer. At 6 min after the initial observation, the acceptor NE abruptly increased in fluorescence. The increase in acceptor NE fluorescence was accompanied by a rapid decrease in donor NE fluorescence as the two NEs approached equilibrium. The acceptor reached 50% of the equilibrium value within ∼2.5 min after the onset of the rapid phase of transfer. Concomitant with the rapid phase of 3XGFP-HDEL transfer, the two NEs could be seen to separate at the site of fusion along a line parallel to the axis of the two nuclei, which is indicative of the widening of the nuclear fusion pore (, 11-min time point). The transfer of 3XGFP-HDEL from donor to acceptor is quantified in based on measurements described in Materials and methods.
Interestingly, in rare examples, such as that shown in , the Spc42p-RFP spot formed from the two SPBs was observed to separate into two spots after the NEs had begun to fuse (t = 8). The two Spc42p-RFP spots remained apart until the 12-min time point when they merged, after which they remained together, even as the NE widened. Thus, the behavior of the Spc42p-RFP in this zygote is consistent with ET, which showed that SPB fusion occurs after NE fusion.
To confirm that the fast phase of the NE luminal filling was caused by NE fusion, we observed 3XGFP-HDEL in matings of karyogamy mutants in which nuclear fusion is blocked. Mutation in the gene causes a block in nuclear congression as a result of defects in the function of the cytoplasmic microtubules (). Nuclear membrane fusion does occur in the rare zygotes of this genotype wherein the nuclei become closely apposed by chance (). In X wild-type matings, the two nuclei did not congress, as demonstrated by the presence of two distinct RFP-labeled SPB dots (). The acceptor NE showed a gradual increase in fluorescence intensity but never a transition to the rapid phase of 3XGFP-HDEL transfer ( and not depicted). The GFP fluorescence in the donor NE remained considerably brighter than the acceptor NE throughout the course of the experiment. Given the dependence of the rapid 3XGFP-HDEL transfer on proteins required for karyogamy and the temporal correlation of this transfer with the onset of NE expansion, we conclude that the rapid phase of transfer is indicative of the onset of NE fusion. Most likely, the rapid phase of transfer corresponds to the dilation of the membrane fusion pore. We surmise that the slow phase of 3XGFP-HDEL transfer is caused by the recycling of HDEL-bearing proteins from the Golgi to the NE/ER of both parents. Although this introduces a small amount of background fluorescence, a sharp transition between the slow and rapid phases could be readily discerned in almost all zygotes.
Transfer of lumenal 3XGFP-HDEL to the acceptor NE/ER is indicative of outer NE fusion, but it does not provide information about the last step of NE fusion, the fusion of the inner nuclear membranes. In principle, this may occur either at the same time as outer membrane fusion as part of a single concerted reaction or as a temporally distinct event. To reveal inner nuclear membrane fusion, we developed an assay based on the transfer of a nucleoplasmic marker from a donor nucleus to an acceptor nucleus. We tested several nucleoplasmic markers, including NLS-tagged fluorescent proteins, fluorescent protein–tagged histones, and fluorescent protein–tagged poly (A) polymerase. NLS-CFP and -YFP exhibited high levels of background cytoplasmic localization as well as rapid transfer into unfused nuclei, presumably because of rapid shuttling between the nucleus and cytoplasm ( and our unpublished data). Fluorescent protein–tagged histones exhibited very low cytoplasmic background and no observable transfer between unfused nuclei. However, fluorescent chromatin from the two haploid nuclei remained segregated as distinct domains in the newly formed diploid nucleus (unpublished data), which obscured attempts to determine a precise time for the completion of nuclear fusion. Ultimately, fluorescent protein–tagged poly (A) polymerase (Pap1p) was found to be suitable for our studies. Fluorescent protein–tagged Pap1p showed both a low cytoplasmic background and a uniform distribution within the nucleus, presumably because its diffusion is not limited by stable association with nuclear polymers ().
We used FRAP to determine whether there were fundamental differences in the rate of diffusion between 3XGFP-HDEL and fluorescent protein–tagged Pap1p fusions that might cause differences in their observed rates of transfer. In this technique, a small region of the cell containing the fluorescent protein was bleached, and the bleached area was monitored for recovery of the fluorescent signal. The rate of diffusion can be measured from the rate of fluorescence recovery. For these experiments, we performed FRAP analysis on 3XGFP-HDEL and Pap1p-GFP. Pap1p-GFP was used instead of the Pap1p-RFP or Pap1p-mCherry (monomeric Cherry fluorescent protein) constructs used in the time course experiments because the RFPs were difficult to bleach with available laser lines. However, GFP and RFP diffusion rates have been previously shown to be very similar (). Thus, any substantial differences in the behaviors of fluorescent protein–tagged Pap1p and 3XGFP-HDEL would be the result of differences in the Pap1p and HDEL portions of the chimeras.
= 0.16 s [ = 18] vs. 0.22 s [ = 20], respectively). Thus, any delay observed for fluorescent protein–tagged Pap1p relative to 3XGFP-HDEL would likely underestimate the true difference. The recovery of Pap1-GFP and 3XGFP-HDEL averaged 97% and 96%, respectively, taking into account the reduced pool of unbleached fluorescent protein.
After establishing that there were no substantial differences in the diffusion rates of the marker proteins, we performed time-lapse microscopy of zygotes in which both proteins were present. Cells of one mating type were labeled with 3XGFP-HDEL and mated with cells of the opposite mating type labeled with Pap1p-RFP or Pap1p-mCherry (). A total of 47 zygotes were imaged, with images taken every 1 min, 30 s, or 20 s. In each, the fluorescence intensity was measured on equivalent areas of the donor and recipient and corrected for background fluorescence using a region outside the cell. Photobleaching during the course of the experiment was corrected by fitting the measurements to a first-order exponential.
In the zygote shown in , the shift from the slow phase to the rapid phase of 3XGFP-HDEL transfer occurred at ∼5 min, reaching a half-maximal transfer within 2 min (). In contrast, the first time point in which a considerable amount of Pap1-RFP fluorescence was observed in the recipient nucleus was at 6 min. Extrapolation of the initial slope to the x axis suggested that transfer initiated at ∼5.5 min, reaching half-maximal transfer at ∼7 min. Subsequent experiments in which images were acquired every 20 or 30 s provided greater temporal resolution of the two fusion events (). Out of the 47 zygotes examined, 32 exhibited clear temporal separations between the initiation times of marker protein transfer, and five showed no measurable delay (). The remaining 10 zygotes were not interpretable because the transfer had begun before beginning the observation, transfer did not occur during the observation period, or the shift from the slow to rapid phase of 3XGFP-HDEL transfer was not sufficiently pronounced to determine the time of initiation. As expected, the number of zygotes in which a temporal delay could not be discerned was dependent on the density of time points (3/13 for 1 min, 1/10 for 20 s, and 1/14 for 30 s experiments). The median time for all 37 interpretable zygotes was 30 s (40 s for the 10 zygotes imaged at 20-s intervals). Two of the zygotes examined using 20-s time points showed unusually long delays in the time of initiation of Pap1p-mCherry transfer relative to 3XGFP-HDEL transfer (440 and 620 s). Because the more rapid time point experiments necessitated increased exposure to the excitation light, these may represent photodamaged cells. When these two cells were omitted from the analysis, the mean time delay for the 20-s interval image sets was 31.25 s and for all image sets was 29 s ( = 35; SD = 17). We conclude that inner NE fusion occurs after outer NE fusion in two distinct membrane fusion events. Taking these data together with the ET, we conclude that nuclear fusion occurs in three steps, culminating in the formation of a diploid nucleus with a single fused SPB.
t
h
i
s
s
t
u
d
y
,
b
o
t
h
E
T
a
n
d
l
i
v
e
c
e
l
l
m
i
c
r
o
s
c
o
p
y
w
e
r
e
u
s
e
d
t
o
e
x
a
m
i
n
e
t
h
e
d
e
t
a
i
l
s
o
f
n
u
c
l
e
a
r
f
u
s
i
o
n
d
u
r
i
n
g
y
e
a
s
t
c
o
n
j
u
g
a
t
i
o
n
.
E
T
r
e
v
e
a
l
e
d
s
e
v
e
r
a
l
n
u
c
l
e
a
r
m
o
r
p
h
o
l
o
g
i
e
s
c
o
n
s
i
s
t
e
n
t
w
i
t
h
n
u
c
l
e
i
p
o
i
s
e
d
a
t
v
a
r
i
o
u
s
i
n
t
e
r
m
e
d
i
a
t
e
s
t
a
g
e
s
o
f
n
u
c
l
e
a
r
f
u
s
i
o
n
,
i
n
c
l
u
d
i
n
g
o
u
t
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
,
i
n
n
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
,
S
P
B
s
a
t
e
l
l
i
t
e
/
b
r
i
d
g
e
f
u
s
i
o
n
,
a
n
d
S
P
B
f
u
s
i
o
n
.
I
n
o
n
e
e
x
a
m
p
l
e
,
o
u
t
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
w
a
s
o
b
s
e
r
v
e
d
w
i
t
h
o
u
t
i
n
n
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
,
s
u
g
g
e
s
t
i
n
g
t
h
a
t
o
u
t
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
p
r
e
c
e
d
e
s
i
n
n
e
r
n
u
c
l
e
a
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
.
I
n
a
d
d
i
t
i
o
n
,
c
o
n
s
i
s
t
e
n
t
w
i
t
h
t
h
e
h
y
p
o
t
h
e
s
i
s
t
h
a
t
n
u
c
l
e
a
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
p
r
e
c
e
d
e
s
S
P
B
f
u
s
i
o
n
,
c
o
m
p
l
e
t
e
n
u
c
l
e
a
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
w
a
s
o
b
s
e
r
v
e
d
i
n
z
y
g
o
t
e
s
t
h
a
t
s
t
i
l
l
h
a
d
d
i
s
t
i
n
c
t
S
P
B
c
e
n
t
r
a
l
p
l
a
q
u
e
s
.
A
p
p
a
r
e
n
t
i
n
t
e
r
m
e
d
i
a
t
e
s
i
n
S
P
B
f
u
s
i
o
n
w
e
r
e
s
e
e
n
i
n
z
y
g
o
t
e
s
w
i
t
h
e
x
p
a
n
d
e
d
n
u
c
l
e
a
r
f
u
s
i
o
n
p
o
r
e
s
,
i
n
d
i
c
a
t
i
n
g
t
h
a
t
S
P
B
f
u
s
i
o
n
o
c
c
u
r
s
c
o
n
s
i
d
e
r
a
b
l
y
l
a
t
e
r
t
h
a
n
n
u
c
l
e
a
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
.
B
y
l
i
v
e
c
e
l
l
m
i
c
r
o
s
c
o
p
y
,
i
n
i
t
i
a
t
i
o
n
o
f
t
h
e
t
r
a
n
s
f
e
r
o
f
a
l
u
m
i
n
a
l
N
E
m
a
r
k
e
r
w
a
s
d
e
t
e
c
t
e
d
∼
3
0
s
b
e
f
o
r
e
t
r
a
n
s
f
e
r
o
f
a
n
u
c
l
e
o
p
l
a
s
m
i
c
m
a
r
k
e
r
,
i
n
d
i
c
a
t
i
n
g
t
h
a
t
o
u
t
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
p
r
e
c
e
d
e
s
i
n
n
e
r
m
e
m
b
r
a
n
e
f
u
s
i
o
n
.
C
o
l
l
e
c
t
i
v
e
l
y
,
t
h
e
s
e
d
a
t
a
d
e
m
o
n
s
t
r
a
t
e
t
h
a
t
b
u
d
d
i
n
g
y
e
a
s
t
n
u
c
l
e
a
r
f
u
s
i
o
n
o
c
c
u
r
s
i
n
a
t
l
e
a
s
t
t
h
r
e
e
d
i
s
t
i
n
c
t
s
t
a
g
e
s
.
All strains used in this study are listed in . Yeast media was prepared, and general methodology was followed as described previously (; ). Genomic integrations of GFP, monomeric RFP (mRFP), and mCherry fluorescent tags at the C terminus of and were conducted using previously published vectors and methods (). R.Y. Tsien (University of California, San Diego, San Diego, CA) provided the mCherry and mRFP fluorescent tags, S. Clark (Princeton University, Princeton, NJ) provided mRFP and mCherry kanMX constructs, and N. Erdeniz (Princeton University) provided the SS-3XGFP-HDEL construct. The -mRFP was fully functional and supported normal rates of growth and nuclear fusion. In brief, primers were designed to amplify the integration cassette, adding ∼40 base pairs of homology to the chromosomal region of interest at either end of the fragment to facilitate integration. PCR products were pooled and gel purified. Then, the strain of interest was transformed. Transformants were checked for proper integration using a diagnostic PCR. Existing GFP integration cassettes were modified for mRFP (pMR5484) and mCherry (pMR5597) integration by the same method using the same primers. During the course of the study, RFP constructs were replaced with mCherry constructs because of the improved stability of this red fluorescent marker ().
Individual liquid cultures of MATα and MATa strains were prepared and mated together as described previously (). For ET, mating mixtures were mated on a nitrocellulose filter disk for 2.5 h at 30°C. Cells were prepared for electron microscopy as reported previously (; ). In brief, mating cells were resuspended in liquid medium, collected by centrifugation, frozen under high pressure, and subsequently freeze substituted in acetone containing 2% osmium tetroxide and 0.1% uranyl acetate at −90°C for 3 d. Cells were subsequently warmed to −20°C for 12 h, rinsed in acetone, warmed to room temperature over 2 h, and embedded in epoxy resin. Serial semithick sections (200–300 nm) were cut using a microtome (Reichert Ultracut-E; Leica), collected on Formvar-coated slot grids, and stained with aqueous uranyl acetate and Reynolds lead citrate. 15-nm colloidal gold particles (Sigma-Aldrich) were applied to both surfaces of the sections to use as fiducial markers during image alignment.
ET was conducted as described previously (). In brief, sections were placed in a high-tilt specimen holder (Gatan), and images were recorded using a Tecnai TF20 or TF30 intermediate voltage electron microscope (FEI) operated at 200 kV or 300 kV, respectively. Using a rotating sample holder, electron microscopy images were captured every 1° over a ±60° range using a CCD camera (2K by 2K; Gatan). To collect dual-axis datasets, the grid was rotated 90°, and a second tilt series was collected. Tomography reconstructions were calculated, displayed, and analyzed using the IMOD software package (). The single-axis tomograms were aligned to each other and combined to create a single tomogram (). Objects such as microtubules, outer and inner NEs, and SPBs were modeled on the tomography reconstructions using the same software. 21 tomograms and models of wild-type mating yeast cells were generated. Nuclear fusion pore dilation was measured using IMOD software. Measurements were taken between the outer envelopes at the narrowest point in the zone of fusion.
For live cell microscopy, mating mixtures were prepared as described previously (). However, instead of transferring mating mixtures to a nitrocellulose filter, cells were immediately transferred to a 2% agarose pad (in synthetic complete media) on a microscope slide and incubated at RT for 1.5 h before data collection.
FRAP analysis on 3XGFP-HDEL and Pap1-GFP was performed using a confocal system (LSM510; Carl Zeiss, Inc.) housed in the Microscopy Core Facility of Princeton University (Department of Molecular Biology). Five prebleach images were collected followed by a four to six iteration photobleach and immediate collection of 45 postbleach images (30–100 ms apart depending on the experiment). 3XGFP-HDEL was bleached using 488- and 514-nm laser lines at 95% power with the bleach set at 100%. Pap1-GFP was bleached using the 488-nm laser line at 95% power with the bleach set at 80%. The half-time of recovery was determined using a MATLAB program written by T. Gregor (Princeton University, Princeton, NJ). The mean recovery for Pap1-GFP and 3XGFP-HDEL (97% and 96%, respectively) was determined by measuring the fraction of total nuclear fluorescence in the ROI before and after the photobleach. For example, in the experiment shown in , the total nuclear fluorescence at t = 0 was ∼4,000 U, which was reduced to ∼2,200 U by bleaching. The total fluorescence of the ROI was ∼1,100 U before the bleaching, which recovered to ∼600 U. Thus, the ROI contained ∼27.5% of the total nuclear fluorescence before the bleach and ∼27.3% of the total nuclear fluorescence after recovery, indicating nearly 100% recovery.
Videos show slices through a tomogram (Video 1), a 3D model of a wild-type zygote (Video 2), and transfer of 3X-GFP-HDEL during nuclear fusion in live cells (Video 3). Table S1 lists all tomographic reconstructions ordered by stage. Online supplemental material is available at . |
Mitochondria synthesize only a small number of proteins in their matrix. 99% of the ∼1,000 different mitochondrial proteins are produced on cytosolic ribosomes and are imported into the organelle (; ; ; ). The classic pathway of protein import into mitochondria involves N-terminal presequences on the precursor proteins (). The presequences target the proteins to receptors of the translocase of outer mitochondrial membrane (TOM) complex. After translocation through the TOM channel, the preproteins are directed to the TIM23 (presequence translocase of inner mitochondrial membrane) complex. The presequence translocase-associated motor (PAM) completes preprotein translocation into the matrix. Here, the mitochondrial processing peptidase removes the presequences, and the proteins are folded to their mature forms.
However, many mitochondrial precursor proteins are not synthesized with cleavable presequences but possess internal targeting signals. Although the TOM complex functions as the central import site for most precursors, the subsequent transport of proteins to the four mitochondrial compartments is mediated by different machineries. Three main import pathways for noncleavable precursor proteins have been defined (). The precursors of outer membrane β-barrel proteins are transferred by the Tim9–Tim10 chaperone complex to the sorting and assembly machinery (SAM) complex of the outer membrane (, ; ). Multispanning proteins of the inner membrane like the metabolite carriers also use the Tim9–Tim10 chaperone complex to traverse the intermembrane space and are inserted into the inner membrane by the TIM22 (carrier translocase of inner mitochondrial membrane) complex (; ; ). Many proteins of the intermembrane space contain cysteine motifs and are imported and oxidized by the mitochondrial intermembrane space assembly system (; ; ; ).
The protein translocases in the four mitochondrial compartments do not function as independent complexes but cooperate in a dynamic manner. This includes transient contacts between translocases located in different compartments and the involvement of protein complexes that have previously been thought not to be related to protein biogenesis, such as the respiratory chain and mitochondrial morphology components.
Cleavable preproteins are guided into mitochondria by a chain of sequential binding sites for presequences, including the receptor domains of Tom20 and Tom22, the channel formed by Tom40, and the intermembrane space tail of Tom22 (; ; ; ). Tom40 is not simply a passive channel but recognizes the presequences and participates in the selection of precursors for the subsequent sorting pathways (). It has been a long-standing question of how preproteins are transferred from the TOM complex to the TIM23 complex of the inner membrane. Models ranged from a permanent TOM–TIM connection to two fully independent translocase complexes.
The identification of Tim50 and Tim21 as new subunits of the TIM23 complex revealed a dynamic mechanism of TOM–TIM cooperation (; ; ). In the absence of preproteins, Tim50 keeps the inner membrane channel formed by Tim23 in a closed state (). As soon as preproteins emerge from the Tom40 channel, Tim50 binds to them and stimulates interaction of the presequence with the intermembrane space tail of Tom22. Thus, remarkably, a Tim protein helps a preprotein in transit to make contact with the trans-side of the TOM machinery. Subsequently, Tim21 binds to Tom22, representing a direct but transient interaction between TOM and TIM during protein import (, stage 2). Presequences and Tim21 compete for binding to the intermembrane space tail of Tom22. Thereby, Tim21 induces a release of the presequence from Tom22 and promotes transfer of the preprotein to the next stage, insertion into the inner membrane (; ). Presequence binding to Tim23 completes the chain of binding sites for preproteins on the way from TOM to TIM ().
Transport of presequences through the TIM23 complex is driven by the inner membrane potential Δψ, which performs a dual role. It activates the channel protein Tim23 and drives translocation of the positively charged presequences by an electrophoretic mechanism (). Now, a decision has to be made about the further pathway of the preprotein: either lateral sorting into the inner membrane or complete transport into the matrix. Recent studies showed that two different modular forms of the presequence translocase exist (; ). The core of both forms consists of Tim50, Tim23, and Tim17, whereas the presence of additional subunits or partners depends on the import route of the preprotein in transit (). The sorting form of the TIM23 complex, which is responsible for lateral release of proteins into the inner membrane, contains Tim21 but not the import motor PAM (, stage 3a), whereas the matrix transport form of TIM23 lacks Tim21 but is associated with the multicomponent PAM machinery (, stage 3b).
Several preproteins carry a hydrophobic segment behind the presequence (, preprotein type a). This sorting signal stops translocation across the inner membrane, and the protein is released into the lipid phase by the motor-free TIM23 complex (, stages 3a and 4a; ). Surprisingly, Tim21 was found to recruit a supercomplex of the mitochondrial respiratory chain consisting of the bc complex and cytochrome oxidase (, stage 3a; ). What could be the function of a direct association between TIM23 and the respiratory chain? Lateral sorting into the inner membrane can be driven by the electrochemical potential as the only external energy source without a requirement for the ATP-dependent motor PAM. Indeed, upon lowering of the overall electrochemical potential of the inner membrane, TIM23 complexes that are in the direct vicinity of the respiratory chain are still competent in preprotein insertion, whereas other transport processes across the inner membrane are diminished (). We envisage two possibilities. The proton motive force may be higher in close proximity to a proton-pumping complex, or protons may be directly translocated to the TIM23 complex and facilitate preprotein transport.
However, the majority of presequence-carrying preproteins are completely translocated into the matrix. To perform this task, the TIM23 complex associates with PAM, which consists of several modules (, stage 3b; ; ). Mitochondrial heat-shock protein 70 (Hsp70 [mtHsp70]) is the central component of PAM. This molecular chaperone binds unfolded preproteins in an ATP- regulated manner. Four membrane-bound cochaperones, Tim44, Pam18, Pam17, and Pam16, interact with the TIM23 complex and coordinate the function of mtHsp70 directly at the TIM channel. Tim44 provides a binding site for mtHsp70, whereas the J protein Pam18 (Tim14) stimulates the ATPase activity of mtHsp70. Pam16 (Tim16) regulates the activity of Pam18, and Pam17 is required to organize the Pam18–Pam16 module (; ; ). Finally, mitochondrial GrpE (Mge1) promotes release of the nucleotides from mtHsp70, completing the motor reaction cycle. Thus, PAM is a multistep motor that involves a coordinated action of membrane-bound and soluble proteins to promote the unfolding of preproteins and drive them into the matrix (; ).
Collectively, the TIM23 complex functions at a junction of protein import. Three partner complexes interact with TIM23 in an alternating manner: the TOM complex in early transfer from outer membrane to inner membrane, the respiratory chain for promoting sorting into the inner membrane, and PAM for translocation into the matrix. Tim21 alternates between binding to TOM and the respiratory chain (), whereas Tim17 is involved in the switch between inner membrane sorting and PAM binding (). We propose that cooperation of the TIM23 complex with its partner complexes involves more than one interaction site in each case. For the TOM–TIM connection, Tim50 was found to cooperate with Tim21 (), whereas for the coupling to respiratory chain and PAM, further interacting partners have to be defined in future studies. As the TIM23–respiratory chain interaction is impaired but not blocked by the deletion of Tim21 (), the existence of at least one more interaction site is apparent.
Most mitochondrial membrane proteins with several transmembrane segments (multispanning proteins) are synthesized without cleavable presequences. Two major classes are the β-barrel proteins of the outer membrane and the metabolite carriers of the inner membrane (). To prevent aggregation of the hydrophobic precursors, chaperones operate at several stages of the biogenesis pathway. For transfer from cytosolic ribosomes to the Tom receptors, chaperones of the Hsp90 and Hsp70 classes bind to the precursors (, stage 1; ; ). The receptor Tom70 possesses a specific binding site for the chaperones, and, thus, the precursor–chaperone complex docks onto Tom70 and delivers the substrate. Tom70 oligomerizes in the presence of substrate such that several Tom70 molecules bind to one precursor polypeptide and prevent aggregation during transfer to the Tom40 channel (; ).
The intermembrane space is an aqueous compartment, and hydrophobic proteins would aggregate here. Therefore, the intermembrane space contains a soluble translocase, the Tim9–Tim10 complex, which binds to the precursors of carrier proteins as soon as part of the polypeptide chain has traversed the Tom40 channel (). Tim9–Tim10 forms a hexameric TIM chaperone complex that protects the hydrophobic segments of precursors from aggregation (; ; ). The carrier precursors do not cross the outer membrane as linear polypeptide chain–like cleavable preproteins but are translocated through Tom40 in a loop formation (, stage 2). Precursor release from TOM requires an active TIM chaperone complex, indicating a close cooperation of both translocases (; ; ). The intermembrane space contains a second TIM chaperone complex, the Tim8–Tim13 complex, which is homologous to the Tim9–Tim10 complex and interacts with a subset of hydrophobic precursor proteins (; ).
The Tim9–Tim10 chaperone delivers the carrier precursors to the TIM22 complex. This involves a rearrangement of the chaperone at the surface of the inner membrane. Tim12, a small Tim protein peripherally bound to the TIM22 complex, associates with Tim9 and Tim10 in a ternary complex, and so these small Tim proteins become membrane bound (, stage 3b; ; ). The TIM22 complex contains three integral membrane proteins: Tim54, Tim22, and Tim18. Tim22 is the channel-forming protein and mediates protein insertion into the inner membrane in a membrane potential– driven manner (). It is not known which of the three integral subunits binds the small Tim proteins. We speculate that Tim54, with its large domain in the intermembrane space, functions as a docking site for small Tim proteins at the carrier translocase.
The TIM chaperone complexes cooperate with a third membrane translocase, the SAM complex of the outer membrane (). Upon translocation via the Tom40 channel, the precursors of β-barrel proteins bind to Tim9–Tim10 or Tim8–Tim13 and are transferred to SAM (, stage 3a; ; ). Precursor insertion into the outer membrane is initiated by Sam50 (Omp85/Tob55), the central component of the SAM complex (; ; ). It is not yet known whether Sam50 provides a direct interaction site for the TIM chaperones.
In summary, the soluble TIM chaperone complexes, Tim9–Tim10 and Tim8–Tim13, provide a shuttle system between TOM and the membrane insertases TIM22 and SAM and, thus, ensure that precursors are kept in a translocation-competent conformation.
The outer membrane SAM complex contains three core components: the channel-forming protein Sam50, Sam37, and Sam35 (; ; ; ; ). Sam50 is a β-barrel protein itself. The lateral opening of a β-barrel protein is energetically unfavorable, as many hydrogen bonds would have to be broken. We envisage that the β-barrel precursors, which are delivered by the TIM chaperones, may insert between several Sam50 molecules and, thus, have access to the lipid phase (, stage 3a). The exact function of Sam35 and Sam37 is not yet known. They likely participate in the insertion and lateral release of precursor proteins.
Sam50 is homologous to Omp85/YaeT of Gram-negative bacteria, implying a conserved mechanism of β-barrel insertion in mitochondria and bacteria (; ; ; ). However, the partner proteins of Sam50 and Omp85/YaeT are not homologous to each other. In addition, as the lipid composition of bacterial and mitochondrial outer membranes differs considerably, it is likely that the mitochondrial assembly machinery was originally derived from the bacterial one but underwent substantial changes during evolution.
Further characterization of the SAM pathway revealed an unexpected connection to the machinery that maintains mitochondrial morphology. A fourth subunit found in a fraction of SAM complexes turned out to be the morphology protein Mdm10 (). Mdm10 is required to assemble β-barrel precursors, in particular the precursor of Tom40, into functional complexes. Mdm10 not only associates with the SAM complex but also with two further morphology proteins, Mdm12 and Mmm1, to form a different complex (; ). Remarkably, this mitochondrial distribution and morphology (MDM) complex is also required for the β-barrel assembly pathway of the mitochondrial outer membrane at a stage after the SAM core components (, stages 3a to 4a). The MDM complex possibly mediates the cooperation of both mitochondrial membranes because MDM proteins are enriched in punctate structures near contact sites of outer and inner membranes (; , , ). It should be emphasized that the majority of proteins that were reported to function in the maintenance of mitochondrial morphology are not involved in the assembly of β-barrel proteins (). Only a subset of morphology proteins associating with the SAM complex or the MDM complex perform a primary function in protein assembly. As their function involves the biogenesis of the TOM complex (i.e., assembly of the main entry gate of mitochondria), a defect of these morphology proteins leads to a defect of the TOM complex and, consequently, to a defect in the import of genuine morphology components.
We are just beginning to understand how the interplay between TOM, SAM, and MDM complexes is organized. Tom7, a small subunit of the TOM complex, plays a second role outside the mature TOM complex. Tom7 regulates the association of Mdm10 with the SAM complex in an antagonistic manner. Upon deletion of Tom7, the amount of Mdm10 at the SAM complex is increased, and the assembly of Tom40 is accelerated (). Thus, Tom7 has two functions. It is a subunit of the mature TOM complex and acts as a negative regulator of the assembly pathway of Tom40. We suggest that the biogenesis of outer membrane β-barrel proteins involves a dynamic cooperation of TOM, TIM chaperones, SAM, and MDM to ensure an efficient and regulated transfer of precursor proteins.
We suggest a new level of organization of the mitochondrial protein import machinery. Although the initial characterization of protein transport led to the identification of numerous components and their presence in stable translocase complexes, we have reviewed here that the translocases are highly dynamic machineries. Depending on the sorting signals present in precursor proteins, the translocases undergo modular rearrangements and transiently interact with each other. Importantly, this involves a dynamic interaction between transport complexes located in different mitochondrial compartments, such as the TOM–TIM23 connection, the TIM23–PAM interaction, and the cooperation of TIM chaperones of the intermembrane space with translocases of both outer and inner membranes. The cooperation not only involves the known translocases but also complexes that have not been related to protein import so far, such as the respiratory chain and the MDM complex.
The dynamic nature of the protein import machinery is also reflected in increasing evidence that transport components perform two or more functions or interact with alternating partners. We outlined the examples of Tom22, Tim50, Tim21, and Tim17 in the TIM23 reaction cycle, the cooperation of Tim9–Tim10 with three different translocases, and the dual role of Tom7 as TOM subunit and regulator of Mdm10. Seeing this growing list, we speculate that import components that play more than one function are much more common than anticipated. Multifunctionality of an import component may be the rule, not the exception.
The rapid increase in knowledge of the cooperation of preprotein translocases suggests that future studies will reveal more dynamic interactions between translocases, be it for preprotein transfer or for regulatory purposes. For example, the three stages defined for the TIM23 reaction cycle likely represent only snapshots that are accessible to our current experimental tools. It is conceivable that the switch between inner membrane sorting and motor binding occurs in several intermediate steps (e.g., for preproteins, which possess a sorting signal but also contain folded domains that require the unfolding power of PAM). We speculate that TOM, SAM, and MDM may be organized in transient, larger assemblies. Moreover, the inner membrane contains machineries for the export of mitochondrially encoded proteins from the matrix (; ; ). It will be interesting to see whether these export machineries cooperate with the TIM import machineries.
A cooperation of machineries and components located in different compartments of mitochondria is not only important for protein biogenesis but also for tethering mitochondria to the cytoskeleton, for fusion and fission of the mitochondrial membranes, and for apoptotic processes (; ; ; ). Thus, a characterization of the mechanisms, which coordinate and regulate the activities of both mitochondrial membranes and the two aqueous compartments, will be a major challenge toward a molecular understanding of this highly dynamic cell organelle. |
xref
italic
#text
We measured nuclear and cellular volumes in individual cells of the fission yeast using a nuclear membrane marker Cut11-GFP and Nomarski optics (). In vegetatively growing wild-type cells, the ratio of nuclear to cell volume (N/C ratio) was found to be 0.080 ± 0.013 (), in agreement with measurements from electron microscopic tomography (). As in budding yeast, there is a linear correlation between cell size and nuclear size (r = 0.68; ). The N/C ratio and the linear correlation were unchanged in two round mutants, indicating that cell volume and not cell length correlates with nuclear size ( and ; ; ). To determine if the N/C ratio remained constant in cells of different sizes, we measured nuclear and cell volumes in spores, small cells (), large cells (using the cell cycle mutants , , and a switch-off strain), and wild-type cells (; ; ; ). We found a very strong positive correlation between nuclear and cell volume (r = 0.97, = 2136, for correlation coefficients (r) of individual distributions; see ). Despite a 35-fold difference in average cell size, the N/C ratio was remarkably constant, varying only between 0.076 ± 0.013 and 0.089 ± 0.017 (). Taking the extremes of single cells, the actual span in cell size between the smallest spore and the largest cdc13 switch-off cell was 160 fold. Even cells of different sizes generated by nitrogen starvation or changes in ploidy had broadly similar N/C ratios (). To investigate if disruption of the cytoskeleton has effects on nuclear size, we treated cells with Latrunculin A to depolymerize actin patches and cables and with MCB, which depolymerizes microtubules. These treatments did not cause obvious changes in the N/C ratio (unpublished data). During the course of those experiments we discovered that in vegetatively growing cells the volume of the nucleolus remained in a similar proportion to the nucleus (0.24 ± 0.06, = 31, r = 0.77), despite the wide range of cell sizes (Fig. S1, available at ). We conclude that nuclear and nucleolar sizes are strongly coordinated with cell size.
We next tested N/C ratios in fission yeast cell cycle mutants that were arrested in G1 or G2 phase with either 1C or 2C DNA content, and in diploid cells, which have twice the DNA content of the haploid cells. Examination of the plots () demonstrated that cells of similar size but containing nuclei with different DNA contents had similar N/C ratios. More dramatically, the switch-off strain produced some cells with a single nucleus of a 2C DNA content and others with a 32C DNA content as determined by FACS and DAPI staining (; ). Despite the 16-fold difference in DNA content, the relative nuclear sizes were the same and N/C ratios were indistinguishable (). We conclude that in growing cells, the DNA content of a nucleus has no direct effect on nuclear size. Rather, ploidy determines cell size at division and in turn the sizes of the subsequent daughter cells determine the sizes of their nuclei. This result is consistent with recent work examining haploid and diploid budding yeast in both G1 and G2 phase (; ).
If overall cell size determines nuclear size, then the gradual increase in cell size during the cell cycle should be linked to a gradual increase in nuclear size. To test this prediction we monitored the N/C ratio using time-lapse microscopy of living wild-type cells proceeding through the cell cycle (Video 1, available at ). As expected, nuclear and cell volume increased concurrently, maintaining an almost constant N/C ratio throughout the cell cycle (). Fission yeast undergoes a rapid mitosis, with no breakdown of the NE. Measurements of nuclear volume and surface area just before and immediately after mitosis established that the combined volumes of the two daughter nuclei were approximately equal to the volume of the undivided parental nucleus. In contrast, the combined surface areas of the two daughter nuclei were much increased (). These results indicate that rapid expansion of the NE during mitosis is necessary to maintain the N/C ratio the same in the two daughter cells. A more flexible ruffled NE was observed in cells undergoing closed mitoses, which could contribute to the rapid expansion of the NE (unpublished data). Consistently, an expansion of the NE has also been reported before and after mitosis in budding yeast, where equal amounts of DNA are partitioned into two nuclei of different size ().
The experiments described above indicate that nuclear size is proportional to cell size. We next generated multinucleated cells using a cytokinesis mutant (; ) to test two contrasting models regarding the role of nuclear position in growth control. First, nuclear growth may depend on nuclear position and the proportional amount of cytoplasm surrounding a nucleus; second, nuclear growth may be similar throughout the cell and independent of nuclear position. After two mitoses at the restrictive temperature, cells contain four nuclei of equal size that are unevenly distributed within the cell, allowing their direct comparison in different subcellular positions. Each nucleus was attributed a proportional cytoplasmic volume, defined by the midpoint between neighboring nuclei and/or cell ends. We found that more closely spaced inner nuclei grew slower than the outer nuclei that are surrounded by a larger proportional cytoplasmic volume. During interphase, the proportional N/C ratio of single nuclei was gradually adjusted until they reached a size close to the proportional cytoplasmic volume (). We propose that the local cytoplasmic environment is an important determinant of nuclear size, and further tested this hypothesis in three experiments. To distinguish between effects linked to the proportional cytoplasmic volume and the proximity to an actively growing cell tip, we first artificially displaced nuclei by centrifugation altering the distribution of nuclei within the cell. Interestingly, the nucleus that was surrounded by a larger cytoplasmic sub-volume grew faster than the others (). Second, we confirmed the hypothesis using a branching mutant, which creates an additional growth zone at the middle of a cell (; ). In agreement with our proposal, nuclei residing in a larger cytoplasmic domain became even larger, and previously small nuclei that were close to an additional growth zone increased in size (). Lastly, we observed that in cells containing 16 or 32 nuclei, the sizes of the nuclei varied up to 10-fold depending on the amount of cytoplasm in their vicinity (). This result is also consistent with the cytoplasmic volume determining nuclear size.
We next artificially perturbed N/C ratio to investigate how rapidly nuclear growth responds to changes in the N/C ratio. Centrifugation of multinucleated cells followed by septation generates cells with increased and decreased N/C ratios (; ; Video 2, available at ). In cells that had a twofold elevated N/C ratio, nuclear growth was arrested and only resumed once a threshold of ∼1.5-fold the normal N/C ratio had been attained (). In contrast, in cells that had a reduced N/C ratio, rapid nuclear growth restored an almost normal N/C ratio, usually within less than 1 h (). Together, these experiments establish that nuclear and cell growth are not directly coupled and that nuclear size is causally dependent upon cell size. These perturbation experiments further demonstrated that during vegetative interphase growth nuclei can grow faster than the cell, but are unable to contract, even at high N/C ratios.
Nuclear growth in general can be either driven by an increase in nuclear volume or by an increase in NE surface area, or by a combination of both. Whereas growth of the NE requires the availability and the targeting of newly formed membrane components from the ER, growth driven by volume increase would involve either nucleocytoplasmic transport or diffusion of smaller molecules through nuclear pores and sequestering within the nucleus. Two sets of experiments suggest that the NE expansion is a result rather than the cause of nuclear volume increase. First, it has been shown that NE-ER over-proliferation is not sufficient to increase nuclear size, but instead leads to an accumulation of NE sheets around the nucleus (; ). Second, when blocking nuclear export of a subset of proteins for 90–150 min using leptomycin B (LMB), a specific inhibitor of the exportin crm1, nuclear size and the N/C ratio increase by 50% (; Fig. S2, available at ). This suggests that nucleocytoplasmic transport directly or indirectly alters nuclear size control, and contrasts with data from budding yeast, where 5–30 min of treatment with LMB had shown no obvious effect on nuclear size (). The differences in the results may be due to the more extended time course of drug treatment in our experiments. We further tested if the distribution of nuclear pores influences the N/C ratio. Cells deleted for and marked with the nucleoporin Nup107-GFP have less evenly distributed nuclear pore complexes (), but the N/C ratio is not affected (unpublished data).
It is possible that nuclear volume could be controlled by some surrogate, such as amount of RNA or protein, numbers of ribosomes, or membrane content. In motoneurons and hepatocytes, cell size and nuclear size both correlate with the cellular RNA/DNA ratio, the expression of ribosomal genes, and general transcription rate (; ). Future studies will be required to dissect the molecular basis of nuclear size control in fission yeast.
A similar general cellular control that regulates nuclear growth in response to the amount of cytoplasm surrounding the nucleus may influence nuclear growth in other eukaryotes. However, differences in the cellular differentiation state and organismal developmental stage or the presence of a nuclear lamina, add more layers to N/C ratio control. Although we have shown that DNA content does not directly influence nuclear size, it might set a minimum to the size of the nucleus as suggested by the nucleoskeletal theory (; ), especially in small cells such as spores. For example, whereas wild-type spores have an N/C ratio of 0.076 ± 0.016 (see ), spores have a 20% smaller cell size but only 8% smaller nuclei, indicating that a minimal nuclear size may have been reached (, = 136, N/C = 0.089 ± 0.017).
Nuclear size regulation could be influenced by several cellular functions such as nucleocytoplasmic transport, lipid metabolism, or ribosome biogenesis. It is important that the biochemical mechanisms underlying NE growth take into account the global cellular control we have described here, which so precisely relates growth in nuclear volume to growth of the cell. Because membrane-bound organelles are an essential part of the function and architecture of the eukaryotic cell, understanding how nuclear growth is regulated is likely to be informative about how the growth of other membranous structures within the cell are coordinated with changes in cell growth and differentiation.
strains are described in . Strains were generated by genetic crosses and tested by segregation of markers or PCR.
Standard media and methods were used (; ). Unless indicated differently, all yeast strains were grown in YE4S medium at 25°C. Temperature-sensitive strains were shifted from 25 to 36.5°C for the time indicated. For nitrogen starvation, exponentially growing cells were washed and resuspended in EMM-N and grown for 12 h before microscopy. For sporulation, strains were pregrown in YE4S. The mixed cells were plated on SPA plates and incubated for 2 d at 25°C before microscopy.
strains were grown in YE4S to OD 0.2–0.4 and mounted on agarose pads (1.4% agarose in YE4S). For time-lapse microscopy, coverslips were sealed with VALAP (Vaseline, Lanolin, Paraffin; 1:1:1). Population based microscopy was performed at 23–25°C on a microscope (Axioplan 2; Carl Zeiss, Inc.) equipped with a CoolsnapHQ camera (Roper Scientific). Data were acquired using the 100× PlanFluar NA 1.45 objective taking 12 z-sections with 0.3-μm spacing. Time-lapse microscopy was performed on a microscope (Axiovert 200; Carl Zeiss, Inc.) with a spinning-disk confocal head (UltraView; Perkin-Elmer), a cooled CCD camera (Orca ER; Hamamatsu), and the 63× PlanApo NA 1.4 objective at 32°C (wild type) or 36°C ().
Images were acquired in MetaMorph (MDS Analytical Technologies) and analyzed in ImageJ (W. Rasband; National Institutes of Health, Bethesda, MD). Projections of the fluorescence channel were combined with DIC images. Cells were measured by hand assuming simple geometries, and volumes were calculated based on axial symmetries (cell: rod; nucleus: prolate ellipsoid). Statistical analyses were performed in Excel and KaleidaGraph. Box-and-Whisker plots represent the distribution, the boxes delimiting the median, the first and third quartiles, and whiskers marking 5th and 95th percentiles. The Pearson product moment correlation coefficient (r) was used to describe linear correlations.
Protocol was modified from . PN10417 was grown on EMM-4S plates for 2 d and inoculated into EMM-N at 2 × 10 cells/ml and starved for 12 h at 25°C. Cells were then collected by centrifugation and resuspended in YE4S+Thiamine at a concentration of 10 cells/ml and were grown at 32°C. FACS samples were collected every 2 h, and aliquots mounted on agarose pads for live microscopy or fixed for quantitative DAPI staining and morphometric analysis.
15 ml of cells (10-h time point) were fixed for 20 min at 32°C with 3% formaldehyde, washed 2× with PEM and resuspended in PEMS. Cells were permeabilized by 1-h digestion with Zymolyase (0.5 mg/ml 100T in PEMS) washed for 10 min with PEMS containing 0.5% Triton X-100 at 4°C. Cells were washed with PEM and PBS and stained with 4,6-diamidino- 2-phenylindole in PBS. For microscopy cells were washed with PBS, air-dried on a glass slide, and mounted in 90% glycerol containing 1 μg/ml phenylenediamine.
