Protein phosphatase 5 is a negative regulator of separase function during cortical granule exocytosis in C. elegans

Mutations in the Caenorhabditis elegans separase gene, sep-1, are embryonic lethal. Newly fertilized mutant embryos have defects in polar body extrusion, fail to undergo cortical granule exocytosis, and subsequently fail to complete cytokinesis. Chromosome nondisjunction during the meiotic divisions is readily apparent after depletion of sep-1 by RNAi treatment, but much less so in hypomorphic mutant embryos. To identify factors that influence the activity of separase in cortical granule exocytosis and cytokinesis, we carried out a genetic suppressor screen. A mutation in the protein phosphatase 5 (pph-5) gene was identified as an extragenic suppressor of sep-1. This mutation suppressed the phenotypes of hypomorphic separase mutants but not RNAi depleted animals. Depletion of pph-5 caused no phenotypes on its own, but was effective in restoring localization of mutant separase to vesicles and suppressing cortical granule exocytosis and cytokinesis phenotypes. The identification of PPH-5 as a suppressor of separase suggests that a new phospho-regulatory pathway plays an important role in regulating anaphase functions of separase.


Introduction
The maintenance of the cohesin complex that holds newly replicated sister chromatids together through metaphase and its timely dissolution at the onset of anaphase are crucial for a successful mitotic cell division. The widely conserved protease, separase, cleaves the Scc1 (also known as Rad21) subunit of the cohesin complex, thereby allowing sister chromatids to be pulled apart towards the spindle poles at anaphase. Before the metaphaseto-anaphase transition, separase is held in an inactive conformation by an inhibitory chaperone, securin. Securin is ubiquitylated by the anaphase promoting complex or cyclosome (APC/C) and subsequently degraded by the proteasome, leading to the activation of separase at anaphase onset.
In addition to the canonical function of separase in cohesin cleavage, new evidence points to other anaphase functions for separase during the meiotic divisions. In murine oocytes, separase is required for the successful extrusion of polar bodies by an unknown mechanism (Kudo et al., 2006). The Caenorhabditis elegans separase, sep-1, was shown to have a role in the exocytosis of specialized vesicles known as cortical granules (CGs), which occurs during the first meiotic division (Bembenek et al., 2007). In addition to localizing to chromosomes and meiotic spindle, the SEP-1 protein can also be found on CG vesicle membranes before their fusion with the plasma membrane during anaphase of meiosis I.
Separase also regulates membrane trafficking during mitosis in C. elegans (Bembenek et al., 2010). SEP-1-depleted embryos fail to complete cytokinesis. Separase localizes to the ingressing cytokinetic furrow and regulates the incorporation of RAB-11positive vesicles into the plasma membrane at the furrow. In human cell lines, both securin and separase regulate membrane traffic and protein secretion (Bacac et al., 2011). A genome-wide RNAi screen in a Drosophila cell line identified separase as a candidate required for constitutive protein secretion and Golgi organization (Bard et al., 2006). Therefore, separase controls membrane trafficking events in addition to its well-conserved role in cohesin cleavage.
The regulation of separase in membrane trafficking events in both mitotic and meiotic cells is currently under intensive investigation in a number of laboratories. Separase is inhibited by phosphorylation and Cdk1-cyclinB binding (Gorr et al., 2005) and thus securin is not the sole regulator of separase. Phosphorylation also regulates securin (Agarwal and Cohen-Fix, 2002) and the substrates of separase (Alexandru et al., 2001;Rogers et al., 2002). Although the canonical regulatory pathway is sufficient to ensure the timely onset of anaphase, phospho-regulation of the separase pathway could be crucial for the coordinated regulation of cohesin cleavage and membrane trafficking.