The method was adapted from . Nuclei of arrested cells were displaced by 4-min centrifugation at 16,000 in a microcentrifuge (5415 D; Eppendorf).
Figure S1 shows the relation between volumes of the nucleolus, nucleus, and cell for growing cells (wild-type and ). Figure S2 shows that treatment with LMB for 90 or 150 min increases proportional nuclear size in wild-type and cells. Video 1: time-lapse analysis of growing fission yeast illustrates that the N/C ratio is constant throughout the cell cycle. Time points of a selected field are shown in A. Video 2 shows how the cell responds to artificial changes of the N/C ratio for small cells containing two nuclei. Selected time points are shown in B. Online supplemental material is available at . |
Although endocytosis represents the prevailing route for the internalization of proteins into cells (; ; ), additional pathways have been suggested that only partially use or might even bypass the endocytic machinery. These pathways apparently facilitate the internalization of polybasic peptides such as magainin-2 and buforin (, ; ), the HIV-1–based Tat protein, and the third helix of antennapedia from (penetratin), which mediates the transmigration of antennapedia through the plasma membrane and nuclear pore complex (; ). Although the mechanisms supporting cellular incorporation might vary among the different types of cell- penetrating peptides, membrane lipids could participate in several of the proposed pathways (). Lipids are indeed capable of directly supporting the membrane passage of proteins as shown for the release of (apo)cytochrome from mitochondria (; ).
Distinctive properties of the lipid architecture of the cell membrane include lateral microdomains enriched in cholesterol and (glyco)sphingolipids, which constitute a platform for signal transmission and the internalization of microorganisms (rafts and caveolae; ; ). Another key feature of the plasma membrane is represented by the transverse asymmetry of lipids. Under specific biological conditions, this asymmetry is attenuated, as exemplified by the surface exposure of phosphatidylserine (PS), which provides a recognition signal for the phagocytosis of apoptotic cells () and establishes a catalytic surface for proteases implicated in blood coagulation (). Under the same conditions, the exposure of phosphatidylethanolamine (PE) is enhanced (; ). However, in contrast to the well-known functions of PS exposure, the functional meaning of PE externalization is largely unknown.
We found that the cellular internalization of the serpin protein C inhibitor (PCI) is crucially supported by plasma membrane PE, which enables its rapid targeting to the nucleus both in vitro and in vivo. Our findings position PCI as an eminent candidate for the nuclear supply of cargo. On the basis of the crystal structure of PCI, a hydrophobic cavity has been characterized as the binding site for PE, which is recognized by specific lipids of conical morphology. Our findings indicate cell surface PE as a mediator for the cell membrane translocation of proteins and suggest that this requires the ability of the lipid to foster formation of transient nonbilayer domains within the membrane.
The lipid structure of the plasma membrane can be dynamically regulated by the selective insertion of extracellular lipids via specialized proteins such as scavenger receptor class B type I, which transfers cholesterol and phospholipids into cells (; ), CD14, a transporter for phosphatidylinositol (), and the fatty acid carrier CD36 (). These also include unknown proteins mediating the cellular import of PE (). To identify the latter proteins, we separated the supernatants from activated blood platelets by gel filtration. Then the phospholipid transfer activities of the fractions were measured by determining the exchange of fluorescent-labeled phospholipids between donor and acceptor vesicles. Among the fractions analyzed, fraction 23 caused the strongest stimulation of the intervesicular transfer of fluorescent-labeled PE (). Because activated platelets secrete the PE- binding serpin PCI (; ), we analyzed whether fraction 23 contained this protein. PCI could indeed be detected in the active fraction (). The serpin was additionally recovered in the total supernatant (). To evaluate whether PCI was principally capable of mediating the PE transfer through aqueous media, the serpin was added to suspensions of labeled donor vesicles and unlabeled acceptor vesicles. Isolated PCI was found to transfer fluorescent-labeled PE to the acceptor vesicles within short time periods (), which documents that the serpin supports the intermembrane exchange of PE. Moreover, PCI markedly augmented the incorporation of extracellular fluorescent-labeled PE into platelets () and several other cells (human monocytes, neutrophils, and umbilical vein endothelial cells; not depicted). However, PCI failed to enhance the incorporation of fluorescent-labeled phosphatidylcholine (PC), sphingomyelin (, SM), and PS (not depicted). The enhanced PE uptake by the activated platelets was almost completely abolished by the anti-PCI antibody (Fig. S1, available at ). Next, resting platelets were incubated with small unilamellar vesicles (SUV) containing 5% of total lipids on a molecular basis (mol%) of PE (reflecting the percentage of PE in human lipoproteins) and, additionally, [C]PE + [H]PC. In the absence of PCI, the [C]PE/[H]PC ratio remained unchanged compared with the ratio of the vesicles alone. In contrast, PCI augmented the [C]PE/[H]PC ratio, indicating that the serpin enhances [C]PE transfer into platelets (). Hence, PCI promotes phospholipid import at physiological concentrations of extracellular PE. Insertion of PE into the platelet cell membrane could influence the activity of the prothrombinase complex, which is known to be stimulated by cell surface PE (). Platelets were challenged with thrombin, a relatively weak stimulator of the prothrombinase activity, which nevertheless induces a robust secretion of PCI. Thrombin formation by the prothrombinase complex was profoundly enhanced when the PCI-mediated PE transfer into the platelets was operating (). In contrast, PCI failed to increase the thrombin generation when the lipid donors were devoid of PE. Control experiments verified that no prothrombinase activity was elicited by the PE vesicles in the absence of the platelets (unpublished data). The anti-PCI antibody abrogated the thrombin formation, which indicates that the secreted PCI is a responsible factor (). Additional PCI further promoted the thrombin formation in the presence of the PE-containing donors but not with pure PC vesicles (). Our findings characterize PCI as a selective lipid transporter for PE that supports the cellular import of this phospholipid.
When visualizing the interaction of biotin-PCI with live HL-60 cells, we found that several cells rapidly incorporated the protein. After 10 min, PCI accumulated in the submembraneous region (). Heparin cofactor II (HC-II), a serpin with sequence similarity to PCI, was not incorporated (see ; not depicted). After longer incubation times, PCI was increasingly targeted to the nucleus. We indeed observed accumulation of PCI in the perinuclear space and intranuclear compartments by immuno EM (). Time-dependent accumulation of PCI was also seen in isolated cytosolic and nuclear fractions (Fig. S2, available at ). The nuclear targeting of PCI might be facilitated by its H helix, because GFP-linked variants of PCI lacking the basic amino acids of this helix failed to enter the nucleus (unpublished data). To explore whether the fast cell entry and nuclear targeting of the serpin was also operating in the intact organism, we injected biotin-labeled PCI into the tail vein of mice. The protein was allowed to circulate for 30 min in the vascular system. Thereafter, the total pool of leukocytes was prepared from the murine blood and cells were visualized by confocal microscopy. Thereby substantial costaining of PCI with the nuclear marker was noted in the granulocyte fractions (), which indicates that the protein had been integrated into the nuclei. We thus conclude that the cell entrance of PCI enables its rapid targeting to intranuclear compartments both in vivo and in vitro.
To quantify the cellular internalization of PCI, neutrophils were first exposed to [I]PCI followed by the addition of a ligand dissociation buffer to exhaustively remove any proteins bound to the extracellular surface (). Within the first 2 min of incubation at 37°C, a large proportion of PCI accumulated within the cells (). The incorporation then proceeded at a slower rate. The quantity of PCI internalized at 6°C amounted to 2/3 of the uptake at 37°C (). In contrast to the PCI internalization, the PE transferase activity of PCI was clearly diminished at low temperatures. Indeed, PCI failed to increase the [C]PE/[H]PC ratio at 6°C (). To further address the participation of endocytic routes, CHO cells were transfected with the K44A mutant of dynamin2 (), which prevents the budding of vesicles necessary for internalization of cargo via several major endocytic routes. Intrusion of Alexa 488–PCI into K44A-transfected cells was only slightly lowered compared with the uptake observed in the wild-type cells (). Also, transfection with the AP180 variant AP180-C selectively repressing the clathrin-mediated pathway barely reduced PCI incorporation (unpublished data). A major proportion of PCI incorporation thus persists under conditions suppressing the classic endocytic pathways. Inclusion of heparin into the cell suspensions to prevent interaction of PCI with the cell surface () substantially lowered the cell entry of PCI (). In contrast, the internalization was largely unaffected by receptor-associated protein, which blocks the function of the low-density lipoprotein receptor–related protein (LRP/α macroglobulin receptor), a mediator of the uptake of Tat (). Heparin also abrogated the PCI-mediated cellular import of [C]PE, as is evident from its ability to prevent the PCI-elicited increase of the [C]PE/[H]PC ratio ().
To evaluate the contribution of PE for the cell entrance of PCI, the cells were preincubated with duramycin, a cyclic peptide that specifically interacts with this phospholipid (; ). Duramycin dose dependently suppressed incorporation of PCI at 37°C (). The peptide also diminished cell entry of PCI at 6°C (Fig. S3 A, available at ), suggesting that the pathway supporting PCI internalization was still functioning at this temperature. Then, before the addition of duramycin, cell surface proteins were excessively cleaved by trypsin/proteinase K. Even so, the capacity of the PE-binding peptide to abrogate the PCI incorporation was maintained (Fig. S3 B). Although PE is a regular component of the external leaflet of the cell membrane (), its cell surface content increases under conditions of PS exposure. PCI can interact with both aminophospholipids (). To specify the interaction of PCI with PE in the presence of PS, we used fluorescence correlation spectroscopy to calculate the diffusion coefficients of PCI bound to lipid vesicles and for free PCI (). The ratio of bound/free PCI amounted to 2.9 ± 0.2 in the case of the PE/PC/PS vesicles, whereas it was 1.5 ± 0.1 for the PC/PS vesicles (mean ± SD), indicating that the preferential interaction of PCI with PE persists in the presence of PS. To further evaluate the participation of PE for the internalization of PCI, we added lipid vesicles enriched with high concentrations of PE. This decreased the PCI internalization to an appreciable extent (). Cell entry of the serpin remained unchanged, with vesicles containing exclusively PC. In the presence of extracellular PE, heparin was unable to further diminish the protein incorporation (). The latter finding establishes that heparin and extracellular PE abrogate the same route of PCI internalization. PCI thus elementarily necessitates PE for its passage through the cell membrane.
PCI is expressed in several body cavities () and hence macrophages poised to clear invading microorganisms from these compartments are continuously exposed to the serpin. Resting macrophages exhibit increased aminophospholipid exposure, which is enhanced during phagocytosis (). We found that the rapid internalization of PCI by THP-1 macrophages (unpublished data) was associated with an increased phagocytosis of (). The effect was preserved when the phagocytes were first allowed to incorporate PCI and then exposed to the bacteria. In contrast, preincubation of the bacteria with PCI before the addition of THP-1 cells reduced phagocytosis. PCI stimulated phagocytosis of in cells in which the incorporation of the serpin was restricted to extranuclear regions of the cells, suggesting that the nuclear localization of PCI is not required for its ability to increase phagocytosis. Quantification by flow cytometry confirmed that PCI fosters the cell entry of bacteria (). Stimulation of phagocytosis was almost identical when PCI was removed after preincubation, whereas preincubation of PCI with bacteria suppressed their uptake (). This shows that stimulation of phagocytosis by PCI requires a direct interaction of the serpin with the cell membrane. Pretreatment of the macrophages with PCI in the presence of duramycin, which also inhibited PCI internalization into THP-1 cells (not depicted), decreased cell entry of (). In contrast, basal uptake of bacteria (absence of PCI) was unaffected by duramycin (). Hence, the specific membrane partitioning of PCI facilitates bacterial phagocytosis.
Because several of the translocating peptides were previously shown to generate transmembrane pores (magainin-2 and mastoparan X; ; ), we investigated whether PCI elicited pore formation. As a positive control, the peptide mastoparan X rapidly evoked nucleation of pores in large unilamellar vesicles (LUV; ), as was evident from the leakage of intravesicular carboxyfluorescein (CF). In contrast, PCI did not elicit the formation of leakage sites for CF (), even after prolonged incubation (up to 120 min; not depicted). To evaluate whether PCI altered the membrane barrier properties for a small hydrophilic molecule, we monitored the membrane passage of the cation praseodymium (Pr; ). By using P nuclear magnetic resonance (NMR), two peaks originating from the phosphate groups of the outer and inner monolayer phospholipids could be discerned (). In LUV containing solely PC, the peak pattern was unchanged by prolonged incubation with the cation. However, the addition of PCI markedly reduced the height and broadened the peak representing the inner leaflet phospholipids when long chain PE was present (). This demonstrates a direct interaction of Pr with the inner monolayer phospholipids. As a negative control, no changes in the inner monolayer peak were seen without PCI. PCI thus generates transient and subtle changes in the bilayer structure that result in the formation of an aqueous pathway within the membrane.
To evaluate whether PCI itself is inserted into the membrane, we added trypsin into the vesicle core. Whereas the serpin remained largely uncleaved by the intravesicular protease in pure PC vesicles, it was clearly degraded when the membrane contained long chain (unsaturated) PE (). In contrast to PCI, the integrity of HC-II was fully conserved in the presence of the intravesicular trypsin, irrespective of the lipid composition of the vesicles (). In a complementary procedure, PCI was first incubated with the lipid vesicles and subsequently treated with extravesicular trypsin. In the presence of PE, part of the PCI remained refractory toward cleavage by the extravesicular trypsin (). In contrast, PCI was completely degraded in suspensions of the pure PC and PC/PS vesicles. The ability of PCI to penetrate into the bilayer thus crucially requires PE.
Partitioning of PCI into the lipid bilayer could be facilitated by a specific binding to PE (e.g., via interactions with the ethanolamine head group) and/or changes in the physicochemical properties of the membrane. In the latter case, nonbilayer domains, which are particularly favored by the conical shape of PE, could be involved (; ). When the LUV were prepared from a short chain species of PE with a cylindrical morphology (di-14:0-PE), the protein was entirely degraded by extravesicular trypsin, similar to the findings obtained with the pure PC vesicles (). However, in LUV consisting of a lipid with a conical configuration but devoid of the ethanolamine head group DAG, PCI was partially resistant toward cleavage by the protease (). This suggests that the conical shape of long chain PE is essential for the membrane penetration of PCI. To further substantiate the extent of membrane insertion of the serpin, we exploited the ability of proteinase K to excessively degrade the serpin (). When the protease was integrated into the aqueous core of vesicles containing long chain PE, PCI was cleaved into several high molecular mass degradation products (), confirming that PCI protrudes into the vesicle core. In LUV supplemented with DAG, very similar fragmentation products were apparent. Conversely, after inclusion of the protease into vesicles only consisting of PC, the integrity of the protein was fully conserved (). This confirms that the serpin failed to reach the vesicle core in the presence of PC alone. Overall, our findings suggest that the conical shape of long chain PE is essential for the membrane penetration of PCI, which adopts a location that completely spans the bilayer.
Molecular docking studies based on the crystallographic structure of PCI () were undertaken to characterize the structural components of PCI enabling its distinctive interaction with phospholipids. In view of the flexibility of the entire PE molecule, we chose a rigid fragment of the saturated hydrocarbon chain (C11 fragment) for the initial docking analysis because saturated fatty acyl chains are esterified to the C1 atom of the glycerol backbone of (diacyl-)PE under biological conditions (Fig. S4, available at ). Docking of the C11 fragment onto the PCI structure revealed 276 clusters with energies (kJ/mol) better than −15 among a total of 25,242,070 calculations. The two best clusters were nearly identical in position and energy (both better than −28 kJ/mol) and were significantly separated from the other clusters, with the next best hit at −24.3 kJ/mol. The best-docked C11 fragment fit into a capped hydrophobic pocket along helix D (D′ channel; ). The orientation of the saturated chain was thus fixed. Consequently, the possible positions of the unsaturated chain, canonically esterified to the C2 atom of PE, were spatially limited to the same region. Of the top 10 clusters, two provided a portion of the possible binding site for the unsaturated hydrocarbon chain (9th and 10th best energies). It was found to be best accommodated by a second hydrophobic channel running along helix H (H′ channel), in close proximity to the PCI-binding site for heparin ( and S4, helix H). Moreover, in view of the locations revealed for the acyl chains, the position of the head group of PE was also suggested. The entire PE molecule was built on the two hydrocarbon chains placed by the docking studies and subjected to energy minimization. In the final model (), the ethanolamine group is found to be located in close proximity to Leu78, whereby the formation of an H bond between the N atom of the ethanolamine group and the main chain oxygen of Leu78 could be enabled. Space limitations within this region principally impair the interaction of PCI with phospholipids containing larger head groups than PE such as PC.
To explore the role of the hydrophobic pocket for the PE-dependent membrane insertion of the serpin, PCI residues potentially implicated in the lipid binding were mutated. L78 was substituted by Trp to prevent the proposed interaction of the protein with the ethanolamine head group. Consequently, the variant remained completely undigested in the presence of the intravesicular proteinase K, oppositely to the native protein (). Substitution of V374, a component of the H′ channel suggested to mediate the binding of the sn-2 fatty acyl chain of PE, by Trp also conferred resistance toward cleavage by the intravesicular protease (). Both amino acid exchanges thus suppress the ability of PCI to partition into the vesicle core. In contrast, mutations of residues R35 and T341, both being unrelated to the proposed lipid binding site, did not perturb the PCI penetration into the vesicle center (unpublished data). A qualitatively different digestion pattern was noted when three Ala were inserted at position 376 (). This insertion was implemented to restitute the loop between the last two β strands specifically truncated in PCI while attenuating its hydrophilic properties. Part of the PCI species was accessible by the intravesicular enzyme, as was evident by the appearance of digestion products (), arguing for its protrusion into the vesicle core. To further detail the extent of membrane penetration, the variant was exposed to extravesicular proteinase K. Thereby, the mutant remained almost entirely refractory toward proteolytic attack (). In contrast, under the same conditions, the complete loss of the substrate and limited cleavage into a single high molecular mass degradation product was noted for the native protein (). The differential digestion pattern for the wild-type PCI in the presence of the intra- versus extravesicular protease () excluded a passage of the enzyme across the bilayer. The findings thus suggest that the 376AAA insertion strongly facilitates the membrane translocation of PCI. This conclusion was enforced by experiments with intact cells. After a 10-min incubation period, the labeled 376AAA variant was found to be rapidly internalized by HL-60 cells (). In particular, during the short exposure, a noticeable proportion of the mutant was incorporated into the nucleus (), in contrast to what had been observed with the native serpin (). Overall, the findings establish that the branched hydrophobic cavity in between helices H and D is of basic importance for the membrane insertion and transmigration of PCI.
We demonstrate that the internalization of a new cell-penetrating protein, the serpin PCI, is essentially supported by the plasma membrane lipid PE. Lipids are known to contribute to the interaction and insertion of proteins into intracellular and bacterial membranes (; ). However, their contribution to the protein insertion into the plasma membrane and consequently the protein internalization into cells is less well documented. Our findings now reveal that cell surface–exposed PE is a major mediator of the protein translocation across the plasma membrane. The essential contribution of PE is corroborated by the ability of the PE-binding peptide duramycin to abrogate protein uptake, the failure of the serpin to cross the cell membrane in the presence of extracellular PE, and the selective PE requirement for membrane penetration of PCI. Interaction of PCI with PE is preceded by docking of the serpin to negatively charged surface molecules such as glycosaminoglycans. This suggests that two basic steps, initial cell surface docking and subsequent interactions with PE, mediate the membrane insertion and cell entry of the protein. Importantly, we show that the distinct conical shape of PE (; ) represents a major driving force for the membrane translocation of PCI.
PE is the eminent plasma membrane lipid with a cone-type morphology because the concentrations of similarly shaped lipids (DAG and phosphatidic acid) are considerably lower. Mechanistically, DAG also efficiently supported the membrane penetration of PCI, whereas the protein intrusion was prevented by cylindrical PE species. This suggests that the conical shape of PE is of greater relevance for the interaction with PCI than specific interactions with the ethanolamine head group. Despite the bilayer-spanning location of PCI, the protein was not entirely transferred to the vesicle core, suggesting that its incorporation into the cell interior requires additional membrane components. Moreover, although transmembrane pores were not implicated in the PCI translocation, which is consistent with the inability of the conical lipids to support the pore formation (), the serpin was found to elicit subtle perturbations of the membrane organization, as indicated by the leakage of ions from the lipid vesicles. This indicates that the PCI–PE interaction induces the formation of aqueous compartments within the membrane. Lipids of cone-type morphology are principally capable of generating areas of negative curvature in membranes (). This can facilitate the formation of intramembrane nonbilayer domains (inverted micelles), which might assist in the formation of aqueous compartments within the membrane. At present, it is unclear how the PCI–PE interaction mediated by the docking of the lipid to a distinct hydrophobic groove in PCI in the molecular system (see last paragraph of Discussion) relates to the alterations in physicochemical properties by the cone-shaped lipids that allow PCI transfer across the membrane. Hence, more in-depth analyses will be required in future studies to clarify whether these two observations are indeed interrelated. Overall, our results indicate that the membrane passage of PCI is crucially driven by lipids of a distinct morphology that generate an aqueous translocation route for the protein via the generation of nonbilayer domains.
The lipid-dependent internalization of cell-penetrating PCI principally agrees with a model theoretically proposed for the internalization of the translocating peptides (). However, a contribution of PE and its peculiar shape for the cellular uptake of proteins has never been revealed. Cone-shaped lipids faciliate membrane fusion and budding processes (), which are required for the formation of phagosomes. Consequently, the enhancement of bacterial phagocytosis induced by the PCI–PE interaction as observed here could be enabled by the stimulation of phagosome maturation. The ability of PCI to foster removal of pathogens suggests its participation in host defense and thus overlaps with the functions of other cell-penetrating proteins implicated in innate immunity. The internalized PCI was found to be rapidly directed to the nucleus, both in vitro and in the intact organism. Our findings position PCI as a promising candidate for the rapid delivery of cargo to the nucleus, including viruses, nucleotides of differing length and structures, and proteins/peptides. Being a component of the human proteome, potential applications of cargo-loaded PCI in humans are not immunogenic. PCI might be exploited in particular for the targeting of cargo into cells with increased surface exposure of PE (), including capacitated sperm () and others.
PCI is shown to selectively extract PE from lipid particles, consecutively inserting the lipid into the plasma membrane of various cells. PCI thus belongs to a new class of proteins promoting the cellular uptake of selective lipids, including those specific for phosphatidylinositol () and sphingoymyelin (). The latter proteins do not share structural homologies with PCI, a situation known from the intracellular lipid transfer proteins (). During the transfer, the phospholipid would have to be largely protected from the contact with water, as revealed for scavenger receptor class B type I, the prototype mediator of the cellular phospholipid import (). Based on the velocity, extent, and temperature dependence of the PE import, we assume that a principally similar process allows PCI to ferry PE into the cells. Only a limited change in the conformation of the protein–lipid complex might be required to switch between the extra- and intramembrane modes of the PCI–PE interaction. Indeed, the protein is most likely integrated into aqueous compartments during the membrane translocation. Nonetheless, at low temperatures, the lipid transferase activity of PCI is suppressed, whereas the PE-dependent PCI internalization is largely maintained. This indicates that the PE transfer is energy independent, whereas the membrane translocation of PCI exhibits a limited energy requirement, a property common to several cell-penetrating proteins. Whereas PCI facilitates PE enrichment on the cell surface by virtue of its lipid transferase activity, PE accumulation is further assisted by the delayed transbilayer movement of the lipid (). We observe that PE insertion accelerates the activity of the prothrombinase complex, thereby amplifying the formation of thrombin, a major coordinator of blood coagulation. Once internalized, PCI could potentially also support the intracellular movement of PE, such as between the cell membrane and membranes of intracellular organelles. Overall, our findings characterize PCI as the first transferase specifically fostering the cell import of PE.
Using the crystal structure of PCI as a template, we revealed that the selective interaction of PCI with PE is facilitated by a dual-channelled hydrophobic groove that is generated by the shortened A helix and a truncated loop between strands 3 and 4B. The H′ and D′ channels arising therefrom accommodate the extensions of the fatty acyl chains of the phospholipid with sharp precision. Indeed, because the H′ channel is not capped, it can house hydrocarbon chains of up to 22 C atoms, the longest unsaturated fatty acids esterified to PE in mammals. Moreover, the restricted size of the structure in which the phospholipid headgroup is buried markedly favors interactions with conical phospholipids. Hydrogen bonds between the ethanolamine group and specific amino acids, suggested to permit the folding of bacterial lactose permease (), might additionally ease the interaction with the phospholipid, as exemplified by the functional contribution of Leu78. Together, this structural constellation is supposed to enable the higher affinity of PCI for PE over PC. The docking and energy minimization studies were corroborated by the mutagenesis of single amino acids designed to specifically remodel the hydrophobic cavity. In particular, the insertion of a triple Ala sequence on top of the H′ channel substantially promoted the intrusion and membrane translocation of the serpin. The insertion, which reconstitutes the truncated loop intercalated between strands 3 and 4B, is likely to strengthen the interaction with the fatty acyl chains of the phospholipids by virtue of its specific location and enhanced hydrophobicity. In conclusion, the hydrophobic pocket appears to represent a unique structure for accommodating lipids capable of supporting the membrane passage of proteins.
Human platelets were isolated from citrated blood as described previously (). For isolation of neutrophils, the buffy coats from human blood anticoagulated with citrate were incubated with microbeads coupled to anti-CD15 antibodies for 15 min at 8°C. The suspensions were applied onto the positive selection column (Miltenyi Biotech) and the neutrophils were eluted with buffer. The purity of the neutrophil suspensions was >95% and the viability of the cells was 98% (trypan blue exclusion). CHO cells were transfected with pcDNA3 wild-type dynamin2 and its dominant-negative K44A mutant (provided by S. Egan, The Hospital for Sick Children, Toronto, Canada). Cells were cultured in Iscove's modified Dulbecco's medium containing 20% FBS (HL-60) or DME supplemented with 10% FCS (CHO).
For preparation of recombinant PCI, the cDNA of human PCI was amplified from a human liver cDNA library using a forward (5′-CCGCTCGAGCATATGCACCGCCACCACCCCCGGGAG-3′) and reversal primer (5′-CGGGATCCATCATCAGGGGCGGTTCACTTTGCCAAG-3′). Mutagenesis was performed using the QuickChange site-directed mutagenesis kit (Stratagene). The following mutants were generated: V374W, T341R, L78W, R35A, and 376AAA. All mutations were confirmed by DNA sequencing. After transformation of with the plasmids, 0.4 mM IPTG was added to induce the expression of the proteins. Bacterial supernatants were applied to a TALON Metal Affinity Resin column (Clontech Laboratories, Inc.) and the fractions eluted from the column were analyzed by SDS-PAGE. Urinary PCI (uPCI) was purified as described previously ().
For the determination of the intervesicular phospholipid exchange, py-labeled donor vesicles and unlabeled acceptor vesicles were coincubated for different time periods, the suspensions were passed over an anion exchange column (Bio-Rad Laboratories), and the fluorescence was determined in Triton X-100 (Sigma-Aldrich) solubilisates of the acceptor vesicles as described previously (). The vesicle preparations and incubations were performed in the presence of 10 nM butylated hydroxytoluene and under argon or nitrogen. The transfer of py-labeled phospholipids into the cells was analyzed as described previously (). Under principally similar conditions, the cells were incubated with lipid donor vesicles containing traces of [C]PE (1-palmitoyl, 2-linoleoyl species) and [H]cholesterol (both from GE Healthcare).
For analysis of phagocytosis of into THP-1 macrophages, the cells (10/well) were seeded in 24-well plates in 500 μl RPMI-1640 supplemented with 10% heat-inactivated FCS, 1% Hepes, 500 U/ml penicillin, and 500 μg/ml streptomycin and differentiated by the addition of 10 nM PMA. After 16 h, the cells were fed with fresh, PMA-free medium and cultivated for two more days. The macrophages were incubated with Alexa 488–labeled uPCI or FITC alone or together with their nonfluorescent counterparts in serum-free medium. Some samples were treated with duramycin (Sigma-Aldrich) for 15 min before the start of the incubations. The samples were visualized with a confocal laser scanning microscope (LSM 510 META). The images were taken by using a 63× Plan-Apochromat oil objective (Carl Zeiss, Inc.). Alexa 488 and FITC were irradiated at 488 nm and detected via a 505 longpass filter (Carl Zeiss, Inc.). The images obtained were analyzed and quantified using LSM Image Browser software (Carl Zeiss, Inc.).
HL-60 cells previously incubated with or without 90 nM of biotin-labeled PCI were washed and fixed with 4% paraformaldehyde/5% glutaraldehyde for 60 min at room temperature. Before embedding in LR white medium resin (Pelco), the HL-60 cells were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer and then rinsed in 0.1 M phosphate buffer. Ultrathin serial sections of 150 nm were cut and mounted on gold grids. Nonspecific binding sites on the sections were blocked with 5% BSA in PBS, and grids were incubated for 120 min at 37°C with 10 nm of gold-labeled goat anti-biotin IgG (Biocell Laboratories, Inc.). The sections were then counterstained with 3% uranyl acetate and lead citrate and examined with a transmission electron microscope (JEM TEM 1200 EXII; JEOL Ltd.).
4 μg of biotin-labeled recombinant PCI (rPCI) was injected into the tail vein of anaesthetized mice. After 30 min, blood was drawn from the periorbital sinus into 0.38% sodium citrate, immediately diluted by 1:10 with erythrocyte lysis buffer (155 mM NHCl, 10 mM KHCO, and 0.1 mM EDTA, pH 7.4), and centrifuged at 400 for 20 min. The pellets were washed three times with decreasing volumes (1.5, 0.5, and 0.1 ml) of erythrocyte lysis buffer. The leukocytes obtained were processed for confocal microscopy as described previously.
THP-1 cells were differentiated with PMA as described in Confocal microscopy. Subsequently, the cells were washed in serum-free medium, mixed with 2 × 10 FITC-labeled (Vybrant Phagocytosis Assay kit; Invitrogen) in a final volume of 500 μl, and incubated for 2 h at 37°C in the presence or absence of 100 nM uPCI, 50 μM cytochalasin D, or 10% (vol/vol) human serum. In the preincubation experiments, either macrophages or bacteria were separately preincubated with 100 nM uPCI in serum-free medium for 60 min at 37°C and washed once before being added to the phagocytosis assay. After 2 h, noninternalized bacteria were rinsed away (after rinsing fluoroscence images were taken) and loosely attached were removed by trypsinization. The phagocytes were collected by centrifugation, resuspended in ice-cold PBS, and analyzed by flow cytometry (FACSCalibur; BD Biosciences).
uPCI was labeled with I (GE Healthcare) by adding 100 μCi of the isotope to 50 μl of 0.25 M sodium phosphate buffer, pH 7.0, followed by the immediate mixing with 10 μl of 5 mg/ml chloramine T solution. 5 min after addition of 2.5 μg PCI, the reaction was stopped by inclusion of 50 μl of a saturated sodium ascorbate buffer, pH 7.0, and the labeled protein was purified from the unreacted I via passage over a Sephadex-G10 column (GE Healthcare). After incubation with the labeled PCI, the cells were washed once with PBS and incubated for 30 min at 4°C with ligand- dissociation buffer (0.05% trypsin, 0.5 M EDTA, and 50 μg/ml proteinase K in PBS; ). The suspensions were centrifuged, the extracellular media were discarded, and the radioactivity of the pellets was determined. Alexa fluorescence intensity was measured in detergent-treated pellets (2% Triton X-100).
The activity of the prothrombinase complex was determined as described previously ().
HL-60 cells were incubated with 100 nM PCI at 37°C, and the cell pellet was resuspended in nuclear extraction buffer 1 (10 mM Tris-HCl, 10 mM KCl, 0.5% IGEPAL CA-630, and protease inhibitor cocktail Complete; Roche), incubated for 30 min on ice, and lysed by sonication. The supernatants obtained from the homogenates were centrifuged and the resulting supernatants were used for separation of the cytosolic proteins. The pellets of the homogenates containing the nuclei were resuspended in extraction buffer 2 (10 mM Tris-HCl, 0.5M NaCl, and Complete), incubated for 30 min on ice, and centrifuged for 30 min, and the supernatants were recovered. After 10% SDS-PAGE, the proteins were transferred onto polyvinylidene fluoride membranes (Millipore), blocked overnight with 5% dry milk, and incubated for 60 min with anti-PCI antibody. After incubation for 45 min with horseradish peroxidase–conjugated anti–rabbit IgG, the proteins were visualized.
LUV were prepared from 10 mg of total lipids per milliliter in a buffer composed of 20 mM Hepes and 150 mM NaCl, pH 7.5, and supplemented with 0.5 mg/ml trypsin or 0.1 mg/ml proteinase K (intravesicular proteases). The various lipids (1-palmitoyl, 2-linoleoyl species of PC and PE, dimyristoyl-PE, dioleoyl-DAG, and PS) were dispersed by vortexing with glass beads. They were passed several times through an extruder (100 or 200 nm filters; LiposoFast; Avestin) until the solution was clear. The vesicle suspensions were subsequently dialyzed. Then the vesicle suspensions were incubated with PCI for 60 min at 37°C. The vesicle preparations and incubations were performed in the presence of 10 nM butylated hydroxytoluene under argon or nitrogen. Incubations with the intravesicular proteases were performed in the presence of 2 mM of the protease inhibitor PMSF. Where indicated, the preparation of the LUV was performed before the proteases, and trypsin or proteinase K were added after the incubations with PCI (extravesicular proteases).
The lipids were dispersed in a buffer composed of 10 mM Hepes, 100 mM KCl, and 1 mM EDTA, pH 7.4, containing 60 mM CF (mixed isomers; Sigma-Aldrich). The vesicles were sized using 200-nm polycarbonate filters and separated from the extravesicular CF by dialysis. The fluorescence signal was recorded after addition of mastoparan X and PCI at 37°C.
. To determine the fluorescence signal for the maximally releasable amount of CF (
), the vesicles were solubilized using the detergent CE.
Giant unilamellar vesicles (GUV) were prepared from 1-palmitoyl, 2-linoleoyl species of PC and PE together with 5 mol% dipalmitoyl-PS. For determination of the translational diffusion coefficient D, 10 mol% of DiD-C (Invitrogen) was added. The dispersed lipids were transferred to a coverglass chamber (Lab-Tek; Nalge Nunc), the formed GUV were allowed to settle for 20 min, and 9 nM Alexa 488–PCI was added. The fluorescence correlation spectroscopy measurements were performed in the autocorrelation mode with the ConfoCor 2 system (Carl Zeiss, Inc.) using either the 488-nm line of an Ar-ion Laser or a 633-nm HeNe Laser (at 50–60 μW). The excitation light was reflected by a dichroic mirror (HFT 488/633; Carl Zeiss, Inc.) and focused onto the sample. The emitted fluorescence light was split by a second dichroic mirror (NFT 635; Carl Zeiss, Inc.), passed through two filters (505–550 nm bandpass and 650 nm longpass filter; Carl Zeiss, Inc.), and detected in two separate channels. In each case, out of plane fluorescence was reduced by a pinhole of 90-μm diameter. For analysis of the membrane interaction of PCI, the fluorescence correlation spectroscopy focal spot was positioned near the center of the upper membrane of the GUV. For characterizing the diffusion properties of PCI in aqueous solution, the focus was moved to a position 20 μm above the upper membrane. The data were evaluated by Levenberg-Marquardt nonlinear least squares fitting to the appropriate model of the autocorrelation function.
LUV of a diameter of 200 nm were prepared by the extrusion method (see Translocation of PCI…) and Pr was added to a final concentration of 5.7 mM. The chemical shifts of the P signals from the phospholipids present in the outer and inner leaflet of the LUV were registered. NMR spectra were recorded with a spectrometer (400S; Varian Medical Systems).
Based on the crystal structure of cleaved PCI (), docking studies for PE binding were performed with the program DockVision. Because of the extreme flexibility of the PE ligand, the docking studies were initially performed on a straight 11-carbon chain fragment (generated using CORINA; ) meant to represent a possible conformation of the fully saturated hydrocarbon chain of PE. The maximum floating radius was set at 100 Å to ensure full coverage of PCI, and a total of 10,000 trials with a schedule of six conditions of 500 steps each was used in docking. The model of the PE–PCI complex using the 1-palmitoyl, 2-linoleoyl species of PE was built in XtalView () and energy minimized using CNS (one cycle of 2,000 steps; ).
The mean values given are ± SD. The determinations were compared by one-way analysis of variance. Differences of P < 0.05 were considered to be significant.
Fig. S1 shows that inhibition of PCI decreases PE import into activated platelets. Fig. S2 indicates that uPCI added to HL-60 cells is incorporated into nuclear fractions of those cells. Fig. S3 shows that the PE-dependent internalization of PCI is also detected at low temperatures and under conditions of extensive proteolysis of cell surface proteins. Fig. S4 provides results from molecular docking analyses indicating the presumed PE binding site in PCI. Online supplemental material is available at . |
To maintain genome stability, DNA replication occurs only once in each cell cycle. This is achieved by the cell cycle–dependent licensing control that permits assembly of prereplication complexes (pre-RCs) at replication origins only once per cell cycle (; ; ; ; ). On exit from metaphase, the minichromosome maintenance (MCM) 2–7 complex is recruited to replication origins to establish pre-RCs. Once replication is activated by the Cdks and Cdc7/Dbf4 kinase, pre-RCs are disassembled and the reassociation of MCM proteins with origins is not permitted until the completion of mitosis.
Regulation of Cdt1 activity during the cell cycle is critical for the prevention of reassembly of pre-RCs within a single cell cycle (; ). When cells enter S phase, Cdt1 is ubiquitinated and degraded (; ; ; ), and this cell cycle–regulated Cdt1 degradation is important for the prevention of rereplication. In the cell free system, the addition of Cdt1 in G2 extracts induces rereplication, which is enhanced in the absence of geminin (; ; ; ). In mammalian cells, overexpression of Cdt1 promotes substantial rereplication in certain tumor cell lines (; ; ).
Rereplication of the genome, or even a segment of it, could initiate gene amplification and cause chromosomal translocation and loss, contributing to the tumorigenesis process. It has been shown that cell cycle checkpoints are activated when rereplication is induced (). In yeast, rereplication leads to accumulation of extensive double-stranded breaks (DSBs) and activates checkpoints (; ). In the cell free system, a caffeine-sensitive phosphorylation of Chk1 occurs when rereplication is induced (). In mammalian cells, overexpression of Cdt1 or the loss of geminin activates the G2/M checkpoint or induces apoptosis (; ; ). Recently, by using egg extracts, it was demonstrated that extensive rereplication, but not a single round of rereplication, leads to head-to-tail collision of rereplicating forks resulting in DNA fragmentation and checkpoint activation (). This finding suggests an important mechanism underlying checkpoint activation when uncontrolled rereplication is induced. However, other mechanisms that do not require extensive rereplication likely exist. In mammalian cells, when licensing control is disrupted by Cdt1 overexpression, intriguingly, extensive rereplication only occurs in certain tumor cell lines (; ; this paper). In primary cell lines and several tumor cell lines, significant rereplication is not induced but checkpoints are activated.
Two protein kinases, ataxia telangiectasia mutated (ATM) and ataxia telangiectasia and Rad3 related (ATR), play critical roles in sensing abnormal DNA structures and initiating signal transduction in the checkpoint pathways (; ). Although related, ATM and ATR respond to DNA damage in overlapping but distinct pathways. ATM responds mainly to DSBs, whereas ATR is activated by various kinds of DNA damage, including UV and hydroxyurea treatment, which stall replication forks. ATR activation requires DNA replication, which is different from ATM activation (; ; ).
In this study, we demonstrated that in mammalian cells, the ATR-mediated S phase checkpoint is activated at the onset of rereplication before the appearance of DSBs. This ATR-mediated S phase checkpoint acts directly to inhibit rereplication so that significant rereplication is prevented in the cell lines with an intact ATR pathway. We also demonstrated that uncontrolled DNA unwinding initiated by MCM proteins from relicensed origins leads to single-stranded DNA (ssDNA) accumulation, revealing an important mechanism underlying checkpoint activation in the absence of extensive rereplication. Unlike ATR, ATM does not play critical roles in the prevention of rereplication but coordinates with ATR to activate the G2/M checkpoint. Finally, we identified important effector proteins, including RPA2 and retinoblastoma protein (Rb), downstream of ATR to mediate the inhibition of rereplication. We propose that specific defects in the ATR-mediated S phase checkpoint pathways present in certain tumor cell lines contribute to their susceptibility to rereplication.
In mammalian cells, when licensing control is disrupted by Cdt1/Cdc6 overexpression, it is intriguing that substantial rereplication occurs only in certain tumor cell lines but not in primary cell lines and several other tumor cell lines (; ). Although p53 is involved (), the loss of geminin-induced rereplication occurs in a p53-independent manner (; ), suggesting that pathways other than p53 are also important for preventing rereplication.
Consistent with previous findings (; ), we observed that overexpression of Cdt1 by adenovirus infection was sufficient to induce DNA rereplication in H1299, HeLa, and U2OS tumor cells, but failed to cause overt rereplication in many other cell lines, including primary cell lines IMR90 and WI38 and tumor cell lines T98G and A549 (). By using a higher titer of Ad-Cdt1 viruses, Cdt1 overexpression levels in these rereplication-susceptible cell lines were similar to those in U2OS and H1299 cell lines ( and not depicted). Co-overexpression of Cdc6 with Cdt1 also failed to induce overt rereplication in T98G, A549, and IMR90 cells, although it enhanced Cdt1-induced rereplication in U2OS and H1299 cells (Fig. S1, available at ; and not depicted).
Although extensive rereplication was not observed in T98G, A549, and IMR90 cells, checkpoints were activated after Cdt1 overexpression. Chk1 and 2 were both phosphorylated upon Ad-Cdt1 infection in an ATR- and ATM-dependent manner, respectively (), although the phosphorylation levels in IMR90 and A549 were reduced compared with H1299 and U2OS cells, where rereplication was achieved extensively (, left; and not depicted). These data suggest that both the ATM–Chk2 and ATR–Chk1 pathways are activated when Cdt1 is overexpressed, even in the absence of extensive rereplication, although overt rereplication may amplify damage signals and enhance checkpoint activation.
The critical role of cell cycle checkpoint activation is to maintain genome stability. The observation that ATR and ATM were activated by Cdt1 overexpression in the cell lines without extensive rereplication prompted us to examine whether checkpoint activation in fact prevented rereplication. Because Cdt1 overexpression–induced ATR and ATM activation was observed in rereplication-susceptible cell lines such as U2OS and H1299 (), we anticipated that defects allowing rereplication in these cell lines were present at the effector pathways downstream of ATR and/or ATM. We first examined rereplication levels in U2OS cells when the expression of ATR or ATM was suppressed by short hairpin RNA (shRNA). If the susceptibility of U2OS cells to Cdt1-induced rereplication was indeed caused by a defect in one branch of the ATR and/or ATM pathways, inactivation of ATR and/or ATM would influence multiple pathways, thereby leading to more severe rereplication after Cdt1 overexpression. We used a low titer of Ad-Cdt1 viruses, which did not induce obvious rereplication in U2OS cells. Interestingly, when ATR-shRNA was expressed, robust rereplication was induced after Ad-Cdt1 infection, whereas inhibition of ATM did not lead to more rereplication (). The Cdt1 expression levels were similar in all the cell lines that were examined (, bottom). Consistently, inhibition of endogenous ATR kinase activity by overexpressing kinase-dead (KD) Flag-ATR in U2OS cells () also led to higher levels of DNA rereplication after Ad-Cdt1 infection (). As an alternative approach, we inactivated DDB1 and Cdt2, the components of a Cul4 ubiquitin ligase that are required for Cdt1 ubiquitination and degradation (; ) by shRNA in U2OS cells. Silencing the expression of DDB1 or Cdt2 leads to detectable accumulation of Cdt1 (1.5- to 2-fold more) and rereplication (). Inactivation of ATR by shRNAs before the depletion of DDB1 or Cdt2 significantly enhanced the rereplication levels compared with vector-infected, ATR-proficient cells. Collectively, these data suggest that ATR, but not ATM, plays an important role in the suppression of DNA rereplication.
We then tested whether ATR is required to prevent rereplication in the cell lines that are resistant to Cdt1-induced rereplication. Strikingly, when ATR activity was inhibited in IMR90, WI38, A549, and T98G cell lines by shRNAs, a significant amount of cells underwent rereplication after Cdt1 or Cdt1/Cdc6 was overexpressed ( and not depicted). Therefore, ATR acts as a critical barrier in these cell lines to resist Cdt1-induced rereplication. In contrast, inhibition of ATM activity did not disrupt the cellular control that prevents rereplication (). These results uncover a critical function of ATR in the prevention of rereplication, especially when licensing control is impaired.
Chk1 and 2 are two effector protein kinases downstream of ATR and ATM, respectively. We suppressed the expression of Chk1 and 2 in multiple cell lines and examined the levels of rereplication after Cdt1 overexpression. Inactivation of Chk1, but not Chk2, significantly increased the rereplication levels in U2OS cells and also led to rereplication in A549 cells, which were normally resistant to Cdt1-induced rereplication (). These data support the finding that the ATR–Chk1 pathway, not the ATM–Chk2 pathway, is involved in the suppression of rereplication.
During DNA rereplication, both DSBs and ssDNA are accumulated (; ; ; ; ). We observed that DNA lesions were also accumulated in the cell lines where Cdt1 overexpression failed to induce significant rereplication. In every cell line tested, RPA1 and 2 were loaded onto chromatin when Cdt1 was overexpressed, suggesting that ssDNA accumulated on chromatin (). RPA2 was also phosphorylated on chromatin, revealed by the presence of slower migrating forms of RPA2 (). The presence of ssDNA was confirmed by BrdU immunostaining under nondenaturing conditions (). Likewise, we also detected H2AX phosphorylation, a common marker for indicating DSBs (), in the cell lines without obvious rereplication, including IMR90, T98G, and A549 cells ( and not depicted). These data suggest that Cdt1 overexpression leads to the generation of both ssDNA and DSBs even in the absence of extensive rereplication.
To investigate how DNA lesions were generated when Cdt1 was overexpressed, we performed a time course analysis to monitor the accumulation of ssDNA and DSBs. When a lower titer of Ad-Cdt1 was used, a slower kinetics of Cdt1 overexpression and Chk1 activation was observed than when a higher titer of Ad-Cdt1 was used (Fig. S2, available at ). Therefore, to clearly demonstrate the temporal order of ssDNA and DSB accumulation and checkpoint activation, we used the low titer of Ad-Cdt1 or Ad-vec (5 × 10 pfu/ml, MOI = 25) to infect U2OS cells. At the 12-h time point (4 h after significant Cdt1 expression), although no obvious rereplication was detected (, bottom), Chk1 was already phosphorylated at the ATR phosphorylation site S317 (, left). At an approximately similar time, RPA chromatin loading was observed (, right), suggesting that ssDNA was generated. Interestingly, Chk2 phosphorylation at ATM phosphorylation site T68 was not detected until the 24-h time point, when rereplication was initially detected. At the same time point, H2AX was phosphorylated, suggesting that DSBs were likely generated relatively late. When ATR was inactivated in U2OS cells by shRNA, the kinetics of Chk1, Chk2, and H2AX phosphorylation was altered. Chk1 phosphorylation was delayed because of the critical role of ATR to phosphorylate Chk1, whereas ATM-mediated Chk2 and H2AX phosphorylation as well as RPA chromatin loading occurred earlier (Fig. S3). This is consistent with the idea that rereplication is more profound in ATR-deficient cells, which leads to more severe DNA lesions and checkpoint activation. These data also suggest that the temporal order of the accumulation of phosphorylated species of Chk1 and 2 and H2AX observed in ATR-proficient cells () is not likely caused by different sensitivities of the specific antibodies.
The observation of ssDNA accumulation at early stages after Cdt1 overexpression suggests that ATR is likely activated by the same S phase checkpoint pathway as when replication forks are stalled by DNA-damaging agents. Indeed, we observed that Rad17 was phosphorylated on chromatin, and Rad9 chromatin loading was promoted when Cdt1 was overexpressed in multiple cell lines, including the cell lines without significant rereplication (). In addition, Cdt1 overexpression– induced Chk1 phosphorylation was diminished when the expression of ATR-interacting protein (ATRIP) or Rad17 was inhibited by shRNA (not depicted). Meanwhile, inhibition of Rad17 or ATRIP expression caused more severe rereplication in U2OS cells and led to rereplication in A549 cells, which were otherwise resistant to Cdt1-induced rereplication (). These data demonstrate a critical role of the ATR-mediated S phase checkpoint pathway in both sensing the loss of licensing control and preventing rereplication, which highlights a new aspect of the S phase checkpoint in the maintenance of genome stability.
U2OS cells were synchronized by a double thymidine block. 6 h after releasing, nocodazole was added and after another 6 h, almost all cells exited from S phase, as revealed by the cell cycle profile and low BrdU incorporation (Fig. S4 A, available at ; and not depicted). At this time point, cells were infected with Ad-Cdt1 or Ad-vec (). 24 h after adenoviral infection, both attached and round-up cells were collected after manually shaking off. Phosphorylation of Cdk1 at Tyr15 and cyclin A expression were detected in attached cells, and the mitotic marker phosphohistone H3 was present in shaken-off cells (Fig. S4 B), suggesting that a significant number of attached cells are in G2 phase, which is consistent with previous findings (). Cdt1 overexpression was predominantly observed in attached cells (Fig. S4 B), indicating that detached mitotic cells either were difficult to infect or expressed Cdt1 at low levels. Attached cells were harvested for FACS analysis after Ad-Cdt1 or Ad-vec infection and a significant amount of rereplication was induced after Ad-Cdt1 infection (). This Cdt1- induced rereplication was prevented when adenoviruses encoding geminin, p27, or a Cdc7 KD mutant (Cdc7KD and Cdc7-D196N; ) were used to infect cells before Ad-Cdt1 infection (), suggesting that Cdt1-mediated licensing and the kinase activity of Cdk and Cdc7 are required for rereplication induced after S phase. Cdt1-induced rereplication was also inhibited when cells were treated with aphidicolin, an inhibitor of DNA polymerase α (; ).
Similar to asynchronized cells, checkpoint was activated when Cdt1 was overexpressed after S phase (, bottom). Overexpression of geminin, p27, and Cdc7KD before Cdt1 prevented ssDNA accumulation and attenuated checkpoint activation (). Interestingly, after aphidicolin treatment, although rereplication was prevented to similar levels as when p27 or Cdc7KD was overexpressed (), ssDNA accumulation shown by RPA1 chromatin loading was still evidently accompanied by checkpoint activation (), although the level of checkpoint activation was reduced. These data suggest that the initiation of DNA rereplication, which requires Cdks and Cdc7 activity, is important for generating ssDNA and activating checkpoint. Because aphidicolin inhibits DNA polymerase α and thus likely arrests replication/rereplication forks close to replication origins, continuous DNA synthesis and extensive rereplication seem to not necessarily be required for ssDNA accumulation and checkpoint activation, although they may possibly enhance these events.
When Cdt1/Cdc6 is overexpressed, MCM proteins are reloaded onto chromatin, but reloaded MCM proteins may not coordinate properly with other replication proteins. For instance, MCM-mediated DNA unwinding may exceed the availability of polymerases that can be used to synthesize DNA when origins are refired. This may cause functional uncoupling of MCM helicase and DNA polymerases leading to the accumulation of ssDNA, which has been described previously when DNA polymerase α is inactivated or the replication fork is stalled (; ). If this model is correct, the MCM helicase activity that is required for DNA unwinding at the onset of rereplication would be needed for ssDNA accumulation and checkpoint activation upon Cdt1 overexpression.
When p27 or Cdc7KD was overexpressed before Cdt1/Cdc6 overexpression, we noticed that the loss of ssDNA accumulation and checkpoint activation was accompanied by a decrease of MCM2 phosphorylation, as demonstrated by a decrease of the faster migrating band (the phosphorylated MCM2 species) on SDS-PAGE gel (). Because this cell cycle–regulated MCM2 phosphorylation is believed to be important for the helicase activities of the MCM complex (), this finding supports the idea that MCM-mediated DNA unwinding is involved in generating ssDNA.
To more directly examine whether DNA unwinding at relicensed origins was required for generating ssDNA, we inhibited MCM helicase activity by overexpressing an MCM7 mutant (K387A) carrying a mutation at its conserved ATPase motif that is required for replication helicase activity (; ). Immunoprecipitation showed that MCM7-K387A interacted with other endogenous MCM proteins (not depicted) and that overexpression of this mutant prevented cells from entering S phase, suggesting a dominant-negative effect of this mutant (Fig. S5, available at ). As shown in , overexpression of the MCM-K387A mutant before Cdt1 overexpression in synchronized cells that had completed S phase inhibited rereplication and significantly reduced RPA chromatin loading and Chk1 phosphorylation. These data strongly suggest that MCM-mediated DNA unwinding from relicensed origins is required for generating ssDNA and activating the checkpoint when Cdt1 is overexpressed.
We have demonstrated that the ATR-mediated checkpoint can be activated by the initial steps of rereplication, which acts to prevent further rereplication. What is the mechanism by which activated ATR inhibits rereplication? We observed that RPA2, which plays essential roles in replication initiation and elongation (; ), was phosphorylated when Cdt1 was overexpressed () in an ATR- but not ATM-dependent manner (, left). Similarly, phosphorylation of MCM2 at S108, a site phosphorylated by ATR upon UV and hydroxyurea treatment (), was also observed after Cdt1 overexpression (, right).
Multiple damage-inducible and ATR-dependent phosphorylation sites have been identified at the N terminus of RPA2 (; ; ). Cdt1 overexpression–induced RPA2 phosphorylation was completely abolished in the HA-tagged RPA2 mutant carrying S/T-to-A substitutions at the ATR-dependent phosphorylation sites S4/S8/S11-13/T21/S33 (, right; ). Meanwhile, rereplication was also significantly enhanced in the cell line expressing the HA- RPA2-phospho mutant with endogenous RPA2 silenced by shRNA compared with the cell line expressing the HA- RPA2 wild type (, left). These results suggest that ATR-mediated RPA2 phosphorylation plays a direct role in suppressing DNA rereplication. Because MCM2 S108 is not the only ATR site and not all ATR phosphorylation sites on MCM2 have been identified (), a direct role of MCM2 phosphorylation by ATR in the suppression of rereplication is currently difficult to analyze.
Overexpression of Cdt1 or Cdt1/Cdc6 can readily induce rereplication in H1299 and U2OS cells. Because checkpoint activation is intact in these two cell lines, the defects in suppressing rereplication are probably present at downstream steps. As described previously, the susceptibility of H1299 cells to rereplication is caused by a loss of p53 function (). However, the U2OS cell line carries wild-type p53, and its genetic basis for permitting rereplication is not clear.
A defect in Rb dephosphorylation was observed in U2OS cells during progression from mitosis into G1 (). Upon DNA damage, Rb dephosphorylation is induced (; ; ; ). To examine whether a dephosphorylation defect of Rb in U2OS cells might contribute to Cdt1-induced rereplication, we used a U2OS cell line that expressed constitutively active Rb under the control of tetracycline (; ). In this Rb mutant, 14 of the 15 conserved Cdk consensus sites are mutated. Cdt1 overexpression–induced rereplication was significantly inhibited when the expression of hypophosphorylated Rb was induced (, top). These data suggest that a defect in U2OS cells affecting Rb dephosphorylation is likely responsible for permitting Cdt1-induced rereplication. Consistent with an important role of Rb in the suppression of rereplication, when the expression of Rb was inhibited by shRNA in T98G and A549 cells, substantial rereplication was observed after Cdt1 overexpression (, bottom).
To assess whether overexpression of Cdt1 leads to Rb dephosphorylation and thus inhibition of rereplication, we monitored Rb dephosphorylation by using the antibody G99-549, which specifically recognizes the hypophosphorylated state of Rb after Cdt1 overexpression (; ). In both IMR90 and A549 cell lines, Rb dephosphorylation was induced when Cdt1 was overexpressed (). Interestingly, when ATR expression was inhibited by shRNA, accumulation of hypophosphorylated Rb species was diminished after Cdt1 overexpression. Collectively, these data suggest that ATR-mediated checkpoint activation induces Rb dephosphorylation, which is involved in the inhibition of rereplication induced by Cdt1 overexpression.
Although both the ATM and ATR pathways are activated when Cdt1 is overexpressed, only the ATR pathway is necessary for the suppression of rereplication. What is the biological role of ATM activation? Previous papers have shown that Cdt1 overexpression activates the G2/M checkpoint by arresting cells before mitosis and that ATR contributes to this arrest (; ). As illustrated in , after Cdt1 overexpression, cells expressing ATM-shRNA were also partially impaired in G2/M arrest as were ATR-shRNA–expressing cells, whereas simultaneous inhibition of both ATM and ATR further increased the number of cells that escaped G2/M arrest. Thus, ATR activation at an early stage after Cdt1 overexpression directly inhibits DNA rereplication and ATM is subsequently activated, which acts synergistically with ATR to arrest cells in G2/M.
t
h
i
s
s
t
u
d
y
,
w
e
d
e
m
o
n
s
t
r
a
t
e
d
t
h
a
t
t
h
e
A
T
R
-
m
e
d
i
a
t
e
d
S
p
h
a
s
e
c
h
e
c
k
p
o
i
n
t
p
r
o
v
i
d
e
s
a
p
r
o
t
e
c
t
i
o
n
m
e
c
h
a
n
i
s
m
b
e
y
o
n
d
l
i
c
e
n
s
i
n
g
c
o
n
t
r
o
l
t
o
p
r
e
v
e
n
t
r
e
r
e
p
l
i
c
a
t
i
o
n
a
n
d
m
a
i
n
t
a
i
n
g
e
n
o
m
e
s
t
a
b
i
l
i
t
y
.
W
e
a
l
s
o
d
e
m
o
n
s
t
r
a
t
e
d
t
h
a
t
i
n
m
a
m
m
a
l
i
a
n
c
e
l
l
s
,
o
n
e
i
n
i
t
i
a
l
s
i
g
n
a
l
t
o
t
r
i
g
g
e
r
c
h
e
c
k
p
o
i
n
t
a
c
t
i
v
a
t
i
o
n
b
y
C
d
t
1
o
v
e
r
e
x
p
r
e
s
s
i
o
n
i
s
s
s
D
N
A
a
c
c
u
m
u
l
a
t
i
o
n
,
w
h
i
c
h
i
s
c
a
u
s
e
d
b
y
u
n
c
o
n
t
r
o
l
l
e
d
M
C
M
-
m
e
d
i
a
t
e
d
D
N
A
u
n
w
i
n
d
i
n
g
,
w
h
e
r
e
a
s
D
S
B
s
a
r
e
g
e
n
e
r
a
t
e
d
s
u
b
s
e
q
u
e
n
t
l
y
.
T
h
i
s
f
i
n
d
i
n
g
s
u
g
g
e
s
t
s
a
n
o
v
e
l
m
e
c
h
a
n
i
s
m
t
o
a
c
t
i
v
a
t
e
t
h
e
c
h
e
c
k
p
o
i
n
t
a
t
t
h
e
i
n
i
t
i
a
t
i
o
n
o
f
r
e
r
e
p
l
i
c
a
t
i
o
n
b
e
f
o
r
e
u
n
c
o
n
t
r
o
l
l
e
d
r
e
r
e
p
l
i
c
a
t
i
o
n
o
c
c
u
r
s
.
U2OS, H1299, A549, T98G, and IMR90 cells were grown in DME supplemented with 10% fetal bovine serum. To induce the expression of Flag-ATR wild type or Flag-ATR-KD in the established U2OS cell lines (), 1 μg/ml doxycycline was added to the media for 24 h. U2OS cells with tetracycline-regulated constitutive active Rb (provided by S. Mittnacht, Institute of Cancer Research, London, UK; ) were maintained in medium, supplemented with 2 μg/ml tetracycline, 300 μg/ml G418, and 0.5 μg/ml puromycin. Rb expression was induced by culturing cells in tetracycline-free medium for 24 h.
Stable U2OS cell lines expressing an HA-tagged RPA2 or phospho-RPA2 mutant were generated by retroviral infection using pBabe vector as described previously (). Silencing of endogenous RPA2 in these cells was conducted by two rounds of retroviral infection using a pMKO vector () that expresses the RPA2 shRNA target sequence located in the 3′ untranslated region of the mRNA, followed by drug selection (). Other retroviral infection was also performed accordingly.
Cells were lysed in NETN (150 mM NaCl, 1 mM EDTA, 20 mM Tris-Cl, pH 8.0, and 0.5% NP-40) containing protease and phosphatase inhibitors (Sigma-Aldrich). For chromatin isolation, cells were washed with PBS, resuspended in CSK buffer (10 mM Pipes, pH 6.8, 100 mM NaCl, 300 mM sucrose, 3 mM MgClB, 1 mM EGTA, 50 mM Na-F, 0.1 mM Na-orthovanadate [Sigma-Aldrich], 0.1% Triton X-100 [Sigma-Aldrich], and protease inhibitors), and incubated on ice for 10 min. Cytoplasmic proteins were separated from nuclei by low speed centrifugation at 1,300 for 5 min. Isolated nuclei were washed once in CSK buffer and lysed in solution (3 mM EDTA, 0.2 mM EGTA, 1mM DTT, and protease inhibitors). After centrifugation (1,700 for 5 min), pellets were resuspended in CSK buffer. 2× SDS loading buffer was added and samples were boiled for 10 min.
shRNA retroviral plasmids were constructed by inserting annealed shRNA oligos into a pMKO retroviral vector (). The shRNA sequences used were ATM (GCACCAGTCCAGTATTGGCTT and AACATACTACTCAAAGACATT), ATR (CGAGACTTCTGCGGATTGCAG and AACCTCCGTGATGTTGCTTGA; ), ATRIP (AAGGTCCACAGATTATTAGAT; ), Chk2 (CAGTGTCCACTCAGGAACTCT), Chk1 (AAGCGTGCCGTAGACTGTCCA and AAGTACTCCAGTTCTCAGCCA), claspin (CCTTGCTTAGAGCTGAGTCTT and GGAAAGAAAGGCAGCCAGATT; ), Rad17 (CAGACTGGGTTGACCCATCTT; ), DDB1 (TGAGTGCTTGACATACCTTGA), and Cdt2 (GTTCCTGGTGAACTTAAACTT; ). Sequence information for Chk1 and 2 shRNA was provided by A. Maclaren, J. Scorah, and C. McGowan (The Scripps Research Institute, La Jolla, CA).
The Cdc7KD was generated by mutating D196 to N in the conserved kinase motif of Cdc7 ().
Adenoviruses encoding GFP, human Cdt1, Cdc7KD, Cdc6, and MCM7-K387A were generated by using the AdEasy system (). In brief, the target cDNAs were constructed into a pAd-track-CMV shuttle vector and in vivo recombination was performed by transforming the pAd-track-CMV shuttle vector together with the pAd-Easy-1 adenoviral vector into a BJ5813 competent cell by electroporation. The recombinant adenoviral plasmids were transfected into 293 cells to produce corresponding recombinant adenoviruses. Large-scale purification of viruses was obtained by CsCl density gradient centrifugation. Recombinant adenoviruses encoding p27 and geminin were provided by S. Reed (The Scripps Research Institute, La Jolla, CA) and J. Cook (University of North Carolina at Chapel Hill, Chapel Hill, NC).
The following antibodies were used in this study: Cdt1 was produced as described previously (); the phospho-Chk1 (S317) antibody was obtained from R&D Systems; phospho-Chk2 (T68) was obtained from Cell Signaling Technology; Chk1, Chk2, Ku70, Rad17, Rad9, and geminin antibodies were obtained from Santa Cruz Biotechnology, Inc.; p27, actin, RPA1, RPA2, ATM, and ATR antibodies were obtained from EMD; γ-H2AX and phosphohistone H3 antibodies were obtained from Millipore; ATRIP and phospho-MCM2 (S108) antibodies were obtained from Bethyl Laboratories, Inc.; Rb, cyclinA, cyclinE, MCM2, and DDB1 antibodies were obtained from BD Biosciences. Cdc7 antibody was provided by T. Tsuji and W. Jiang (The Burnham Institute for Medical Research, La Jolla, CA).
Cells were collected and fixed with ice cold 70% ethanol overnight at 4°C. After washing with PBS, cells were stained with 15 μg/ml propidium iodide solution containing 38 mM sodium citrate and 10 μg/ml RNase A and analyzed on a flow cytometer (Becton Dickinson) using Cellquest software (BD Biosciences). For BrdU incorporation analysis, cells were pulse labeled with 10 mM BrdU for 1 h. After fixation with ice cold 70% ethanol for 1 h, cells were resuspended in 2 mM HCl and incubated for 20 min at room temperature. After centrifugation, cell pellets were resuspended in 0.5 ml of 0.1 M NaBO, pH 8.5, and after a PBS wash, cells were stained with 50 μl of antibody solution (30 μl PBS containing 0.5% Tween-20/0.5% BSA + 1 μl mouse anti-BrdU monoclonal antibody) for 1 h at room temperature, followed by 30-min staining with FITC-conjugated secondary antibody in the dark before propidium iodide staining.
Fig. S1 shows that overexpression of Cdt1 or Cdt1/Cdc6 induces DNA rereplication in certain tumor cell lines. Fig. S2 shows that Cdt1 overexpression and Chk1 phosphorylation were detected earlier when a higher titer of Ad-Cdt1 was used. Fig. S3 shows a time course analysis of checkpoint activation and RPA chromatin loading in ATR-deficient or -proficient cells after Cdt1 overexpression. Fig. S4 shows that U2OS cells were synchronized to G2/M. Fig. S5 shows that overexpresion of the MCM7-K387A mutant prevents S phase entry. Online supplemental material is available at . |
A large percentage of the population, in particular numerous aged individuals, patients with diabetes or cancer, or people treated with anti-inflammatory steroids, suffer from chronic, nonhealing wounds (; ). Therefore, there is a strong need to develop strategies for the improvement of the repair process. This requires a thorough understanding of the underlying molecular and cellular mechanisms. A powerful approach to reach this goal is the identification and functional characterization of genes, which are regulated by skin injury and which are therefore candidate regulators of the repair process. Because the gene expression profile of the most malignant tumors resembles the profile of healing skin wounds (), wound-regulated genes may also be important targets for the development of novel and efficient therapeutics for the treatment of cancer. Therefore, we used differential display RT-PCR and microarray analysis to identify genes that are regulated by skin injury in mice (; ). Interestingly, many of the identified injury-regulated genes encode enzymes, which detoxify reactive oxygen species (ROS), or transcription factors, which regulate these genes (; ; ; ; ). Because large amounts of ROS are produced in early skin wounds by invading inflammatory cells as a defense against bacterial infection (; ), the expression of ROS-detoxifying enzymes by cells in the wound tissue may be an important mechanism to protect inflammatory and resident cells from ROS toxicity.
One of the wound-regulated genes encodes peroxiredoxin 6 (Prdx6). Peroxiredoxins comprise a family of six enzymes that catalyze the reduction of hydrogen peroxide and a broad spectrum of organic peroxides. Prdx1–5 have two reactive cysteines and they use thioredoxin and/or glutathione as a substrate (; ; ). By contrast, Prdx6—also designated 1-Cys-peroxiredoxin—has only a single redox-active cysteine (). This cytosolic enzyme was reported to use glutathione () or ascorbate () as reducing agent. In addition, Prdx6 displays phospholipase A activity ().
Recent studies revealed an important function of Prdx6 in the cellular stress response. Thus, overexpression of Prdx6 in different cell types protected from ROS-induced cytotoxicity (; ), whereas antisense-mediated knockdown of this enzyme enhanced the sensitivity to oxidative stress (; ; ). Prdx6 knockout mice were more sensitive to systemic treatment with the oxidative stress-inducing agent paraquat (). They also showed increased lung injury and mortality in response to hyperoxia (), and their hearts were more vulnerable to ischemia-reperfusion injury ().
Recent studies from our laboratory also suggest an important role of Prdx6 in the skin. Initially, we identified Prdx6 as the product of a keratinocyte growth factor target gene in cultured keratinocytes (). In vivo, overexpression of Prdx6 was found in the hyperproliferative epidermis of mouse skin wounds and of psoriatic patients as well as in cells of the wound granulation tissue (; ). To determine the consequences of enhanced expression of Prdx6 in keratinocytes, we recently generated transgenic mice overexpressing this enzyme in the epidermis. Interestingly, the enhanced levels of Prdx6 protected keratinocytes from UVA and UVB toxicity in vitro and in vivo and accelerated wound closure in aged animals (). These results revealed that Prdx6 is rate-limiting in keratinocytes under stress conditions and suggested an important role of this enzyme in the skin. To address this question we performed UV irradiation and wound-healing studies with Prdx6 knockout mice (), and we identified important functions of endogenous Prdx6 in UV protection and in blood vessel integrity in wounded skin.
Before we challenged the knockout mice we confirmed the loss of Prdx6 in the skin by Western blot analysis of whole skin lysates (). No compensatory up-regulation of other cytosolic peroxiredoxins (Prdx1, Prdx2) and of the secretable Prdx4 was observed as determined by RNase protection assay (). A detailed histological analysis of tail and back skin did not reveal any obvious abnormalities in the Prdx6 knockout mice (shown in for tail skin), demonstrating that Prdx6 is dispensable for skin morphogenesis and homeostasis.
We next determined the role of endogenous Prdx6 in the response to physiological doses of UV. Upon irradiation with UVA, the number of apoptotic sunburn cells was significantly higher in the epidermis of UVA-irradiated Prdx6 knockout mice compared with wild-type controls ( = 10; ). This result was confirmed by staining of UVA-irradiated skin sections with an antibody against p53, a protein, which is stabilized and activated upon DNA damage ( = 10; ). Enhanced apoptosis in Prdx6-deficient epidermis was also observed after UVB irradiation as demonstrated by staining of UVB-irradiated skin for cleaved caspase-3 ( = 5; ) and by analysis of sunburn cells in hematoxylin/eosin (H/E)-stained sections ( = 5; P = 0.056; unpublished data). These findings demonstrate that endogenous Prdx6 protects keratinocytes from UVA- and UVB-induced DNA damage and subsequent apoptosis.
We then subjected the Prdx6 knockout mice and their wild-type littermates to full-thickness excisional wounding. Although Prdx6 expression is particularly strong in keratinocytes of healing skin wounds (), no obvious delay in wound reepithelialization was observed in the knockout mice (), and this was also confirmed by morphometric analysis (Fig. S1, available at ). We could not detect any histological abnormalities in the wound tissue up to d 3 after injury (). However, the granulation tissue of Prdx6 knockout mice was characterized by severe hemorrhage at d 5 after wounding (). This time point correlates with the phase of extensive wound angiogenesis. Hemorrhagic areas were present throughout the wound tissue, but were particularly large below the hyperproliferative epithelium (). They were seen in male and female knockout mice, but the phenotype was more pronounced in males (unpublished data). Hemorrhage was also occasionally observed in heterozygous animals (unpublished data). At d 8 after wounding, the hemorrhage was mostly resolved and only a few small affected areas remained (). Hemorrhage was no longer detectable at d 14 after wounding (). At this time point wounds in mice of both genotypes were fully healed. Despite the hemorrhage, no difference in the rate of wound closure, or in wound size and area of hyperproliferative epithelium was detected (Fig. S1). Furthermore, the rate of keratinocyte proliferation was unaltered (Fig. S1), and the inflammatory response was not obviously affected as determined by immunohistochemical staining for macrophages and neutrophils (unpublished data) and by expression analysis of the pro-inflammatory cytokine interleukin-1β (IL-1β) ().
To unravel the reason for the severe hemorrhage, we stained the wound sections with an antibody against the endothelial cell-specific protein MECA-32 and determined the number and size of blood vessels in the wounds. However, we did not find a difference between knockout and wild-type animals (). Furthermore, maturation of blood vessels, which is reflected by the presence of surrounding smooth muscle cells, was also unaltered as determined by costaining for the endothelial cell marker PECAM-1 (CD31) and α-smooth muscle actin (α-SMA; ). Consistent with these findings, expression of several genes encoding proteins involved in angiogenesis, e.g., vascular endothelial growth factor-A (VEGF-A), angiopoietins-1 and -2, and transforming growth factor β1 (TGF-β1) were normally expressed in unwounded and wounded skin of Prdx6 knockout mice (). These findings suggest that the observed hemorrhage is not due to impaired angiogenesis or vessel maturation. Rather, defects in the endothelial cells themselves appeared likely. This hypothesis was confirmed by analysis of semi-thin sections from the wound tissue and in particular by electron microscopy. In the wounds of control mice, open and well-structured blood vessels had formed (), whereas wounds of Prdx6 knockout mice showed strongly damaged blood vessels (), and vacuoles were present in the cytoplasm of endothelial cells. Furthermore, damaged blood vessels were surrounded by extravasated erythrocytes, reflecting the hemorrhage (), and erythrocytes were also present in the wound epidermis of the knockout mice (), but not of wild-type controls (). A semi-quantitative analysis of the semi-thin sections revealed that more than 50% of the vessels were defective in the superficial granulation tissue, adjacent to the wound epidermis. In the middle of the granulation tissue 10–50% of the vessels were affected, whereas less than 10% of the vessels were damaged in the lowest part of the wound ().
The ultrastructural appearance of the endothelial cells suggested that some of them are apoptotic. This was confirmed by double-immunofluorescence staining with antibodies against PECAM-1 and cleaved caspase-3. At d 5 after wounding, apoptotic endothelial cells were detected in some blood vessels in the granulation tissue below the hyperproliferative epidermis, and their number was significantly increased in the knockout mice compared with wild-type controls (). This result is consistent with the histological appearance of the endothelial cells in semi-thin sections, although the percentage of morphologically damaged endothelial cells was much higher compared with the percentage of cleaved caspase-3–positive cells. This suggests that at least some of the damage may be reversible and does not lead to cell death.
In addition to the endothelial cells, some fibroblasts and granulocytes in the granulation tissue of Prdx6 knockout mice also appeared damaged ( and unpublished data). However, ∼90% of all cleaved caspase-3 positive cells in the wound were also PECAM-1 positive, demonstrating that endothelial cells are particularly affected (unpublished data).
The endothelial cell damage was still detectable at d 8 after wounding, but to a much lesser extent (Fig. S2, B and D; available at ). This is also reflected by the reduced hemorrhage at this later stage of healing (). Endothelial cells from vessels, which were already surrounded by perivascular cells (Fig. S2 D), appeared less damaged compared with those of immature vessels (Fig. S2 B).
We next determined if the extent of hemorrhage correlates with the severity of oxidative stress at the wound site. For this purpose we performed oxyblot analyses with lysates from pooled wounds of different animals to determine the levels of oxidized proteins, which are characterized by the presence of carbonyl groups. In three independent experiments with different wound lysates, 5-d wounds of knockout mice had a consistently higher protein carbonyl content compared with wild-type controls (). Protein oxidation was more pronounced in male animals, consistent with the more severe phenotype seen in Prdx6 knockout mice. In addition, staining with an antibody against nitrotyrosine, which reflects formation of the toxic peroxinitrite from nitric oxide and superoxide anions, revealed a significantly increased number of nitrotyrosine positive cells in wounds of Prdx6 knockout mice (). Collectively, oxidative stress is obviously enhanced in wounds of Prdx6-deficient animals, and the elevated levels of ROS are likely to damage endothelial cells.
To determine if the lack of Prdx6 in endothelial cells is indeed deleterious under conditions of oxidative stress, we studied the consequences of siRNA-mediated Prdx6 knock-down on the survival of cultured human umbilical vein endothelial cells (HUVEC) in response to ROS treatment. Efficient knock-down was achieved with three different Prdx6 siRNAs, whereas random siRNAs had no effect (). The knock-down of Prdx6 did not affect cell viability under normal culture conditions, but Prdx6 siRNA treatment strongly reduced the viability of HUVEC cells upon treatment with either hydrogen peroxide or glucose oxidase in comparison to random siRNA-treated cells (). These results were reproduced in 9 independent experiments with two different siRNA sequences against Prdx6 (#79, #81) and in four independent experiments with siRNA #80. Based on these in vitro results and the hemorrhage observed in wounds of Prdx6 knockout mice we conclude that endogenous Prdx6 is required for survival of endothelial cells under conditions of oxidative stress.
Finally, we determined if the loss of Prdx6 in endothelial cells is solely responsible for the phenotype or if enhanced levels of ROS produced by Prdx6-deficient inflammatory cells also contribute to the endothelial cell damage. The latter possibility was suggested by the finding that macrophages from Prdx6 mice have higher levels of intracellular ROS (), and at least hydrogen peroxide can be released into the surrounding extracellular space and possibly damage endothelial cells. To address this question we generated bone-marrow chimeric mice. Wild-type and Prdx6-deficient female mice were subjected to 950 rad gamma-irradiation to destroy the hematopoietic cells in the bone marrow. The irradiated mice subsequently received bone marrow cells from either wild-type or knockout donors. Male mice were used as bone marrow donors to allow subsequent identification of hematopoietic cells through the presence of the Y chromosome. 6 wk after irradiation and bone marrow transfer the fur of the treated mice had turned gray (Fig. S3 A, available at ). At this time point successful reconstitution of the hematopoietic system was verified by semi-quantitative PCR of blood cell DNA using primers specific for the Y-chromosome DNA (Fig. S3 B). 10 wk after irradiation this result was confirmed for DNA from bone marrow (Fig. S3 C). To estimate the efficiency of bone marrow transplantation, we performed real-time PCR with DNA from blood cells at wk 6 after bone marrow transplantation. In all female recipients the amount of Y chromosome DNA reached 70% or more of the amount present in DNA from male mice (Fig. S3 D).
Chimeras were subjected at wk 10 after bone marrow transplantation to full-thickness excisional wounding. This late time point was chosen to ensure full reconstitution of the hematopoietic system at the time of wounding. Nevertheless, the bone marrow transplantation procedure delayed the wound healing process in mice with all genotype combinations, possibly due to irradiation-induced tissue damage and/or enhanced stress of the mice. Therefore, only a small area of the clot was replaced by highly cellular granulation tissue, and angiogenesis was hardly detectable in 5-d wounds. This time point was therefore too early to detect damaged blood vessels (unpublished data). We subsequently repeated the experiment and analyzed the wounds at d 7 after injury. Western blot analysis of wound lysates confirmed that wild-type hematopoietic cells had infiltrated the wound tissue of Prdx6-deficient recipient mice. Thus, Prdx6 protein was detected in lysates of 7-d wounds from transplanted Prdx6 knockout mice, which had received bone marrow from wild-type mice, whereas Prdx6 protein was not detectable in normal back skin lysates of these mice (). The severe hemorrhage in the granulation tissue and the presence of extravasated erythrocytes in the wound epidermis were confirmed in wounds of Prdx6 knockout mice, which had received Prdx6- deficient bone marrow (; ko/ko; = 7). In these mice, the number of apoptotic endothelial cells was significantly enhanced compared with wild-type mice, which had received wild-type bone marrow (data not shown). Mild hemorrhage was observed in most Prdx6 knockout mice that had received wild-type bone marrow (; ko/wt; = 8) and in most wild-type mice with Prdx6-deficient bone marrow (; wt/ko; = 5). In both treatment groups, the number of apoptotic endothelial cells was also increased, although the difference was not statistically significant (unpublished data). As expected, hemorrhage was absent or very mild in wild-type mice with wild-type bone marrow (; wt/wt; = 6). The intensity of the hemorrhage phenotype in the different treatment groups was confirmed by immunostaining for fibrin. In all mice, fibrin staining was seen in the middle of the wound, where the original clot was not yet replaced by mature granulation tissue (unpublished data). However, in most of the mice, which had Prdx6- deficient inflammatory cells or resident cells or both, intermediate or strong fibrin staining was also seen in the mature granulation tissue underneath the hyperproliferative epithelium (). This reflects the necessity to rapidly seal damaged vessels with a blood clot in order to avoid excessive bleeding. By contrast, fibrin deposition was absent or much less pronounced in wild-type mice with wild-type bone marrow (). At the ultrastructural level (Fig. S4, available at ), damage of endothelial cells as well as morphological abnormalities in some fibroblasts and granulocytes were detected in all animals with Prdx6-deficient cells (Fig. S4, C–F). Again, the most striking phenotype was observed when both hematopoietic and resident cells lacked Prdx6. Collectively, this in vivo experiment revealed that the presence of Prdx6 is required in inflammatory cells as well as in resident cells for blood vessel integrity in wounded skin (shown schematically in ).
#text
HUVEC (Clonetics) were seeded on collagen A (10 μg/ml, Biochrom AG) coated dishes and cultured in endothelial growth medium EGM-2 (Clonetics). Synthetic siRNAs dissolved in RNase-free PBS were combined 1:1 to a final concentration of 50 μM, incubated at 50°C for 2 min in annealing buffer (25 mM NaCl, 5 mM MgCl), and cooled down to room temperature. 3500 HUVEC/well were seeded into 96-well plates. 24 h later they were transfected with Prdx6 or random siRNAs (30 nM each) for 48h. Protein lysates were analyzed by Western blotting for the levels of Prdx6 and β-actin.
For siRNA lipid complex formation, 10 μl growth medium without any additives but with 10 mM Hepes and siRNA (30 nM) were placed into polystyrene tubes and vigorously mixed. After the addition of 10 μl complex medium and the lipid Atufect 01 (1.2 μg/ml; Atugen AG), the components were incubated at 37°C in 95% relative air humidity and 5% CO for 30 min. Thereafter, this mixture was added to the cells, which had previously received 80 μl fresh growth medium.
Cells were grown for 2 d in growth medium including the siRNA, and treated for 7 h with 500 or 750 μM hydrogen peroxide or with 15 or 20 mU/ml glucose oxidase. Cell viability was quantified using the 3-(4,5-dimethylthiazol- 2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) assay ().
Mice were housed and fed according to federal guidelines, and all procedures were approved by the local veterinary authorities (Zurich, Switzerland).
Mice were anaesthetized by intraperitoneal injection of ketamine/xylazine. Two 5-mm full-thickness excisional wounds were made on either side of the dorsal midline (). Complete wounds including 2 mm of the epithelial margins were excised and frozen in liquid nitrogen. Alternatively, wounds were fixed overnight in 95% ethanol/1% acetic acid or in 4% paraformaldehyde (PFA)/PBS, followed by paraffin embedding, or directly embedded in tissue-freezing medium. Sections (6 μm) from the middle of the wound were stained with hematoxylin/eosin (H/E), by the Masson trichrome procedure or with the Masson-Goldner staining kit (Merck), or used for immunofluorescence/immunohistochemistry. Only littermates or at least age-matched animals of the same sex were used for direct histological comparison.
Mice were anaesthetized, shaved, and irradiated with 60 J/cm UVA or with 100 mJ/cm UVB as described by . 24 h later they were killed and the tissue was fixed in 4% PFA. To identify apoptotic “sunburn” cells PFA-fixed paraffin sections were stained with H/E. Alternatively, frozen sections were analyzed by immunofluorescence with an antibody against cleaved caspase-3 (see below). The number of apoptotic keratinocytes per mm of basement membrane was determined by two (UVB) or three (UVA) independent investigators. For this purpose cells were counted in 10–15 independent microscopic fields per mouse or along the entire epidermis present on the section.
Female mice (9–11-wk old) were irradiated (950 radγ) and subsequently 5–7 × 10 bone marrow cells that were flushed from tibiae and femurs of wild-type and knockout male donor mice were intravenously injected. Wild-type and knockout mice received bone marrow from either wild-type or knockout male mice. After transplantation mice were maintained in a laminar flow environment. They received sterilized food and water, which was supplemented with antibiotics (Borgal, 0.1%) for 2 wk. 6 wk after transplantation genomic DNA from blood cells was tested by semi-quantitative PCR for the presence of the Y chromosome () and by Real-time PCR using SYBR green (Applied Biosystems) and primers described elsewhere (). Assays were performed twice in duplicate and evaluated by the Δct-method as described by the manufacturer (Applied Biosystems). 10 wk after transplantation the wound healing experiment was performed. After sacrifice, bone marrow was taken, and the presence of the Y chromosome in DNA of bone marrow cells was verified by semi-quantitative DNA.
Mice were lethally anesthetized with pentobarbital (700 mg/kg) and perfused with 4% PFA in PBS. Wounds were kept overnight in fixation solution, rinsed, and stored in PBS. Before embedding, the wound samples were treated with 2% OsO for 2 h. After washing, they were stained in 1% uranyl acetate, dehydrated through series of graded ethanols and embedded in araldite resin. Semi-thin sections (500 nm) were cut with a glass knife and stained with methylene blue. Ultra-thin sections (30–60 nm) were processed with a diamond knife and placed on copper grids. Transmission electron microscopy was performed using a 902A electron microscope (Carl Zeiss, Inc.).
RNA isolation and RNase protection assay were performed as described (; ). All protection assays were performed at least in duplicate with different sets of RNAs from independent experiments. A murine cDNA fragment (nt 310–555; Accession no. 12805152) and a murine cDNA fragment (nt 257–576; Accession no. 7948998) were used as templates. Other templates were described previously (; ; ; ; ).
Preparation of protein lysates and Western blot analysis were previously described (). A mouse monoclonal antibody against β-actin (Sigma-Aldrich) and a rabbit polyclonal antibody against Prdx6 () were used. Oxidized proteins were detected using the OxyBlot assay kit (Chemicon) according to the manufacturer's instructions.
Sections were incubated overnight at 4°C with the primary antibodies diluted in PBS containing 3% BSA or 12% BSA and 0.025% NP-40. After three washes with PBS/0.1% Tween 20, they were incubated for 1 h with the Cy2- or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, Inc.), washed again and mounted with Mowiol (Hoechst). For immunohistochemistry biotinylated secondary antibodies (Jackson ImmunoResearch Laboratories, Inc.) were used, followed by counterstaining with hematoxylin. We used a mouse monoclonal antibody against α-SMA coupled to FITC (Sigma-Aldrich), rat monoclonal antibodies against MECA-32 or PECAM-1 (CD31) (both from BD PharMingen), a goat polyclonal antibody against pan-keratin (AbCam), and rabbit polyclonal antibodies against cleaved caspase-3 (Cell Signaling), fibrinogen/fibrin (DAKO), or nitrotyrosine (BIOMOL). For p53 immunohistochemistry PFA-fixed paraffin sections in combination with a polyclonal antibody against p53 (Novocastra) were used.