In C. elegans, the sep-1 gene is essential for embryonic viability partly because of its role in chromosome segregation (Siomos et al., 2001). Previously, the only allele known was e2406ts, a temperature-sensitive allele. Further analysis revealed that embryos with this allele have a significant reduction in the rate and overall number of CG exocytic events (Bembenek et al., 2007). Just after fertilization, chitin synthesis is initiated (Johnston et al., 2010). Next, during anaphase I, the contents of the CGs are released and assembled to form the impermeable eggshell (Sato et al., 2008). Hence the reduction of cortical granule exocytosis (CGE) in the sep-1 mutant causes eggshell defects that result from insufficient proteoglycan secretion from CGs (Olson et al., 2006). The precise function of SEP-1 on these vesicles has yet to be determined. Importantly, homozygous sep-1(e2406ts) embryos show minimal chromosome segregation defects (Bembenek et al., 2007), although RNA interference (RNAi)-treated embryos do display this expected defect (Siomos et al., 2001). Furthermore, the SEP-1 protein in sep-1(e2406ts) mutants fails to localize to vesicles, yet still localizes to the spindle (Bembenek et al., 2007). These data suggest that sep-1(e2406ts) specifically affects the membrane trafficking functions of sep-1, making it a 'separation-of-function' allele.
Here we report the characterization of two novel sep-1 alleles that produce defects in CGE (Bembenek et al., 2007) similar to the sep-1(e2406ts) mutant. Using the advanced genetics available in C. elegans, we identified mutations in the pph-5 gene that function as suppressors of the embryonic lethality of separase mutants. The proteins encoded by these new separase alleles did not have the same subcellular localization as wild-type SEP-1 and these new sep-1 alleles were not equally suppressed by mutations in the pph-5 gene. Interestingly, the pph-5 mutants restored the localization of SEP-1 mutant proteins to CGs. The pph-5 mutants displayed no developmental defects in an otherwise wild-type background. Our work defines a novel regulatory mechanism involving the PPH-5 phosphatase, which regulates the CGE and cytokinesis activities of separase during anaphase. Given the numerous roles that phosphorylation plays in the regulation of separase activity and substrate susceptibility, our discovery of PPH-5 as a crucial regulator of separase reveals new insights into the phospho-regulation of the separase pathway in the early embryo of C. elegans.

Results
Characterization of mutant phenotypes of the novel sep-1 alleles We identified two novel alleles of the C. elegans sep-1 gene, ax110 and ax521, from a collection of mutants isolated in a temperature-sensitive embryonic lethal screen (Golden et al., 2000). Both these alleles failed to complement sep-1(e2406ts); sequencing revealed that ax110 and ax521 are point mutations in separase that result in the amino acid changes H738P and C232Y, respectively (Fig. 1A). These sep-1 alleles exhibited a recessive non-conditional maternal-effect lethal (Mel) phenotype that was fully penetrant at the non-permissive temperature of 24˚C (Table 1, lines 2 and 3) and .95% penetrant at 16˚C (data not shown). Unlike sep-1(e2406ts), these two alleles did not cause sterility at any temperature. DAPI staining of embryos revealed a failure to extrude polar bodies (Fig. 1C); the resulting polyploid embryos continued to undergo sequential rounds of DNA replication without cytokinesis (Fig. 1E,F). The terminal phenotype was arrest as multinucleated one-cell embryos ( Fig. 1C,E,F). Interestingly, all three alleles caused minimal chromosome segregation defects during the meiotic divisions ( Fig. 2), unlike sep-1 RNAi (Siomos et al., 2001), suggesting that cohesin cleavage during the meiotic divisions is normal. However, during mitosis (Fig. 1E,F; Fig. 2), lagging chromosomes and chromosome bridges were observed, but to a lesser degree than with sep-1 RNAi. We believe that the chromosome segregation defects that we observed during mitosis in mutants with these alleles are due to the incorporation of abnormal chromatin, derived from failed polar body extrusion, into the metaphase plate. In cases where chromatin from failed polar body extrusion did not incorporate into the spindle, nondisjunction was much less severe ( Fig. 2E-H; supplementary material Movie 1). Given that the sep-1 mutations caused minimal meiotic chromosome segregation defects, but were much more penetrant for polar body extrusion and cytokinesis defects than strict loss of function by sep-1 RNAi ( Fig. 2Q-T), we conclude that these three sep-1 alleles are probably separation-of-function alleles.