Tissue sections were examined at room temperature by light microscopy using an Axioskop2 photomicroscope (Carl Zeiss, Inc.) and 10×/0.30, 20×/0.50, or 40×/1.30 Plan-Neofluar objectives (all from Carl Zeiss, Inc.). Images were acquired with an AxioCam HRc camera and AxioVision 4.1 software package set (Carl Zeiss, Inc.) and grouped with Adobe Photoshop or Illustrator (Adobe Systems).
The following information is available online as supplemental information: Fig. S1 shows a quantitative analysis of the wound healing process in Prdx6-deficient mice and wild-type controls, including the area of hyperproliferative wound epidermis, wound size, wound closure, and cell proliferation in the wound epidermis. Fig. S2 shows electron micrographs of wound vessels in Prdx6-deficient mice and control littermates at d 8 after injury. Fig. S3 demonstrates the successful generation of bone marrow chimeric mice as demonstrated by semi-quantitative and quantitative PCR for the Y-chromosome using DNA from blood and bone marrow cells. Fig. S4 shows electron micrographs of 7-d wounds of bone marrow-transplanted mice. Fig. S5 shows H/E stained sections from lung and skin of E13.5 Prdx6 embryos and wild-type littermates. Online supplemental material is available at . |
Endochondral ossification is the process by which most long bones of the body are formed (). During endochondral ossification, mesenchymal cells first aggregate to form condensations (). The cells in the center of these condensations differentiate into chondrocytes, forming the cartilaginous template, whereas the undifferentiated cells at the periphery form the surrounding perichondrium (). After formation of the cartilage template, the innermost chondrocytes differentiate into hypertrophic chondrocytes and the cells of the inner layer of the perichondrium differentiate into osteoblasts (; ), forming a bone collar around the cartilaginous core (). The hypertrophic cells secrete a distinct ECM that gradually becomes calcified (). This specialized matrix allows for vascular invasion from the bone collar and the entry of osteoclasts and osteoblasts that degrade the mineralized cartilage matrix and deposit bone (). Apoptosis of hypertrophic cells and the deposition of a matrix rich in type I collagen (Col I) results in two opposing growth plates that allow for longitudinal bone growth in both directions. This process is in contrast to the craniofacial skeleton bones that are formed by intramembranous ossification, where mesenchymal cells directly differentiate into bone without an intermediate cartilage template.
In this paper, we have identified site-1 protease (S1P) as a new player involved in regulating endochondral ossification. S1P (also known as membrane-bound transcription factor protease, site 1) is a proprotein convertase and a key member of the regulated intramembrane proteolysis pathway involved in the unfolded protein response and cholesterol homeostasis (). A role for S1P in cartilage development was shown through the study of the zebrafish mutant (), which has both lipid and skeletal abnormalities. S1P plays a critical role in the processing of the sterol regulatory element binding proteins (SREBP-1a, -1c, and -2; ). SREBPs are membrane-bound transcription factors in the ER and regulate cholesterol and fatty acid biosynthesis and uptake. When cholesterol levels are high, SREBP is retained in the ER membrane as a complex with the sterol-sensing protein SREBP cleavage–activating protein (SCAP) and the retention protein INSIG (insulin induced gene). When cholesterol levels drop, the SREBP–SCAP complex dissociates from INSIG and translocates to the Golgi bodies, where SREBP is sequentially cleaved by S1P and S2P to release the amino-terminal domain of SREBP containing the basic helix-loop-helix leucine zipper region. The basic helix-loop-helix leucine zipper region translocates to the nucleus to bind to cis-acting sterol responsive elements. In a similar fashion, S1P is also involved in the activation of other ER membrane-bound proteins such as activating transcription factor 6 (ATF6; ; ), old astrocyte specifically induced substance (OASIS; ), and cAMP-responsive element binding protein H (CREBH; ), which are transcription factors for the unfolded protein response target genes.
To elucidate the role of S1P in all aspects of skeletal development, we created cartilage-specific S1P knockout mice (S1P) using a promoter. S1P mice die shortly after birth and exhibit severe chondrodysplasia. The cartilage matrix is abnormal in S1P mice with defects in Col II protein secretion and assimilation into the matrix, and endochondral bone formation is completely absent. This is the first example of a defect in a regulated intramembrane proteolysis enzyme that affects cartilage development and endochondral ossification in mice. Deletions of various matrix metalloproteinases (MMPs), such as MMP13 or MMP9 (; ), thought to be important in bone morphogenesis did not abolish endochondral ossification. Thus, S1P is a unique enzyme that plays an integral role in skeletal development.
To produce mice lacking S1P in the cartilage, the Col2-Cre recombinase was used to delete exon 2 of S1P. S1P;Col2-Cre mice were bred with S1P mice to generate conditional knockout S1P;Col2-Cre (S1P) mice. S1P mice exhibit severe chondrodysplasia () with deformed limbs, a protruding tongue (), and a distended belly and die during or shortly after birth. All structural elements are present in S1P mice but are considerably smaller than in wild-type (WT) mice (). shows a detailed analysis of the skeletal elements. The smaller skull in S1P mice, caused by a shortened chondocranial base coupled with foreshortened mandibles and upper jaw bones, appears to be responsible for the protrusion of the tongue. The skull bones show increased sensitivity to potassium hydroxide (used for clearing the tissue during skeletal staining) that resulted in the dissolution of the bones in the back of the skull (, arrow), which remain intact in the WT. The chest cavity is small and presses the internal organs into the abdominal cavity, giving the belly a distended appearance. The smaller chest cavity probably causes acute neonatal respiratory distress and subsequent lethality. The skewed orientation of the limbs, particularly the hind limbs, suggests anomalous articular joint development. Heterozygotes (S1P;Col2-Cre) appear normal and do not show any phenotype (unpublished data).
To confirm that the phenotype in S1P mice is caused by a lack of S1P, we analyzed the expression of the gene in chondrocytes in different developmental stages by in situ hybridization (). is abundantly expressed in the chondrocytes in the developing WT forelimb at all stages (). The heterozygotes show drastic reduction in expression in chondrocytes but show normal bone development, with expression in bone and surrounding tissues intact (). As expected, the homozygous knockout (S1P) does not show any expression in skeletal tissues (). These data suggest that the mutant phenotype could be directly attributed to the efficient knockout of both copies of in the cartilage in S1P mice.
As S1P mice die during or shortly after birth, we conducted a histological analysis of various skeletal elements at different embryonic stages. S1P mice lack an organized hypertrophic zone and endochondral ossification, as shown for the humerus at different stages (). At embryonic day (E) 13.5, no major differences are observed in the morphology of the humerus between S1P and WT mice. At E13.75 (7 h after E13.5), the WT humerus increased in size and the chondrocytes in the center initiated hypertrophy. S1P mice, however, showed only a slight increase in the size of the humerus at E13.75. The transition from proliferating to hypertrophic cells appears to be initiated but does not progress to hypertrophy, though a change in morphology of the cells at the center of the humerus is noticeable. At E14.5, the humerus in WT mice has considerably increased in size and hypertrophy is completely established. In S1P mice, however, the transition from proliferating to hypertrophic cells is still incomplete at E14.5. At E15.5, endochondral bone is observed in WT mice with the establishment of a growth plate. However, at E15.5, S1P mice still do not exhibit an organized hypertrophic zone. Notably, there is no endochondral ossification or growth plate, and, consequently, the humerus is small compared with the WT. In the absence of endochondral bone, however, there is an exuberant growth of the bone collar (, arrow).
At E18.5, it is clear that S1P mice suffer from a lack of endochondral bone formation and a structured growth plate, rather than a delay. At this stage, the exuberant bone collar often impinges into the diaphysis and bisects it. A lack of endochondral ossification is also observed in the hind limbs, sternum, ribs, and spine (Fig. S1, available at ). Although a hypertrophic zone is not seen in S1P mice, they have the ability to make hypertrophic cells, which is more distinguishable in later developmental stages (Fig. S2). In summary, a lack of organized endochondral ossification is observed in all skeletal elements in S1P mice because of the disruption of S1P activity.
Because of a lack of endochondral ossification in S1P mice, we examined Ihh signaling, which is known to be an essential regulator of chondrocyte proliferation and hypertrophic differentiation. Ihh's function is thought to be dependent on lipophilic modifications critical for the spatially restricted localization of Ihh signaling (). In the growth plate, Ihh is distributed in a gradient that is necessary for coordinated chondrocyte proliferation and differentiation (). As cholesterol and fatty acid synthesis is likely to be affected in chondrocytes because of the absence of S1P, it could lead to defects in Ihh signaling. A defect in Ihh signaling is associated with a decrease in chondrocyte proliferation (). Thus, to analyze the nature of Ihh signaling in our model, we labeled proliferating chondrocytes with BrdU at E14.5 (). Notably, at E14.5, analysis of the proliferating BrdU-labeled cells revealed that both WT and S1P mice exhibit nearly identical rates of proliferation (). Moreover, WT and S1P proliferating chondrocytes displayed the ability to exit the cell cycle, as seen by a lack of BrdU incorporation as they progressed toward hypertrophic differentiation ().
The null mutant is also characterized by a lack of expression of its receptor and transcriptional target () and (; ). PTHrP and its receptor PTHrP-R play important roles in the control of hypertrophic differentiation (; ). Therefore, we analyzed the expression of and in S1P mice by in situ hybridization. Expression of both () and () is seen in S1P mice, though the expression of is not completely similar to WT mice and appears slightly reduced. Ihh is also required to establish the bone collar by regulating the differentiation of the perichondrial mesenchymal cells into osteoblast progenitors (). Consequently, null mutants do not form a bone collar. Therefore, we studied bone collar formation at E13.5 by analyzing the expression of (), a major structural protein of the bone matrix specifically expressed by maturing osteoblasts. Notably, both WT and S1P mice displayed a timely expression of in the developing bone collar at E13.5 (). Collectively with and expression, these data suggest that Ihh signaling is intact in S1P mice.
Hypertrophic differentiation begins with the proliferating chondrocytes exiting the cell cycle and beginning to express and to form the prehypertrophic cells, which exhibit a decreased expression of . The fully differentiated hypertrophic chondrocytes are large, no longer express , and are characterized by the expression of and . In S1P mice, hypertrophic differentiation is initiated as chondrocytes exit the cell cycle and no longer incorporate BrdU (). However, under higher magnification, this region in E14.5 S1P mice does not show the same morphology as in the WT (). In the WT humerus (), there is a clear definition of the hypertrophic zone. Above the hypertrophic zone, proliferating chondrocytes are seen organized in groups of columnar chondrocytes. Between these groups, there is an abundant matrix that stains a deep red. In S1P mice (), however, columnar organization of the cells is absent. Chondrocytes are present in a random configuration and are more densely packed. Although a few enlarged cells are present, a distinct hypertrophic zone is absent.
To analyze hypertrophic differentiation in S1P mice, we performed a series of in situ hybridizations on E14.5 humerus serial sections. First, we analyzed the expression of . Runx2 is a powerful positive regulator of chondrocyte maturation into hypertrophy, and null mice exhibit an absence of both endochondral and intramembranous ossification because of an arrest in osteoblastogenesis (; ). By in situ analysis, S1P mice showed an expression pattern for that is similar to the WT in that expression is seen in the perichondrium and presumptive prehypertrophic and hypertrophic cells (). To better define the prehypertrophic and hypertrophic regions, we analyzed the expression of and in prehypertrophic cells and , which is expressed only by hypertrophic cells. In WT mice, the - () and –expressing () prehypertrophic cells and the -expressing () hypertrophic cells form well-defined regions, with the prehypertrophic cells always lying proximal (from the humerus head) to the hypertrophic cells. In S1P mice, however, such well-defined regions are absent. Zones expressing (), (), and () largely overlap each other. Thus, even though , , and are expressed, which indicates hypertrophic differentiation, the characteristic hypertrophic cell morphology and organization is missing in S1P mice.
These results indicate that the loss of S1P does not prevent the molecular program necessary for hypertrophic differentiation. Rather, the loss of S1P appears to affect cell morphology and may affect matrix production. Therefore, we analyzed the expression of the matrix genes and () and and in the E14.5 humerus (). At E14.5, chondrocyte hypertrophy is clearly established in the WT, characterized by the expression of () and the lack of () and () expression in hypertrophic cells. In S1P mice, however, expression of () and () is maintained into the zone of -expressing cells (). Thus, in S1P mice, expression of hypertrophic markers is seen concomitantly with expression of markers for proliferating cells. However, the high levels of and expression seen in the columnar cells in WT mice (, brackets) are absent in S1P mice. Exit from the cell cycle and the expression of and suggest hypertrophic differentiation. But the lack of characteristic hypertrophic morphology and the persistence of , , , and into the -expressing zone suggest that organized transitions between various cell types are missing; the cells do not respond cohesively to differentiation signals and are therefore unable to organize themselves into structurally discrete zones. Finally, we checked for expression of at E14.5 in S1P cells. is normally expressed by the terminal hypertrophic cells in the growth plate (). WT mice showed considerable expression of () in the terminal hypertrophic chondrocytes. However, S1P mice exhibit a dramatic decrease in expression in the humerus at E14.5 (). This decrease is not a reflection of delay in differentiation but is maintained at E15.5 in S1P mice (), which suggests a defect in terminal hypertrophic differentiation. However, expression was observed in the perichondrium of S1P mice at E15.5 presumably in the osteoblasts in the exuberant bone collar. These data suggest that S1P is required for the completion of hypertrophic differentiation and organization of the hypertrophic zone.
As demonstrated in the preceding section, hypertrophic differentiation remains incomplete in S1P mice. Therefore, we analyzed posthypertrophic differentiation events. Hypertrophic differentiation is usually accompanied by the mineralization of the ECM, allowing for vascular invasion in a VEGF-dependent manner followed by the removal of hypertrophic cells by apoptosis (). Extensive mineralization of the ECM, as analyzed by von Kossa staining, was observed in both the E15.5 tibia () and humerus () of S1P mice, providing a prelude to vascular invasion. In S1P mice, however, extensive abnormal mineralization of the chondrocytes in the humerus and to some extent in the tibia is also seen. Abnormal mineralization is often associated with an increase in apoptosis, which leads to the release of apoptotic bodies (). Using a TUNEL assay, we studied apoptosis in the forelimbs of WT and S1P mice at E13.75, 14.5, and 15.5. A considerable increase in chondrocyte apoptosis is observed in both the humerus () and scapula () in S1P mice as compared with the WT at E15.5, but not at E13.75 or 14.5. Next, we performed immunohistochemistry (IHC) for the platelet/endothelial cell adhesion molecule 1 (PECAM-1) antigen to analyze vascular invasion of the mineralized matrix. The WT humerus showed normal vascular invasion demarcating the bone from the nonvascular cartilaginous ends (). PECAM-1 staining is absent within the humerus in S1P mice, suggesting a lack of vascular invasion (); however, it is seen in the bone collar, suggesting a defect only in endochondral bone formation.
Vascular invasion takes place in a VEGF-dependent manner and is needed for the invasion of osteoblastic progenitors and osteoclasts (). Osteoblasts synthesize MMP13 (collagenase) and osteoclasts synthesize MMP9 (gelatinase). The cooperative activity of these two MMPs breaks down the mineralized matrix of the hypertrophic cells and replaces it with bone matrix that is rich in Col I (). Interestingly, both WT and S1P mice demonstrated considerable and expression (Fig. S3, available at ). The presence of -positive osteoclasts in S1P mice was further confirmed by tartrate-resistant acid phosphatase–positive cells in the bone collar (Fig. S3). Because of the lack of endochondral bone development in S1P mice, we analyzed whether S1P mice have a defect in osteoblast formation. Osteoblastogenesis would be expected to be normal in the mutants because of the proper onset of -expressing bone collar () and the expression of (). To extend these findings, we analyzed the expression of , the bone-specific , and , which is vital to bone morphogenesis, in S1P mice (Fig. S3). Expression of , , and is observed in S1P mice, though only in the bone collar because of the lack of endochondral bone. These observations indicate that S1P mice have the ability to produce normal osteoblast progenitors and that the lack of endochondral bone is caused by the inability of the cartilage matrix to support osteoblast invasion.
Several lines of evidence suggest that matrix abnormalities are primarily responsible for the S1P phenotype. First, chondrocytes are more densely packed in S1P skeletal elements. Second, Ihh signaling is intact, as indicated by and expression, although expression appears to be slightly disorganized. Third, it is unlikely that the reduced expression seen in S1P mice is responsible for the lack of endochondral bone, as the targeted inactivation of in mice does not prevent endochondral bone formation (; ). Finally, the increase in chondrocyte apoptosis in S1P mice that is suggestive of an abnormal matrix is also seen in transgenic mice lacking Col II (). S1P may thus play a role in the deposition of a normal cartilage. To assess abnormalities in the matrix, we studied the properties of the Col II protein, which is a major cartilage component. First, we attempted to extract the Col II protein from E17.5 WT and S1P cartilage by pepsin digestion after extraction of the proteoglycans by 4 M Gu-HCl. After pepsin digestion, the extract was concentrated and equal concentrations of protein from each sample were subjected to SDS-PAGE and analyzed by Western blotting for Col II (using the IIF antibody that detects the triple helical domain of Col II) and Col X protein (). We were able to extract the full-length Col II from the WT head (, lane 3) and the rest of its skeleton (, lane 1). Notably, we were not able to extract the full-length Col II from any of the S1P skeletal elements (, lanes 2 and 4). In some preparations, some lower molecular weight proteins with immunoreactivity to the IIF antibody are seen (unpublished data). But we were unable to extract comparable amounts of the full-length Col II protein in any of our attempts. By Western blotting analysis of twofold dilutions of the pepsin-digested material, we determined that we were extracting ∼64-fold less of the full-length Col II protein from the S1P mice as compared with WT (unpublished data). In contrast, we were able to extract comparable amounts of the Col X protein from both WT and S1P mice.
These data suggested that the Col II protein is less abundant or absent in S1P mice. However, we were able to detect the full-length Col II procollagen in the medium of cultured chondrocytes from both E18.5 WT and S1P mice by Western blotting (). Furthermore, by IHC analysis, Col II protein was detected in the humerus of E18.5 WT () and S1P () mice. We also tested for the presence of the Col IIA protein (using the IIA antibody that detects exon 2), which is normally detected in the matrix surrounding chondroprogenitor cells and early immature chondrocytes. The Col IIA isoform was detected in both WT () and S1P mice (). In the WT, detection of Col IIA protein was strongest at the location of the early, immature chondrocytes close to the articular surface. Surprisingly, in S1P mice, Col IIA was detected throughout the length of the humerus (). Therefore, we analyzed whether alternative splicing of the procollagen mRNA from to was normal in S1P mice. For this analysis, we performed in situ hybridization using highly specific 24-mer oligonucleotide probes spanning the exon 1 and 2 junction for analysis and the exon 1 and 3 junction for the analysis. Splicing of both and mRNA was, however, found to be normal in S1P mice (Fig. S4, available at ). These data suggested that Col II synthesis was taking place in S1P mice, although the turnover of Col II is abnormal, as total Col II and Col IIA was detected throughout the length of the humerus. These data also suggested that the organization of the cartilage ECM may be abnormal. Therefore, to assess the structural organization of the collagens in the ECM, we first determined the birefringent properties of the Col II fibrils in both WT and S1P mice. A marked decrease in the birefringence of the collagen network was observed in the E16.5 humerus in S1P mice () as compared with the WT (), which suggests a poorly organized collagen fibrillar network in the cartilage of S1P mice. This reduced birefringence in S1P mice is similar to that seen in mice with inactivation of the gene (). These data together suggest that one of the molecular defects caused by a lack of S1P is the abnormal organization of the Col II protein, which results in an abnormal cartilage that prevents vascular invasion and endochondral ossification in S1P mice.
To pursue this observation further, we performed double-labeled immunofluorescence analysis with the IIA and IIF antibodies () to study the organization and localization of the Col II protein (). In the WT, the secretion and incorporation of Col IIA into the matrix is seen as a well-formed and organized green lattice network. Localization of the collagen triple helical domains is seen as red immunofluorescence, which in the WT is seen as a distinct red lattice network with hardly any yellow colocalization signals from the two antibodies, suggesting that the red lattice is almost entirely made of Col IIB protein. As expected in the WT, the presence of Col IIA is stronger in the matrix surrounding the early immature chondrocytes () and the presence of Col IIB is stronger around the more mature columnar chondrocytes (). In S1P mice (), the organization of the Col IIA lattice network appears normal, though it is also prominent in the matrix surrounding mature chondrocytes (). However, in the matrix surrounding early chondrocytes (), well-formed organization of Col IIB fibrils (red) is considerably reduced and the majority of the signal appears to be caused by colocalization (yellow), suggesting that both antibodies are detecting primarily Col IIA and very little Col IIB. In the matrix surrounding more mature chondrocytes (), the lattice network is primarily green with very little incorporation of Col IIB (red signaling) into the lattice network of the ECM. Most of the Col IIB protein (, arrows) is seen trapped inside the cell. The lack of matrix accumulation can be seen more clearly in cross sections of the ribs (at E18.5) in S1P mice (). Again, primarily colocalization signals (yellow) and the substantial absence of matrix when compared with the WT () is observed, which showed abundant matrix (red) with yellow colocalization signals restricted to the rim of the rib.
These data suggested that in S1P mice, although there is no problem in the expression and secretion of Col IIA and its incorporation into the matrix, the expression and/or secretion of Col IIB is abnormal and results in a matrix with considerably reduced levels of Col IIB and therefore an inability to extract Col II from the matrix (). To understand the mechanism for this defect, we performed electron microscopic analysis of tibia () and humerus (not depicted) from E15.5 WT and S1P mice. In the WT (), the cartilage from the proliferative zone showed the presence of well-formed and abundant, homogenous Col II fibrils surrounding normal chondrocytes. In S1P mice, however (), the distribution of collagen fibril density is irregular and considerably reduced. The ER is drastically enlarged, irregular in shape, and fragmented and also often found filled with granular, homogenous, or heterogenously stained material that also sometimes appears to be crystalline. Similar observations were seen in the chondrocytes/ECM of the articular, hypertrophic, and joint cartilage regions in S1P mice. These data suggest the induction of ER stress in chondrocytes in S1P mice, presumably caused by the excessive demand for complex matrix protein synthesis required for cartilage deposition, and a requirement for S1P activity to alleviate ER stress during matrix protein synthesis and deposition.
In this paper, we have established the importance of S1P to cartilage deposition and endochondral ossification. Null mice with a loss of S1P in all tissues were found to be embryonic lethal at a very early stage and therefore did not provide a suitable avenue to study its contribution to cartilage and bone development (). Our studies on S1P mice have allowed us to focus on the requirement of S1P in chondrocyte differentiation and the genesis of a normal endochondral skeleton. Cartilage-specific disruption of S1P in S1P mice results in total lack of endochondral bone in all skeletal elements that exhibit longitudinal bone growth through endochondral ossification.
The lack of endochondral bone seen in S1P mice is also seen in () and null mice (; ). In both these mutants, endochondral bone development is hampered because of a defect in osteoblastogenesis. Both and null mice also suffer from varying degrees of an abnormal chondrocyte differentiation program (; ). In S1P mice, we initially expected that the phenotype could have arisen because of a defect in the Ihh signaling pathway. However, our data suggest that S1P mice have an intact Ihh signaling pathway, as indicated by the expression of and in S1P skeletal elements. However, the mislocalization of –expressing cells leaves open the possibility that S1P has an effect on the activity level and/or diffusion of Ihh. The morphological phenotype that sets S1P mice apart from and null mice is that S1P mice exhibit an exuberant bone collar showing normal osteoblastogenesis. Elimination of S1P results in an uncoupling of normal endochondral bone formation from cortical bone formation. Thus, like Ihh and Runx2, S1P is a major positive regulator of endochondral ossification. Expression of is observed in the perichondrium in the WT. However, in S1P mice, S1P activity would be expected to be absent in the perichondrium because of active Col2-Cre activity in these cells. These data suggest that S1P is not necessary for cortical bone formation. Mice with double knockouts of the transcription factors L-Sox5 and Sox6 also exhibit a lack of endochondral ossification with development of a thick cortical bone (). However, it is unlikely that this pathway is affected in S1P mice, as they exhibit abundant expression of L-Sox5 and Sox6 (unpublished data).
The lack of endochondral bone in S1P mice appears to be caused by an inability of the blood vessels to invade the abnormal, mineralized cartilage in spite of apparently adequate expression. According to the tenets of organogenesis, where development is guided by tissue interactions, the developmental history of a tissue plays an important role in its ability to respond to instructive cues (). Thus, it would be expected that an abnormal cartilage would constrain its ability to respond to instructions from the perichondrium and/or vascular tissue. The genesis of a normal matrix is necessary for proper hypertrophic differentiation and endochondral ossification. Our data suggest that the defects in hypertrophic differentiation and endochondral ossification can be traced to an abnormal matrix in S1P mutants. In S1P mice, the chondrocytes are densely packed and randomly oriented with very little matrix between them. The increased chondrocyte apoptosis seen in S1P mice is also highly indicative of an abnormal cartilage that is unable to support chondrocyte survival. This is similar to the increased apoptosis seen in transgenic mice lacking Col II (). The absence of a proper cartilage could hinder not only the supply of nutrients to the avascular ECM () but also have profound effects on integrin-mediated pathways and distribution of growth factors (). The inability to extract normal amounts of full-length Col II from cartilage, the decreased birefringence of Col II network, the lack of the Col IIB lattice network, and the considerable reduction in collagen fibrillar density as seen by electron microscopy all attest to the fact that the cartilage is abnormal in S1P mice. S1P mice also exhibit abnormal spine development in that the intervertebral discs are not well formed and lack the gelatinous nucleus pulposus in the center. This is also seen in transgenic mice with a null mutation in the gene ().
An abnormal cartilage ECM could mask angiogenic signaling by VEGF in spite of the adequate expression that is seen in S1P mice. Given the abnormal characteristics of the cartilage matrix, even if VEGF is not masked, the MMPs may not be able to process and break down the alien matrix in S1P mice. An inability to degrade the matrix is suggested by the persistence of Col IIA in the S1P cartilage, as Col IIA is seen throughout the length of the humerus. These observations are highly suggestive of a lack of structurally intact cartilage matrix in S1P mice that is normally a forerunner for endochondral bone development. Some of the phenotypes seen in S1P mice also bear resemblance to the phenotypes seen in mutants for other matrix proteins. For example, a lack of columnar hypertrophic chondrocytes, disorganized growth plates, chondrodysplasia, and incomplete nucleus pulposus are seen in mice lacking perlecan (; ), Agc1 (; ), and/or link protein (). These observations clearly attest to the importance of a properly organized matrix for chondrocyte differentiation. However, these mutations exhibit only reduced or delayed endochondral ossification. Thus, it is noteworthy that only the disruption of the gene () results in a complete lack of endochondral ossification as seen in S1P mice. These data suggest a convergence of S1P activity with a property of Col II in relation to the cartilage matrix.
Thus, our paper shows that S1P is essential to endochondral ossification. The much accepted role of S1P is that of a proprotein convertase required to activate the transcription factors SREBPs, ATF6, OASIS, and CREBH, the latter three playing vital roles during ER stress response. Given that the knockout of the SREBP molecules has not exhibited any known defect in bone morphogenesis in mice that escape embryonic lethality (; ), it seems unlikely that the phenotype in S1P mice is manifested through SREBPs. Furthermore, in the zebrafish phenotype, knockdown of SCAP (the lipid sensor) results only in strong lipid phenotypes with normal cartilage (). These data suggest that the abnormal cartilage phenotype in S1P mice is mediated by lipid-independent pathways. Given the engorgement and fragmentation of the ER in S1P mice, it seems likely that S1P is required to activate ER stress signaling (ERSS) in response to the hyperproduction of complex cartilage matrix proteins seen in chondrocytes. Unlike other studies (), it is noteworthy that the ER stress observed in S1P mice is not caused by the introduction of any mutant matrix protein but by the inability of the chondrocyte to respond to ER stress because of a lack of S1P activity. Based on these data we propose that the ER stress response is part of the vital protein processing machinery in normal chondrocytes and is essential for normal matrix production (). In a normal chondrocyte differentiation program, chondroprogenitor cells do not require ER stress response because of the relatively low level of protein synthesis and secrete Col IIA into the matrix. However, on differentiation into chondrocytes, to alleviate the stress caused by the high demand for complex matrix protein synthesis, the chondrocytes initiate ERSS, a mechanism that requires S1P to activate ATF6, OASIS, or CREBH (currently the three known ER stress signal transducers that require activation by S1P). This allows for the deposition of normal cartilage followed by endochondral ossification. In S1P mice, chondroprogenitor cells do not show any defect in the secretion of Col IIA either because of the level of Col IIA synthesis or the different dynamics of Col IIA secretion. However, upon differentiation into chondrocytes and a switch to the alternatively spliced form Col IIB and the concomitant increase in protein synthesis and secretion, S1P mice are unable to initiate the ER stress response because of an inability to activate the appropriate transcription factors in the absence of S1P. This results in the engorgement and fragmentation of the ER followed by poor Col IIB secretion into the matrix, which results in the matrix acquiring the properties typically seen on disruption of the gene. Other cell types, such as the periosteal cells, which lack S1P but do not require the same ER stress response, are able to synthesize bone, as seen in S1P mice. Thus, chondrocytes in this respect are like the differentiating plasma cells, which turn on the unfolded protein response to alleviate ER stress and optimize antibody secretion (), or exocrine glands, where the ER stress response is necessary for the full development of the secretory machinery ().
In the zebrafish S1P mutants, ER stress and unfolded protein response were reported to be normal based on in situ hybridization to two mRNA sequences known to be downstream of ATF6. In light of recent discoveries, this assessment may have been misleading. First, the analysis in mutants involved the in situ analysis of the full-length mRNA and not the specific splice form necessary for ER stress response (; ). Second, the presence of immunoglobulin heavy chain binding pretien (BiP) chaperone does not ensure an adequate ERSS in chondrocytes, and BiP can be regulated by other transcription factors such as NF-Y (). In preliminary experiments, we also detect BiP expression in both WT and S1P mice (unpublished data). Lastly, besides ATF6, at least two other ERSS transcription factors, OASIS and CREBH, are regulated by S1P. Collectively, these observations suggest the requirement of a new specialized pathway for regulation of matrix protein synthesis in chondrocytes. We are currently investigating these possibilities using our mouse model to provide a more complete picture of the utility of S1P in ER stress response, matrix synthesis, and endochondral bone formation.
To generate S1P mice, S1P mice (mice homozygous for the S1P allele in which the exon 2 of S1P is floxed, provided by J.D. Horton, University of Texas Southwestern Medical Center, Dallas, TX; ) were bred with Col2-Cre (where the Cre recombinase is under the influence of the cartilage-specific promoter, limiting its expression primarily to cartilage; a gift of D. Ornitz, Washington University School of Medicine, St. Louis, MO; ) transgenic mice of the C57BL/6J strain to produce S1P;Col2-Cre mice (mice heterozygous for the S1P allele and positive for the Col2-Cre transgene). The S1P;Col2-Cre mice were bred with S1P mice to generate S1P mice. Genotypes were verified by PCR analysis of tail-derived DNA.
The isolation of collagens was done from E17.5 WT and S1P mice using a modification of previously published protocols (). In brief, E17.5 embryos were skinned and eviscerated, and the head was processed separately from the rest of the body. The spine/limbs and head were frozen and crushed, and the proteoglycans were extracted with 4 M Gu-HCl in 0.05 M Tris-HCl, pH 7.5, containing 0.01 M EDTA and a 1× complete protease inhibitor cocktail (Roche), for 36 h at 4°C. The cartilage residues left after proteoglycan extraction were washed with cold water and digested with 1 mg/ml pepsin in 0.5 M acetic acid for 24 h at 4°C. The digest was then concentrated with centricon30 (Millipore) to ∼250 μl, dialyzed against 0.05 M Tris-HCl, pH 7.5, buffer containing 0.4 M NaCl and 5 mM EDTA, and stored at −20°C. Equal concentrations of protein preparations (determined by Bio-Rad Laboratories protein assay) from WT and S1P mice were separated by SDS-PAGE followed by Western blotting for Col II (using the previously described rat antibody IIF against bovine Col II triple helical domain; ) and Col X proteins (EMD). The molecular weight marker used in SDS-PAGE was BenchMark prestained protein ladder (Invitrogen).
To study the biosynthesis and secretion of the Col II protein in vitro, primary cultures from WT and S1P mice were established. Chondrocytes were isolated from the sternum and ribs of WT and S1P mice as described previously (). In brief, rib cages were taken from WT or S1P mice and treated with 2 mg/ml pronase (Roche) at 37°C for 30 min followed by 3 mg/ml collagenase (Roche) treatment in the presence of 8% CO for 1.5 h at 37°C or until all soft tissues are detached from the cartilage. The cartilage was then separated from the soft tissues and treated again with 3 mg/ml collagenase for 4 h at 37°C in the presence of 5% CO. The digest was filtered through a cell strainer to remove undigested bony parts and the cells were then pelleted by centrifugation, washed with 1× PBS, and plated at a density of 2.5 × 10 per cm in DME with 10% heat-inactivated serum, 2% penicillin/streptomycin (Sigma-Aldrich), and 0.25 mM sodium ascorbate. 50 μl of media from the WT and S1P chondrocyte cultures was harvested at regular intervals and separated by 7.5% SDS-PAGE (Bio-Rad Laboratories), and the type II procollagen was detected by Western blotting using the IIF antibody.
Whole-mount skeletal staining of embryos by alcian blue and alizarin red was performed as described previously (). For histological and in situ hybridization analyses on sections, embryonic tissues were collected at various embryonic time points and processed and sectioned as described previously (). Histological analyses were done primarily by Safranin O, Fast green, and hematoxylin staining. Detection of mineralization was performed by staining with 1% silver nitrate as per the von Kossa method, followed by counterstaining with Methyl green. In situ hybridization analyses were performed using S-labeled riboprobes as described previously (). All in situ probes with the exception of Agc1 were provided by F. Long (Washington University School of Medicine) and have been described by and references therein. The Agc1 in situ probe (a gift of E. Vuorio, University of Turku, Turku, Finland) covering the IGD sequence of the mouse Agc1 has been described previously (). The full-length S1P and the S1P exon 2 in situ probes were derived from a full-length cDNA clone (American Type Culture Collection). For BrdU analyses, pregnant females were injected with BrdU as described previously (), and proliferation of chondrocytes was analyzed using a kit (Invitrogen). For PECAM-1 IHC, embryos were fixed using 4% formaldehyde followed by 30% sucrose infiltration. 10-μm cryostat sections were derived from tissues embedded in OCT (Tissue-Tek; Thermo Fisher Scientific) and IHC was performed using a monoclonal rat anti–mouse PECAM-1 antibody (BD Biosciences). Detection of PECAM-1 was done using HRP-conjugated anti–rat IgG and DAB. Methyl green was used for counterstaining. Detection of apoptosis was performed by a TUNEL assay using the in situ cell death detection kit (Roche) according to the manufacturer's instructions. Nuclei were counterstained with DAPI and sections were examined by fluorescence microscopy. Tartrate-resistant acid phosphatase–positive cells were stained using standard procedures with methyl green as counterstain. Study of the birefringence of collagen fibers by polarized light microscopy was done on 5-μm-thick paraffin-embedded sections as described previously (). Detection of total Col II protein by IHC was done on 5-μm-thick paraffin-embedded sections at E15.5 and 18.5 stages using the IIF antibody and procedure described previously (), with the exception that detection was done using an HRP-conjugated goat anti–rat secondary antibody and a DAB substrate (Invitrogen). Detection of Col IIA protein was done similarly with the previously described rabbit antisera against recombinant human type IIA-GST (IIA; ; ) and an HRP-conjugated goat anti–rabbit secondary antibody. Double-labeled immunofluorescence with IIF and IIA antibodies was performed as described previously () on paraffin embedded sections, with the exception that the IIA and IIF antibodies (and their corresponding secondary antibodies) were used sequentially with the IIA antibody and its secondary antibody applied first. The secondary antibodies used were goat anti–rabbit Alexa Flour 488 (Invitrogen) and goat anti–rat Alexa Fluor 546 (Invitrogen). Images for double immunofluorescence were collected with a 60×, 1.4 NA oil immersion objective using either a scanning laser confocal microscope (for E18.5 ribs cross sections; MRC600; Bio-Rad Laboratories) mounted on a microscope (Eclipse E800; Nikon) and LaserSharp 2000 software (Bio-Rad Laboratories) or a camera (2000R Fast1394; Retiga)and Q Capture Pro (for limbs; QImaging). All other sections (in situ/histology) were viewed with a microscope (BX51; Olympus) using a 10× objective (all in situ analysis) or 20 and 40× objectives (for stained sections or IHC), and images were captured with the digital camera (DP70; Olympus) using DP controller software (Olympus). For in situ analysis, pictures of hybridization signals were hued red and superimposed on toluidine blue–counterstained images using Photoshop (Adobe).
Fig. S1 shows the lack of endochondral ossification in the femur, ribs, sternum, and spine of S1P mice. Fig. S2 shows the abnormal hypertrophic differentiation in S1P mice. Fig. S3 shows normal and expression and normal osteoblastogenesis in S1P mice. Fig. S4 shows normal mRNA splicing in S1P mice. Online supplemental material is available at . |
Changes in the regulation and expression of key molecules of the actin cytoskeleton contribute in a major way to differences between metastatic and non-metastatic cancer cells (). Cofilin, and its regulatory proteins, are differentially expressed and regulated in metastatic cancer cells and the activity status of the cofilin pathway is directly correlated with metastasis in mammary tumors (, , ,). Cofilin is a small ubiquitous protein (∼19 kD) that is able to bind both monomeric (G) and filamentous (F) actin (; ). By severing actin filaments, cofilin increases the number of filament ends for polymerization and depolymerization (; ; ). Cofilin can be regulated through different upstream effectors; LIM 1 and 2, and TES 1 and 2 kinases phosphorylate cofilin on the serine 3 residue, thus rendering it inactive (; ; ); whereas type 1, 2A, 2B, slingshot, and chronophin phosphatases dephosphorylate cofilin (; ; ; ; ). In addition, cofilin is inhibited when bound to phosphatydalinositol-4,5-bisphosphate (PIP2) (, ). In vivo studies suggest that in metastatic tumor cells, PLC-mediated hydrolysis of PIP2 can release cofilin from this complex, thereby activating it (; ).
Cofilin is a key player in regulating the dynamics of the actin cytoskeleton of migrating cells both in vivo and in vitro. Cultured cells with low cofilin levels show defects in actin polymerization and depolymerization (, ; ). The direct stimulation of cofilin by uncaging in vivo demonstrates that the local activation of cofilin causes free barbed end formation, and can define the site of actin polymerization, protrusion, and cell direction (). Consistently, PLC and cofilin are required for the initiation of protrusions toward an EGF gradient (, ). All of these considerations suggest that cofilin is a key regulator of directional migration, chemotaxis, and cell polarity.
The ability of cofilin to affect actin dynamics specifically at the leading edge is amplified by two factors. First, cofilin activity is spatially restricted at the leading edge because this is the only region in crawling mammalian cells where filaments are not saturated with tropomyosin; filaments containing tropomyosin are resistant to cofilin severing, thereby concentrating the activity of cofilin to the leading edge (; ). Second, cofilin severs actin filaments to generate free barbed ends which elongate forming newly polymerized filaments that are preferred by Arp2/3 complex for dendritic nucleation. Cofilin stimulates Arp2/3 activity by producing these filament ends (). This synergistic amplification of Arp2/3 complex's dendritic nucleation activity by cofilin has been observed both in vitro () and in vivo (). This synergy may explain the location and the bias of the dendritic nucleation activity of Arp2/3 complex toward the barbed ends of newly polymerized filaments (; ; ).
In this study, we investigated the effects of inhibiting the expression of cofilin in mammary tumor cells. Our goal was to determine the cofilin knockdown phenotype in enough detail to understand the role of cofilin in protrusion dynamics, including the effects of cofilin on stability and directionality of lamellipodial protrusions in metastatic cells in particular.
Cofilin is the dominant isoform compared with the actin depolymerizing factor (ADF) in several cells (), including MTLn3 cells (, ). To suppress cofilin expression we used siRNA that specifically targets cofilin, but not ADF. Using this sequence, Western blot analysis showed that the cofilin levels were lowered by an average of 95% in MTLn3 cells, as compared with control (oligofectamine only) treated cells at 36 h after transfection (); reversion to normal levels was observed by 96 h (). Because the antibody used for Western blotting in these experiments recognizes both cofilin and ADF, the small amount of protein remaining in cells after the siRNA treatment is consistent with the small amount of ADF relative to cofilin. We tested two other siRNA sequences as negative controls: a scrambled version of the cofilin siRNA, and a non-silencing bacterial siRNA, which does not have target sequences in MTLn3 cells. Using Western blotting and immunofluorescence, both siRNAs did not cause any change in the cofilin expression levels, whereas the cofilin siRNA significantly reduced cofilin expression ().
Control, scrambled, and bacterial siRNA treated cells (, and d, respectively) showed similar morphology to that of untreated cells, whereas cofilin siRNA treated cells (referred to hereafter as CF KD) exhibited an elongated morphology (, b). Using FACS analysis, we also tested the viability of the control and CF KD cells and found that at 36 h after siRNA transfection, the majority of cells were viable (unpublished data). Therefore, all further experiments reported in the paper were performed 36 h after siRNA transfection, when maximum suppression of cofilin expression was observed (). It is important to note that suppression of cofilin did not affect the expression of other factors that are important in the regulation of the actin cytoskeleton, such as WAVE 2 and N-WASP expression (Fig. S1, C and D; available at ).
#text
MTLn3 and MTC cells (rat mammary adenocarcinoma cell line) were maintained in α-MEM (GIBCO BRL) with 5% FBS, as described previously (; ). For the Rac rescue experiment, 18 h after administration of the cofilin siRNA, cells were transfected with the GFP-RacQ61L plasmid (a gift from Dr. Klaus Hahn; University of North Carolina, Chapel Hill, NC) for 18 h before cells were analyzed (36 h in total after siRNA transfection).
The non-silencing siRNA (AATTCTCCGAACGTGTCACGT), designed against a bacterial target sequence and the cofilin siRNA designed against (AAGGTGTTCAATGACATGAAA), were purchased from QIAGEN. The scrambled siRNA (AAGGTGTCTAATGACATAAAG), purchased from Ambion, was generated by scrambling the cofilin siRNA and blasted to assure the absence of any possible target sequence. Cells were transfected with the siRNA in the presence of oligofectamine (Invitrogen). The transfection was terminated after 4 h by using α-MEM containing 15% FBS. For the DB KD, both siRNAs were added simultaneously to the cells. The p34 siRNA was designed against the following sequence (AAGGAACTTCAGGCACACGGA). All experiments in the paper were performed 36 h after siRNA transfection.
Whole cell lysates were prepared as described previously (). Anti-p34 antibody (Upstate Biotechnology), anti-Arp3 antibody (Santa Cruz Biotechnology), and anti-actin antibody were purchased from Sigma-Aldrich. Anti-cofilin (chicken IgY) antibody was raised against purified recombinant full-length rat cofilin. For N-WASP and WAVE2, specific monoclonal Ig rabbit antibodies were used (generous donation from Dr. Tadaomi Takenwa's Lab; Tokyo University, Tokyo, Japan). GTPases were pulled down from MTLn3 lysates, and their activity was detected using Western blotting as described ().
Cells were fixed and stained using rhodamine-phalloidin. Actin analysis was performed as described previously (). In brief, cells were double-blind scored and subdivided into three categories: cells exhibiting normal F-actin, prominent F-actin structures (thicker stress fibers); and F-actin aggregates (cells containing F-actin aggregates). Control cells were analyzed in comparison to representative control cells, and CF KD cells were scored in comparison to representative CF KD cells.
Time lapse video microscopy images were processed using Image J software. Cells were either traced in ImageJ, or using DIAS. Cell shape () and motility parameters (–
) were then calculated using DIAS, as described in (), and plotted in Excel. For , unwrapping was done by conversion from polar to rectangular coordinates (centroid of the cell being the origin). The unwrapped points are plotted with (X1,Y1) coordinates, where X is the point of the boundary of the shape and Y is the angle of the vector connecting the centroid to the boundary point.
The length/width ratio (L/W) was calculated using ImageJ, as follows: length was scored as the Feret's diameter of the cell [the longest distance between any two points along the boundary] (, black line); width was scored as the secondary axis of the best fit ellipse of the circled cells (, white line).
The micropipette assay was performed as described previously (). Time-lapse series were taken using 20× objectives on a microscope and analyzed using Image J. Cells were divided into two areas, front and the back of the stimulated cells, defined in reference to shape of the elongated-CF KD cells. The front was designated as the leading edge of the elongated cell, while the back refers to the elongated tail of the cell. In case of control cells, which lack distinctive elongated front and back polarity, polarized cells were chosen to be consistent with the shape of the CF KD cells and also compared with control cells chosen at random. The EGF upshift assay was performed as described previously (). In brief, cells were starved in L15 medium (GIBCO BRL) supplemented with 0.35% BSA for 3 to 4 h. Cells were then stimulated with a bath application of 5 nM EGF (Invitrogen), treated at 37°C.
Fig. S1: analysis of the cofilin siRNA shows high efficiency in suppressing cofilin levels with specificity. Fig. S2: the directionality and persistence of CF KD cells can be rescued by reexpression of Human cofilin plasmid. Fig S3: perturbing cofilin levels by overexpression of LIMK in MTLn3 changes their motility behavior. Fig S4: cofilin knockdown suppresses EGF-induced protrusion, but does not affect the presence of the lamellipodium. Fig. S5: the distribution of myosin II isoforms in cofilin knockdown cells. Video 1: control MTLn3 cells exhibit random walking behavior. Video 2: cofilin siRNA knockdown cells show a directional motility behavior. Video 3: expression of the WT(ΔH) cofilin plasmid rescues the directional walking behavior characteristic of the cofilin siRNA knockdown cells. Video 4: upon EGF stimulation, control MTLn3 cells exhibit lamellipod extension and accumulation of F-actin at the leading edge of the lamellipod. Video 5: cofilin siRNA knockdown suppresses EGF-induced F-actin assembly and protrusion in MTLn3 cells. Video 6: expression of the WT(ΔH) cofilin plasmid rescues the suppression of the EGF-induced F-actin assembly and protrusion observed in the cofilin knockdown cells. Online supplemental material is available at . |
Palytoxin is a potent marine toxin that impairs ATPase activity of the Na/K pump and simultaneously increases cation permeability of mammalian cells (for reviews see ; ). Two lines of evidence initially established that palytoxin interacts with the Na/K pump itself. First, palytoxin elicited ouabain-sensitive cation flux in yeast cells (which lack endogenous Na/K pumps) only after expression of both Na/K-ATPase α and β subunits, and not of either subunit alone (). Second, incorporation of in vitro–translated Na/K-ATPase α and β subunits into lipid bilayers permitted palytoxin to open cation channels with a unitary conductance of ∼10 pS (), similar to that of the relatively nonselective channels opened by palytoxin in the surface membrane of mammalian cells (; ). Furthermore, scanning cysteine accessibility measurements (for review see ; see also ) have demonstrated that several positions along the fourth, fifth, and sixth transmembrane helices of the Na/K pump are water accessible in palytoxin-bound “pump-channels.” Amino acids at some of those positions are homologous to ion-coordinating residues in crystal structures of the related Ca-ATPase (; ), and had been assigned cation-binding roles in the Na/K pump on the basis of mutagenesis results (; ). It appears, therefore, that not only is the Na/K pump the target of palytoxin, but the ion pore of the palytoxin-bound pump-channel comprises at least part of the pathway normally negotiated by the pumped Na and K ions.
Additional evidence suggests that current flow through the ion channels opened by palytoxin is controlled by two gates in series that normally alternately regulate access to the ion-binding sites deep within the Na/K pump (; ). Thus, the open probability of palytoxin-bound pump-channels can be modified by replacing extracellular Na ions with K ions, or by altering the intracellular [ATP]. Because these are the physiological ligands of the Na/K pump that drive the conformational changes that underlie its Na/K transport cycle, their influence on the gating of palytoxin-bound pump-channels supports the use of palytoxin as a tool for probing the Na/K pump's ion translocation pathway.
Here we present the results of experiments to further characterize that pathway in Na/K pumps of squid giant axons. Extracellular, but not intracellular, application of palytoxin was shown long ago to depolarize squid axons by rendering them leaky to Na and other small monovalent cations (; ). We first verified that palytoxin-bound pump-channels in outside-out patches excised from the neurons that give rise to the squid giant axons displayed characteristics comparable to those observed in mammalian cells. We then exploited those characteristics to examine both unidirectional and net Na ion movement through palytoxin-bound pump-channels in voltage-clamped, internally dialyzed squid giant axons. Although it is well known that, in its E1P conformation, the Na/K pump is capable of binding up to three Na ions and of occluding them inside the protein core (for review see ), evidence suggests that palytoxin stabilizes an E2P-like conformation (; ; ) that, during the normal transport cycle, occludes two K or Na ions (; ). Interestingly, we found that the Ussing flux ratio exponent, ', was close to 1.0 for palytoxin-bound pump-channels in all experimental conditions tested. We speculate how the characteristic side-by-side arrangement of the ion-binding sites in this family of P-type ATPases might result in an ion pathway that, when both gates are open, behaves like an ion channel rarely occupied by more than one ion.
Neuronal cell bodies of the squid, were dissociated following and . In brief, the giant fiber lobe dissected from the stellate ganglion was incubated for 40–45 min in filtered sea water containing 10 mg/ml of protease Type XIV (Sigma-Aldrich). After several washes with fresh sea water, the lobe was placed in culture medium and the cells were dispersed by gentle mechanical disruption with a fire-polished glass pipette. Dispersed cells were seeded onto glass coverslips that had been immersed in ethanol, flamed, and treated with poly-L-lysine overnight (1 mg/ml in 0.15 M Trizma, pH 8.5). The cell culture medium was Leibovitz L-15 (GIBCO BRL) supplemented with (in mM) 263 NaCl, 4.6 KCl, 9 CaCl, 50 MgCl, 2 HEPES, 2 L-glutamine, 6% FBS (GIBCO BRL), 50 U/ml penicillin, and 50 mg/ml streptomycin; pH 7.8 with NaOH. Cells were incubated at 17°C and used for recording within the first week of culture.
Currents in outside-out patches () were recorded at room temperature (19–23°C) using pipettes pulled from borosilicate glass (PG52151–4; WP Instruments) with a horizontal puller (P90; Sutter Instruments) and with tips coated with Sylgard. When filled with 500 mM Na solution, the pipette had a resistance of 1.5–12 MΩ for macroscopic, or 7–40 MΩ for single-channel current recordings. Currents were recorded with an Axopatch 200B amplifier, filtered at 1 kHz (8 pole Bessel), and sampled at 5 kHz using a Digidata 1322A A/D converter and pClamp 9 software, and also continuously acquired in parallel at lower rates with a Minidigi 1A and Axoscope 9 (Axon Instruments, Inc.). Single-channel current records were digitally filtered at 100 Hz and baseline drifts were subtracted using Clampfit 9.
External solutions used for measuring single-channel conductance and evaluating concentration dependence of palytoxin action in patches contained (in mM) ∼520 or ∼560 sulfamic acid, 460 or 500 NaOH (as specified in figure legends), 10 CaCl, 40 Mg(OH), and 10 HEPES; NaOH was replaced with equimolar KOH or -methyl--glucamine (NMG) in Na-free solutions. The internal (pipette) solution contained (in mM) 470 sulfamic acid, 500 NaOH, 10 phenylpropyltriethylammonium (PPTEA)-SO, 10 MgCl, 5 Tris-ATP, 10 EGTA, and 20 HEPES (for most single-channel recordings, 9 MgCl was replaced with 9 Mg-sulfamate to improve signal-to-noise ratio). For determination of biionic reversal potentials, external solutions contained (in mM) 540 sulfamic acid, 550 NaOH or tetramethylammonium (TMA)OH, 1 CaCl, 2 MgCl, and 10 HEPES, and the internal solution contained (in mM) 520 sulfamic acid, 550 NaOH, 10 PPTEA bromide, 2 MgCl, 1 Tris-ATP, 10 EGTA, and 10 HEPES. Other external and internal Na (or TMA) concentrations were obtained by isoosmotic replacement of Na (or TMA) sulfamate with sucrose. The osmolality of all solutions in patch recording experiments was 970–1,000 mOsm kg, their pH was 7.4, and all external solutions contained 0.2 μM TTX.
The standard external solution contained (in mM) 400 Na isethionate, 75 Ca sulfamate, 0.1 Tris EDTA, 5 Tris HEPES (pH 7.7), 1 3,4-diaminopyridine, and 0.2 μM TTX. In Na-free external solutions, Na isethionate was replaced with 400 mM NMG sulfamate or 400 K sulfamate; the Na and NMG solutions were mixed to vary [Na], and external solutions containing both Na and K were obtained by mixing or by adding K from a 0.5M K-sulfamate stock solution. The internal dialysate contained (in mM) 100 Na HEPES, 20 NMG HEPES, 50 glycine, 50 PPTEA-SO, 5 dithiothreitol, 50 BAPTA NMG, 10 Mg HEPES, 5 MgATP, 5 NMG phosphoenolpyruvate, and 5 NMG phosphoarginine, pH ∼7.5, and Na as needed. The osmolalities of dialysate and extracellular solutions were adjusted to be the same, within 1%, and were ∼930 mOsm kg.
Simultaneous measurements of membrane current and Na efflux in voltage-clamped, internally dialyzed giant axons were as previously described (; ). A 23-mm segment of squid giant axon, cannulated at both ends, was threaded with a cellulose acetate capillary rendered porous to low molecular weight solutes to allow internal dialysis, a blackened Pt wire for passing current, and a 3 M KCl–filled glass capillary electrode for recording voltage. The voltage difference between this internal electrode positioned in the middle of the axon segment and an external flowing 3 M KCl reference electrode positioned near the solution outflow was controlled by a stable, low noise voltage clamp (). The experimental chamber consisted of two lateral pools containing the cannulated axon ends and a central pool, physically separated by grease seals. The central pool was superfused with the external experimental solution kept at 21–22°C. The end and central pools were kept at the same potential by two ancillary voltage clamp circuits to prevent current flow between them. The superfusate from the central pool was collected and the efflux of Na across the axon membrane in the experimental region was measured by either liquid scintillation or gamma counting. Where shown, data points and error bars indicate the mean ± SEM.
The supplemental material is available at . Fig. S1 demonstrates that TTX-sensitive current and Na efflux are equivalent in a squid axon under zero-trans conditions, confirming that they emanate from the same controlled area of axon membrane. Fig. S2 illustrates simultaneous measurements of current and Na efflux flowing through palytoxin-opened Na/K pump-channels during four consecutive applications of palytoxin to a single axon, at three different membrane potentials.
We first examined characteristics of palytoxin-bound squid Na/K pump-channels in membrane patches excised from enzymatically dissociated giant-fiber lobe neurons, the neurons in the stellate ganglion whose axons fuse to form the giant axons. In outside-out patches, with 500 mM Na on both sides of the membrane and 5 mM ATP in the pipette, exposure to picomolar concentrations of palytoxin allowed individual Na/K pump-channel transformation events to be detected as tiny inward current steps at negative membrane potentials (). Because palytoxin remains tightly bound, the pump-channels continued to open and close long after patches were washed free of unbound toxin (). This allowed measurements of current steps at several voltages () and construction of single-channel current–voltage curves (e.g., ). Single pump-channel conductance measured in this way in five patches averaged 7.4 ± 0.2 pS.
With symmetrical [Na], and ATP in the pipette, exposure of outside-out patches to stepwise increasing concentrations of palytoxin caused a progressive inward current increase at −50 mV, reflecting eventual transformation of every Na/K pump in the patch into a cation channel (). Complete replacement of external Na with NMG after that transformation caused a large outward current shift (), due to the slower permeation of the larger NMG cation through palytoxin-bound pump-channels. Current flowing through the palytoxin-bound pump-channels was estimated by appropriate subtraction of currents averaged over the final 5 ms of 50-ms steps to voltages between −100 and +100 mV (in 20-mV increments), in external Na and in external NMG, before () and after () exposure to palytoxin. The resulting palytoxin-induced current is shown plotted against voltage in . From the large negative shift of reversal potential (Δ = −94 ± 2 mV, = 3) upon replacing external Na with NMG, the relative permeability / was calculated to be 0.024 ± 0.002. The smaller outward current at large positive potentials suggests that extracellular NMG impaired the outward flow of Na ions through palytoxin-bound pump-channels.
of 74 ± 21 pM ( = 6). Current decayed slowly after palytoxin removal (not depicted) and, in Na solution, followed a single exponential time course (τ = 770 ± 260 s, = 3). But current decay upon replacing external Na with K was faster and required two exponentials with time constants τ = 40 ± 10 s and τ = 400 ± 120 s (with fractional amplitudes = 0.55 ± 0.05 and = 0.45 ± 0.05, = 3).
One test for multiple ion occupancy of single-file pores is to examine their reversal potentials under biionic conditions, but with different values for the identical concentrations of the two permeant ions on opposite sides of the membrane. If a single-file channel is occupied by at most one ion, the biionic reversal potential is expected to remain constant, independent of the absolute ion concentrations (; ; ).
Because TMA is somewhat more permeant than NMG in palytoxin-bound pump-channels (), we paired internal Na with external TMA (instead of NMG) to more accurately measure reversal potentials. The reversal potential in nominally symmetrical Na solutions ([Na] = 55, 165, or 550 mM) averaged 0.1 ± 0.7 mV ( = 11). We determined biionic potentials in each patch as the reversal potential shift (Δ) on quickly switching from the Na-containing external solution to one containing the same concentration of TMA. shows examples of palytoxin-induced currents activated in outside-out patches initially exposed to symmetrical solutions containing 550 mM Na (A) or 55 mM Na (B). The corresponding, roughly linear, palytoxin-induced current–voltage curves, with reversal potentials close to 0 mV (, filled symbols), were obtained by subtraction of currents recorded during voltage-step episodes marked b and c, and f and g, respectively. Equivalent palytoxin-induced current–voltage curves at the same internal Na concentrations, but after switching to external solutions containing 550 mM () or 55 mM () TMA, were obtained by subtracting currents from episodes a and d, and e and h, respectively, which yielded reversal potentials negative to −50 mV. On average, Δ upon replacing external Na with TMA () was the same at ion concentrations of 55 mM (−58 ± 2 mV), 165 mM (−57 ± 3 mV), and 550 mM (−57 ± 1 mV), indicating that / was concentration independent over this range, with a mean value of 0.102 ± 0.004 ( = 11).
To accurately compute unidirectional Na influx from measurements of net electrical current and unidirectional Na efflux, measured current and tracer flux must derive from the same area of membrane. In voltage-clamped internally dialyzed giant axons treated with veratridine and bathed in Na-free external solution, we have previously demonstrated equivalence of the TTX-inhibited steady net outward current and the TTX-sensitive unidirectional Na efflux, in experiments lasting ∼15 min (). This is the expected result for electrodiffusive Na ion flow through Na-selective channels under zero-trans conditions, and so it confirmed that current and flux were measured from the same population of channels. An experiment showing that the same holds true for measurements of longer duration (∼40 min), similar to those needed in the palytoxin experiments described below, is illustrated in Fig. S1 (available at ).
As established for Na ion movement via TTX-sensitive channels, confirms that under zero-trans conditions, unidirectional Na efflux and net current flowing through palytoxin-bound pump-channels are also equivalent. When this axon, held at −20 mV, with ∼100 mM internal Na and in Na-free, NMG-containing external solution, was exposed (at c) to 10 nM palytoxin, outward current (continuous noisy black trace) and Na efflux (determined at 1.5-min intervals; blue line) increased promptly and in parallel; their time courses and amplitudes (after scaling by Faraday's constant) were the same, as expected if they derive from the same population of palytoxin-bound Na/K pump-channels.
Like ouabain's action to shut palytoxin-bound Na/K pump-channels in mammalian cells and accelerate dissociation of palytoxin (), in squid axon, too, withdrawal of palytoxin and addition of 10 μM ouabain (at d in ) resulted in rapid, near parallel decreases of both outward current and Na efflux. The ouabain-induced decrement of Na efflux was slightly larger than that of the current, and both reached smaller values than observed before palytoxin. These differences possibly reflect abolition by ouabain of not only the palytoxin-induced signals but also a low level of electrogenic Na/K pump activity (with Na efflux/current ratio of 3/1) supported by contaminating K in the nominally K-free and Na-free, but 400 mM NMG-containing, external solution (). That this axon was indeed capable of producing changes in Na efflux and current of the expected relative magnitudes upon Na/K pump activation, by addition of 10 mM external K, is shown by comparison of the increments of Na efflux and current initiated at a. The ratio of the pump-mediated changes in flux and current (ΔΦ/Δ) calculated from these increments (measured as in ) was 2.94, near the value of 3.0 expected for 3 Na:2 K exchange. The red trace shows the recorded change in current multiplied by 3.0, to allow direct visual comparison with the simultaneous change in Na efflux. This close agreement with expectation further corroborates that measured flux and current originate from the same area of axon membrane.
A subsequent ∼20-min re-exposure to 10 nM palytoxin (beginning at e), with concomitant withdrawal of ouabain, failed to transform any pumps into ion channels, presumably because they were protected by persistently bound ouabain under these conditions (but see and , below).
At high (400 mM) external Na, responses to 20 nM palytoxin applied to an axon, held at 0 mV, and with ∼100 mM internal Na, appeared larger () than expected from the results in Na-free NMG external solution (e.g., ). In the axon of , before its accelerating action was precipitously interrupted by addition of 100 μM ouabain, palytoxin activated >35 (compared with ∼15) pmol cm s of Na efflux within 10 min, and a net current (≥15 μA cm) that was larger than predicted, assuming independence. The smaller response in likely reflects partial block of palytoxin-bound pump-channels by external NMG, as observed in outside-out patches (). A further contrast with is the order-of-magnitude slower reversal of palytoxin action upon its withdrawal in the presence of saturating concentrations of ouabain (1 mM was no more effective than 100 μM; ). The palytoxin pump complex is evidently stabilized at high external [Na] (compare ).
Measurement of the unidirectional ion flows through an ion-selective pore, in both directions, and estimation of the Ussing flux ratio exponent, ', provides information about the average number of ions inside the pore at any instant (; ; ). We previously measured ' for TTX-sensitive Na channels in squid axons and obtained a value of 0.97 ± 0.03 (). For such Na-selective channels, simultaneous measurement of unidirectional Na efflux (Φ) and net current () allows calculation of unidirectional Na influx (Φ) by difference and, hence, determination of the ratio of unidirectional fluxes. The flux ratio exponent ' can then be computed fromwhere [Na] and [Na] are the internal and external Na concentrations, is the membrane potential, and , and have their usual meanings.
Although palytoxin-bound pump-channels select poorly among small monovalent cations (; ; ; ), under the conditions of these experiments, Na ions are the principal charge carriers. This is because Na ions are ∼50-fold more permeant than NMG (; ) or Ca (; ), the other major cations present. The observed correspondence between palytoxin-induced outward current and unidirectional Na efflux into Na-free external solution () supports the essentially Na-selective behavior of palytoxin-bound pump-channels in these conditions. So, on the assumption that Na was the charge-carrying species, we used the measured palytoxin-elicited increments in unidirectional efflux (ΔΦ) and net current (Δ) to calculate the palytoxin-induced increment in unidirectional Na influx (ΔΦ),and then substituted the ratio ΔΦ/ΔΦ for Φ/Φ in to compute the flux ratio exponent ' for the channels opened by palytoxin.
and ∼100 mM Na
and the axon held at 0 mV, an efflux/influx ratio of 0.25 (100/400) is expected if the Ussing flux ratio exponent were 1.0. An initial estimate of ' can be garnered as follows. The peak increase in inward current caused by palytoxin was 16.0 μA cm (equivalent to 165.8 pmol cm s), and the peak increase in Na efflux was 40.4 pmol cm s, giving a calculated peak change in influx equivalent to 206.2 pmol cm s. Thus the peak response gives an efflux/influx ratio of 0.20, expected (via ) for ' ∼ 1.14. As ' must be independent of the fraction of open pump-channels, a similar calculation can be repeated for each pair of palytoxin-induced current and flux measurements, made at 1.5-min intervals, during the activation or deactivation time course. In practice, all data pairs within a given time period were simultaneously least-squares fit to yield a single estimate for '. The period (marked by red traces, see below) was chosen to allow reliable extrapolation of baseline (e.g., straight blue lines in ) values of current and Na efflux and to ensure that palytoxin-induced increments in current and flux were of significant magnitude. The procedure illustrated in inset was used to synchronize the current and efflux records, to account for the (imprecisely known) delay between experimental chamber and the fraction collector that collected radioactive samples. The least-squares fit was repeated for successive 5-s increments of assumed delay, to find the delay time that minimized the sum of squared deviations for the estimate of '; in the inset of this occurred for a time delay of 4.70 min and gave ' = 1.16. This method of synchronization was employed for all estimations of ' reported here.
and Fig. S2 is indicated by the red traces, which represent the measured efflux data, transformed using and and the best-fit ' to generate a “computed current,” and plotted over the measured current after synchronization by the least-squares delay. The superposition illustrates the goodness of fit.
In , to more clearly display small changes, the flux ordinate () is expanded relative to that for current ().
and the axon held at 0 mV, raising Na
from 0 to 400 mM (at a) resulted in transactivation of Na efflux (reflecting, in part, pump-mediated Na/Na exchange) and an inward net current shift (attributable to electrogenic Na flow through unidentified, non-TTX-sensitive, pathways). Addition (at b), followed shortly after by removal (at c), of 20 mM K
then temporarily converted pump-mediated activity to electrogenic Na/K exchange (whose maximum rate exceeds that of Na/Na exchange via the pump; ), resulting in a temporary further increment of Na efflux accompanied by a, similarly temporary, small pump-mediated outward current shift.
(at d, in ) elicited rapid increases in inward current and in Na efflux, which stopped upon palytoxin withdrawal (at e) and simultaneous addition of 1 mM ouabain.
with 400 mM K
(at f), still in 1 mM ouabain, then greatly accelerated recovery from palytoxin. That switch to K was marked by an initial, near instantaneous, increase in inward current owing to the greater permeability of palytoxin-bound pump-channels to K than to Na (by ∼20% in mammalian Na/K pumps; , ). K
-induced closure of a large fraction of palytoxin-bound pump-channels was then reflected in rapid decreases in Na efflux and in inward current ().
(at g), still with 1 mM ouabain, did not reopen the pump-channels that had been closed by K, confirming that palytoxin had dissociated from those closed channels (). Instead, the switch back from K to Na caused a slowing of the decline of Na efflux, and a decrease in inward current largely due to the smaller permeability to Na than to K of residual open pump-channels. The eventual near-complete recovery from the action of palytoxin is shown by the return of the current (after i) to the extrapolated baseline level. The switches between Na- and NMG-containing solutions (at h and i) revealed a small component of transactivated Na efflux that was not mediated by the pump (ouabain was still present), and a somewhat enhanced non-TTX-sensitive conductance.
Using the least-squares fitting procedure described for , the segment of palytoxin-induced flux and current data marked by the red trace in gave an estimate for ' of 0.96.
By exploiting the action of external K to close pump-channels and accelerate dissociation of palytoxin, repeated estimations of ' could be made for the same population of pump-channels under different conditions. shows 1 of 11 experiments in which membrane potential was varied, in this case from 0 to −20 mV.
, at 0 mV, the switch to 400 mM Na
(at a) caused the usual small increases in net inward current and Na efflux (as in ). Activation of electrogenic Na/K pumping by addition of 20 mM K
(at b) then also elicited the usual increment in Na efflux accompanied by an outward current shift. Pump inhibition by 100 μM ouabain (at c) in the presence of 20 mM K
caused a reduction of Na efflux (by 16.4 pmol cm s) and an inward current shift (of 0.52 μA cm). These specific pump-mediated signals yield an ΔΦ/Δ ratio of 3.04, close to the 3.0 expected for stoichiometric 3 Na:2 K transport (). For similarly treated axons that contributed ' values included in , ouabain-sensitive ΔΦ/Δ ratios averaged 3.09 ± 0.13, = 5, supporting the identical origins of flux and current signals in all of these experiments.
solution (at d), temporary exposure to 10 nM palytoxin (e to f) induced somewhat delayed increases in inward current and Na efflux, despite the previous application of ouabain. This contrasts with the lack of effect of 10 nM palytoxin when applied after ouabain in Na-free external solution (). The effects of palytoxin were reversed (as in ) by readdition of 100 μM ouabain (at g) and switching the extracellular solution to Na-free 400 mM K without ouabain (at h). The palytoxin-induced increases in current and efflux were fitted during the period indicated by the red trace, and yielded an estimate for ' of 1.14.
Following a second exposure to 400 mM K
solution (from j to k) to complete the recovery from palytoxin action (signaled by the return of current and flux levels to near extrapolated baseline values; blue lines), membrane potential was stepped from 0 to −20 mV (at l). This caused a small inward holding current shift but no measurable change in Na efflux. At −20 mV, brief application of 10 nM palytoxin (from m to n) caused a larger increase in inward current, but smaller increase in Na efflux, than seen earlier at 0 mV. Nevertheless, the value of ' estimated from the palytoxin-induced flux and current changes at −20 mV was 1.14, identical to that determined at 0 mV. Switching to Na-free 400 mM K
solution (o to p) again led to almost complete recovery of Na efflux and membrane current. Fig. S2 shows an experiment similar to that of but with four applications of palytoxin, at holding potentials of −20, 0, −40, and again 0 mV.
We used the same protocol, transient exposure to palytoxin with termination of its action by ouabain, and recovery speeded by external K, to examine the influence of [Na] on the flux-ratio exponent ' of palytoxin-bound pump-channels.
, ' was estimated first in 400 mM Na
at a holding potential of −20 mV, then in 400 mM Na
at −40 mV, and finally in 100 mM Na
at −40 mV, yielding values for ' of 1.14, 1.06, and 1.11, respectively.
summarizes estimates for ' under various experimental conditions. In all cases ' was close to 1.0, and the values from the different conditions did not differ significantly from each other.
. Statistically, the overall average value of ' (1.05 ± 0.02, = 28), and the average for 400 mM Na
at 0 mV (1.07 ± 0.02, = 13), both differ significantly from 1.0 (P < 0.01). But the difference is small (5–7%) and could reflect systematic experimental error.
Consistent with findings on mammalian Na/K pumps, in which withdrawal of ATP decreased the apparent affinity for palytoxin action () and lowered the open probability of palytoxin-bound pump-channels (), palytoxin action on Na/K pumps in squid axons dialyzed with nucleotide-free solution was slower (, compare with and ).
solution without nucleotides for ∼80 min before the start of the records illustrated. Addition of 10 mM K to the Na-free external solution (a to b) then elicited no measurable K
-activated Na/K pump current or Na efflux.
solution (at c), a high concentration of palytoxin (25 nM; d to e) slowly increased inward current and Na efflux, yielding an estimate for ' of 1.03, indistinguishable from the values obtained in the presence of 5 mM ATP. In the nominal absence of ATP, even low concentrations of external K (20 mM at f, 50 mM at g) were effective in accelerating closure of palytoxin-bound pump-channels, also in accord with results on mammalian Na/K pumps (, ). Palytoxin caused similar slow increases in current and flux in all three such experiments without ATP.
sup
#text |
In all organisms, post-transcriptional modifications play an important role in the maturation and function of cellular RNAs, especially stable non-coding RNAs (). The human ribosome is estimated to contain over 200 modified nucleotides and these fall primarily in functionally important regions of the rRNAs (,,). In eukaryotes and archaea, rRNAs and other non-coding RNAs are modified by two classes of RNA-guided modification enzymes: C/D and H/ACA RNPs (). C/D RNPs methylate the 2′--hydroxyl group of ribose rings in target nucleotides (,). H/ACA RNPs isomerize target uridine residues to pseudouridines by base rotation (,). These modification enzymes are comprised of a set of three or four core proteins and a cognate guide RNA that determines the target nucleotide by base pairing with the substrate RNA ().
Some key aspects of the mechanism of H/ACA RNP function have been well defined (,). Seminal studies revealed that the substrate recognition site is formed by juxtaposing two antisense sequences within an internal loop of the conserved hairpin structure of the guide RNA (B) (,). This loop that comprises the substrate recognition site is termed the pseudouridylation pocket. The antisense elements recognize substrate sequences flanking the target uridine, resulting in placement of the uridine to be modified at the apex of the pseudouridylation pocket. It is quite clear based on sequence and structure homology, and mutational analysis that Cbf5 is the pseudouridine synthase (,). The functions of the other three proteins, Gar1, Nop10 and L7Ae (or Nh2p in eukaryotes) are not established, but are known to be essential for the function of the complex (,).
Our laboratory and the Branlant laboratory successfully reconstituted and characterized functional H/ACA RNPs using components from and , respectively (,). These studies established that the four core proteins and a guide RNA are necessary and sufficient for full activity . We found that both Cbf5 and L7Ae interact directly with the guide RNA in the absence of other proteins. The remaining proteins, Gar1 and Nop10, bind to independent sites on Cbf5.
L7Ae belongs to a family of proteins that interact with RNA kink (k)-turns (). The k-turn binding proteins also include components of the ribosome, proteins involved in the assembly of spliceosomes, mRNA binding proteins and components of the RNase P and MRP complexes that function in tRNA and rRNA processing (). L7Ae appears to be important for the kinetics of pseudouridylation by the H/ACA RNP (). The primary interaction of L7Ae within the H/ACA RNP is with the k-turn of the guide RNA; no substantial interaction with the other proteins is observed in the absence of the RNA (). Moreover, L7Ae binding is not required for association of the other three proteins with the guide RNA, though it may enhance their binding (,). L7Ae binding sites are located either near (canonical k-turn) or overlapping (non-canonical k-turn) the apical loop of archaeal H/ACA RNAs (,). The essential role of L7Ae in H/ACA RNP function was not apparent, but seemed likely to be accomplished through its interaction with the RNA component.
Mutational analysis in combination with RNA–protein-binding assays indicated that Cbf5 requires several important elements of the guide RNA for its interaction, including sequences in the apical loop, pseudouridylation pocket and box ACA (B), suggesting that Cbf5 may interact with these regions of the RNA (). A subsequent crystal structure of the H/ACA RNP (including the four proteins and a guide RNA) indicates that in the context of the complete complex, Cbf5 interacts with box ACA and nucleotides in the lower stem, and to a lesser extent with the apex of the pseudouridylation pocket (). Similar interactions were mapped in RNA footprinting studies with yeast Cbf5 (). No interaction of Cbf5 with the apical loop was observed in the holoenzyme ().
In this work, we have examined the impacts of Cbf5 and L7Ae, both individually and in combination, on a guide RNA by enzymatic and chemical footprinting. The influences of the proteins on the RNA footprinting patterns substantiate and clarify the RNA–protein interactions predicted by the previous mutational analysis and observed in the crystal structure of the full complex (,). In addition, the results indicate that L7Ae plays an important role in formation of the pseudouridylation pocket (i.e. substrate recognition site). Finally, we observed an interaction of Cbf5 with the apical loop of the RNA that is disrupted by the binding of L7Ae. Our results indicate that RNA remodelling events triggered by binding of specific components of the H/ACA RNP govern the ability of the RNP to function in target RNA recognition and nucleotide modification.
Cbf5 and L7Ae genes were amplified by PCR from genomic DNA and sub-cloned into a modified version of pET21D expression vector as previously described (). The resultant recombinant proteins containing N-terminal 6 × histidine tags were purified by affinity chromatography on Ni-NTA agarose (Qiagen), eluted with buffer A (20 mM sodium phosphate, pH 7.0, 1 M NaCl, 350 mM imidazole) and quantified using BCA protein assay (Pierce). Prior to use in RNA-binding assays, the proteins were dialysed against 40 mM HEPES–KOH, pH 7.0, 1 M KCl (or K-acetate).
The single hairpin, H/ACA RNA Pf9 was transcribed from PCR-amplified DNA product containing a SP6 promoter using SP6 RNA polymerase (Epicentre Biotechnologies) as previously described (). RNA was gel purified by electrophoresis through a 15% polyacrylamide/7 M urea gel. Purified RNA was ethanol precipitated and washed with 70% ethanol. Purified RNA was dephosphorylated with calf intestinal alkaline phosphatase according to the manufacturer's protocol (Ambion). The dephosphorylated RNA was P labelled with T4 polynucleotide kinase (Ambion) and [γ-P]ATP (7000 Ci/mmol, MP Biomedicals). 5′-end labelled RNA was then gel purified as described above.
Reconstitution of RNP complexes was performed as described previously (). Briefly, 5′-end radiolabelled RNA (0.05 pmol) was incubated in buffer B (20 mM HEPES–KOH, pH 7.0, 500 mM KCl, 1.5 mM MgCl, 5 μg tRNA) alone or with various concentrations of protein in a final volume of 20 μl for 1 h at 65°C. RNP complexes were analysed on an 8% non-denaturing polyacrylamide gel and visualized by autoradiography.
P-end labelled Pf9 RNA (0.05–0.1 pmol) was incubated in the absence (free RNA) or presence (RNPs) of increasing concentrations of purified Cbf5 or L7Ae proteins for 1 h at 65°C in buffer B (described above) in a final volume of either 20 or 50 μl. For ribonuclease cleavage, the reactions were initiated by addition of 0.1 or 0.2 U RNase T1 (Sigma), or 1 or 2 ng RNase A (Sigma) and incubated for 15 min at 37°C. The enzymatic reactions were stopped by extraction with phenol/chloroform/isoamyl alcohol. Hydroxyl radical footprinting experiments were performed essentially as described (). Briefly, the cleavage reactions were initiated by adding freshly prepared 18 μM ethylenediaminetetraacetic acid (EDTA) iron (III) sodium salt dihydrate (Aldrich), 2 mM sodium ascorbate (Sigma) and 0.14% (v/v) HO (Sigma). The reactions were carried out at 65°C for 30 s and stopped by addition of 1 mM thiourea (Aldrich) followed by phenol/chloroform/isoamyl alcohol extraction. For lead (II) footprinting, the reactions were carried out in a modified buffer B where the KCl was substituted with 200 mM K acetate. Lead cleavage was performed essentially as previously described () with 15 mM Pb(II) acetate (Merck) freshly prepared in sterile water. The reactions were performed at room temperature for 10 min and were stopped by adding EDTA to final concentration of 20 mM before ethanol precipitation. As sequence markers, RNA alkaline hydrolysis ladders (cleavage after each nucleotide) were generated by incubating RNA with 5 μg tRNA in 50 mM sodium carbonate at pH 9.5, 1 mM EDTA for 5 min at 90°C. RNase T1 ladders (ΔT1) (cleavage after each guanosine) were generated by incubating the RNA in 20 mM sodium citrate at pH 4.5, 1 mM EDTA, 7 M urea for 10 min at 50°C. For both enzymatic and chemical probing reactions, the treated RNA samples were then ethanol precipitated in the presence of 0.3 M sodium acetate at pH 5.2 followed by washing with 70% ethanol. The dried RNA pellets were resuspended in RNA loading dye [10 M urea, 2 mM EDTA, 0.5% (w/v) SDS, 0.02% (w/v) each bromophenol blue and xylene cyanol]. The cleavage products were separated on 15 or 20% polyacrylamide (acrylamide:bis ratio 19:1) 7 M urea-containing gel and visualized by autoradiography.
To assess the interactions of Cbf5 and L7Ae with H/ACA guide RNA Pf9, we analysed the two RNA–protein sub-complexes (Cbf5-Pf9 and L7Ae-Pf9) by hydroxyl radical nucleotide protection assays (). Hydroxyl radicals cleave the RNA backbone independent of RNA sequence or secondary structure (,). Thus, in the absence of the proteins, cleavages were observed at all ribose moieties (A, lanes 4, 11, 16). Protection of a ribose from hydroxyl radical cleavage upon addition of a protein generally indicates a direct association with the protein (). RNA–protein complex formation (with 5′ end-labelled Pf9 RNA and purified recombinant proteins) was verified by gel shift analysis (). The majority of the Pf9 is shifted into RNA–protein complexes at 2 μM Cbf5 and 1 μM L7Ae. Both proteins provided some global protection of the RNA, however, as can be seen in A, distinct RNA protections were observed with increasing Cbf5 or L7Ae concentrations (see regions indicated with blue and green bars).
B shows the protection results in the context of a secondary structure model of Pf9 RNA that is based on the well-defined, functional features of the H/ACA RNA family (). The pseudouridylation pocket of the H/ACA RNA is the bipartite target recognition site, established and bounded by the upper and lower stems. The predicted pseudouridylation pocket of Pf9 is complementary to sequences that flank 16S rRNA U910 (i.e. nts 905–917), and consistent with this model, we have confirmed that U910 is modified in rRNA extracted from (Marshburn,S., Terns,R. and Terns,M., unpublished data). Box ACA, the signature sequence element, is located 3′ of the lower stem. The k-turn of Pf9 is found within the upper stem, adjacent to the apical loop. Canonical k-turns are helix-bulge-helix structures that produce an ∼120° bend between the axes of the two adjacent RNA helices (). The bulge of a k-turn is bounded by two G–A base pairs that terminate the first helix and a G–C base pair that initiates the second helix. The motif generally includes several flanking base pairs (B).
In multiple studies, L7Ae and its close homologues have been found to interact directly with sequences in the k-turn of partner RNAs (,,), and as expected, L7Ae provides strong protection on both strands of the k-turn (green shading, ). In particular, the k-turn-binding proteins are consistently found to contact nucleotides in the bulge of the k-turn (,,) and L7Ae's protection of Pf9 includes the bulge ().
The interaction of Cbf5 with H/ACA RNAs is less well studied. Previous gel shift analysis suggested that box ACA, the pseudouridylation pocket and sequences in the apical loop of the RNA, may be involved in the interaction; alterations in these elements affect the stability of the Cbf5–guide RNA complex (). In the crystal structure of a complex that includes Nop10, Gar1 and L7Ae as well as Cbf5 and a guide RNA, contacts were observed between nucleotides in box ACA, the lower stem and the pseudouridylation pocket of the RNA and Cbf5 (). As can be seen in (blue shading), Cbf5 provides extensive protection of Pf9 from hydroxyl radical cleavage. Cbf5 significantly reduces cleavage of the 3′ strand of the lower stem, the 5′ strand of the pseudouridylation pocket and the 5′ half of the k-turn and the apical loop. Additional weak protection was observed along the 3′ strand of the pseudouridylation pocket and the upper stem. The 5′ strand of the lower stem and the 3′ single-stranded region that contains box ACA were not assessed in this experiment due to resolution limitations. The results suggest that Cbf5 interacts directly with the lower stem, pseudouridylation pocket, apical loop and 5′ strand of the k-turn in the absence of other proteins.
The observed protections were specific to the individual proteins with the notable exception of the 5′ strand of the k-turn (). Interestingly, the results reveal strong protection of the 5′ strand of the k-turn by both Cbf5 and L7Ae, suggesting that both proteins interact with this region of the RNA in the sub-complexes. Direct interactions of Cbf5 and L7Ae with the 5′ side of the k-turn would be expected to be mutually exclusive.
Cbf5 and L7Ae do not interact directly, either independently () or in the context of the fully assembled complex (), and therefore should not directly influence the interaction of the other protein with the guide RNA; however, we were interested in the possibility of effects translated through the RNA (for example, via changes in RNA structure). Moreover, we were interested in examining the footprint in the presence of both proteins on the 5′ side of the k-turn (where both proteins were observed to bind, ). To test for any impact of one protein on the interaction of the other protein with the RNA, we examined the hydroxyl radical footprints of combinations of Cbf5 and L7Ae on Pf9 (). In these experiments, the RNA was mixed with increasing concentrations of one protein in the presence of a constant concentration of the other protein in final amounts that promote nearly complete incorporation of the RNA into complexes (, lanes 10–14).
For the most part, the guide RNA protection pattern in the presence of both proteins (, lanes 6 and 12) appears to be the simple sum of the patterns obtained with the individual proteins (, lanes 2 and 8), however there are several interesting exceptions. Examination of multiple experiments and exposures revealed two regions where greater protection is observed than would be expected from the individual protections (, turquoise shading). In the 3′ strand of the k-turn, several nucleotides (nts 39, 40, 44 and 45) are partially protected by L7Ae, and not significantly protected by Cbf5, but are nearly completely protected in the presence of both proteins (A, lanes 6 and 12, and indicated in B). Similarly, little protection of sequences in the 5′ strand of the upper stem was observed with either L7Ae or Cbf5 binding, however cleavage of most nucleotides in this region (,,) is reduced by the combination of the two proteins (). These increased protections likely reflect enhanced interaction of the proteins with these regions when the other protein is bound to the RNA.
In contrast, protection of nucleotides in the apical loop by Cbf5 (nts 30 and 31) is lost upon introduction of L7Ae (A, lane 8 versus lane 9, purple shading). The de-protected nucleotides are part of a larger, contiguous Cbf5-binding site that also includes the 5′ strand of the k-turn, the region where both Cbf5 and L7Ae interact with the RNA (). The loss of the Cbf5 protection pattern suggests that L7Ae disrupts or prevents the interaction of Cbf5 with the 5′ k-turn/apical loop region. This is consistent with the absence of an interaction between the k-turn/apical loop of the RNA and Cbf5 in the crystal structure of the full complex ().
Box ACA is the signature sequence of the H/ACA guide RNAs, and RNA–protein-binding studies and crystal structure data indicate that Cbf5 specifically binds and recognizes this family feature (,,). We examined the individual impacts of Cbf5 and L7Ae on the 3′ half of the RNA including box ACA by lead (II) acetate cleavage. Lead (II) acetate induces cleavage preferentially at single-stranded and dynamic regions of RNAs such as bulges and loops (,). As expected from the predicted secondary structure, the 3′ strand of the lower stem of Pf9 is inaccessible to lead-induced cleavage in the absence of proteins (see lack of cleavage between 3′ pseudouridylation pocket and ACA in A, lanes 5, 12, 16). However, the 3′ strand of the upper stem was unexpectedly sensitive to cleavage (indicated with red arrowheads in A, lane 12).
Binding of Cbf5 to Pf9 RNA results in substantial protection of the conserved ACA sequence as well as the nucleotide immediately upstream (, blue shading). Protection of most of the lower stem could not be assessed in this experiment (because this region is already insensitive to lead-induced cleavage); however, we observed that Cbf5 also provides some protection to the 3′ strand of the pseudouridylation pocket. In contrast, L7Ae does not protect these regions (, lanes 13–15). The addition of L7Ae to the RNA results in reduced cleavage of the 3′ strand of the k-turn and also of the upper stem outside of the k-turn motif (, green shading). L7Ae also produced increased sensitivity to lead-induced cleavage in nucleotides in the 3′ half of the pseudouridylation pocket (, yellow shading).