To more fully characterize the mitotic phenotypes of the new sep-1 alleles, we analyzed embryos expressing GFP::histone representation of the C. elegans SEP-1 protein denoting the locations of the amino acid substitutions encoded by the mutations ax110, ax521 and e2406ts, as well as av103, the intragenic suppressor of sep-1(e2406ts). The regions with corresponding homology to the separase ortholog pfam group (peptidase C50) and the catalytic histidine and cysteine residues are indicated. The truncated and frame-shifted protein predicted to be expressed from the sep-1(ok1749) allele is also shown. The shaded box represents the out-of-frame residues that are encoded before the premature stop codon. (B,C) In a live sep-1(+) embryo (B) only the first polar body (indicated by a white arrow) was stained by DAPI (yellow), whereas in a live sep-1(ax110) embryo (C), the polyploid zygotic DNA stained with DAPI. The embryos expressed GFP::histone to reveal zygotic DNA (green). (D-F) Methanol-fixed embryos stained with anti-tubulin antibodies (green) and TOTO-3 (red). (D) A two-cell wild-type embryo, (E) a 'four-cell' polyploid ax110 embryo and (F) a 'two-cell' polyploid ax521 embryo. The arrowheads indicate successful chromosome separation at anaphase. Scale bars: 10 mm.
H2B. At 25˚C, 83% (n55/6) of sep-1(ax110) embryos showed cytokinesis failures whereas only 50% (n53/6) of the embryos showed chromosome bridges during mitotic anaphase ( Fig. 2E-H,Q,R). Similar observations were made with sep-1(ax521) embryos (data not shown). By contrast, sep-1 RNAi caused a higher incidence of more severe chromosome nondisjunction but fewer cytokinesis failures ( Fig. 2I-L,Q,R) (Bembenek et al., 2010), consistent with the hypothesis that these new alleles are separation-of-function alleles. The homozygous sep-1(ax110 and ax521) mutant progeny developed into fertile adults at all temperatures (16-25˚C) in contrast to mutants with the conditional allele e2406ts, where homozygous animals developed into sterile uncoordinated (Stu) adults at temperatures above 20˚C (Table 1, lines 4 and 5) (Siomos et al., 2001). A deletion allele, sep-1(ok1749), which removes a 1.2 kb region from within intron 3 to within exon 6 ( Fig. 1) was predicted to cause aberrant splicing into sequences within exon 6. This would result in a frameshift that would encode a truncated SEP-1 protein without a protease domain. This protein is unstable as it was not detected by western blot analysis (supplementary material Fig. S1). Homozygous sep-1(ok1749) animals died during larval development or developed into Stu adults.
Loss of separase function causes a defect in CGE and eggshell formation (Bembenek et al., 2007;Siomos et al., 2001). Therefore, we tested whether the novel sep-1 alleles also confer an eggshell phenotype. Live embryos were isolated in buffer containing DAPI and examined by fluorescence microscopy. In post-meiotic wild-type embryos, only the first polar body was DAPI labeled because it resides outside the permeability barrier of the C. elegans eggshell (Fig. 1B). Eggshell defects disrupt the permeability barrier, allowing DAPI labeling of zygotic chromatin. The zygotic chromatin of sep-1(ax110 and ax521) embryos was accessible to DAPI, indicating eggshell defects ( Fig. 1C; Table 2, lines 1-3).
To further assay the eggshell phenotype, we quantified the number and rate of CG exocytic events during anaphase I using an assay in which the incorporation of FM2-10 dye into the embryo plasma membrane is monitored at the sites of CG fusion (Bembenek et al., 2007). Both the new sep-1 mutants showed a significant decrease in the total number and overall rate of CGE events (Fig. 3A,B,E,G) and had an increase in the time interval between the first and last CGE event ( Fig. 3C; supplementary material Movies 2-4) when compared with wild-type embryos. Interestingly, although all the sep-1(ax110) embryos failed to extrude polar bodies, none (n50/7) of them displayed nondisjunction during the first meiotic division (Fig. 2S,T). Therefore, we have identified two novel separation-of-function alleles of sep-1 that are capable of promoting meiotic chromosome segregation, but fail to promote CGE.