The unexpected sensitivity of the 3′ side of the upper stem of Pf9 to lead-induced cleavage () led us to further probe the secondary structure of the RNA in the absence of proteins. We performed partial enzymatic digestions of 5′ end labelled Pf9 RNA using RNase T1 and RNase A (). RNases T1 and A cleave accessible phosphodiester backbones following un-base-paired guanines (Gs) and pyrimidines (Cs and Us), respectively. Adenines (As) are not subject to analysis. Base-paired regions are resistant to both enzymes. The results are summarized schematically in the context of the predicted secondary structure in B.
Under the experimental conditions analysed, in the absence of proteins, accessible regions of the RNA include the 3′ tail, the pseudouridylation pocket, the apical loop and the bulge of the k-turn motif as expected (B, orange shading). Consistent with the predicted secondary structure, nucleotides in the lower stem are inaccessible to the single-stranded nucleases. In addition, nucleotides in this region are susceptible to RNase V1, a nuclease specific for double-stranded regions (data not shown). However, we found that the upper stem is sensitive to single-stranded nuclease digestion [see strong cleavages in red boxed region (B) and indicated by red arrowheads (A)]. The results suggest that the upper stem of Pf9 RNA is not stably structured under these experimental conditions.
Although the upper stem of the RNA does not appear to be firmly established in the absence of proteins ( and ), the interaction of L7Ae with sequences on both sides of the k-turn motif () strongly implies the existence of a helix within the upper stem of Pf9 RNA in the presence of the protein. Moreover, the L7Ae-induced resistance of the upper stem beyond the k-turn motif (i.e. outside the region where L7Ae has been shown to directly contact RNA) to lead-induced cleavage () suggests the formation of the stem in the presence of the protein. To further investigate the impact of L7Ae on the secondary structure of the RNA, we also analysed partial enzymatic digestions in the presence of the protein ().
The strong ribonuclease T1 and A cleavages observed in the ‘upper stem’ both within and outside the k-turn motif in the absence of the protein (red arrowheads, A, lanes 3 and 5) are significantly reduced upon L7Ae binding (A, lanes 4 and 6). At the same time, L7Ae increases cleavage of the apical loop (B, yellow shading). The interaction of L7Ae with the k-turn is well documented (,,). No extensive interactions outside the k-turn have been described. The strong protection that we observe in the upper stem, as well as the increased sensitivity in the apical loop, is consistent with formation of the upper stem upon L7Ae binding.
In this work, we examined the arrangements of the RNAs and proteins in a series of sub-complexes of the H/ACA RNP by chemical and enzymatic footprinting. The combined approach has the potential to detect both physical protein interaction sites and effects on RNA configuration, and we found evidence for both types of impacts in this study. The results provide detailed insight on steps in the assembly of the complex and essential roles of the proteins.
A significant amount is known about the sites of RNA–protein interaction within the fully assembled H/ACA guide RNP from the crystal structure of the holoenzyme (using a modified Afu 46 guide RNA) (). In the Cbf5–guide RNA and L7Ae–guide RNA sub-complexes that we examined here, we observed footprints consistent with the well-established interaction of L7Ae with the k-turn (,,) and with the contacts observed between Cbf5 and the guide RNA in the holoenzyme crystal structure (). Our results support the extensive interaction of Cbf5 with the guide RNA from box ACA through the lower stem to the pseudouridylation pocket [observed in RNA-protein-binding assays (,), holoenzyme crystal structure () and RNA footprinting of the eukaryotic complex ()]. The results indicate that the interactions of both Cbf5 and L7Ae with the guide RNA in the fully assembled enzyme are established in the Cbf5–guide RNA and L7Ae–guide RNA sub-complexes.
In addition, however, we found evidence of an interaction between Cbf5 and the guide RNA that is unique to the sub-complex. In the absence of other proteins, Cbf5 protects the 5′ strand of the k-turn and apical loop from hydroxyl radical cleavage, an effect that generally reflects a physical interaction (). Moreover, previous studies showed that mutation of this region of the RNA weakens the binding of Cbf5 (), supporting the existence of the interaction and suggesting that the interaction is important in formation and stability of the sub-complex. In the crystal structure of the holoenzyme, Cbf5 is not found in proximity with the apical loop and the 5′ strand of the k-turn (). Our results indicate that L7Ae successfully competes for the site and displaces Cbf5 ( and ). Accordingly, L7Ae is found in close proximity with this region of the RNA in the crystal structure of the holoenzyme (). [The specific equivalent L7Ae–RNA interactions could not be compared as the guide RNA used in the crystal structure differs significantly from Pf9 in this region (non-canonical k-turn) and its structure is also incomplete in this region ().]
Because the intermediates in the assembly and function of the H/ACA RNP have not been precisely defined, it is not yet clear what role the newly identified Cbf5–guide RNA interaction may play in H/ACA RNP assembly or function. In eukaryotes, evidence indicates that three of the four core proteins, including Cbf5, assemble on the H/ACA RNA at the site of transcription (and that association of Gar1 occurs at a later point in the temporal and spatial assembly pathway) (,,). Among the core proteins, Cbf5 shows the strongest association with the H/ACA RNA genes, suggesting that Cbf5 could be the first of the H/ACA RNP proteins to associate with the newly made guide RNA in yeast (). Thus, the Cbf5 interactions defined here may provide for the initial recognition of the guide RNA and subsequent complex assembly. It is also possible that a sub-complex lacking L7Ae is involved in the function of the H/ACA RNP (for example, as a step in substrate release).
Our studies also revealed a substantial effect of L7Ae on the guide RNA configuration beyond the k-turn with significant implications for proper establishment of the target recognition site. Previous studies had shown that the structure of the k-turn motif itself is dynamic in the absence of protein and that formation of the kink (i.e. 120° bend from linear) is induced by the binding of L7Ae and related proteins (). Our results indicate previously undescribed effects on the RNA beyond this region (). Our data from both partial enzymatic hydrolysis () and lead-induced cleavage () indicate that the upper stem of the guide RNA is not stably formed in the absence of proteins under the solution conditions used in our work. However, upon addition of L7Ae the upper stem nucleotides become resistant to single-stranded nucleases () and lead-induced cleavage (), strongly suggesting L7Ae-induced formation of the upper stem. The observed increase in the sensitivity of nucleotides in the apical loop and pseudouridylation pocket to single-stranded nucleases () and lead-induced cleavage () upon L7Ae binding are also consistent with formation of the upper stem, which defines these loops. While L7Ae provided strong protection of the upper stem against enzymatic cleavage (), this region was not significantly protected from hydroxyl radical cleavage outside of the k-turn motif (), providing further evidence that the observed protection of this region from enzymatic cleavage reflects induction of base pairing (rather than steric interference). Importantly, the upper stem establishes the pseudouridylation pocket—the H/ACA RNP target RNA-binding site.
Our results reveal that L7Ae plays a significant role in substrate binding and placement in the archaeal H/ACA RNP via formation of the pseudouridylation pocket. These findings may explain the positive impact that L7Ae appears to have on formation of substrate-containing H/ACA RNP complexes and on activity of the complex (,). In addition, analysis of a crystal structure of a sub-complex of the H/ACA RNP with a substrate RNA recently obtained by Hong Li's laboratory suggests that both pseudouridylation pocket formation and positioning of the substrate uridine in the Cbf5 active site are defective in the absence of L7Ae (Liang,B., Xue,S., Terns,R., Terns, M. and Li, H., submitted for publication). At the same time, this and several other recent studies describe guide–substrate RNA interactions in the absence of L7Ae (,). It is most likely that the difference reflects the high concentrations of molecules used in these structural studies (,). In the cell, substrate capture likely depends on a well-formed pseudouridylation pocket established by L7Ae.
Given that the pseudouridine synthase Cbf5 can interact directly with the guide RNA, which has the capacity to capture and present the substrate, it was previously not clear why L7Ae should be needed. Our findings indicate that the importance of L7Ae in the function of the H/ACA RNP is in remodelling the guide RNA to form the substrate-binding site. In eukaryotes, the H/ACA RNP protein homologous to L7Ae is Nhp2, a protein with less well-defined RNA binding properties (), and it remains to be determined whether Nhp2 will also play a role in definition of the substrate-binding site in the eukaryotic H/ACA RNPs. However, L7Ae is also a component of C/D RNPs and the ribosome in archaea (), and our findings suggest that L7Ae and other k-turn-binding proteins could play a similar role in important alterations of RNA structure beyond the k-turn in other complexes as well. |
Telomeres are nucleoprotein structures that protect the ends of eukaryotic chromosomes (). Telomerase is a ribonucleoprotein complex (RNP) that can add short DNA repeats onto telomeres and thus compensate for the losses caused by incomplete replication or degradation (). The essential core components of this specialized enzyme are telomerase RNA (TER) and telomerase reverse transcriptase (TERT), which copies a small portion of TER, the template, onto the telomere's 3′-end. TERs are highly divergent in sequence and length even among closely related species. Based on phylogenetic covariation, secondary structure models were predicted for ciliates (,), vertebrates () and species (). However, only limited similarity in the general architecture of these models was observed (,).
The proposed secondary structure for ciliate TERs consists of four base-paired regions in most tetrahymenine ciliates ( and species) (,), denoted by Roman numerals I–IV (). All species have an additional helix V (), while hypotrichous ciliates ( and ) and lack helix II (,). Nucleotides in the apical loop of stem IIIb can potentially form four base pairs outside of the stem, thus forming a PK (). PK structures were later proposed for vertebrates () and yeast () TERs and found to be important for telomerase function ().
The role of the PK in TER has been studied by several groups. Autexier and Greider studied telomerase activity reconstituted from micrococcal nuclease-treated endogenous telomerase fractions and transcribed TER (). They found that mutations disrupting stem IIIa and even deleting most of the PK (76–99Δ; see inset for the numbering scheme) only moderately affected telomerase activity. Licht and Collins () reconstituted telomerase activity in rabbit reticulocyte lysate from both TER and TERT expressed , thus eliminating possible interference by partially digested endogenous TER. In this system, disrupting either of the two PK stems (70–86Δ or 77–98Δ) reduced telomerase activity more significantly. Further studies showed that the PK and stem IV are important for telomerase activity and cooperate in providing the repeat addition processivity—the ability to synthesize multiple telomeric repeats (). Disrupting stem IIIa impaired the assembly of the telomerase RNP complexes and reduced the telomere length and telomerase activity assayed in partially purified cell extracts (,).
The high-resolution structure of the human TER PK has been solved by NMR (). This structure revealed an extended triple-helical segment with base triples in the major groove of stem S2 and minor groove of stem S1, and a Hoogsteen UA pair in the junction. The triple helix within stem S2 consists of three consecutive AUU base triples, where uracils from the third strand form Hoogsteen hydrogen bonds with adenines from the duplex. A molecular model with five consecutive AUU base triples, based on extensive mutational analysis, has been proposed for the PK of the budding yeast TER (). Importantly, disrupting the triplex part of the PK abolished telomerase function in cells, while triple compensatory mutations forming pH-dependent GCC triples partially restored it.
In the current work, we analyzed PK sequences in 28 species of ciliates, belonging to seven different genera ( and ). Based on the lengths of the stems and loops, we classified them into six morphologically different groups and calculated molecular models for representatives of each group. All ciliate sequences can potentially form at least one base triple in stem 2 of the PK. In all structures except that of , these are the conventional AUU base triples, while the latter contains unconventional GCG and AUA triples.
TER sequences, determined by a number of groups (,,,), were used as available from the GenBank () with the following exceptions. The sequences of and were not available in the GenBank and were acquired instead from (). For (pseudogene B), and (strain ‘JR Preer stock C-101’), some discrepancies were found between the sequences in the GenBank and the original publication; the published versions of these sequences were used (). The TER sequence of (strain ‘RB1’) has not been published but is available from the GenBank (accession number AJ132318). GenBank accession numbers of all sequences used are given in Supplementary Data. Altogether, we analyzed 30 TER sequences from 28 ciliate species ( has different TER sequences in two strains, and both the functional TER gene and a pseudogene are sequenced in ).
Molecular models of PK structures were calculated using the miniCarlo program () on an SGI Octane R12000 computer essentially as described previously (). miniCarlo is an internal coordinates-based program for molecular mechanics calculations of nucleic acids. The program assumes fixed values of bond lengths and fixed idealized geometries of aromatic bases. The set of internal coordinates includes generalized helical parameters that define relative positions of bases of nucleic acids in space. Flexible sugar rings are calculated using a one-parameter model (pseudorotation phase angle). A specialized chain-closure algorithm is used to calculate coordinates of the sugar-phosphate backbones connecting adjacent nucleosides (). Conformational energy is calculated using an empirical force field optimized for nucleic acids (,); hydration effects are modeled with the distance-dependent dielectric constant, and nucleotides are assumed electroneutral to take into account the shielding effect of counterions. Molecular graphics representations were prepared with the UCSF Chimera (,) and MIDASPlus programs (). The atomic coordinates of all calculated models are available from the authors upon request.
A shows a schematic representation of a generic PK, and B shows the structure of the AUU base triple. Standard notations for the PK stems S1 and S2 and loops L1, L2 and L3 () will be used below instead of the IIIa and IIIb more commonly used in the ciliates telomerase literature (compare with ). With few exceptions, loop L2 has a length of zero in the PKs of the ciliate TERs.
A search of the GenBank and published literature resulted in 30 TER sequences from 28 different species of ciliates. The PK elements were located with the help of multiple sequence alignments using CLUSTALW () and with the help of published alignments and secondary structures (,,). The PK sequence of contains a stretch of two AU base pairs in stem S2 and two uridines in loop L1 (). The strand direction of loop L1 is parallel to the adenines’ strand. Therefore, there is a potential for forming a mini-triplex with two AUU base triples (,), similar to the proposed human and PK structures (,). Inspection of PK sequences in other ciliates showed that two such triples could potentially form in most species. In and species and in , there is only one such AUU triple in each PK. The only PK that cannot form any AUU triples is that of , which does not have any uridines in loop L1. Representatives of all morphologically different PK folds are listed in and their modeling is described below. A list of all PK folds of ciliates is given in Supplementary Data.
When the first structure of an RNA PK was characterized by NMR, the authors noted that residues of loop L1 spanning the major groove of stem S2 could potentially form base triples (). Later, a GCC triple with a protonated cytosine was observed in a crystal structure of the frameshifting PK from the beet western yellow virus () and was shown to contribute to the stabilization of the PK (). An NMR structure of the human TER PK determined by Theimer () revealed an extended triple-helical region, which includes three consecutive AUU major-groove triples in stem S2. A model with a similar motif in stem S2, including five consecutive AUU triples, was proposed for the PK element in TER (). The TER sequence of another budding yeast, , also appears compatible with the common structural motif of a PK that includes AUU triples (unpublished data). Although the PK formation in ciliate TERs was suggested as early as 1991 (), the high-resolution structure of this element is still unsolved; the dynamic nature of this element in ciliates () has probably hindered its structural analysis. To overcome this obstacle, we undertook a computer modeling approach. By calculating molecular models of six representative folds, we demonstrated that all available ciliate TER sequences are capable of forming base triples in their PK elements. In 20 out of 30 available sequences, including that of , there are two AUU triples in this motif. In 9 sequences, including those of and all and species, there is only one AUU triple, and in one species, , three unconventional purine–purine–pyrimidine base triples can potentially form. Despite significant variations in the PK sequences of ciliate TERs, the morphological parameters are balanced in such a way that can accommodate base triples in all PK folds. An example of this delicate balance is demonstrated in the TER sequences with the suboptimal value of parameter 2. In these structures, the presence of a conserved GU wobble pair compensates for the suboptimal 2 value by reducing the size of the gap in junction 2.
Such a remarkable conservation of this unusual structure suggests that it contributes a conserved telomerase function. In , mutations disrupting the triplex or shifting the alignment of the third strand abolished or severely impaired telomerase function. Even a relatively minor alteration of the triplex structure by substituting GCC for AUU triples affected the fidelity or the processivity of the template copying , i.e. they led to nucleotide misincorporations and truncated telomeric repeats (). However, the exact role of this structural element and the molecular mechanism involved remain unclear. A possible role of base triples could simply be the stabilization of the PK structure. However, there are other ways to do so, such as to lengthen its stems or replace AU with GC base pairs. The remarkable conservation of base triples in vertebrates, yeast and ciliates TER PKs, despite the widely different lengths of the PK stems () argues against such an explanation.
Base triples could be specifically recognized by one of the telomerase proteins, such as TERT or one of the regulatory proteins, or base triples could be required for stabilization of a different binding site. Conflicting experimental data supporting or contradicting this possibility were published over the years. PK sequences were found to be required for the binding of Est2 (the yeast TERT) to the TER (,). In , the PK element appeared to be dispensable for the binding of TERT to TER , at least in some experiments (). However, mutant TER sequences with disrupted stem S1 failed to form active telomerase RNP (). Secondary structure probing suggested that the PK region in naked TER was rather dynamic and unstable (). The reconstitution of TER and TERT into an active complex caused a significant stabilization of the PK fold (). This is also consistent with footprinting data: while the AAUU sequence of stem S1 and the adenines of stem S2 were accessible for modification by dimethyl sulfate in naked RNA , they were protected from methylation (). Interestingly, the single-stranded CAAA sequence of loop L3 was also protected from methylation , suggesting that a protein may bind to this site or that these residues are involved in the minor-groove base triples.
It is possible that a relatively low stability of the PK is a specific feature of ciliates and may be connected to the relatively short length of stem S1 (4 bp). Indeed, our attempts to reconstitute the PK from two separate RNA strands were unsuccessful (unpublished data), even though similarly designed constructs formed stable dimers for the sequences (). This may also explain why the high-resolution structure for the PK has not yet been solved, despite the relatively short size of this RNA. It is not clear if this relatively low stability has a functional role, e.g. in facilitating a conformational switch, as proposed for the human PK (). The short size of stem S1 is universally conserved in ciliates (Supplementary Table S1), supporting this notion. There is a precedent for a short RNA stem (exactly 4 bp) being important for function—the nucleocapsid protein-driven maturation of the dimerization initiation site in HIV-1 RNA ().
Similarly to ciliates, the human PK was also suggested to be dynamic (). The idea of the human TER PK serving as a conformational switch was introduced based on the fact that two alternative folds, the PK and a hairpin, have similar stabilities in the context of short RNA constructs (,). NMR structures of both the PK and the hairpin conformations have been solved (,). However, mutations disrupting the intra-loop base pairing in the hairpin structure did not affect the activity of telomerase reconstituted either or (), casting doubt on the biological relevance of the hairpin structure. The role of such a conformational switch, if any, is unknown. One may hypothesize that it regulates the transition between different stages of the telomerase reaction cycle, enabling the processive synthesis of multiple telomeric repeats from a single template. The contribution of the PK (in cooperation with stem IV) to the processivity of the telomerase () is consistent with this hypothesis. In , stem S2 with 5 predicted base triples is expected to be more stable while a hairpin conformation was not detected by UV melting experiments (), arguing against a conformational switch that involves the unwinding of stem S2 in this species. Instead, a different type of a conformational switch is theoretically possible, involving a one base shift in the register of the third strand of the triplex (). However, there is no experimental support so far for the relevance of such a conformational switch. Unlike ciliate and vertebrate telomerases, and telomerases lack the ability to synthesize multiple repeats without dissociating from the telomeric substrate, at least (,). It is possible that the lack of repeat addition processivity is related to the absence of an alternative hairpin conformation in yeast. Another possible role for the low stability of the ciliate PK may be related to the biogenesis pathway of the telomerase RNP. TER undergoes structural rearrangement following the binding of the telomerase protein p65, which in turn enables the binding of TERT (). It is possible that such a structural change requires the dynamic nature of the PK. Since p65 homologs or a similar folding pathway have not been found in yeast or vertebrates, such a role for the low stability of the PK may be specific for ciliates. Additional studies are needed to uncover the role of this unusual PK element with major-groove base triples, which is common to all telomerase RNAs examined.
p
p
l
e
m
e
n
t
a
r
y
D
a
t
a
a
r
e
a
v
a
i
l
a
b
l
e
a
t
N
A
R
O
n
l
i
n
e
. |
The installation of various aryl groups at the -position of 2′-deoxyadenosine was accomplished by the direct copper catalyzed arylation of 2′-deoxyadenosine () with the corresponding aryl iodide (or bromide) in dimethylsulfoxide in the presence of copper (I) iodide (catalyst), ethylenediamine (ligand) and potassium phosphate (base) and sodium iodide (in the case of arylbromide) to give (B). Dimethoxytritylation of in the presence of 4,4′-dimethoxytrityl chloride in pyridine afforded intermediates in 80–92% yields. Phosphitylation of the 3′-OH of in dichloromethane in the presence of diisopropylethylamine generated the phosphoramidites in 85–93% yield. The coupling of into DNA oligonucleotides was quantitative. Detailed characterization of the synthesis products and oligonucleotides are provided in the Supplementary Data section.
methylation of adenosine was accomplished by following Hobartner's procedure to yield intermediate (). Selective protection of the 2′-OH of with -butyldimethylsilylchloride was accomplished by adopting Beigelman's strategy to give intermediate (). Dimethoxytritylation of in the presence of 4,4′-dimethoxytrityl chloride in pyridine afforded intermediate . Phosphitylation of the 3′-OH of in dichloromethane in the presence of diisopropylethylamine generated phosphoramidite (B) in 82% yield. Detailed characterization of the synthetic products and oligonucleotides are provided in the Supplementary Data section.
The ligation substrates consist of two oligonucleotides. The oligo containing the recognition residue at its 5′ end residue (e.g. -phenanthren-dA for Ψ or dG for mA) is referred to as the ‘floater’, the other oligo substrate is referred to as the ‘anchor’.
A 30-mer model RNA (5′-AUCCGCUGUAACGCGAGCAAUGCCUGGUA, X = U, Ψ, A from Dharmacon Research, Inc., X = mA was synthesized as described above) was used to identify the recognition residue for RNA modifications (). The optimized ligation reactions were carried out with 0.15 µM 30-mer RNA with or without Ψ or mA modifications, 0.5 µM floater and 0.38 µM of 5′-P-labeled anchor in 66 mM Tris–HCl, pH 7.6, 0.5 mM ZnCl, 10 mM DTT, 66 µM ATP, 15% DMSO and 0.25 U/µl T4 DNA ligase (USB Inc.). All components were mixed and incubated at 16°C for 16 h, and the ligation products separated on denaturing polyacrylamide gels containing 7 M urea.
The analysis of Ψ in the rRNA present in yeast total RNA was performed as follows. A total of 0.4 µM anchor and 0.5 µM 5′-P-labeled floater first hybridized with yeast total RNA in 20 mM Tris–HCl, pH 7.5, 50 mM NaCl, 0.4 µg/µl total yeast RNA, plus 10 nM model 30-mer RNA, its 30 nM P-labeled floater, and 60 nM anchor as the control for ligation efficiency and loading. Hybridization was performed by placing the tubes in a 95°C heat block for 1 min, followed by immediately placing the heat block at 4°C for 45 min before placing on ice for 5 min. Following hybridization, the ligation reaction was initiated by the addition of a 2× ligation mixture containing 132 mM Tris–HCl, pH 7.5, 1 mM ZnCl, 20 mM DTT, 132 µM ATP, 30% DMSO and 1 U/µl T4 DNA ligase. The ligation proceeded at 27°C for up to 120 min. In order to remove excessive background of the unreacted, 5′-P-labeled oligos (the P-label is always in the middle of the ligation product), 5 µl aliquots of the ligation mixture were treated with 0.1 U/µl calf-intestine alkaline phosphatase (Boehringer-Mannheim) and 0.1 U/µl RNase H (Epicenter Technologies) at 37°C for 10 min. Alkaline phosphatase was used to reduce the background derived from the unligated P-oligo substrates, and RNase H was used to reduce the interference of rRNA in the subsequent gel analysis. The reaction was quenched with the addition of an equal volume of 9 M urea/50 mM EDTA. The mixture was boiled for 2 min and rapidly cooled on ice prior to its loading on denaturing polyacrylamide gels containing 7 M urea.
Recently, we reported a ligation-based method for the detection and quantification of 2′--methyl modifications in RNA that can overcome the limitations of the current methods (). Our approach uses the RNA as a template to direct the enzymatic ligation of two adjunct oligodeoxynucleotides (B blue strands) designed to form Watson–Crick base pairs with sequence elements immediately upstream and downstream of the modification site (B, open circle). The residue that opposes the modification site in the template/substrate ternary complex (herein referred to as the ‘recognition residue:’ B, blue circle) influences the ligation efficiency. An oligonucleotide bearing an appropriate recognition residue for a 2′--methylnucleotide (2′-Me-A, G, C or U) ligates with an efficiency that depends significantly on the presence or absence of the modification. The ligation yield then correlates with the fraction of RNA bearing the modification. Because the ligation reaction for each modification site produces a DNA oligonucleotide of unique sequence, we can adapt this approach to a microarray platform to enable analysis on a genome-wide scale (for example, all known Ψ-modification sites in rRNA). The central challenge in extending this approach to other modification types therefore involves the identification of recognition residues for specific modification types.
In previous work, we screened empirically a collection of 24 oligonucleotide pairs to identify recognition residues for all four 2′--methyl nucleotides (2′-Me-A, C, G, U) within RNA (). We found that when the 2′-Me-nucleotide opposes the complementary 2′-deoxynucleotide, T4 DNA ligase works much less efficiently compared to when the corresponding unmodified (2′-OH) nucleotide opposes the recognition residue. In hindsight, we could have rationalized these findings by the distortion in A-form helix backbone geometry caused by the 2′H–2′Me (modified RNA) base pair as compared to the 2′H–2′OH (unmodified RNA) base pair.
Previously, we established that the reaction catalyzed by T4-DNA ligase can detect and quantitate methylation of the 2′-hydroxyl group at specific sites in RNA if the substrates contained the appropriate recognition residue. In that work, we identified the recognition residue by screening a collection of oligonucleotide ligation substrates bearing commercially available nucleoside analogs at the recognition position. An analogous screen for pseudouridine using a larger collection of oligonucleotide substrates failed to identify a suitable recognition residue. In the current work, we used available structural information pertaining to duplexes containing the modified or unmodified nucleosides together with the principles of molecular recognition to identify recognition residues that enable ligation-based detection and quantitation of Ψ and mA modifications in RNA. As pseudouridylation and base and hydroxyl group methylation represent the most subtle post-transcriptional RNA modifications, our ligation-strategy appears robust in its ability to discriminate rather subtle chemical changes in RNA. We expect that our approach will be generally applicable for many other, if not all, RNA modifications.
The major advantages of the molecular recognition/enzymatic ligation approach to study RNA modifications include the ability to quantify the extent of modification at multiple defined positions simultaneously in the same biological sample and its adaptability to high-throughput study of modifications at all sites. Because the ligation reactions generate DNA oligonucleotides of unique sequence in yields that reflect the fraction of biological RNA modification at each site, microarray technologies can provide a well-established platform for high-throughput analysis. With this technology in hand, we can address questions that require information about the global landscape of RNA modifications within cells. For example, we can begin to assess how the 46 Ψ modifications in yeast rRNA work synergistically for ribosome function, and how the extent of modification at these sites changes as a function of physiological state.
p
p
l
e
m
e
n
t
a
r
y
D
a
t
a
a
r
e
a
v
a
i
l
a
b
l
e
a
t
N
A
R
O
n
l
i
n
e
. |
Type II topoisomerases are ubiquitous ATP-dependent enzymes that catalyze the transport of one DNA segment through a transient double-stranded break in a second segment (). Their role in the cell is to control the supercoiling of DNA and to untangle the catenanes that arise during replication or recombination (). Bacteria express two type IIA topoisomerases, DNA gyrase (Gyr) and DNA topoisomerase IV (Topo IV) with distinct and unique functions (,). Gyr introduces negative supercoils into DNA, maintaining bacterial chromosomes in an underwound state to promote compaction and unwinding (,). In contrast, Topo IV is an efficient decatenase, separating tangled daughter chromosomes following replication (). Both Gyr and Topo IV are heterotetramers, comprising two subunits each of GyrA and GyrB, and ParC and ParE, respectively (). The GyrA and ParC subunits are homologous in their N-terminal domain that carries the DNA breakage-reunion function, but diverge in their C-terminal regions that also contribute to DNA binding (). The GyrB and ParE proteins contain the ATPase site involved in energy transduction and are closely homologous, in both primary sequence and structure.
Transient double-stranded DNA breakage by Gyr and Topo IV proceeds via an enzyme–DNA intermediate termed the ‘cleavage complex’ involving a 4-bp staggered break and covalent attachment of GyrA or ParC subunits to each 5-prime (5′) DNA-end through a phosphotyrosine linkage (,). Protein–DNA linkage preserves the energy of the DNA phosphodiester bond and allows resealing of the DNA backbone by attack of the 3-prime (3′) OH-ends of the broken DNA. The breakage-reunion equilibrium is overwhelmingly in favor of the intact nucleic acid chain to avoid fortuitous lethal events . However, the cleavage complex can be stabilized by drugs such as quinolones producing DNA breaks, which can be detected after enzyme denaturation with sodium dodecyl sulfate. Generally, DNA cleavage is sequence-selective and is characterized by covalent bonding of the nucleic acid to GyrA or ParC (). Identification of DNA breakage site specificity is important for the physiological functions of bacterial topoisomerases and for their roles as quinolone targets.
Cleavage determinants have been recently determined for Gyr and Topo IV from a Gram-positive bacterial species, , in the presence of various clinically relevant quinolones (). The cleavage consensus sequences were GNG(G/c)(A/c)G*GNNCtTN (C/a) for Gyr and G(G/c)(A/t)a*GNNCt(T/a)N(C/a) for Topo IV (where the asterisk denotes the cleavage site, capital letter indicates preferred base, lower case letter denotes unfavored base and N indicates no base preference). Similar cleavage preferences were observed among the quinolones tested. Even though the analysis was extremely comprehensive, some aspects remain to be answered (): (i) The consensus sequence obtained with both enzymes is strikingly symmetric: does this property derive from actual enzyme requirements or does it arise from asymmetric sites as a consequence of analyzing just one strand of the double helix? (ii) Base prevalence was analyzed independently at each position: we do not know yet if the full consensus sequence is required for efficient DNA cleavage or if only part of it is necessary. (iii) If the entire consensus sequence, which spans a 12-base-long DNA tract, is needed for efficient DNA processing, this relatively long DNA segment should be statistically represented. Statistical analysis would hence benefit from using a more assorted DNA template repertoire.
For Gram-negative bacteria, site-specific DNA breakage details are much less exhaustive. Even for topoisomerases, the data are scattered and discordant; for Gyr the most extensive analysis to date involved only 19 sites induced in on plasmid pBR322 and generated a 20-bp consensus (); for Topo IV just one study has been performed, which mapped a weak 2-bp consensus ().
This study aimed to address the following points: (i) investigation and comparison of the cleavage consensus sequence exhibited by topoisomerases IIA from Gram-positive () and -negative species () on a large selection of DNA templates; (ii) comparison of cleavage sites generated by Gyr and Topo IV enzymes; (iii) assessment of the minimal sequence requirements and (iv) symmetric/asymmetric properties for effective DNA recognition and scission by bacterial topoisomerases in the presence of quinolones.
We found discrete preferred sequences for each enzyme analyzed. In particular, consensus sequences were very similar when comparing different enzymes (i.e. Gyr and Topo IV) from the same bacterial source, but partly different using enzymes from diverse origins (i.e. and ). The chemical nature of the fluoroquinolones used to stabilize the DNA/enzyme complex in each case did not play a significant role. We proved that symmetry was a crucial property of the recognized DNA sequence and that an extended G/C-rich sequence improved efficient DNA processing. Specific bases inside the 4-bp staggered break were required, possibly because of their interaction with quinolones, however, an important role in enzyme recognition was also played by bases outside the enzymatic cleavage site, Finally, we provided evidence that the peculiar consensus sequence suggested by our study adopts an unusual double-stranded DNA conformation which could trigger enzyme-DNA recognition.
Ciprofloxacin and Gemifloxacin were provided by GlaxoSmithKline (Verona, Italy). Stock solutions were made in mQ-grade water and diluted to the working concentration in the desired buffer. Buffer components were purchased from Sigma-Aldrich (St. Louis, MO, USA). Electrophoretic reagents, dNTPs and polymerase were from Amersham Biosciences Europe (Freiburg, Germany). [γ-P]ATP was from Perkin Elmer (MA, USA); T4 polynucleotide kinase and DNA ligase were purchased from Invitrogen (Paisley, UK). Restriction enzymes were purchased from New England Biolabs (MA, USA).
Topo IV and Gyr from were purified as previously described (,,). Topo IV and gyrase from were purified as described in (,).
Plasmid pBluescript II KS(+) and SV40 DNA were purchased from MBI Fermentas (MD, USA) and Invitrogen (Paisley, UK), respectively. Primers were purchased from Eurogentec (Liege, Belgium). Primers for sequencing analysis used to amplify 239-bp PCR fragments on SV40 DNA template are described in (). The amplified fragments encompass ∼27% of the complete SV40 sequence. Consensus oligos cloned in pBluescript II KS(+) are: AGGG SYM F: 5′-GCC GGG ATC CTT ACT TCA TGA ATT ATGA TAA TAC CTT AAG AAT TCG CGC-3′ and its complementary (AGGG SYM R); AGGG N-SYM F: 5′-GCC GGG ATC CTT ACT TCA TGA ATT ATGA TAA TAC CTT AAG AAT TCG CGC-3′ and its complementary (AGGG N-SYM R); GGAGGG F: 5′-GCC GGG ATC CTT ACT TCA TGA ATT TAA TAC CTT AAG AAT TCG CGC-3′ and its complementary (GGAGGG R); GGATGG F: 5′-GCC GGG ATC CTT ACT TCA TGA ATT TAA TAC CTT AAG AAT TCG CGC-3′ and its complementary (GGATGG R); GGAGTG F: 5′-GCC GGG ATC CTT ACT TCA TGA ATT TAA TAC CTT AAG AAT TCG CGC-3′ and its complementary (GGAGTG R); GGAGGT F: 5′-GCC GGG ATC CTT ACT TCA TGA ATT TAA TAC CTT AAG AAT TCG CGC-3′ and its complementary (GGAGGT R). Consensus sequences are underlined.
All consensus sequence oligos (57 bp) contained BamHI and EcoRI restriction sites at their 5′- and 3′-end, respectively, for insertion into EcoRI- and BamHI-cut pBluescript II KS(+). Consensus oligos contained an additional AflII site flanking the 3′-end of BamHI site to facilitate screening of mutant colonies. Restricted oligos and plasmid were purified with Nucleotide Removal kit and Gel Extraction kit (Qiagen, CA, USA), respectively, and ligated at 16°C overnight. Ligation mixtures were transformed into competent DH5α cells. Bacteria were grown in the presence of IPTG and X-Gal, and white colonies were selected to confirm the presence of plasmid with the inserted oligo of interest [p(consensus oligo)Bluescript II KS(+)].
For primer labeling, 100 pmol of primer were incubated with 2 μl (10 μCi/μl) of [γ-P]ATP and 10 units of T4 polynucleotide kinase in 50 mM Tris–HCl (pH 7.5), 7 mM MgCl and 10 mM DTT, at 37°C for 30 min. After incubation, DNA was purified through MicroSpin G-25 columns (Amersham Biosciences Europe). The labeled primers were used in PCR to amplify 239- or 215-bp fragments of SV40 or each p(consensus oligo)Bluescript II KS(+), respectively. Each PCR reaction was prepared by mixing 200 μM dNTPs, the labeled primer, 100 pmol of the cold primer, 50 ng of template DNA and 5 units of polymerase in PCR reaction buffer [10 mM Tris–HCl (pH 9.0), 50 mM KCl and 1.5 mM MgCl] to a final volume of 100 μl. PCR cycles for both SV40 and p(consensus oligo)Bluescript II KS(+) templates were 94°C for 30 s, 55°C for 30 s and 72°C for 30 s (30 cycles). DNA fragments were then purified with a QIAquick PCR purification kit (Qiagen, CA, USA).
Topo IV from was reconstituted with 0.45 μg ParC and 1.7 μg ParE in reaction buffer [40 mM Tris–HCl (pH 7.5), 6 mM MgCl, 10 mM DTT, 200 mM potassium glutamate, 50 µg/ml bovine serum albumin (BSA)]. Reaction buffer for Gyr from (1U/μl) was 35 mM Tris–HCl (pH 7.5), 24 mM KCl, 4 mM MgCl, 2 mM DTT, 6.5% glycerol, 0.1 mg/ml BSA. Topo IV from was diluted to a final concentration of 0.22 mg/ml ParC and 0.44 mg/ml ParE in dilution buffer [50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM DTT, 40% (w/v) glycerol, 1 mM EDTA]. Reaction buffer consisted of 40 mM Tris–HCl, pH 7.5, 20 mM KCl, 6 mM MgCl, 1 mM DTT, 10 mM spermidine, 0.5 mg/ml BSA. Gyr from was diluted to 0.35 μM (0.1 U/μl) in dilution buffer [50 mM Tris–HCl (pH 7.5), 100 mM KCl, 2 mM DTT, 50% glycerol, 1 mM EDTA]. Reaction buffer consisted of 7 mM Tris–HCl (pH 7.5), 5 mM KCl, 0.8 mM MgCl, 0.4 mM DTT, 1.3% (w/v) glycerol, 0.1 mg/ml BSA. One unit of Gyr is defined as the amount of enzyme that supercoils 0.5 μg of relaxed pBR322 DNA in 30 min at 37°C; one unit of Topo IV is the amount of enzyme that relaxes 0.4 μg of supercoiled pBR322 DNA in 30 min at 37°C.
For DNA cleavage assays, reconstituted topoisomerase enzymes were incubated with labeled PCR product DNA in 25 µl reaction mixtures in the presence of different drug concentrations. After a 30-min incubation at 37°C, 1 µl of 10% SDS and 2 µl of a 20 mg/ml stock of proteinase K were added, and incubation continued for 30 min at 45°C. Samples were precipitated with ethanol, resuspended in formamide gel loading buffer [95% formamide, 200 mM EDTA pH 8.0, 0.1% (w/v) xylene cyanol and 0.1% (w/v) bromophenol blue] and heated at 90°C for 2 min. Cleavage products were resolved in 8% polyacrylamide, 7 M urea gels, alongside DNA markers obtained with the dideoxynucleotide chain termination protocol. Briefly, G, C, A and T dideoxy-nucleotide ladders were obtained by incubating 1 μl of labeled primer, 4 μl of ddNTP/dNTPs extension/termination mixture (), 50 ng template DNA, 50 mM KCl, 1.5 mM MgCl, 10 mM Tris–HCl (pH 7.5), 5 U DNA polymerase in a volume of 10 μl at 95°C for 30 s, 54°C for 30 s, 72°C for 1 min for 25 cycles and at 95°C for 30 s, 72°C for 2 min for 10 cycles. Reactions were purified by ethanol precipitation. Gels were then transferred to Whatman 3MM filter paper, dried under vacuum at 80°C, and bands were visualized by phosphorimaging analysis (Molecular Dynamics, Amersham Biosciences Europe). Quantification was performed by ImageQuant software (Molecular Dynamics). Each cleavage site was quantified as the ratio of the intensity of the cleavage band over the intensity of the corresponding band/position in the control lane (enzyme without drug) (intensity ratio = IR). Only sites with IR > 40 were considered in the statistical analysis.
The probability () of deviation from expectation of each nucleotide at each position around the cleavage site (positions −20/+20) was calculated as previously described in (,).
Reference, sample1, sample2 and control oligos were dissolved in 10 mM Na.Cacodylate () to a final concentration of 90 μM per residue. Circular dichroism spectra from 220 to 320 nm were recorded at 4°C using 10 mm path length cells on a Jasco J810 spectropolarimeter equipped with a NESLAB temperature controller and interfaced to a PC100. Observed ellipticities were converted to mean residue ellipticity [θ] = deg cmdmol (Molar Ellipticity).
To analyze and compare DNA sequence determinants of fluoroquinolones and prokaryotic topoisomerase complexes, we used (i) two clinically approved fluoroquinolones, i.e. ciprofloxacin and gemifloxacin (). Ciprofloxacin, a second-generation fluoroquinolone, was used as a reference compound with assessed activity against Gram-negative and -positive bacteria; gemifloxacin was included as a new IIIb generation fluoroquinolone with increased activity against Gram-positive species. (ii) Both Gyr and Topo IV from and were employed as representative of Gram-negative and -positive species, respectively. (iii) SV40 DNA was used as a generic nucleic acid template to avoid bias in species-specific enzyme/DNA recognition.
Six 239-bp long SV40 segments were 5′-end P-labeled and amplified by PCR reaction to obtain significant coverage of the whole SV40 sequence and processing of a total of 1434 bp. The forward or reverse strand were labeled and analyzed independently to gain information on the symmetry of enzyme-mediated DNA cleavage. All DNAs were incubated with - or -derived Gyr or Topo IV in the presence of ciprofloxacin or gemifloxacin, and the cleavage patterns were analyzed by loading reaction mixtures on denaturing urea-polyacrylamide gels to separate DNA fragments (A). For each enzyme/drug combination, we were able to collect an adequate amount of cleavage sites (i.e. around 30–50 per each strand). Site number was similar between ciprofloxacin and gemifloxacin with enzymes, while it was in favor of gemifloxacin with proteins. Sequences of cleavage sites were aligned at the point of the phosphodiester bond break in the 5′ → 3′ orientation, and bases immediately 5′ and 3′ to the cut were numbered −1 and +1, respectively.
The deviation of base distribution from the expected SV40 DNA base frequencies was evaluated at each position (±20 from the cleavage site) by χ analysis: a core region of non-random base composition was found between positions −4 and +8. Probabilities of the observed base frequency deviations from expectation at each position were assessed for all cleavage sites on the forward strand, and non-complementary sites on the reverse strand, and vice versa, for each enzyme/drug combination (statistical analysis for Topo IV/ciprofloxacin from and Gyr/gemifloxacin from , positions −4/+8, are reported in B and C, as examples). The complete list of favored and disfavored bases (−LogP > 2 and −LogP < −2, respectively) is reported in .
For the enzymes, a strong and consistent preference was observed for positions −1G, +1G, +2G, +3C, +5C (). Position +4 also exhibited a consistent preference for C (6 out of 8 enzyme/drug/DNA strand combinations), but with a slightly lower statistical significance (1.5 < −LogP < 2). Positions −2/−4 and +6/+8 indicated a very mild preference for G and C, respectively (0.5 < −LogP < 1.5). No essential cleavage differences were found between ciprofloxacin and gemifloxacin. As for the two enzymes, the requirements for Topo IV appeared to be somewhat more stringent than for Gyr, especially for positions −1 and +5. Positions outside the −4/+8 interval did not show distinct preferences. Disfavored bases were much less regularly repeated: inside the 4-bp staggered break only position +2A showed an indication of non-preference ().
A thorough statistical analysis for the enzymes has already been reported by Leo (). Nonetheless, we performed cleavage analysis on SV40 DNA substrate to: (i) compare results obtained with and enzymes using the same drugs and DNA template, (ii) compare results with the same enzymes on different DNA templates [i.e. SV40 and genome ()] and (iii) address the issue of consensus sequence symmetry by analyzing both cut DNA strands. Only one drug was tested with each enzyme (ciprofloxacin with Topo IV and gemifloxacin with Gyr) because it had already been shown that fluoroquinolone identity does not play a critical role in DNA recognition (). In accordance with the previous analysis (which used −LogP > 3 and −LogP < −3 as significant), we found that highly preferred bases for both Topo IV and Gyr were −2A and +6T; +1G and +4C, and −1G (gyrase only) were also significantly represented. In addition, +3C and +5C were favored (at −LogP > 2) (). No steady preference was observed for other positions, even though positions −4, −3 and +7, +8 showed a mild inclination for G and C, respectively (0.5 < −LogP < 1.5). Non-preference data were again less stringent: consistency and significant –LogP values (−LogP < −2) were found just for –2T and +6A ().
Our statistical analysis indicated a clear-cut consensus sequence that was in part different between - and -derived enzymes. In particular, GGGCCC sequence at the −1/+5 interval was obtained with the Gram-negative enzymes, with enhanced preference for positions −1 and +5, +2 and +3. On the opposite, the most preferred bases with proteins were −2A and +6T. Positions –1/+5 were GGGCCC as with topoisomerases, but this preference was less statistically represented. In neither case, a major difference was noticed between Topo IV and Gyr or between ciprofloxacin and gemifloxacin.
To answer these questions, we focused on the enzymes and their preferred sequences: we cloned the −1/+5 GGGCCC consensus sequence into a plasmid vector and subsequently amplified a 215-bp DNA fragment by PCR. Sequences upstream (−8/−2) and downstream (+6/+13) the consensus sequence were arbitrary chosen to be A/T rich (see Materials and Methods section). A 215-base long nucleic acid was employed to achieve optimal topoisomerase processivity: in fact DNA Gyr is reported to protect as much as 150 bp on binding to DNA (,).
Cleavage analysis was performed on the symmetric and non-symmetric sequences in the presence of Topo IV or Gyr and ciprofloxacin or gemifloxacin. Data for the Topo IV/ciprofloxacin combination are shown in . Cleavage was indeed obtained corresponding to −1G of the inserted consensus sequences in the symmetric oligonucleotide (lane 3, AGGG SYM oligo, ), but this was relatively mild compared to cleavage sites on other portions of the tested sequence. In contrast, no cleavage was observed at −1G in the non-symmetric consensus sequence (lane 3, AGGG N-SYM, ). It should be noted that all cleavage bands were shared between the symmetric and non-symmetric DNA. Similar results were gained with the remaining enzyme/drug patterns (data not shown).
While it is now clear that symmetry is crucial to improve topoisomerase/fluoroquinolone-mediated DNA cleavage, the GGGCCC consensus sequence did not stand out as a hot spot for induced DNA scission. The consensus sequence found in the present study was limited to positions between −1 and +5, and the region just outside this interval (i.e. −8/−2 and +6/+13) was arbitrary chosen to be extremely A/T rich (14 out of 15 bases were A or T, see Materials and Methods section) in the DNA oligonucleotide used for the cleavage assay. We hypothesized that these flanking non-random regions could decrease the cleavage efficiency of the consensus sequence. In fact, our current statistical analysis on both and topoisomerases showed a preference for −4G, −3G, −2A/G, +6T/C, +7C, +8C, which was not considered before because at a low statistical significance (i.e. −LogP < 2) (B and C). In addition, it had previously been demonstrated that enzymes tested on a different DNA template (), preferred G/C-rich tracts outside the cleavage site (positions −4, −3 and +8). We hence introduced this additional information in our current consensus sequence, obtaining the GGAGGGCCCTCC DNA stretch (named GGAGGG oligo) spanning positions −4/+8.
The new four extended and modified consensus oligonucleotides were inserted into plasmid vectors, amplified to obtain 215 bp DNA tracts and processed as described above. It should be noted that DNA tracts outside positions −4 and +8 remained unchanged compared to the previous AGGG SYM oligo. Cleavage experiments with Topo IV and Gyr are shown in A and B, respectively.
In the presence of the new oligo substrates, unique and extremely strong cleavage bands corresponding to position −1G of the consensus sequences were produced by Topo IV and fluoroquinolones (lanes 3 and 4, A), indicating that A/T at positions −4, −3 and +7, +8 in fact decrease DNA processing. In particular, the −1G cut site was the most intense with both fluoroquinolones in the case of the wild-type consensus-derived oligo. In the mutated sequences, gemifloxacin-induced cleavage was more prominent with GGAGGT and GGAGTG sequences, while ciprofloxacin was more efficient with the GGATGG oligo (compare lanes 3 and 4, all oligo, A). Gemifloxacin induced an extra band at position +2 in oligos GGAGGG, GGAGGT and GGATGG, respectively (lanes 4, * symbol, A). This cleavage site was particularly intense in the case of oligo GGAGGT, where T, instead of G, is present at position +2. An additional band was produced by gemifloxacin at position +5T in oligos GGAGGT and GGATGG, respectively (lanes 4, ÷ symbol, A). A few minor sites were present towards the DNA 3′-end, which were shared by all oligos.
When employing Gyr (B), the results were quite different. Most strikingly, while cutting at position −1G in the consensus sequence was still present; this was not the most intense band along the DNA sequence. As observed for Topo IV, the relative intensity of strand scission at position −1 was higher in the wild-type oligo (GGAGGG) compared to the mutant sequences (lanes 3 and 4, B). Ciprofloxacin-induced cleavage was stronger than gemifloxacin-mediated breakage in the case of oligo GGAGGG and GGATGG, while the reverse was found for oligos GGAGGT and GGAGTG. Additional bands were found both inside (downstream position −1) and outside the consensus sequence (both upstream and downstream). The wild-type oligo showed fewer cleavage bands inside the consensus sequence.
Next, we wondered if the above described peculiar consensus sequence could adopt a definite structure in solution. It is reported that the DNA octamer GGGGCCCC forms in aqueous solution a stable heteronomous DNA duplex where guanine–guanine stacking is A-like, whereas cytosine bases stack in a B-like fashion (). To examine the actual conformation of the consensus sequence primarily recognized by Topo IV (A), we compared the CD spectrum of our sample1 oligo (GGAGGGCCCTCC), with that of the reference oligo (GGGGCCCC) and of a control DNA (GCGACGCGTCGC), which shares the same base composition of the sample1 oligo, but displays a different sequence. In addition, sample1 oligo was compared to sample2 oligo (ATAGGGCCCTAT) where flanking regions were A/T rich such as those used in the cleavage assay of . Under native conditions ( well below , i.e. 4°C), the spectrum of the reference oligo exhibited two strong overlapping positive bands at 285 nm and at 260 nm, and a negative peak at 237 nm (). The CD of our sample1 oligo was essentially superimposable to the reference oligo, showing the same bands with very similar intensities. Conversely, the control and sample2 oligos displayed substantially different CD patterns with one positive band at 285 or 265 nm, respectively, and a negative peak at 255 or 240 nm, respectively. We conclude that the native consensus sequence adopts in solution a peculiar conformation, which closely resembles the heteronomous conformation of the reference oligo [i.e. a combination of B-type and A-type base stacking ()], whereas the control sequence is organized in a more classical B form. Sample2 oligo, on turn, exhibits a spectrum which reflects a lower propensity to adopt the heteronomous conformation thus favoring a B form. This is in agreement with the shift of the CD cross-over towards longer wavelength as an indication of lower A-like character and with the crystal structure of the dodecamer duplex CATGGGCCCATG showing a structural continuum along the transition between A- and B-DNA (,).
Type IIA DNA topoisomerases have concerted and complementary functions in controlling bacterial chromosome topology: Gyr catalyzes the negative supercoiling of closed-circular DNA, thus, it removes the positive supercoils that arise during DNA unwinding; Topo IV relaxes positive and negative supercoils, and unlinks daughter chromosomes at cell division (). In addition, topoisomerases are crucial targets for fluoroquinolone-based therapy. In this instance, Gyr or Topo IV can be alternative selective targets depending on the species of origin: Gyr is often the primary target in Gram-negative species, such as , while Topo IV is principally affected in Gram-positive species, such as (). However, it has also been reported that topo IV and/or gyrase can be alternative intracellular target depending on quinolone structure (,).
To investigate whether this double biased behavior (Gyr versus Topo IV and versus ) could arise from a distinctively specific recognition of DNA sequences by topoisomerases (± quinolones), we comprehensively studied and compared the DNA sequence specificity of Gyr and Topo IV from and in the presence of clinically used fluoroquinolones. In view of the key role played by these proteins, the resulting information would be important from both physiological and therapeutic standpoints.
A consensus sequence for the two topoisomerases from has been recently reported using ParE-ParC gene sequences as DNA templates (). Here, to impartially compare DNA sequence specificity of enzymes from different species, we used non-specific DNA fragments derived from the SV40 genome.
By statistical analysis of preferred bases around the cleavage position (−20/+20) of a large set of sites (over 400), we found that the GGGCCC sequence at positions −1/+5 was favored with both Gyr and Topo IV from the two bacterial species considered. However, this preference was much more prominent for the enzymes. Conversely, positions −2 and +6 were non-randomly selected only with enzymes, which revealed a striking preference for A and T, respectively.
Overall, statistical analysis of consensus sequences reported no major differences between Topo IV and Gyr, although the requirements for the latter appear to be less stringent in the Gram-negative species. By contrast, discrimination of Gram-positive and -negative enzymes apparently resides on different DNA sequence requirements for efficient cleavage, in particular at positions −2 and +6.
The high structural similarity of the N-terminal binding regions of ParC and GyrA from both and species (,) may explain the comparable DNA recognition properties shared by the tested topoisomerases in the presence of the two quinolone drugs. As an added proof, the aminoacidic residues reported to be essential in the protein catalytic activity and implicated in quinolone resistance, hence, likely involved in the concurrent binding of topoisomerases to DNA and quinolones in the cleavage complex (i.e. Arg32, Arg47, His78, His80, Ser83 and Asp/Glu87 in GyrA and corresponding positions in the other topoisomerases) (), are identical in the four tested enzymes. It is intriguing to note that the only consistent difference in the aminoacidic sequence between and N-terminal region of GyrA/ParC subunits is at position 84 of GyrA (and corresponding positions in the other enzymes), where an Ala is substituted with a Ser in the Gram-positive proteins. Since position 84 is just adjacent to other positions involved in enzyme/quinolone/DNA interactions (see above), it could be speculated that the introduction of an H-bond donor site (i.e. Ser) modulates topoisomerases DNA recognition properties, driving sequence specificity towards A/T at positions −2/+6.
The fluoroquinolone chemical nature did not play a major role in determining the consensus sequence, at least for the tested compounds. However, it influenced the extent of cleavage stimulation. In particular, sites number was similar between ciprofloxacin and gemifloxacin with enzymes, while it was clearly in favor of gemifloxacin with proteins. These results are in line with the reported activity of the two drugs on bacterial species: gemifloxacin is more effective than ciprofloxacin on strains, while the activity of the two drugs is similar on bacteria (,). Therefore, our data indicate that, from a pharmacodynamic point of view, the biased activity of the two drugs may rest on their different ability to poison the molecular target, rather than on recognizing a different sequence determinant. This implies that DNA recognition is initially driven by the topoisomerases and then modulated by quinolone stabilization of the cleavage complex.
Our data, performed on SV40 DNA template, differ slightly from those obtained previously with the genome DNA template (). In the first report, the consensus found between positions −4 and +8 were GGAN*GNNCNTNC and GGAG*GNNCNTNC, asterisk at the cleavage site, for Topo IV and Gyr, respectively (), compared to NNAN*GNCCCTNN and NNAG*GNCCCTNN, respectively, in the present work. Noteworthy, several preferences match exactly, whereas others are not observed in one case or the other. This difference depends in part on the level of statistical stringency considered: on lowering the −LogP threshold, +2G and +3C bases become preferred in () and −4G, −3G and +8C register as slightly preferred with both Gram-positive and -negative enzymes in this work (B and C). The combination of different DNA template, enzyme activity and drug concentration could be responsible for the diverse statistical relevance found. In fact, DNA traits consisting of two or more bases can be differently represented within nucleic acids from various origins. For instance, even though SV40 and ParC/ParE DNAs share similar base composition (G/C percentage is 40% in SV40 and 42% in ), GGG and CCC sequences are 27% more frequent in the SV40 genome than in the ParC/ParE genes.
Since each position is independently taken into account during statistical analysis, we next analyzed topoisomerase–quinolone cleavage efficiency on the full DNA consensus sequences found in this work or obtained by combining data from the present and previous research (). Further, mutations at selected preferred positions were also introduced to determine the contribution of each site on the cleavage complex DNA processivity.
We found that the −2/+6 consensus sequence AGGGCCCT environed in G/C-, but not A/T-, rich flanking regions, is able to induce highly efficient DNA cleavage. In fact, the 8-nt long tract AGGGCCCT at positions −2/+6, environed in an A/T-rich region, was able to promote only weak DNA damage (). In this instance, cleavage intensity was similar among the two enzymes and the two quinolones employed. In contrast, addition of G and C bases at positions −4, −3 and +7, +8, respectively (), significantly augmented cleavage, especially with Topo IV (see A and B). The presence of the full consensus sequence was still required: in fact, modification of a single G nucleotide in the −1/+5 region generally decreased cleavage, in the same manner for Topo IV and Gyr. In addition, whereas the intensity of cleavage with the wild-type sequence was similar between ciprofloxacin and gemifloxacin, differences in cutting efficiency were observed at the mutated sequences with the two fluoroquinolones. In particular, positions inside the 4-base staggered break seem primarily implicated in quinolone binding. In fact, changes at positions +2, +3 and +1, +4 more dramatically impaired ciprofloxacin activity, whereas gemifloxacin was more severely affected by nucleotide changes at +1 and +5. Therefore, gemifloxacin requirements seem to be somewhat less stringent than those of ciprofloxacin when modifying the quinolone-interacting region of the consensus sequence. These results are also supported by reports showing an increased binding ability of quinolones to G/C-rich sequences () and site-specific analysis of Gyr with oxolinic acid ().
Our data further suggest that positions −4, −3 and +7, +8, which must be G/C rich, are likely required for efficient DNA recognition and binding by the topoisomerases, since these positions significantly affect cutting intensity depending upon the enzyme, but not upon the drug. The consensus found is particularly effective for Topo IV, less so in the case of gyrase, for which additional features outside positions −4/+8 might play a key role in specificity. Positions −2 and +6 are also probably involved in enzyme recognition. In particular, they help discriminate between proteins from different bacterial sources (e.g. versus ), but not between the two topoisomerases (see above).
Symmetry is an additional feature required for efficient DNA processing, as proven by analyzing cleavage sites on the two complementary DNA strands and by testing symmetric and non-symmetric consensus sequences. This property reasonably arises from the dyadic symmetry of topoisomerase complexes.
The consensus primary sequence found is very peculiar. It is known that selected nucleic acid sequences can confer unusual 3D characteristics to the DNA (,). Indeed, it is reported that Gn/Cn DNA sequences can be crystallized in A/B intermediate conformations (). More interestingly, the GGGGCCCC sequence adopts a partial A-like structure in aqueous solution. In particular, G bases are stacked in a DNA-A form, whereas the Cs maintain a canonical B-form (). Surprisingly, in the reverse sequence CCCCGGGG, both C and G bases fold in a different A-like and less stable fashion (). By circular dichroism analysis, we proved that our consensus sequence GGAGGGCCCTCC folds as the reference GGGGCCCC DNA. A partial A-like conformation is likely to protect G/C-rich regions, which are involved in regulating transcription of many human genes (), against mutations: in fact these are mutated much less than A/T-rich regions (). According to our findings, this particular A/B DNA conformation could also prompt DNA recognition by topoisomerases. It should be noted that the reverse sequence CCCCGGGG, which folds up in a full-A-like structure and hence presents a different environment for enzyme recognition, was not selected by statistical analysis of cleavage sites. Topoisomerases would likely recognize −4, −3, −2 and +6, +7, +8 nt, permitting cleavage at −1 and +5 positions. Quinolones would then be able to bind between the two loose ends with subsequent inhibition of the enzymatic DNA resealing.
Finally, our DNA specificity results using the test quinolones suggest that the therapeutic efficacy against a bacterial species should be related to the frequency of occurrence of G/C-rich sequences in the bacterial genome and to the efficiency of cleavage complex stabilization by the drug. Clearly, the observed clinical outcome is also expected to depend upon the ADME properties of each individual quinolone.
In summary, this work provided evidence that (i) a symmetric G/C consensus sequence at positions −1/+5 allows efficient bacterial topoisomerase-mediated DNA processing; (ii) G/C over A/T-rich flanking sequences (positions −4, −3 and +7, +8) are highly preferred in protein DNA processing, which is particularly efficient for Topo IV; (iii) positions −2 and +6 interact with the enzyme and distinguish from topoisomerases; (iv) positions −1 to +5 participate to drug activity and may modulate cleavage intensity based on the quinolone chemical nature; (v) symmetry of the target sequence is a key trait to promote effective cleavage and (vi) the consensus sequence adopts a heteronomous A/B conformation, which may trigger efficient DNA processing by enzyme/drug complex. |
Histones are a highly conserved group of DNA-binding proteins required for the condensation of DNA in the eukaryotic nucleus. Coordinated regulation is required to ensure stoichiometric o production of the core histones in a time frame within the cell cycle coincident with DNA replication. The majority of histones are therefore regulated in a cell cycle manner and are termed replication-dependent histones (,). During S-phase of the cell cycle there is an accumulation of histone mRNAs, which are then rapidly degraded upon entry into G2. Regulation of histone mRNA levels results from control of histone gene transcription, mRNA 3′-end formation and mRNA degradation ().
In higher eukaryotes, replication-dependent histone mRNAs are non-polyadenylated and the formation of the 3′-ends of the mRNAs contributes to their accumulation during the S-phase of the cell cycle. The 3′-ends of the mRNAs are defined by a highly conserved stem-loop (S-L) structure () and a loosely defined purine-rich element () that lie immediately upstream and downstream of the 3′-end cleavage sites, respectively. A number of proteins including a stem-loop-binding protein (SLBP) (,) and a 3′ to 5′ exonuclease (h3′exo) () interact with the S-L structure while the U7 snRNP interacts with the downstream element (). Communication between the S-L and the downstream element is mediated by a zinc-finger protein (). Recent studies have revealed that components of the general mRNA 3′-end cleavage and polyadenylation machinery are also required for metazoan histone mRNA 3′-end processing (,). The latter finding provides the first indication that common molecular machinery is involved in the processing of both polyadenylated and non-polyadenylated mRNAs.
Histone mRNA biogenesis is also cell cycle regulated in lower eukaryotes, such as fungi and protozoans, however, here the 3′-ends of histone mRNAs are polyadenylated (,,). Both transcription and post-transcription events contribute to the accumulation of histone mRNAs in S-phase (,). Currently, little is known regarding the contribution of the individual steps in mRNA biogenesis towards the cell cycle accumulation pattern of histone mRNAs in lower eukaryotes. In a previous study, we have identified a purine-rich region, termed the distal downstream element (DDE), which lies ∼100 nt downstream of the 3′-end cleavage sites for the H2B-encoding mRNA, HTB1. Using a cell cycle-regulated chimera gene, neo-HTB1 (), which contains the 3′ UTR, 3′-end cleavage sites and downstream sequences of the HTB1 gene fused to an open reading frame of a neomycin phosphotransferase gene, we showed that mutations in the DDE appear to prevent the cell cycle-dependent degradation of neo-HTB1, leading to its accumulation in G2-phase of the cell cycle. (). The location of the DDE suggested that aberrant transcription termination might contribute to this effect although it is not clear how this could lead to transcript accumulation during the cell cycle.
It has been known for sometime that the process of mRNA 3′-end cleavage and polyadenylation is tightly coupled to transcription termination (). Mutations in signal sequences on the primary transcript that control 3′-end cleavage and polyadenylation can reduce transcription termination efficiency (). Conversely, null mutations in genes coding for certain cleavage/polyadenylation factors including the components of cleavage factor IA (CFIA), Rna14p, Rna15p and Pcf11p, result in the disruption of both 3′-end processing and transcription termination () leading to the accumulation of unprocessed read-through transcripts. These unprocessed transcripts are substrates for the nuclear exosome, a multi-subunit complex of 3′ to 5′ exonucleases (), which rapidly degrades the transcripts in a process referred to as nuclear surveillance (). Surprisingly, if the nuclear exosome component Rrp6p is deleted in a strain that also has inactive Rna14p or Rna15p, functional mRNAs can be produced from unprocessed ‘read-through’ transcripts by a ‘chewing back mechanism’ mediated by the core exosome, followed by polyadenylation (). This observation has led to suggestions that this redundant function of the nuclear exosome may play a physiological role in regulating the abundance of specific mRNAs under certain conditions. There is evidence to show that the nuclear exosome component Rrp6 is required for the degradation of normal mRNAs that are retained in the nucleus () and for the autoregulation of specific mRNAs such as Nab2 and Nrd1 (,).
In addition to its role in nuclear surveillance, the nuclear exosome is required for the processing of many stable and unstable RNAs in the nucleus such as rRNAs, snoRNAs and snRNAs (). Processing and/or degradation of these pre-RNAs requires an initial polyadenylation of the substrates, mediated by a complex consisting of the PolyA polymerases Trf4 and Trf5, the putative ATP-dependent RNA helicase, Mtr4p/Dob1p and zinc knuckle-binding proteins, Air1/2p, referred to as TRAMP. The polyadenylated substrates are subsequently degraded by the nuclear exosome. Recently, Reis and Campbell () have shown that loss of Trf4/5p or Rrp6p results in elevated levels of histone mRNAs in suggesting that this may be a preferred pathway for histone mRNA degradation.
To further explore the links between 3′-end processing, transcription termination and the nuclear exosome in regulating histone mRNAs levels, we have examined the role of a number of RNA processing factors and the unique component of the nuclear exosome, Rrp6p, in the biogenesis of histone mRNAs. Our results show that the components of CFIA, Rna14p, Rna15p and Pcf11p are required for the 3′-end cleavage and proper termination of HTB1 mRNA and in the /▵ double mutant, steady-state levels of functional polyadenylated HTB1 mRNA are restored to near wild-type levels. Furthermore, in the absence of Rrp6p (▵), the normal pattern of HTB1 mRNA accumulation in the S-phase is disrupted. Flow cytometry analysis indicates that deletion of leads to a prolonged S-phase and this leads to HTB1 mRNA accumulation. The continued accumulation of HTB1 mRNA in cells is influenced by the interaction of the nuclear exosome with the 3′-end processing machinery. We propose a model to explain the role of the nuclear exosome in the cell cycle regulation of histone mRNAs.
strains used in this study are listed in . Yeast strains were cultured at 22°C or 30°C in YEP (1% and 2% (w/v) yeast extract and bactopeptone, respectively), containing either 2% dextrose (YEPD) or 2% galactose (YEPGal) to an optical density at 660 nm (OD) of between 0.6 and 2.0. Plasmids containing a URA3 gene were introduced into yeast strains by the lithium acetate procedure as previously described (). The transformants were selected on synthetic complete (SC) agar medium without uracil. For neomycin phosphotransferase expression, cells were plated on YEPGal plates containing 200 μg/ml G418.
Strains were grown in YEPD medium to early log phase. α-mating factor was added (final concentration of 2 μg/ml) and the culture was incubated for 3 h at 30°C or overnight at 22°C as indicated in figure legends 3 and 4 respectively. Cell cycle arrest was monitored by microscopic examination of the culture. When 90% of the cells were in G1-phase, they were collected by centrifugation and washed twice with sterile water. The pellets were resuspended in 200 ml of pre-warmed YEPD media and incubated at the temperatures indicated in the figure legends. For the temperature-sensitive strains, cells were incubated at either permissive (22°C) or non-permissive (37°C) temperatures for 1 h after α-mating factor removal. Samples were collected at 10 min or hour intervals and the cell pellets were frozen at −70°C until needed.
Cells were synchronized with α-mating factor as described above. Cell samples were collected at 15 min intervals after removal of α-mating factor, fixed in 70% ethanol and stored at 4°C. The samples were washed in water and resuspended in 50 mM sodium citrate, pH 7.0. The cells were sonicated for 10 s at a setting of 5 amplitude microns on a SoniPrep sonicator (MSE). RNase A in 50 mM sodium citrate was added to give a final concentration of 0.1 mg/ml and the cells were incubated at 50°C for 2 h. Proteinase K was added to give a final concentration of 1 mg/ml and the cells incubated for a further 2 h at 50°C. Propidium iodide was added to give a final concentration of 4 μg/ml. Cell sorting as carried out on a Beckman Coulter Exics XL FACs Analysis machine. The data was analysed using Beckman Coutler Expo 32 ADC software.
RNA extractions, northern blotting and hybridizations were carried out as previously described () but using a slightly modified hybridization buffer [7% sodium dodecyl sulphate (SDS), 2% casein, 0.75 M NaCl, 0.075 M sodium citrate, 50 mM sodium phosphate (pH 7.0) and 0.1% -lauroylsarcosine] at 68°C for 15–20 h Digoxigenin-dUTP (dig-dUTP) labelled DNA hybridization probes were prepared as described previously (). Details of primers are listed in . An alkaline phosphate tagged anti-dig antibody and the CDP-star substrate were used according to manufacturer's instructions for detection of the hybridized probe.
Autoradiographs of northern blots were quantified using the densitometry software (Quantity One, version 4.2.1, Bio-Rad, Richmond, CA, USA). To account for variations in sample loading, the following normalization protocol was adopted. For cell cycle experiments, the 30 min time point following release from α-factor synchronization was used as a reference point (relative value = 1.0) for each probe. The value for each other time point was divided by this reference value to generate a normalized value. The normalized value for each of the time points for endogenous HTB1 was then divided by the corresponding normalized values for actin.
To remove any contaminating DNA, RNA samples (30 μg) were incubated at 30°C for 1 h with RQ DNase (Promega Inc., Madison, WI, USA) according to supplier's instructions. Following digestion, the RNA was purified by adsorbing to silica columns (Sigma Chemical Co., Poole, UK). Reverse transcription reactions (20 μl) were carried out using 2 μg of DNAse-treated RNA as template, a transcript-specific reverse primer and reverse transcriptase (ImProm-II; Promega Inc., WI, USA) as specified by the manufacturer at 42°C for 60 min, followed by 15 min at 70°C. The PCR (25 μl) was carried out using 2 μl of cDNA and Taq polymerase (New England Biolabs, MA, USA) according to manufacturer's instructions. The primers used for cDNA synthesis are shown in and in the figure legends. The cDNA was amplified using 25 cycles of (95°C for 30 s 55°C for 1 min and 72°C for 45 s) followed by final extension step was at 72°C for 7 min.
Real-time quantitative PCR was performed on Rotor-gene RG-3000 (Corbett Research, Canberra, AU) using Green Jumpstart ready mix as described by the manufacturer (Sigma Chemical) in 14 μl reaction volume and 2 μl of template (reverse transcription reaction) using 10 pmol of primers (HTB1_Endo_F and HTB1_Endo_R for endogenous and Actin_F2 and Actin_R2 for actin, see ) and an annealing temperature of 55°C. The values obtained were normalized to that of actin as described in the section above.
To test this assumption, the production of the GAL1 inducible chimeric neo-HTB1 mRNA (,), which contains the coding region of the neomycin phosphotransferase gene fused to sequences for the last 17 amino acids and the downstream 1300 nt of the HTB1 gene including the 3′-end cleavage sites, was examined in the CFIA temperature-sensitive mutants and (A) ().
As shown in B, accumulation of neo-HTB1 mRNA is significantly reduced at the non-permissive temperature in the mutants (lane 2) and (lane 8) and to a lesser extent in (lane 3). The reduction in steady-state levels of mRNAs is consistent with the known phenotype of CFIA mutants and reflects the rapid turnover of unprocessed transcripts by the nuclear exosome (,). Steady-state levels are partially restored when Rrp6p, a component of the nuclear exosome is also deleted in the (B, lane 4) and correctly sized mature WT neo-HTB1 transcripts are now apparent (B, lane 4*). The neo-HTB1 transcripts produced in the and / strains are exported and can produce a functional neomycin phosphotransferase protein as observed by the growth on G418 plates, while under the same conditions, no functional protein is produced in cells (C). Interestingly, compared to its WT isogenic strain W303B (), more functional protein is expressed in / and cells (C). This reflects the general instability of neo-HTB1 in a number of WT strains of at temperatures greater than 30°C (Canavan,R., unpublished data). While neo-HTB1 mRNA levels are less affected by mutations in (Figure IB, lane 3), more functional mRNA is produced in / cells at the non-permissive temperature as indicated by growth on G418 plates (data not shown).
Previous work from our laboratory () has shown that mutations in a region lying ∼100 nt downstream of the 3′-end cleavage sites of the HTB1 gene, termed the DDE, can influence the steady-state levels of the neo-HTB1 mRNA in asynchronous cells as well as during the cell cycle. Specifically, we previously observed that levels of neo-HTB1 mRNA were reduced in asynchronous cells in the mutant pSAC20 while its accumulation pattern during the cell cycle is altered in mutant pSAC21 (see A for mutated bases). The steady-state levels of neo-HTB1 were unchanged in the mutant pSAC15 in asynchronous cells (Campbell and Bond, unpublished data). transcription run off studies revealed that the DDE lies in the region where transcription termination occurs (). This led us to speculate that the DDE region may influence the steady-state levels of neo-HTB1 mRNA through its interaction with the 3′-end processing machinery. To examine this possibility, the levels of the mutant neo-HTB1 mRNAs transcribed from the mutants pSAC20, pSAC21 and pSAC15 were examined in the single mutants -3 and and in the double mutant /Δ. As shown in A (lanes 1—4), the steady-state levels of DDE mutant neo-HTB1 mRNAs are somewhat reduced at the permissive temperature in cells when compared to the levels of the WT neo-HTB1 mRNA. In the / double mutant, the neo-HTB1 mRNA levels are restored (A, lanes 5–8). In the single mutant , the DDE-mutant neo-HTB1 mRNAs appear to be preferentially stabilized relative to the WT neo-HTB1 mRNA (A, lanes 9–12). At the non-permissive temperature, the levels of the WT and the mutant neo-HTB1 mRNAs are reduced in cells (B, lanes 1–4). The DDE mutant neo-HTB1 mRNAs appear more stable than the WT neo-HTB1 mRNA in the mutant background, although it should be noted that the WT neo-HTB1 RNA sample is under loaded (B, lanes 9–12). Additionally, we noted that the steady-state levels of the WT and DDE mutant neo-HTB1 mRNAs could be influenced by the physiological state of the cells (our unpublished data). Taken together, the data indicates that neo-HTB1 mRNAs containing mutations in the DDE appear to be more unstable under conditions where 3′-end processing is inhibited but are stabilized in cells.
We next questioned if the nuclear exosome may be contributing to the cell cycle regulation of endogenous histone mRNAs. The pattern of accumulation of the endogenous HTB1 mRNA was examined in WT and mutant cells. As shown in A, the steady-state levels of the endogenous HTB1 mRNA steadily increase and peak between 60 and 70 min following removal of α-factor and rapidly decrease thereafter. Quantification of the northern blots and by real-time RT-PCR, reveals 4- to 5-fold increase in the steady-state levels of endogenous HTB1 transcript between time 0 and 70 min followed by rapid turnover of the RNA with levels returning to baseline by 110 min (C). In the background, the endogenous HTB1 mRNA appears to accumulate up to 50 min in a manner similar to that observed in the WT strain, however, thereafter, the RNAs levels decrease slightly and remain at the same level for the remainder of the time course (B and C).
The observed difference in the accumulation of HTB1 in the WT and cells may reflect difference in the growth and division rate of the cells. As shown in D, both the WT and mutant show very similar growth rate patterns and display a doubling time of ∼3.3 and 3.4 h, respectively at 30°C. To examine the cell cycle progression in more detail, flow cytometry analysis of cells released following α-factor synchronization was preformed. As shown in E, W303 and cells enter S-phase at the same rate. Cells with a 2 content appear at ∼60 min after release from α-factor, which coincides with the peak in HTB1 mRNA levels. However, there is a slight delay in exit from S-phase in the cells. This delay in exit from S-phase may account for the continued accumulation of HTB1 mRNA in these cells. Taken together, the data suggests that the nuclear exosome Rrp6 contributes to S-phase accumulation of HTB1 mRNA and appears to alter the exit from S-phase.
The nuclear exosome is known to play a role in degrading unprocessed and/or read-through transcripts in the nuclear surveillance pathway as demonstrated in . Therefore the possibility arises that the 3′-end processing of histone mRNAs is differentially regulated during the cell cycle leading to an accumulation of unprocessed transcripts in the G-phase and thus generating substrates for the nuclear exosome. To test this hypothesis, the endogenous HTB1 mRNA levels were examined during the cell cycle in the / mutant.
The cell cycle accumulation pattern of HTB1 mRNA in / and its isogenic parent W303 cells grown at 22°C is shown in A. At this temperature, the doubling time in both strains is ∼6–7 h (data not shown). HTB1 mRNA levels persist for a longer period, up to 4 h in the absence of Rrp6p (A, compare lanes 10–14 with lanes 1–5, 22°C panel). Inactivation of Rna14p by incubation of cells at 37°C for 1 h after removal of α-factor leads to an even greater persistence of HTB1 mRNA with levels remaining high for the following 6 h (B, compare lanes 10–16 with lanes 1–7, 37°C panel). Thus, both Rrp6p and Rna14p appear to act in concert to influence the steady-state levels of HTB1 during the cell cycle.
To determine if 3′-end unprocessed transcripts, which might be potential substrates for the nuclear exosome, differentially accumulate during the cell cycle, reverse transcription reactions were carried out using a series of reverse primers corresponding to sequences immediately upstream or downstream of the 3′-end cleavage sites (A), followed by PCR using a forward primer anchored in the 3′ UTR region of the HTB1 mRNA. In asynchronized / cells grown at the permissive temperature, the majority of the transcripts correspond to correctly cleaved mRNAs (R1 primer) while some unprocessed transcripts are evident extending approximately to the region where the DDE lies (B, 22°C, lanes 3–5). At the non-permissive temperature, we observe stabilized unprocessed pre-mRNA transcripts extending up to 450 nts beyond the 3′-end cleavage sites (B, 37°C, lanes 2–7).
When the same primers were used in RT-PCR reactions using RNA isolated at 1 and 6 h following α factor release of W303 cells grown at 22°C (), transcripts corresponding to correctly cleaved HTB1 (R1 primer) are evident at both time points (C, 22°C panel lanes 3, 10*) and no read-through transcripts are apparent. Based on HTB1 accumulation patterns, these time points correspond to the S-phase and post S-phase of the cell cycle, respectively (). Thus in WT cells, the majority of the transcripts are correctly cleaved in both the S-phase and the post-S-phase of the cell cycle. Products of the correct size were amplified by primers R1 and R2 in the 1 and 6 h samples of W303 cells that were treated at 37°C for 1 h after removal of α-factor (C, 37°C panel, lanes 3, 4, 10, 11*). It should be pointed out that a number of artifactual bands are also evident in all lanes. This most likely reflects the relatively low levels of available substrate for cDNA synthesis using primers R2 through R6. Thus, preincubation of cells at 37°C in the first hour after α-factor release in of itself did not alter the 3′-end processing pattern of HTB1 mRNA during the cell cycle.
Similarly, the accumulation pattern of unprocessed read-through HTB1 transcripts was also examined in synchronized / cells using RT-PCR. At the permissive temperature (22°C), only R1 and R2 primers can reverse prime and amplify products in samples taken 1 and 6 h after alpha factor release (, panel D, 22°C, lanes 3, 4, and lanes 10, 11*). However, when the cells are shifted to the non-permissive temperature for 1 h after removal of α-factor to inactivate Rna14p, transcripts can now be amplified with the reverse primers R1, R2, R3 and R5 (A) in both the 1 and 6 h samples (D, 37°C panel, lanes 3–6, and lanes 10–13). Thus, the efficiency of 3′-end processing does not appear to vary significantly during the cell cycle and read-through unprocessed transcripts only accumulate under conditions when 3′-end processing is impaired (/ at 37°C, D, 37°C panel).
The molecular mechanisms underlying the regulation of histone mRNA levels during the cell cycle in is poorly understood. In higher eukaryotes, the fluctuations in the steady-state levels of histone mRNAs during the cell cycle is mainly controlled by post-transcriptional events, specifically by mRNA 3′-end processing mediated by the interaction of unique protein factors with the conserved S-L structure at the 3′-end of the non-polyadenylated histone mRNAs and by the U7snRNP. Bioinformatic analysis of the yeast genome has not identified homologues of the SLBP nor the 3′ hExo both of which interact with the 3′-end of histone mRNAs in higher eukaryotes, nor have homologues of the U7 snRNP been identified (Bond,U., unpublished data). Since yeast histone mRNAs are polyadenylated but are cell cycle regulated, it appears that an alternative regulatory mechanism must have evolved to control the accumulation of histone mRNAs in the S-phase of the cell and their subsequent turnover in the G2-phase. Previous studies in have indicated that transcriptional regulation plays a major role in the accumulation patterns histone mRNAs during the cell cycle. Specifically, the transcriptional repressors Hir1p and Hir2p act coordinately to repress transcription in the G1-phase and through their ability to recruit components of the SWI/SNF chromatin-remodelling complex, activate expression in S-phase (,). Additionally, post-transcriptional events appear to contribute to the cell cycle regulation as chimera genes containing the 3′ untranslated region of the HTB1 gene can confer cell cycle periodicity to a neomycin phosphotransferase mRNA (). The 5′ untranslated region appears to play a lesser role ().
In this article, we have explored the mechanisms controlling histone mRNA 3′-end formation and cell cycle regulation in . Our analysis using the neo-HTB1 chimeric mRNA in CFIA mutants indicate that the 3′-ends of HTB1 mRNA are generated for the most part through the general 3′-end processing machinery. The steady-state levels of the neo-HTB1 mRNA transcripts are significantly reduced in the temperature-sensitive mutant . Mutations in RNA15 (), had a lesser impact on neo-HTB1 mRNA levels. This same phenotype was observed with a second mutant allele (data not shown). The reduced effect of mutants on HTB1 mRNA levels most likely reflects the fact that, in our hands, these temperature-sensitive mutants are not completely inactivated following heat treatment of the cells at 37°C: while cell growth is significanly reduced at this temperature, some growth is observed suggesting that these mutants only partially inactivate 3′-end processing. It is interesting to note that the differential effect of and mutants on the steady-state level of the neo-HTB1 transcripts was not noted in the analysis of other mRNA transcripts such as the Act1 or Cyc1 (), although in this study the cells were incubated at 37°C for 2 h as opposed to the 1 h treatment used here. This may suggest that histone mRNAs show a lesser requirement for the Rna15p component of CF1A for 3′-end processing.
The neo-HTB1 primary transcripts are inherently unstable when 3′-end processing is inhibited, unlike actin mRNA transcripts which persist under these conditions (). Deletion of in CFIA mutant backgrounds not only restores the levels of neo-HTB1 mRNA but also generates correctly sized transcripts representing fully processed mRNA. This RNA is exported from the nucleus and can be translated to produce functional proteins. Thus, even under conditions where 3′-end processing is inhibited, mature functional transcripts can be generated. Others have also observed similar ‘processing’ of read-through transcripts of a number of mRNAs in an / background (). It was postulated that these processed transcripts result from a trimming back of the read-through transcripts to the correct 3′-end by the 3′ to 5′ exonucleases of the core exosome and subsequent polyadenylation by Poly A Polymerase. An alternative hypothesis is that a sub-optimal 3′-end processing machine can now cleave and polyadenylate the unprocessed transcripts, which have been stabilized through a lack of action of Rrp6p. In either case, it appears that such redundant mechanisms exist to ensure that the production of key transcripts are generated despite adverse conditions. While the precise mechanism is not clearly understood, the data presented here verifies that Rrp6p appears to play a role in the biogenesis of histone mRNAs through its interaction with the 3′-end processing machinery.
Our results also show that the steady-state levels of neo-HTB1 mRNA appear to be influenced by the DDE sequence, which lies in the region of transcription termination of the HTB1 gene, specifically under conditions when 3′-end processing is inhibited or when RRP6 is deleted. Point mutations in the DDE region appears to reduce the stability of neo-HTB1 transcripts when the 3′-end processing is inhibited and conversely transcripts are more stable when RRP6 is deleted. This inverse relationship in mRNA stability in the two mutant backgrounds suggests that mutations in the DDE region of the HTB1 gene may alter the substrate specificity of the exosome thus allowing the mutant transcripts to escape degradation by the exosome or that some level of competition for binding to the DDE may occur between the 3′-end processing machinery and the nuclear exosome. The stability of the WT and DDE mutants can be influenced by the physiological conditions of the cell, particularly by cell density (our unpublished data). It is known that the 3′-end processing machinery communicates with RNA Polymerase II as it reaches the transcription termination site (,). It is possible that the DDE may act as a gateway for various 3′-end processing factors to access the transcription complex, thus establishing communication to effect transcription termination and subsequent processing.
The observation that the deletion of RRP6 alters the pattern of HTB1 transcript accumulation during the cell cycle provides an insight into a possible mechanism of histone mRNA regulation in . Quantification of HTB1 transcripts during the cell cycle by densitometry analysis of northern blots and quantitative real-time RT-PCR, indicates that while HTB1 mRNA levels appear to accumulate similarly in WT and Δ cells, thereafter, the levels decrease to baseline very quickly in the WT cells but in the Δ cells, HTB1 levels while decreasing slightly, persist throughout the remainder of the cell cycle. In fact, examination of HTB1 transcript levels in /Δ at 22°C showed continued accumulation up to 4 h after alpha factor release, while in WT cells, levels return to baseline after 2 h The prolonged accumulation of HTB1 transcripts in Δ cells appears to be exacerbated by inactivation of 3′-end processing as HTB1 transcripts accumulate for an even longer time in the double mutant /Δ at the non-permissive temperature.
Flow cytometry analysis of synchronized cells reveals that the prolonged accumulation of HTB1 in cells appears to result from a delay in exit of cells from S-phase or entry into M-phase. The reason for this delay in cells is currently unclear. Since RRP6 encodes for a 3′ to 5′ exonuclease, it is possible that Rrp6p is required for the decay of HTB1 mRNA at the end of S-phase and in the absence of this trigger, progression into M-phase is delayed. Alternatively, Rrp6p may regulate the levels of other mRNAs that encode for cell cycle checkpoint proteins. Interestingly, Gill () have shown that accumulation of CLB2 mRNA, which encodes for B-cyclin, accounts for the exit from mitosis defect observed in RNase mitochondrial RNA processing (RNase MRP) mutants. The CLB2 mRNA was found to be a substrate for the RNase MRP endoribonuclease. Cleavage in the 5′ UTR region of CLB2 occurs at the end of mitosis, removing the 5′ cap and allows subsequent degradation by the 5′ to 3′ exonuclease Xrn1p. Thus, regulation of check point mRNAs at the level of mRNA stability may be a common theme in cell cycle regulation.