Characterization of SEP-1 localization in mutants
The SEP-1 protein has a dynamic localization pattern during meiosis I (Bembenek et al., 2007). As the maturing oocyte enters M phase, shortly before fertilization, SEP-1 localizes to cortical filaments (Bembenek et al., 2007). After fertilization, SEP-1 was found to localize to both the meiotic spindle and the CGs (Fig. 4A,B; Fig. 5). In mitotic embryos, SEP-1 localized to the spindle matrix during metaphase (Fig. 4C,D). We characterized the localization of the mutant SEP-1 proteins and detected several aberrations (Figs 4,5). Previously, we found that SEP-1 in sep-1(e2406ts) embryos still localized to the anaphase I meiotic spindle, but not to CGs (Bembenek et al., 2007). The SEP-1 protein in sep-1(ax110) animals did not localize to any of these subcellular locations ( It was also present on the spindle matrix in polyploid embryos cycling through mitosis (Fig. 4G,H). Western blot analysis revealed that the SEP-1 protein was produced in these mutants (supplementary material Fig. S1). Additionally, the distal germlines in all mutants contained abundant cytoplasmic SEP-1 that could be reduced upon sep-1 RNAi feeding (data not shown).

Characterization of pph-5-mediated suppression of sep-1 lethality
We used the temperature-sensitive sep-1(e2406ts) allele to perform an EMS mutagenesis screen for secondary suppressor mutations at the non-permissive temperature of 20˚C. Although most studies have shifted sep-1(e2406ts) animals to 25˚C (Bembenek et al., 2010;Bembenek et al., 2007;Siomos et al., 2001), we used 20˚C, the lowest temperature at which embryonic lethality is fully penetrant. Three independent suppressors were recovered from screening 100,802 haploid genomes. Suppressors av101 and av102 mapped to linkage groups (LG) V and LG III, respectively, whereas av103 was identified as an intragenic sep-1(e2406ts) animals at their non-permissive temperature of 20˚C layed a brood of embryos that fail to hatch (line 4), but are suppressed by pph-5(av101 or tm2979) at this temperature (lines 9 and 23). The smd-1 gene was used as a negative control, as it has no obvious RNAi phenotype (lines 12-15).
Numbers in the second and third columns reflect the total progeny counted after pooling several experiments (with a minimum of five mothers per strain). Unsuppressed strains were maintained at 24˚C as heterozygotes balanced over the hT2 balancer chromosome. Homozygous L4 sep-1 mutants were moved to new plates and embryonic viability and lethality were monitored daily.  (Bembenek et al., 2010). Stronger sep-1(RNAi) treatments completely prevented chromosome segregation (data not shown). (M-P) Loss of pph-5 activity did not rescue the loss-of-function phenotypes of sep-1(RNAi) embryos (n56; N,O,S) and cytokinesis failures were observed in 33% of the embryos, similar to the sep-1 RNAi phenotype in wild-type embryos (n56; P,T). In sep-1(ax110);pph-5(av101) embryos, no chromosome segregation defects (n57; S) or cytokinesis failures were observed (n56; T) during either meiosis or mitosis. Scale bar: 10 mm. missense mutation (L556F) in sep-1(e2406ts) (Fig. 1A). The av101 allele could suppress sep-1(e2406ts) at 20˚C (Table 1, compare lines 4 and 9), but not at 24˚C (Table 1, lines 5 and 11). Interestingly, av101 could suppress the sep-1(ax110) mutation at all temperatures (Table 1, compare lines 2 and 7), but not sep-1(ax521) at any temperature (Table 1, lines 3 and 8, and data not shown). Curiously, av101 suppressed sep-1(e2406ts) semidominantly, but behaved as a recessive suppressor with sep-1(ax110) (supplementary material Table S1). This semi-dominant suppression of sep-1(e2406ts) by av101 suggests that the different sep-1 alleles are sensitive to the dose of the protein mutated in av101 animals. The allele av102 failed to suppress either of these new sep-1 alleles (data not shown). Neither av101 nor av102 was able to suppress sep-1 RNAi or sep-1(ok1749) (data not shown). We also observed equivalent rates of chromosome nondisjunction and cytokinesis failures in wildtype and av101 embryos treated with sep-1 RNAi, suggesting that av101 does not suppress sep-1 RNAi at the cellular level ( Fig. 2I-P,Q,R). Because these sep-1 mutants have reduced chromosome segregation defects, it is probable that the new suppressors regulate the exocytic role of separase. The remainder of this report will focus on the characterization of the av101 allele.
The semi-dominant nature of av101 suppression thus suggests that even a modest reduction in PPH-5 function is sufficient to suppress some sep-1 mutant phenotypes. This hypothesis is supported by our observation that expression of PPH-5 protein was reduced but not completely abolished by RNAi feeding (supplementary material Fig. S2), even though this method suppressed both sep-1(e2406ts and ax110) mutants.  The presence of SEP-1 at cortical filaments, the meiotic spindle, cortical granules, and the mitotic spindle matrix was scored in each strain and the embryonic cell cycle stage was determined independently by inspecting the chromosome configuration with DAPI staining. The number of positively stained cells and the total number of cells examined for a particular cell cycle stage are shown. *The cortical granules of sep-1(ax110); pph-5(av101) were labeled with SEP-1 antibodies, however, the overall intensity was reduced compared with wildtype embryos. sep-1(ax110) animals bearing the pph-5(av101) (av) or pph-5(tm2979) (tm) suppressor mutations. Each lane was loaded with protein equivalent to 50 gravid hermaphrodites. Actin was used as a loading control, and the asterisk indicates a non-specific cross-reactive band. Full-length and internally deleted PPH-5 proteins are expected to migrate at 56 and 50 kDa, respectively.

CGE and SEP-1 localization defects are suppressed by pph-5 mutations
We next investigated whether PPH-5 restores the membrane trafficking functions of the separase mutants. With regard to the DAPI permeability and CGE quantification assays, sep-1(ax110); pph-5(av101) embryos were indistinguishable from wild type, demonstrating that eggshell formation and CGE were no longer defective ( Fig. 3; Table 2). In addition, sep-1(ax110); pph-5(av101) embryos were fully rescued from chromosome nondisjunction defects and cytokinesis failures during mitosis (Fig. 2Q,R). Polar body extrusion was also fully rescued (Fig. 2S). Disruption of pph-5 had no effect on the eggshell and CGE defects observed in sep-1(ax521) mutant embryos; such double mutant embryos were still permeable to DAPI and failed to undergo CGE (data not shown).
Next we determined whether PPH-5 affected the localization of separase. Indeed, in sep-1(ax110); pph-5(av101) embryos, the localization of SEP-1 to the meiotic spindle and cortical filaments was restored (Fig. 4M). SEP-1 was also detected at the CGs during the first meiotic anaphase although at a reduced intensity compared with that in wild-type embryos ( Fig. 4N; Fig. 5). We observed similar restoration of SEP-1 localization patterns when sep-1(ax110) was suppressed by pph-5(tm2979) or pph-5(RNAi) (data not shown). The subcellular localization of mutant SEP-1 in sep-1(ax521) embryos was not affected by pph-5(av101) (Fig. 5,  data not shown). Surprisingly, although the sep-1(ax110); pph-5(av101) embryos appeared to develop normally, SEP-1 localization to the mitotic spindle matrix was never observed (Fig. 4O,P; Fig. 5) but instead appeared cytoplasmic. The localization of wild-type SEP-1 was not affected in pph-5(av101) embryos ( Fig. 4Q-T; Fig. 5). These data reveal that PPH-5 suppresses separase mutations by restoring its function during vesicle trafficking and not by a bypass mechanism. This is consistent with the fact that these sep-1 alleles preferentially affected vesicle functions of separase more than chromosome segregation functions. This observation might account for the ability of pph-5 mutations to suppress the hypomorphic sep-1 alleles but not strict loss of function of sep-1. Therefore, PPH-5 is a negative regulator of the exocytic function of separase, influencing the localization of separase to vesicles at the appropriate times during cell division.