Recently Reis and Campbell () have identified a role for the nuclear poly (A) polymerases TRF4 and TRF5 complex in the regulation of bulk histone mRNA levels in yeast cells. Trf4/5p, together with Air1p, Air2p and Mtr4p, form the TRAMP complex, which has a specific function in the processing and/or degradation of rRNAs, snoRNAs and snRNAs. The TRAMP complex appears to ‘tag’ its substrates for processing and/or degradation by the nuclear exosome by adding a short polyA tail onto the 3′-end of these RNAs (,). Histone mRNA levels are elevated in Trf4/5 double mutants and likewise in strains bearing a deletion of , the HHF2 core histone transcript level was at least twice as much of the wild-type level. The level of HHF2 mRNA appears to come down at the end of S-phase but appears to have increased 4-fold in the subsequent S-phase. This is reminiscent of the suboptimal HTB1 transcript degradation we observed here. Taken together, the data provide clear proof for a role for the nuclear exosome in the regulation of histone mRNAs in yeast cells.
How might the nuclear exosome regulate histone mRNA levels during the cell cycle? Much of our understanding of how the nuclear exosome recognizes its intended targets comes from the analysis of its known substrates. First, the nuclear exosome is required for the 3′-end processing and transcription termination of snoRNAs and is recruited to the transcription termination sites via the RNA-binding proteins Nab3 and Nrd1 (). Secondly, microarray analysis has revealed a small group of mRNAs whose steady-state levels are altered by deletion of RRP6. These include the autoregulated mRNAs Nrd1 and Nab2 (). Thirdly, as shown previously and above, 3′-end unprocessed read-through transcripts are substrates for the nuclear exosome (,,). A common theme among these exosome targets is the requirement for Nrd1/Nab3-binding sites. From the accumulated data, the model emerging is one in which the exosome complex is recruited to its substrate via its specific association with the RNA-binding protein Nrd1 (). Substrates are then degraded in a 3′ to 5′ direction unless they encounter any additional Nrd1 recognition sites upstream which would block the progress of the exosome ().
This model can also explain the ‘processing’ and subsequent polyadenylation of read-through unprocessed neo-HTB1 transcripts seen in / strains. It is likely that this processing to a specific location reflects the presence of an Nrd1-binding site, which prevents further degradation by the exosome. In the presence of Rrp6p, complete degradation of the transcripts is observed suggesting that Rrp6p may add processivity to the core exosome or alternatively may prevent binding of Nrd1 to the transcript. It has been suggested that Nrd1 and the exosome may play a ‘back-up’ role in biogenesis of some mRNAs to ensure the production of mature mRNAs under certain conditions such as when the 3′-end processing machinery is inactivated (). This suggests that the 3′-end processing machinery and the nuclear exosome may compete for the same pre-mRNA substrate and that physiological conditions in the cell will dictate the outcome of the competition. Extending this model, the data presented here suggests that HTB1 primary transcripts are recognized by the 3′-end processing machinery and the nuclear exosome in S-phase facilitating proper processing of the transcript. However, in G2-phase, the nuclear exosome interaction is favoured over the 3′-end processing machinery leading to an increase in the mRNA turnover. Whether this alteration in substrate specificity is due to changes in Nrd1 binding or alterations in 3′-end processing is currently unclear. However, our results showing that 3′-end processing is not differentially regulated during the cell cycle leads us to favour the former mechanism over the latter.
Thus, Rrp6 may contribute to histone mRNA biogenesis through its role in the nuclear surveillance pathway or through its interactions with the nuclear export pathways where it ensures that only correctly processed and functional mRNA can leave the nucleus (). Alternatively, histone mRNAs may be retained in the nucleus in G2-phase thus evoking degradation of the transcripts by the DRN pathway (degradation of mRNA in the nucleus) (). Rrp6p, along with nuclear cap-binding protein Cbc1p, has shown to be required for this pathway, which degrades mRNAs that are retained in the nucleus (). It is interesting to note that the mammalian orthologue of RRP6, PM/Scl-100 (), associates with the cytoplasmic exosome (). Currently the role of the cytoplasmic mRNA degradation pathways () in the regulation of histone mRNAs has not been explored.
Finally, our finding that the nuclear exosome component Rrp6p contributes to the correct cell cycle accumulation pattern of HTB1 mRNA, acting in conjunction with transcriptional controls to contribute to the steady-state levels of histone mRNAs, provides further options to investigate the mechanisms controlling the regulation of histone mRNA during the yeast cell cycle. |
Ribosome-inactivating proteins (RIPs) are N-glycosidases which cleave the N-glycosidic bond of adenine-4324 in eukaryotic 28S rRNA or adenine-2660 in 23S rRNA (,). This adenine is located in a highly conserved GAGA hairpin within the α-sarcin/ricin loop. Removal of the specific adenine hinders the elongation factor 1-dependent binding of aminoacyl-tRNA and GTP-dependent binding of elongation factor 2 to the ribosome. Thus, protein synthesis is arrested at the elongation step (,). RIPs get access to the ribosome by firstly interacting with ribosomal proteins; for example, trichosanthin (TCS) binds to the acidic ribosomal P proteins (,), ricin A chain (RTA) binds to L9 and L10e () and pokeweed antiviral protein (PAP) binds to L3 ().
RIPs are important biomedicine because they are highly cytotoxic towards human cancer cells, including lymphoma and myeloma. RTA conjugated to monoclonal antibodies anti-CD25 and anti-CD30 is being used to treat Hodgkin's lymphomas (). Saporin is coupled to major histocompatibility complex (MHC) class I tetramers to kill antigen-specific CD8(+) T cells, which are important effector cells responsible for tissue destruction in several autoimmune and allograft-related diseases (). TCS is used to induce midterm abortion, treat ectopic pregnancies and hydatidiform moles, reset menstruation and expel retained placenta (). TCS and PAP have also been shown to possess anti-HIV activities ().
Based on the number of subunits, RIPs are grouped into two classes. Type I RIPs such as TCS and saporin consist of a single polypeptide chain, with molecular weight around 30 kDa. They are actively uptaken by the alpha-2 macroglobulin receptor (α-2-MR) (,), which is widely distributed in different cell types such as macrophages, hepatocytes and follicular cells of the ovary (). Type II RIPs such as ricin and abrin consist of two polypeptide chains linked by a disulphide bridge. Chain A is the catalytic subunit sharing high structural homology to type I RIPs, while chain B facilitates the intracellular delivery of chain A by interacting with carbohydrates on the cell surface (). Both type I and II RIPs are basic proteins, with pI greater than 8. Maize RIP is an unusual RIP, which is either classified as a type III RIP () or considered as an atypical type I RIP (). It is synthesized as a 34 kDa acidic inactive precursor in endosperm, with proper folding and a pI of around 6 (). Its expression is controlled by the regulatory locus (). During germination, this precursor is converted to a two-chain active form by the elimination of 16 aa at the N-terminal region (residues 1–16), 25 aa at the acidic central region (residues 163–189) and 14 aa at the C-terminus (residues 287–300) to generate a two subunit basic protein of 248 aa (). The two subunits of 16.5 and 8.5 kDa are tightly associated without any covalent linkage.
Among the sequences to be removed during the activation of maize RIP, the 25 aa internal region (known as internal inactivation region) is the most crucial, as removal of this region increases the activity by at least 600-fold, whereas removal of the N- or C-terminal region only increases the activity by 6- or 5-fold, respectively (). Deletion of the internal amino acid residues represents a novel mechanism of enzyme activation in plants and resembles processing of certain hormones, such as insulin ().
It is hypothesized that maize RIP can directly inhibit pathogens by inactivating their ribosomes and causing cell death (). The active form is therefore a promising anti-insect and anti-fungal agent. This form has been overexpressed in transgenic rice, wheat and tobacco, for increasing the resistance to plant insects such as larvae of the cigarette beetle (), the tobacco hornworm () and the corn earworm () (). Co-expression of the active form and a rice basic chitinase gene in transgenic rice has also led to increased resistance to sheath blight () ().
To further understand how maize RIP functions, in particular the relationship of the internal inactivation region to the rest of the protein, and to provide the first structural example of a type III RIP, we set forth to reveal the structures of the active (MOD) and inactive forms (Pro-RIP) of maize RIP and analyse the active site pocket.
Maize [Δ1-16, Δ287-300]-Pro-RIP (Pro-RIP-WT in short) and [Δ1-16, Δ163-164, Δ167-189, Δ287-300]-Pro-RIP (MOD-WT in short) were obtained from Prof. R.S. Boston. The numbering of the aa residues is made according to Ref. (). [Δ1-5]-Pro-RIP-WT (Pro-RIP), [Δ1-5]-MOD-WT (MOD), [E207A]-MOD, [E207AV238E]-MOD, [E207DV238E]-MOD and [V238E]-MOD were generated by polymerase chain reaction mutagenesis using overlapping primers and KOD DNA polymerase (Novagen). A methionine residue was added to the N-terminus as start codon. All concerned DNA were cloned into pET3a expression vector and sequenced to ensure that no secondary mutation had occurred. Proteins were overexpressed in strain C41 (DE3) (Novagen) in M9 medium (6 g/l NaHPO, 3 g/l KHPO, 1 g/l NHCl, 4 g/l glucose, 0.5 g/l NaCl, 100 µg/l ampicillin, 2 mM MgSO, 0.1 mM CaCl). Bacterial cells were grown in 37°C until OD 600 reached 0.4–0.6 and 0.4 mM IPTG was added to induce protein expression at 25°C. The cells were harvested after overnight culture by centrifugation at 4°C. Cell pellet was resuspended and sonicated in 20 mM phosphate buffer, pH 7.0 (buffer A). Cell lysate was collected by centrifugation at 4°C and loaded onto a HiTrap CM-FF column (Amersham) pre-equilibrated with buffer A, and eluted using a gradient of 0–0.5 M NaCl in buffer A. Fractions containing the target protein were pooled and dialyzed against buffer A, and loaded to a HiTrap SP column (Amersham) pre-equilibrated with buffer A. The protein was eluted using a gradient of 0–0.5 M NaCl in buffer A. Target fractions were pooled and concentrated to 5 ml for further purification by Superdex 75 gel filtration column (Amersham), which was pre-equilibrated with 20 mM Tris–HCl, 0.1 M NaCl, pH 7.0. Purified protein was concentrated and stored at −80°C.
Rat liver tissue of 250 g was used for ribosome purification (). In brief, the liver tissue was homogenized in ice-cold homogenization buffer [50 mM Tris–HCl (pH 7.6), 25 mM KCl, 5 mM MgCl, 0.25 M sucrose] and was centrifuged at 13 000 for 30 min. The supernatant was filtered through glass wool and the filtrate was centrifuged at 145 000 for 2 h. The pellet was resuspended in a buffer containing 35 mM Tris–HCl (pH 7.8), 25 mM KCl, 10 mM MgCl, 0.15 M sucrose and 6 mM 2-mercaptoethanol. One-tenth of the volume of a freshly prepared solution of 10% sodium deoxycholate was added to release the ribosomes from the microsomal membrane. The resultant suspension was layered over an equal volume of 0.3 M sucrose pad in buffer B [35 mM Tris–HCl (pH 7.8), 600 mM KCl, 10 mM MgCl, 6 mM 2-mercaptoethanol] and then centrifuged at 176 000 for 90 min. The ribosomal pellet was rinsed and resuspended in buffer C [50 mM Tris–HCl (pH 7.8), 50 mM KCl, 5 mM MgCl, 10 mM KHCO, 0.25 M sucrose, 6 mM 2-mercaptoethanol] to 7.5 mg/ml.
Crystals of Pro-RIP and MOD were grown by mixing equal volume of protein solution and buffer in sitting drop at 16°C. The buffer used for Pro-RIP was 2 M ammonium sulphate, 0.25 M Tris–HCl and 0.1 M sodium acetate tri-hydrate, pH 3.5. The buffer for MOD was 0.2 M sodium acetate tri-hydrate, 0.1 M Tris–HCl pH 8.5 and 30% PEG 4000. MOD-Adenine crystal complex was obtained by soaking MOD crystals in the mother liquid containing saturated adenine for 72 h. Diffraction data of Pro-RIP and MOD were collected at 110 K using an in-house Rigaku MicroMax 007 X-ray generator and the synchrotron 6B beamline at Pohang Accelerator Laboratory, Pohang, Korea. Diffraction data were processed by MOSFLM. Phase determination was carried out by molecular replacement using MolRep of the CCP4 program suite. Structure of ricin A chain (PDB ID : 1RTC) with the side chains first eliminated by the Align4MR program () was used as the search model for Pro-RIP. For MOD, the refined Pro-RIP with the internal fragment deleted was used as the search model. The starting model was subject to rigid refinement using CNS and model building was carried out by XtalView. Further refinement was also achieved by CNS. Adenine and water molecules were added after the rest of the structure was well refined. Water molecules with Sigma values less than 1.0 in the 2− map were excluded. The stereochemical quality of the model was assessed by PROCHECK ().
The ribosome-inactivating activity assay was carried out using a rabbit reticulocyte lysate protein synthesis system, with L-[, , 5-H (N)]-leucine as label (NEN Life Science Products). In brief, 0.1 pM to 10 μM proteins and 5 nCi [H]-leucine were incubated with the translation system in triplicate at 30°C for 30 min. Newly synthesized protein was precipitated with 25% trichloroacetic acid and captured by filtration through glass microfibre filters (Whatman).
MTT assay was performed on human choriocarcinoma JAR cell line. In brief, JAR cells were treated with Pro-RIP or MOD for 72 h followed by incubation with MTT (5 mg/ml) for 4 h. DMSO was then added and OD values were measured using a colorimetric microplate reader (Model 3550, Biorad). The result was presented in terms of percentage to control as mean ± SEM. Two-way ANOVA with Bonferroni test as the test for multiple comparisons was used to compare cell viability after treatment with Pro-RIP and MOD.
JAR cells were seeded in a six-well plate (3 × 10 cells/well). Pro-RIP and MOD were labelled by green fluorescence dye F1640 (Roche) and incubated with the cells for 4 h. The cells were harvested and analysed by flow cytometry (BD biosciences). Uptake of the Pro-RIP and MOD was presented as mean ± SEM of percentage to control. One-way ANOVA with Dunnett's Multiple Comparison Test was used to assess the significance of protein uptake as compared to the dye control.
Pull-down assay was carried out to find the interaction between maize RIP and purified rat ribosomes. In brief, Pro-RIP and MOD columns were prepared by immobilizing purified Pro-RIP or MOD to a 1 ml NHS-column (Amersham). Then purified ribosomes were loaded onto the columns. Unbound protein was washed away by 30 ml PBS (1 mM KHPO, 10 mM NaHPO, 137 mM NaCl, 2.7 mM KCl, 0.005% Tween 20, pH 7.4), followed by elution buffer (1 M NaCl in PBS, pH 8.0) to elute the bound proteins. The eluted fractions were analysed by 15% SDS–PAGE. Control experiment was carried out by loading the ribosomes onto uncoupled NHS-Sepharose.
BIAcore 3000 surface plasmon resonance biosensor (Pharmacia Biosensor AB) was used to measure the kinetic parameters of the interaction. Pro-RIP or MOD (1 nM) in 10 mM sodium acetate, pH 5.0 was covalently linked to the dextran on the surface of CM5 sensor chip via primary amino groups using the Amine Coupling Kit (Pharmacia) at a flow rate of 5 µl/min, 25°C. A range of 0–240 nM of ribosomes in PBS were injected at a flow rate of 5 µl/min, 25°C, onto the RIP immobilized sensor chip surface. The binding surface was regenerated by 2 M NaCl between sample injections. Control experiment was carried out similarly on uncoupled sensor chip surface.
The coordinates of yeast ribosome were derived from the cryo-EM structure (PDB ID: 1S1H and 1S1I). The adenine ring of A-2697 (analogous to A-4324 in rat 28S rRNA) was positioned manually to the adenine-binding site of MOD. Pro-RIP was then superimposed to MOD. The model was energy minimized using the program CNS.
Pro-RIP and MOD were purified, with purities greater than 95% and the usual yield was 50 mg/l culture. Variant [E207AV238E]-MOD had the lowest yield but still reached 20 mg/l culture. Crystals of Pro-RIP, MOD and MOD-adenine complex were obtained. The cell parameters and data collection statistics are shown in .
The structures of Pro-RIP (PDB ID: 2PQG), MOD (PDB ID: 2PQI) and MOD–adenine (PDB ID: 2PQJ) were resolved to 2.4, 2.5 and 2.8 Å, respectively. Phase determination was carried out by molecular replacement. The refinement statistics are shown in . Crystals of MOD belong to space group P3(), with three molecules per asymmetric unit. After refinement, the final and values were 0.2188 and 0.2635, respectively. The structure shows a large N-terminal domain (aa 21–230, aa 21 is an added methionine) and a small C-terminal domain (aa 231–283). The former consists of five α-helices and five-stranded mixed β-sheets; the latter is made up of four α-helices. M21-F27 in MOD is flexible. It exists as a β-strand in chain C but shows no secondary structure in chains A and B (). For the MOD crystals soaked in saturated adenine solution, electron density consistent to an adenine molecule was found in the active site of the protein. The space group remained P3() and no major structural change was observed. The final and values were 0.2319 and 0.2934, respectively.
The crystals of Pro-RIP belong to space group P2() with two monomers per asymmetric unit. The final and values were 0.2114 and 0.2299, respectively. The overall structures of Pro-RIP and MOD are very similar, except the presence of a unique internal inactivating region Ala163–Asp189 in Pro-RIP (), which is rich in acidic residues. Clear electron densities were observed in flexible loop regions Ala163–Val170 and Pro177–Ala179, and a long α-helix in Ala180–Ala188. When aligned to type I and type II RIPs, maize RIP showed some significant structural differences (). There are no α-helix B and β-strand 6 (purple) in the large domain and the anti-parallel β-strands 7 and 8 in the small domain of other RIPs (purple) are replaced by a short α-helix.
[E207A]-MOD, [E207AV238E]-MOD, [E207DV238E]-MOD and [V238E]-MOD were constructed and expressed in . [E207A]-MOD and [E207AV238E]-MOD were soluble and purified. On the other hand, [E207DV238E]-MOD and [V238E]-MOD formed inclusion bodies. The purified proteins were assayed for ribosome-inactivating activity. Percentage inhibition of protein synthesis could be fitted to sigmoidal curves (), indicating concentration-dependent inhibition. MOD exhibited the highest protein-synthesis inhibition activity whereas Pro-RIP had the least. Compared to the wild-type MOD, the activity of [E207A]-MOD decreased by about 556-fold. Variant [E207AV238E]-MOD was 9-fold more active than [E207A]-MOD.
The cytotoxicities of Pro-RIP and MOD to choriocarcinoma JAR cells were evaluated by MTT assay. MOD was found to be more effective in reducing the viability of JAR (IC = 0.37 µM) when compared to Pro-RIP (IC > 40 µM) (A).
Uptake of Pro-RIP and MOD by JAR cells was analysed by flow cytometry. JAR cells were incubated with protein labelled fluorescent dye F1640 (Roche) for 4 h. It was found that both Pro-RIP and MOD entered JAR cells with similar efficiency (B).
Pull-down assays were carried out to study the interaction between the two forms of maize RIP and purified rat ribosomes. MOD and Pro-RIP were coupled to NHS Hi-Trap columns, with efficiencies of 85–90%. Purified rat ribosomes were loaded onto the two RIP-coupled NHS columns, which were then washed by PBS. The bound ribosomes were eluted and analysed by SDS–PAGE. Our results showed that only MOD, but not Pro-RIP interacts with ribosomes (A). Rat ribosomes did not interact with uncoupled NHS column.
A BIAcore 3000 surface plasmon resonance biosensor (Pharmacia Biosensor AB) was used to study the kinetic parameters of the interaction. 5100 RU of Pro-RIP and 5200 RU of MOD were coupled onto the surface of CM5 sensor chips via primary amide groups. When 60 nM of purified rat ribosomes at 5 µl/min were loaded onto the RIP-coupled CM5 sensor chips, both association and dissociation responses of MOD were higher than those of Pro-RIP. Within a 240 s window, the association signal rapidly reached 105 RU with a slow dissociation ended at 60 RU. On the contrary, Pro-RIP associated slowly to 40 RU and dissociated sharply to 10 RU (B). Control experiment was carried out similarly on uncoupled sensor chip surface. To determine the binding affinity () of the ribosomes to MOD and Pro-RIP, ribosomes of different concentrations were allowed to interact with the maize RIP-immobilized sensor chip surface. The values of MOD and Pro-RIP were 6.33 ± 0.73 and 500 ± 46.20 nM, respectively. The 80-fold decrease of binding affinity of Pro-RIP was due to the slower association rate and faster dissociation rate.
docking of Pro-RIP to yeast ribosome showed that if the adenine ring of A-2697 (analogous to A-4324 of rat 28S rRNA) is placed properly into the active site pocket, the internal inactivation region would clash with A1174-C1179, G1195-A1220 and G2507-A2511 of the 25S rRNA (). Therefore, this region of Pro-RIP probably sterically hinders the interaction of the protein with the ribosome.
Maize RIP represents a unique class of RIP. It has a prominent internal acidic region spanning Ala163–Asp189, which is absent in type I and type II RIPs. Previously, it was predicted that the internal acidic region would alter the arrangement of key residues in the active-site cleft or disrupt protein folding, thus affecting the catalytic activity (). To shed light on its biochemical function, we set out to elucidate the structures of Pro-RIP and MOD.
Secondary structure analysis of MOD-WT and Pro-RIP-WT by nnPredict program () showed that the N-terminal 5 aa is a flexible loop and crystallization of MOD and Pro-RIP was only achieved by deleting these 5 aa (data not shown). The overall structures of Pro-RIP and MOD are very similar, except the presence of the internal inactivation region Ala163–Asp189 in Pro-RIP (). Each protein has two domains, consisting of five α-helices and five-stranded mixed β-sheet in the large N-terminal domain. The small C-terminal domain is composed of four α-helices, with a bend between helices G and H. This helical bend is conserved among RIPs. The conserved active site residues Tyr94, Tyr130, Glu207, Arg210 and Trp241 are located at the cleft between the N-terminal and C-terminal domains. The tyrosine rings of Tyr94 and Tyr130 in MOD are facing each other, which may facilitate the insertion of A-4324 adenine ring of 28S RNA. In Pro-RIP, the tyrosine ring of Tyr94 assumes a different conformation. It flips towards the adenine-binding site such that it becomes perpendicular to the ring of Tyr130, with its hydroxyl group hydrogen-bonded to the carboxyl group of Gly128 (A). RIPs in Family Poaceae have more tryptophan residues than those in other families. In maize RIP, there are a total of six tryptophan residues of which only Trp241 in the active site is conserved. The only two cysteine residues (Cys51 and Cys206) of maize RIP are 14.37 Å apart, without disulphide-bond linkage.
The overall root mean square deviation (RMSD) among the structures of TCS, PAP, RTA and SO6 is 0.755. However, the value increases to 1.516 when MOD is included, suggesting that maize RIP is more structurally distinct. Compared to other RIPs, the α-helix B and β-strand 6 in the large domain are missing in MOD, and the anti-parallel β-strands 7 and 8 in the small domain are replaced by a short α-helix (). These structural differences, nevertheless, do not significantly diminish the ability of MOD to inhibit protein synthesis , implying that the variable regions are not crucial for the enzymatic activity. In Pro-RIP, the internal inactivating region Ala163–Asp189 is very rich in acidic residues. Our structure shows that the internal inactivating region consists of a flexible loop (Ala163–Ala179) and a long α-helix (Ala180–Ala188) (). Since this fragment is located on the protein surface and is 15.5 Å away from the active site, the conformation of the protein and the active-site cleft should not be affected.
At the C-terminal region, the hydrogen bonds Val280 N to Pro60 O and Val280 O to Leu62 N are conserved among RIPs. In TCS, the corresponding hydrogen bonds are Leu240 N to Pro35 O and Leu 240 O to Leu37 N. Deletion of these hydrogen bonds has been shown to disrupt the folding of TCS. Therefore, the C-terminal region in TCS can only be deleted up to Leu240 for an active variant (). It will be of interest to investigate if the hydrogen bonds orchestrated by Val280 also play a role in the folding of Pro-RIP and MOD.
Our biochemical studies showed that the presence of the internal inactivation region in Pro-RIP affects the cytotoxicity but not the uptake to JAR cells (A and B). Pull-down assay indicated that MOD, but not Pro-RIP, interacts with rat ribosomes (A). Surface plasmon resonance analysis also showed that the binding affinity () of Pro-RIP is about 80-fold higher than that of MOD (B). These indicated that the internal inactivation region of Pro-RIP might obstruct the protein to dock onto the ribosome. Indeed, docking of Pro-RIP to yeast ribosome showed that the internal inactivation region clashes with multiple sites of the 25S rRNA (). It has been reported that the inhibitory activity of Pro-RIP on maize ribosomes is lower than that of MOD (). Therefore, the internal inactivation region may serve the purpose of preventing the maize RIP from attacking its cognate ribosomes .
In TCS and saporin, the ribosome-binding sites are located between the anti-parallel β-sheets 7 and 8 in the C-terminal domain (,). Since the corresponding region in Pro-RIP is replaced by a short alpha-helix, and the internal inactivation region is located on the surface of helices D and E in the N-terminal domain, we predict that the ribosome interaction site on maize RIP may be different from that of TCS and saporin.
In the active sites of many RIPs of dicotyledonous plants, there are two glutamate residues. The carboxyl group of one of the glutamate residues stabilizes the oxocarbenium ion-like transition state. The other glutamate residue serves as a backup when the catalytic glutamate is mutated (,). Interestingly, in the active site pocket of Pro-RIP and MOD, only one glutamate residue (Glu207) is found. The residue corresponds to the backup glutamate that has turned into valine (Val238). This phenomenon is also observed in other known RIPs of the Family Poaceae, such as rice RIP (GenBank: BAB85659) and JIP60 (GenBank: AAB33361). This raises the puzzle of the absence of the backup glutamate in these RIPs. We found that the ribosome-inactivating activity of [E207A]-MOD decreased by about 556-fold (), confirming the importance of this residue. Variants [V238E]-MOD and [E207D-V238E]-MOD were expressed as inclusion bodies, showing the existence of two long side chains in the active site disrupts the folding and/or structure of the protein. On the other hand, the variant [E207AV238E]-MOD was soluble and regained partial ribosome-inactivating activity, indicating that a glutamate residue may be placed in position 238 as a backup for improving the activity.
Compared to other RIPs, the main chains of Ile134-Met144 and Gln225-Thr231 of maize RIP exhibit some significant structural differences. With reference to TCS, these two chains have shifted 3.67 and 3.74 Å inward, respectively (B). In maize RIP, α-helix C (Tyr130-Ile134) is shorter, due to the presence of Gly135 as an α-helix C-cap terminator (). The hydrogen bond between Ile134 O and Lys137 N ensures an inward shift of the fragment Ile134-Met144. Moreover, the hydrogen bonds between Met144 O and Val228 N and Met144 N and Val228 O result in a tight packing of Gln225-Thr231 towards the active site. As a result, the active-site pocket becomes too small to accommodate an extra glutamate residue. Molecular modelling indicated that if Val238 is mutated to glutamate, Glu238 would clash with the side chain of Leu139 and Leu230 (Supplementary A). This is consistent with our finding that MOD variant V238E expressed as inclusion bodies. Furthermore, maize RIP is not prepared to have a glutamate residue in position 238. In TCS, the corresponding Glu189 is stabilized by Arg122 and Gln156 (Supplementary B). On the other hand, the latter 2 aa have become Leu139 and Val203 in maize RIP (Supplementary C).
Interestingly, known bacterial RIPs including Shiga toxin have only one catalytic glutamate in the active site (Supplementary D). In Shiga toxin, the position for the backup glutamate is occupied by Thr200. Previous phylogenetic analysis of representative plant and bacterial RIPs has indicated that maize RIP is more related to the latter (). Hence, it is likely that RIPs having one catalytic glutamate presents a prototype that acquires a second glutamate residue in the active site during evolution.
In conclusion, we have solved the crystal structures of Pro-RIP and MOD, the precursor and the mature form of maize RIP. Our data reveal the structural and functional differences of the two forms and the role of the internal inactivation region for its biological activities. The structure of the active-site pocket also indicates that maize RIP may be an intermediate step of evolution from prokaryotes to higher plants.
p
p
l
e
m
e
n
t
a
r
y
D
a
t
a
a
r
e
a
v
a
i
l
a
b
l
e
a
t
N
A
R
O
n
l
i
n
e
. |
The -activating responsive (TAR) RNA element of HIV-1, located at the 5′ untranslated region of the viral genome, is a 57-nt imperfect stem-loop essential for HIV replication (). TAR regulates transcription by interacting with viral and cellular proteins. The upper part of TAR constitutes the binding site of the viral protein Tat that recruits cellular proteins from the positive transcription elongation factor (P-TEFb) complex (). The TAR element displays a highly conserved 6-nt loop 5′-CUGGGA-3′ () and constitutes a valid target for designing ligands that could inhibit TAR–protein interactions, thus preventing the development of the virus.
The systematic evolution of ligands by exponential enrichment (SELEX) approach was used against TAR to generate high affinity ligands. DNA and RNA aptamers specific for TAR were identified by selection (,). These aptamers, folded as stem-loop, displayed a 5′-ACTCCCAT-3′ (for DNA) or 5′-GUCCCAGA-3′ (for RNA) consensus sequence in the apical loop, partially complementary to the TAR one, leading to the formation of TAR–aptamer ‘kissing’ complexes () (A). Similar motifs regulate gene expression in different organisms. Loop–loop interactions have been shown to regulate the copy number of the plasmid ColE1 (). The dimerization and encapsidation of retroviral RNAs of the Moloney murine leukaemia virus are initiated through the formation of a RNA kissing complex (). The dimerization of the HIV genome is also initiated at a conserved stem-loop structure that forms a kissing complex ().
The loop of the TAR RNA aptamer, termed R06(GA), is closed by selected conserved G and A residues on the 5′ and the 3′ sides, respectively. Mutations of these residues decreased the stability of the TAR–aptamer complex (). To improve the nuclease resistance of the aptamer, chemically modified analogues of R06(GA) were investigated (). Among them, the locked nucleic acid (LNA) has received particular attention (). LNA residues are ribonucleotide analogues containing a methylene linkage between the 2′- and 4′-C of the ribose ring that locks the sugar moiety in a -conformation () (B). This generates the most stable hybrids ever characterized with a Δ of +3 and +10°C per LNA residue upon binding to DNA and RNA, respectively (). Structural studies have pointed out interesting properties of oligonucleotides containing LNA and DNA residues. LNA changes the sugar conformation of adjacent DNA residues to the N-type making DNA/LNA hybrids good mimics of RNA (,). This effect reaches a maximum for oligonucleotides containing <50% of LNA residues. These properties of LNA were considered to generate an aptamer derived from the parent RNA aptamer R06(GA), with improved binding properties ().
Only a limited number of LNA modifications in the aptamer loop are tolerated for generating a ligand recognizing the TAR target (). In a recent work, all possible combinations involving LNA and 2′--methyl residues in the anti-TAR aptamer loop (except closing G, A residues) were tested (). Results pointed out that the affinity of these aptamers for TAR decreased with increasing LNA residues in the loop. Moreover, stability also depends on the location of LNA residues. Three aptamers with one or two LNAs at positions 5 or 5,6 or 5,7 (the G residue on the 5′ side of the loop being numbered 1) display affinities for TAR one or two orders of magnitude higher than the parent RNA aptamer. One of these combinations, containing two LNAs at positions 5 and 6 in the loop was able to inhibit the TAR-dependent expression of a luciferase reporter gene in cultured cells (). On the other hand, no inhibitory effect was observed with an aptamer containing four LNAs in the loop at positions 2, 4, 6 and 7. Preliminary structural studies by nuclear magnetic resonance (NMR) revealed that the position of LNAs in the loop induced structural modifications at the stem-loop junctions ().
To elucidate the origin of the stabilization due to the GA combination and to the introduction of LNAs in the loop, the solution structure of the kissing complex formed between TAR and the LNA/RNA chimeric aptamer, LR06(GA), with LNAs at positions 5 and 6 in the loop was solved by NMR. Structural analysis reveals that the stability results from the formation of a G•A pair that increased the stacking at the stem-loop junction. In addition, the data show that the LNA residues at positions 5 and 6 also provide additional stability compared to the parent RNA aptamer.
Unlabelled LNA/RNA chimeric aptamer LR06(GA) was purchased from Eurogentec (Belgium). Milligram quantities of TAR RNA were prepared unlabelled or C-N labelled by transcription with T7 RNA polymerase from an oligonucleotide template containing a 2′--methyl modification at position 2 (). Unlabelled R06(GA) was also prepared. Two G were introduced at the 5′-end to improve the yield of the transcription (A). Labelled NTPs (nucleotide triphosphate) were purchased from Spectra Stable Isotopes (Columbia, USA). LR06(GA), R06(GA) and TAR were purified as described by Puglisi and Wyatt (). After electroelution and ethanol precipitation, resuspended RNAs were then dialysed for 48 h against the buffer used for NMR experiments. Samples were concentrated by lyophilization and resuspended in 90%/10% HO/DO for experiments involving exchangeable protons and in 100% DO for non-exchangeable protons experiments. Each sample was refolded by heating at 95°C (2 min) and snap-cooled at 4°C. Complexes were formed by addition of TAR to LR06(GA) monitoring the imino region of 1D spectra. At each point of the titration, the sample was heated at 95°C and snap-cooled at 4°C.
NMR experiments were recorded at 500 MHz on a Bruker Avance spectrometer equipped with a BBI (broad band inverse) -gradient probe. NMR data were processed using TopSpin (Bruker) and were analysed using Sparky software packages (). NMR experiments were performed in 10 mM sodium phosphate buffer (pH 6.4). The concentration of unlabelled and labelled RNAs samples ranged from 0.8 to 1 mM. Samples volumes were 280 μl in Shigemi NMR tubes.
H, C and N assignments were obtained using standard homonuclear and heteronuclear methods. NMR data were acquired at 4 or 15°C for exchangeable protons and at 15°C for non-exchangeable protons experiments. Solvent suppression for samples in 90%/10% HO/DO was achieved using the WATERGATE and ‘Jump and Return’ sequences (). The residual HDO resonance in DO was suppressed using low power pre-saturation. Two-dimensional NOESY spectra in 90%/10% HO/DO were acquired with mixing time of 300, 150 and 50 ms. Base pairing was established via sequential nuclear Overhauser effects (NOEs) observed in 2D NOESY spectra at different mixing times. NOESY spectra with mixing times of 50, 150, 200 and 400 ms in DO were acquired at 15°C.
Heteronuclear NMR spectra were measured at 15°C in 100% DO at the exception of the H-N heteronuclear single quantum coherence (HSQC) experiment that was acquired at 4°C in 90%/10% HO/DO (). C or N decoupling during acquisition time was achieved using globally optimized alternating phase rectangular pulse (GARP) composite pulse sequence. The assignment of non-exchangeable proton was achieved using 2D NOESY experiment at different mixing times, 2D TOCSY, 2D COSY-DQF (double quantum-filtered correlated spectroscopy) and 3D NOESY-NOESY (). The assignment of labelled TAR was completed using H-C HSQC, 3D HCCH-TOCSY and 3D NOESY-HMQC (,,). Resonance assignments are reported relative to TSP(3-(trimethylsilyl) propionic acid-d4 sodium salt) and to known chemical shifts.
Classical A-helix angles values were used for G1 to G4, C13 to C16, C1* to C3* and G14* to G16* (A): α = −65 ± 30°, β = 178 ± 30°, γ = 54 ± 30°, ε = −155 ± 30°, ζ = −71 ± 30° as described by Chang and Tinoco () and Kieken (). Torsion angles χ (chi) were derived from the observation of intra-residue H6/8-H1′cross-peak volumes. All bases, except G12 and G11*, were restrained to anti-conformation (χ = −158 ± 30°). Torsion angles values ν to ν were derived from the analysis of COSY-DQF and TOCSY experiments. Nucleotides with no COSY and no TOCSY cross-peaks between H1′ and H2′ protons, i.e. all nucleotides except G11*, were restrained to the ′--conformation (ν = 6 ± 15°, ν = −25 ± 15°, ν = 37 ± 15°, ν = −37 ± 15°, ν = 21 ± 15°). This first set of structural restraints was used for structure calculations. Analysis of converged structures revealed that for U7 to G12 and C7* to A10*, classical A-helix ε angles values were observed. For G8 to G10 and C7* to A10*, classical A-helix ζ angles values were also observed. Thus, for U7 to G12 and C7* to A10* nucleotides, classical A-helix ε angles values (ε = −155 ± 30°) were added and for G8 to G10 and C7* to A10*, classical A-helix ζ angles values (ζ = 71 ± 30°) were also included. This second set of structural restraints was used for structure calculations. These constraints led to higher convergence efficiency with similar converged structures. In addition, in order to validate our structures, two additional calculations were run on the one hand including only ε angles values for U7 to G12 and C7* to A10* nucleotides, and on the other hand including only ζ angles values for G8 to G10 and C7* to A10*. Both structure calculations led to similar results as those obtained without these angle restraints.
Structures were calculated using CNS (cristallography and NMR system) torsion angle molecular dynamics (TAMD) protocol for nucleic acids using NOE and dihedral angle restraints (,). LNA nucleotides (C9* and A10*) were defined by modifying the appropriate nucleotides C and A, and atomic charges were determined according to those previously described (). One hundred structures were generated from two randomized extended strands. The first stage consisted of a high-temperature torsion angle dynamics for 30 ps at 20 000 K with a van der Waals scale factor of 0.1. During the second stage, the molecules were cooled for 40 ps of torsion angle dynamics with a van der Waals scale factor increased from 0.1 to 1. In the third stage, molecules were submitted to a second slow cooling for 35 ps in Cartesian space where the van der Waals scale factor increased from 1 to 4. Finally, 10 Powell cycles of energy minimization of 300 steps each were done. The structures with zero violation on NOE distance (0.2 Å), dihedral (5°) and with the lowest energy were selected. Structures were visualized and were analysed with MOLMOL and PyMOL software packages (,).
Coordinates of 10 converged NMR structures have been deposited in the Protein Data Bank (PDB ID code is 2PN9). Coordinates of converged NMR structures calculated with no ε and ζ angles restraints for the loop–loop helix have also been deposited in the Protein Data Bank (PDB ID code is 2OOM).
Thermal denaturation of complexes was monitored on a Uvikon XL UV/visible spectrophotometer (BIOTEK) equipped with a 10-positions sample holder and a Peltier temperature control accessory. The experiments were performed at 1 μM final concentration of each RNA hairpin in 20 mM cacodylate buffer, containing 0.3 mM (or 3 mM) Mg, pH 7.3 at 20°C, in 200 μl micro-quartz cuvettes. Samples were overlaid with 300 μl of mineral oil to prevent evaporation at high temperature. An initial 30-min equilibrium time at 5°C was included prior to the temperature ramping. Denaturation of the samples was achieved by increasing the temperature at 0.4°C/min from 5 to 95°C and the UV absorbance was followed at 260 nm. The melting temperature () was determined as the maximum of the first derivative of the UV melting curves.
Surface plasmon resonance experiments were performed on a BIAcore™ 3000 apparatus (Biacore AB, Sweden) as described previously (). Briefly, binding kinetics were performed at 23°C in 10 mM phosphate buffer, pH 7.2, containing 20 mM sodium chloride, 140 mM potassium chloride, 0.3 mM magnesium chloride and 0.005% surfactant P20 (Biacore). Samples were injected at 20 μl/min at two different concentrations. Three independent experiments were performed for each run. The kinetic parameters, and , were determined assuming a pseudo-first-order model by direct curve fitting of the sensorgrams using the Bia-evaluation 4.1 software (Biacore) (Supplementary Data, and ). The dissociation equilibrium constant, , was calculated as /.
#text
We investigated the structure of a complex formed between the TAR RNA element of HIV-1 and a LNA/RNA chimera derived from an aptamer identified previously (,,). Two important parameters account for the high stability of this complex.
First of all, LNAs at positions C9* and A10* in the aptamer loop induce an increased stability of the loop–loop helix. We have observed that the C6•G11* pair is more stable than the homologous pair in the RNA–RNA complex TAR/R06(GA). We have also noticed that the conformation of the loop–loop helix of TAR/LR06(GA) differs from the one described by Chang and Tinoco (). The loop–loop geometry of TAR/LR06(GA) is similar to the one of a RNA–RNA dimer recently described (). Indeed, the loop–loop dimer formed by SL1 HIV-1 is close to a canonical A-type helix, except at the stem-loop junctions (). Thus, our structure shows that the introduction of two LNA nucleotides in a loop–loop context allows the hybridization with a RNA partner, retaining a normal Watson–Crick duplex framework and stabilizing adjacent pairs, with a loop–loop conformation close to an A-type helix. The increased stability due to the presence of LNA residues originates from the pre-organized geometry in a -conformation that likely provides an ideal geometry for the complex formation. LNAs would promote stacking interactions stabilizing the complex. This phenomenon has been previously described for 2′-deoxy-2′-fluoro-β--arabinonucleosides (2′F-ANA) (). Thermodynamic parameters were measured on hybrids formed between 2′F-ANA and RNA or DNA. It was shown that the increased stability of the complex appeared to originate in conformational pre-organization of the fluorinated sugars and a favourable enthalpy of hybridization.
The second important structural feature is the formation of the non-sheared G5*•A12* base pair closing the loop of the aptamer. This provides additional stabilization at the helical junction compared to the UA combination present in the complex TAR/TAR*(UA) (). The replacement of a pyrimidine cycle by a purine cycle increased the stacking effect. This accounts for the enhanced affinity of aptamers with G•A base pair for TAR over the rationally designed hairpin with U•A closing the loop () (). Our experimental results are in agreement with the molecular dynamics simulation performed by Beaurain () on two RNA–RNA complexes derived from TAR/TAR*(UA). The authors used the TAR/TAR*(UA) NMR solution structure as the starting structure to model the complex with a GA loop closing combination. They replaced the U by a G and positioned the G and A residues in a Watson–Crick/Watson–Crick base pair according to preliminary NMR results.
Finally, in a previous work, the inhibitory effect of LNA derivatives was tested in a TAR-dependent double-luciferase HeLa cell reporter system (). It was demonstrated that the aptamer LR06(GA) inhibited the TAR-dependent expression of a luciferase reporter gene whereas an aptamer derivative with four LNAs in the loop (LR06) and that binds to TAR with a reduced affinity, had no effect (). Previously, we also demonstrated that TAR/LR06 and TAR/R06(GA) had different structures at the stem-loop junctions. Indeed, in the case of TAR/LR06 U6* and A11 residues are involved in a classical Watson–Crick base pair and the non-sheared base pair G5•A12* is not formed (). Thus, the location of LNAs residues in the aptamer loop induces structural modifications at stem-loop junctions. These results unambiguously demonstrate that subtle structural changes in the loop–loop helix and at the stem-loop junctions of the kissing complex correlate with the observed biological effect.
The TAR RNA element, and more generally RNA structures constitute valid targets for designing agent of potential therapeutic interest (). Aptamers that display a high specificity of recognition expand the virologist's tool base (,). Structural studies will bring additional information that will allow refining the ligands identified by SELEX.
p
p
l
e
m
e
n
t
a
r
y
D
a
t
a
a
r
e
a
v
a
i
l
a
b
l
e
a
t
N
A
R
O
n
l
i
n
e
. |
Subsets and Splits