PPH-5 expression and localization in the C. elegans germline
To determine the subcellular localization of PPH-5 in C. elegans embryos, a GFP::PPH-5 translational fusion was expressed using the germline-restricted pie-1 promoter and 39UTR sequences (Cheeseman and Desai, 2005;Reese et al., 2000) (see supplementary material Fig. S3A for schematic of the transgene). GFP::PPH-5 was cytoplasmic during interphase and became enriched at the spindle matrix in mitosis, similar to what is observed with GFP::SEP-1 and SEP-1 antibody staining (compare Fig. 7B,C with Fig. 4C,D) (Bembenek et al., 2007). The GFP::PPH-5 protein dispersed from the spindle during anaphase (supplementary material Movie 5). Attempts to confirm this pattern with various anti-PPH-5 staining protocols were unsuccessful, despite raising antibodies to many epitopes. Therefore, PPH-5 localizes to similar subcellular locations as separase during cell division, where it functions to regulate separase activity.
To determine whether pph-5 promoter sequences could also drive expression in the germline, we fused them to mCherry::histone H2B (supplementary material Fig. S3B). A line was generated and revealed robust levels of nuclear-localized mCherry signal in all cells at all developmental stages ( Fig. 7D-I, and data not shown). This experiment suggests that the pph-5 regulatory sequences can drive expression throughout development in somatic tissues and the entire germline.

Depletion of PPH-5 in the germline is responsible for suppression
Because pph-5 regulatory elements can drive expression in both the soma and germline, we investigated which tissue was relevant to the mechanism of sep-1 suppression. We combined sep-1(ax110) with null alleles of rrf-1 and ppw-1, which are defective for RNAi in somatic and germline tissues, respectively (Sijen et al., 2001;Tijsterman et al., 2002). Following feeding with pph-5(RNAi) bacteria, sep-1(ax110) rrf-1(pk1417) double mutants produced a viable brood at 24˚C (Table 3, lines 2, 4 and 6). Conversely, the pph-5(RNAi) feeding did not suppress the sep-1(ax110) ppw-1(pk1425) double mutant, suggesting that suppression requires functional germline RNAi (Table 3, lines 9, 13 and 17). This failure to suppress was not attributable to some novel interaction between sep-1 and ppw-1 because genetic suppression was still possible using pph-5(tm2979) ( Table 3, compare lines 17 and 18). We conclude that suppression of sep-1 embryonic lethality is mediated by interfering with the function of germline-derived PPH-5 protein.

Discussion
Characterization of two novel sep-1 alleles with OID phenotypes We have characterized two new alleles of the sep-1 gene in C. elegans. Both sep-1(ax110) and sep-1(ax521) are recessive non- conditional alleles and embryos bearing these alleles have defects similar to those exhibited by sep-1(e2406ts) embryos. Importantly, all three alleles cause limited defects in sister chromatid separation. Instead, these alleles disrupt the function of separase during vesicle trafficking. The amino acid substitutions encoded by e2406ts and ax521 lie well outside the peptidase domain. The ax110 allele harbors a mutation in a non-conserved residue that resides just within this domain. Thus it is probable that these alleles do not inactivate the catalytic activity required for cohesin cleavage. This contrasts with mutant alleles of separase in other organisms in which they disrupt the cohesin cleavage activity of separase (López-Avilés and Uhlmann, 2010). Therefore, these alleles provide a unique opportunity to examine the function of separase in vesicle trafficking.
Although it is well established that the cleavage of cohesin is a conserved function of separase, there is mounting evidence that separase has many roles during cell division, including some that do not require its protease activity (Gorr et al., 2006;Kudo et al., 2006;López-Avilés and Uhlmann, 2010). Gorr et al. have shown that specific anti-separase antibodies, which do not interfere with the cleavage of cohesin but prevent the interaction between separase and Cdk1, inhibit the extrusion of polar bodies when injected into Xenopus oocytes (Gorr et al., 2006). Similar results were reported using separase-null mouse oocytes, which have a polar body extrusion defect that can be rescued by injection of mRNA encoding catalytically inactive separase (Kudo et al., 2006). Whether these data reflect a role for separase in CGE in vertebrate oocytes remains to be determined.
High resolution electron micrographs have revealed that sep-1(e2406ts) embryos fail to form the chitin and lipid-rich inner layers of the eggshell (Bembenek et al., 2007;Benenati et al., 2009). The reduced level of CGE observed in sep-1-deficient animals is a probable cause for these defects in the eggshell, because the CGs contain the raw materials required for assembly of the mature eggshell (Bembenek et al., 2007;Sato et al., 2006). Consistent with this observation, we find that all the SEP-1 mutant proteins fail to properly localize to CGs during anaphase (Fig. 3). Hence, we conclude that these alleles are hypomorphic for a subset of sep-1 functions, including vesicle exocytosis. How ever far removed the role of separase is in CGE from its canonical role in cohesin cleavage, it is likely that it still falls under the regulation of securin and the APC/C, because loss of expression of these genes also disrupts the eggshell and interferes with CG trafficking (Bembenek et al., 2007;Golden et al., 2000;Shakes et al., 2003;Wallenfang and Seydoux, 2000). It will be interesting to determine whether a catalytically inactive version of SEP-1 can rescue these particular phenotypes in C. elegans embryos, as has been shown for the polar body extrusion defects in separase-deficient mouse oocytes (Kudo et al., 2006).

Aberrant SEP-1 localization in mutant embryos
Separase has a dynamic subcellular localization pattern in early C. elegans embryos (Bembenek et al., 2007). The novel sep-1 mutants can be distinguished from one another by the subcellular localization of their respective mutant proteins. sep-1(ax521) mutants had a SEP-1 localization defect that was specific to CGs. A more general defect was observed in sep-1(ax110) animals where SEP-1 failed to localize to the cortical structures, the spindle matrix or the CGs. Therefore, these mutant proteins probably disrupt the function of separase by disrupting its subcellular localization. We speculate that the spindle localization of the SEP-1 protein might reflect its roles in regulation of the anaphase spindle or membrane trafficking dynamics rather than cohesin cleavage. This conclusion is supported by the finding that separase can directly bind DNA in vitro, facilitating its access to and cleavage of cohesin (Sun et al., 2009), which is probably unperturbed with these novel mutant proteins.
Although the sep-1(e2406ts) mutant was used in our suppressor screen, only sep-1(ax110) is fully suppressed (at all temperatures) by pph-5 mutations or RNAi. The sep-1(ax521) allele is not suppressed by loss of pph-5, and sep-1(e2406ts) is only suppressed up to 20˚C. It remains to be determined whether this suppression profile reflects the strength of each of the sep-1 alleles, the pleiotropic nature of each or the structural and/or conformational alteration caused by each of these missense mutations. Interestingly, the two mutations that are not fully suppressed are both located in the large N-terminal domain of separase, well outside the catalytic domain.
Because the synthesis of a mature eggshell is essential to the survival of the embryo (Rappleye et al., 1999), it is not surprising that our screen isolated a secondary mutation that restored CGE and eggshell function. Although the function of SEP-1 at its various localities is not understood, it is important to note that suppression restored the localization of separase to vesicles and the efficient fusion of CGs with the plasma membrane in sep-1(ax110) embryos. This finding suggests that the mechanism of suppression involves restoring the function of a defective SEP-1 protein as opposed to bypassing it altogether, such as by stimulating exocytosis through another pathway. This would also explain why the suppression seems to be specific to certain sep-1 alleles, and does not suppress the effect of sep-1(ax521) or sep-1(RNAi) where separase cannot be restored to vesicles. It will be interesting to determine the mechanism underlying the recruitment of separase to vesicles and how PPH-5 is involved.
Despite our observations that modulating the level of pph-5 activity can have dramatic effects on the survival of two distinct separase mutants, this gene appears to be non-essential. Although the pph-5 alleles behave as hypomorphic mutations, their catalytic domains are intact (Fig. 6). Therefore the question of whether pph-5 is essential for embryonic development remains unanswered until a true null allele becomes available. Nonetheless, the two pph-5 alleles behave as 'silent' suppressors of sep-1 mutants, and thus highlight how beneficial genetic screens are for identifying novel factors that regulate the separase pathway.
Interestingly, mice homozygous for a deletion in the murine protein phosphatase 5 (Ppp5) gene are viable and fertile (Yong et al., 2007). Mouse embryonic fibroblasts from such embryos, however, exhibit defects in the DNA damage checkpoint arrest in response to ionizing radiation. To date, no mutant phenotypes have been reported for the Drosophila or S. cerevisiae protein phosphatase 5 orthologs (Chen et al., 1994).

Phosphatase regulation of the separase pathway
Because pph-5 encodes a phosphatase, we hypothesize that the reduction of pph-5 activity causes a net increase in the phosphorylated subpopulation of one or more target proteins.
This persistent phosphorylation might affect enzymatic activity, protein-protein interactions, or subcellular localizations of proteins in the SEP-1 pathway. There are examples of phosphorylation playing such roles for many components of the separase pathway including cohesin, Cdc27 and separase itself (Huang et al., 2007;Huang et al., 2009;Kudo et al., 2009). Our finding that the reduced activity of pph-5 restores CG localization of SEP-1 in sep-1(ax110) embryos suggests that the subcellular localization of SEP-1 to CGs might be regulated by phosphorylation. This regulation could be direct and phosphorylation of SEP-1 (or an associated protein) could be essential for the movement of SEP-1 to CGs. Alternatively, suppression could involve a more indirect pathway whereby PPH-5 regulates the activity of an upstream regulator of separase. Ultimately, elucidating the mechanisms employed by PPH-5 to regulate the localization of separase to vesicles and its exocytic activity will be a fascinating subject of future studies.
In yeast, the phosphatase PP2A acts as a regulator of the separase pathway (Clift et al., 2009;Riedel et al., 2006). Rec8 only serves as a separase substrate when it is phosphorylated during meiosis (Riedel et al., 2006). When PP2A is targeted to centromeres by shugoshin, it removes phosphates from the neighboring cohesin complexes thereby reducing their suitability as separase substrates and ultimately leads to the protection of centromeric cohesion during meiosis I (Riedel et al., 2006). Perhaps PPH-5 in C. elegans plays an analogous role and removes phosphates from vesicle proteins to reduce their affinity as SEP-1 interacting proteins/substrates. In sep-1; pph-5 suppressed mutants, these phospho-epitopes might persist, allowing a 'compromised' separase more time to recognize and cleave its phosphorylated vesicle substrates.
In mammalian systems, separase is the target of an inhibitory phosphorylation that occurs on serine 1126 (Stemmann et al., 2001), which is in the unstructured region adjacent to the peptidase domain. In vitro experiments have shown that Cdc2-cyclinB and MAPK can phosphorylate S1126 of purified separase (Stemmann et al., 2001). Phosphorylation of S1126 is a prerequisite for separase to bind cyclinB directly, an interaction that mutually inhibits both the protease activity of Sep1 and the kinase activity of Cdk1-cyclinB (Gorr et al., 2005). Although it is unknown whether these regulatory mechanisms are conserved in C. elegans, or whether CDK-1 interactions affect the localization of SEP-1, our data does support the view that the exocytic functions of separase is regulated by an interplay between kinases and phosphatases. Whatever the mechanism, our findings strongly suggest that the exocytic function of separase in C. elegans is influenced by the phosphatase PPH-5; identifying its substrates is key to unraveling the novel regulatory mechanism of separase function in membrane trafficking.

Strains and alleles
Wild-type C. elegans was the Bristol strain N2. All strains were cultured using standard techniques (Brenner, 1974). Temperature-sensitive strains were maintained at 15˚C; all other strains were maintained at 20˚C.
Quantification of cortical granule exocytosis CGE was quantified according to a protocol previously published (Bembenek et al., 2007).

Time-lapse imaging of embryos
The time-lapse images of avIs74; itIs37 embryos were taken on a Nikon Eclipse TE2000U microscope equipped with a 603 1.4 NA Plan Apo objective along with a Spectral Applied Research LMM5 laser merge module, a Yokogawa CSU10 spinning disk unit, and a Hamamatsu C9100-13 EM-CCD camera. Image processing and AVI production were carried out using ImageJ 1.42q software (Abramoff et al., 2004).